TridentX + Z77X-UD5H: Set TRRD=7

[Leeghoofd]

Nope had it enabled and even disabled it still doesn't post further then 15, then shuts down if manual timings set ( in reference to what the bios reads out from the XMP profile )

Dimms post fine at 2666 with 11-13-13-35 2T TRFC 147 on Gene V and Sabretooth... no idea what's up with this board... up to 2400 all is okay, from 2600 divider hell breaks loose.

EDIT

Messed with the timings, started to compare the read outs via MemTweakit. Adjusted manually setting by setting till we got a post. tRRD, tRSSR and tWWSR are the trick !

Issue 1: board bios badly reads out the timings set by user or XMP (see RED)

http://dl.dropbox.com/u/66364457/Z77%20UD5H/rambios.jpg

http://dl.dropbox.com/u/66364457/Z77%20UD5H/rambios2.jpg

Issue 2: culprit to post at high speeds (+2600) is the too tight tRRD setting, needs to be at 6 to post here with the TridentX kit.

Do not touch the tRWDRDD timing, leave on auto. Any value set here results in a no boot.

Super Dino will fix this for sure !!!

[sin0822]

it is set low to better efficiency and performance.

[Massman]

Nice find, Leeghoofd!

it is set low to better efficiency and performance.

That's a stupid reason not to allow 'clocking over DDR3-2600. Don't tell me that one timing is giving you so incredibly much more performance that high MHz should just not work ;).

[Leeghoofd]

Bios department needs to fix these things Steven. I'm not saying the above works for everybody, but at least my UD5H board post stable with all timings set at AUTO, just tRRD to 6 and to increase stability tRRSR to 5 manually adjusted.

Fixes needed :

1) Settings read out should be correct, otherwise there's no use for the given info.
2) Just force tRRD at 6, tRSSR to 5, when selecting 2600 or higher divider and maybe remove the tRWDRDD setting in the bios. Board at least needs to be able to post at 2600 or 2666 divider. Stability is another thing.
3) And why not finally allow vcore read out again via CPU-Z and co...

Imagine you are buying a high end motherboard and iddem ditto specced ram kit and the combo doesn't even post. The XMP profile boots, but voltages ( VTT and co )are mad (G.Skill thing ofcourse ) Do you agree it's not a win win situation. Who do you think will they blame in the end ? The RAM vendor or your company ?

YFI on the ASUS boards we experienced bad samples too ( DLX working only with single channel, etc... ) MSI GD65 has got some issues too with high ram speeds.

As always tweaking can be done manually later ( for those that feel the need to do it )

C'mon Gigabyte you can do it ! Time to fix compatibility iso aiming at efficiency

[pro]

mate easiest way for most users is to enable xmp profile1, then just set any timings you want over the top

but your way also works good, nice tip

[Vivi]

i can set my 2600 XMP profile for my teams and they boot just fine on this UD5H.

Can even go to 2700 on the same timings and XMP and same volts. Maybe i have nice CPU imc or something?

[Massman]

With TridentX?

Might be some very specific SPD issue for that kit, I guess. With the RipjawZ I'm booting 2600 easily as well (although fails with some beta bioses)

[Leeghoofd]

i can set my 2600 XMP profile for my teams and they boot just fine on this UD5H.

That's what XMP profiles are for Vivi :D

My point is I need to adjust manually for eg for the Flares ( as the XMP is 2000c7) I didn't get there unless I did the above settings

If you do this on another board, it might work with just adjusting the main timings to eg 9-12-9-27 with the Flares, the Giga needs more work ;)

[sin0822]

I use some crappy single sided hynix(2000mhz cas9), i can boot at 2666 without touching those tertiary timings, just loading XMP and loosening the first set and some second set of timings, but all 3rd are the same as at stock.

I can ask for above 2600-2666 to have some of the tertiaries maxed out, but Dino would have to make a request like that as it would affect the performance of the boards, however who really goes above that and doesn't use XMP? If if they did i would hope they know how to set their timings, as that is no go zone. one of the main things that the engineers at GB worked on was getting XMP profiles working correctly, so they expect the user to use XMP these days. I mean seriously if I bought a TridentX 2666 kit and i tossed it in and didn't set XMP and i wanted 2666, well then I would just feel like i don't expect it to work, unless somehow the board can read my mind, or has its timings set so lose it would take days to tighten them.

[stasio]

one of the main things that the engineers at GB worked on was getting XMP profiles working correctly

G.Skill F3-17000CL8D-4GBXMD with XMP 8-9-8-24-CR1,on UD3H in Windows 8-9-13-24-CR2 !!.

http://ggts.gigabyte.com.tw/ClientUserFile/OK/1230522/1.jpg
http://ggts.gigabyte.com.tw/ClientUserFile/OK/1230522/2.jpg
http://ggts.gigabyte.com.tw/ClientUserFile/OK/1230522/3.jpg
http://ggts.gigabyte.com.tw/ClientUserFile/OK/1230522/4.jpg

Pi32m can get only with CL.9.
GB need to enhance memory performance at least 50% of Asus settings.

[Lsdmeasap]

tRWDRDD has a major impact on performance, and should NOT/Never be removed or set auto by the BIOS and then hidden/non-user adjustable!!

You can adjust this setting yourself, then it's depending on the ram kit if it's going to boot or not, too low and it'll fail of course.

You can see what auto sets for that on your board/mem by using Memtweakit, note in my image I have edited the program to correctly show which value this is on Z77

http://i.imgur.com/jDEMc.jpg

I've actually found tWCL to be the culprit of no boot at 2600 for me, more than a few times on a few sets of memory. It should be CL -1, but there is times where that fails or the BIOS will set CL -2 (or sometimes it sets -1 but you need -2, ect), so sometimes you need to set manually, and sometimes you need to set this to CL = tWCL or manually set tWCL to CL -1, -2, ect *Edit* Sometimes you may also need to set tWCL manually to less than expected, for example, sometimes for me even at CAS 10 if I set tWCL to -1 (9) it'll fail and give 51 post code, same for CAS 11. In those instances usually when I set tWCL to 7 or 8 it would then boot and work, 8 usually works the best for me. So there is something funky going on at certain CAS levels in Z77, or the BIOS itself, not sure which but I know this setting needs to be set manually most of the time and it's not often what you'd expect to have to set or what you'd expect should/should not work. *

Thanks for the tRRD tip too, I'll see if that helps on any of the kits I've been playing with.

Someone needs to share some of those 2400-2600 G.Skill SPD's, so Samsung users can have a proper XMP to try and work with (Still impossible to set above 2200 for everyone that last I've seen)!!

[Stelaras]

I tried the 2400 g.skill SPD on the Samsung low profile sticks but still can't post higher than 2200 .

Also , all my PSC sticks clock about 50 Mhz lower than Asus even with higher voltage.
For example ... Asus 2600Mhz 8-12-8 @ 1.83V ... Giga 2550Mhz 8-12-8 @ 1.90V
All the available subtimings are the same . Bios & memtweakit

[Lsdmeasap]

There isn't a TRWDR setting that I can see (UD3H), or am I missing it? And I think it's not, which is what gave me the idea to edit the name in Memtweakit once I noticed what it really was. I only left the TRWDR in Memtweakit like that so I would remember where it was in case I needed to add back later for some reason.

If you set TRWDRDD to 8 or above, then you will see poor bandwidth, then lower in windows one at a time and you'll see bandwidth go up dramatically as you get lower.

That's why I said leave and don't hide, it's kinda like tRD on old platforms. Finally Z77 has brought actual memory tweaking performance to DDR3, other settings have good impact as well, unlike many of the DDR3 platforms before Z77.

I have not looked for the exact correlation in Memtweakit for TWRDRDD, not sure if it's even there (Or maybe I looked already and it wasn't, so I didn't change CR)? If it is and you know which it is, I can change the wording for ya if you want, let me know.

Also, if you want a copy of my edited version let me know, I only edited the Z77 timing description, it will still show whatever names are default for other chipsets.

I don't have any G.Skills myself, I've been testing with 8GB kit's of BFR's, and 4GB kits of PSC/BBSE mainly. Would love to push on my Sammy's, but have yet to see anyone post a G.Skill profile so I can try it myself.

Someone please post, like Stelaras, please share! Thanks! I'd love to try myself! If anyone shares please be sure it's from a double sided Samsung kit. Stelaras, was the SPD you tested from a double sided samsung kit, were you positive on that? Just wondering

Also, my comments above about tWCL were discovered on PSC (XDZ.... A3G-A), and that's mainly what I've played with it on so far (Just discovered). But I'm pretty sure with it auto setting incorrectly at times that would affect any kit, unsure if XMP enable would change that behavior or not, don't have any Z77 XMP memory to test with.

I'll be testing your tRRD finding shortly, thanks a lot! That's one setting I probably never thought auto would set too low, so I'm sure it's probably been a big part of memory hassles for me at times too.

*Edit*

Also see above for my edits, after looking back at my notes I see I previously worded the above a little incorrectly.

I also think Gigabyte needs to remove IDE detection from the BIOS routines, A0 shouldn't be normal for the debug code while in windows. Of course it's going to stop there at initialize IDE since there is no IDE, so it should be removed.

What do other Z77 debug coders show in windows? I know I could be wrong as Z77 might be different, but I'd expect for Gigabyte that's the ONLY reason it stalls there. That could also be holding us back on other things, who knows!?!?

I plan to inquire about that myself, just trying to wrap up too many reviews at once right now so I haven't had a chance yet, but I will shortly!

[Lsdmeasap]

I looked into the tRRD thing a little bit, haven't played with 2600 with your info in mind yet so not sure how much if any that will help me there, but here's what I'm seeing so far.

I am not using XMP, all manual/or auto, expert mode Turbo or Extreme (Both = same on this).

At 2133 tRDD Auto = 7 actual (Reflected in MIT status and windows)
At 2400 tRDD Auto = 8 actual (Reflected in MIT status and windows)

So I'd expect 2600 tRRD Auto = Maybe 8 / Probably 9 or above, no way is it going to auto set 4.

What you see on the left of the BIOS is never the actual used setting, it never has been from what I've seen in any Gigabyte BIOS. Maybe it's correct/actual sometimes if you use XMP not really sure on that as I never use it, but I've always told users to ignore the left side values as they're never actual but only what the board used previously at some time in the past or sometimes what auto thinks might work (But also never uses anyway)

So that 4 you are seeing on the left for auto is not actually what the BIOS is attempting to use at 2600 or above, I can assure you of that. Great to hear setting 7 is working for you though, 8+ might help even more too now that I've looked at those auto setting for that value, depending on your kits and other settings, voltages, ect.

But you can use tRRD 4 at 2400, I know that for sure as that's what I used for 2400 in my review (tRRD 4 / tFAW 16-20, they are connected so this also might be a factor for you when setting tRRD). Not sure how low you can take it at 2600 though, my CPU is iffy there on water, and I don't have any Z77 XMP kits (Or any memory meant for Z77 so this may be part of the issue too?). Anyway I've only been able to play with that a little bit for now, until I have some time to get my CPU frozen or better memory, not sure if this CPU will be able to bench there or not.

Hope that helps some!!

[Leeghoofd]

I tested our 3 teritary timings, for me tRSSR is the most important boost in bandwith performance, tRWDRDD and tWRDRDD hardly give anything 200-300MB/s

Biggest influence is tRSSR. lowest usable range at 2666mhz is 4, setting 3 will give similar performance as 7

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/tssrs.png

tRWDRDD sees slight gains in Read and Copy. Value 3 BSODs

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/trwdrdd.png

tWRDRDD similar to tRWDRDD increase, but can set value from 1-7 without any big losses...

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/twrdrdd2.png

Are the UD3H bios and/or the UD5H's settings mixed up ?

[Lsdmeasap]

Very nice testing results! Sorry, I didn't look back at my notes when commenting on performance above, and must have been confused on which settings exactly gave best performance increases, but I know I'm happy to see all the new tertiary timings as it's a great move in the right direction for DDR3! Looks like you are correct about the performance gains and TRSSR

Where are you seeing tRWDR in the BIOS? Does UD5H have this and UD3H not? In Memtweakit, on UD3H anyway, TRWDR is TRWDRDD that is why I changed the name of the setting. Haha, I just read your last comment!! I'm lost too, are you seeing TRWDR in the BIOS or not? It's not in UD3H BIOS at all, and TRWDRDD directly changes to anything I set it at in Memtweakit, so I'm certain that is what it is. I think we are both referring to the same setting, only you are saying what Memtweakit calls it, and I'm saying what the BIOS name for that setting is... Right?

[Lsdmeasap]

OK, I just looked in UD5H BIOS S4 see below, and there is no TRWDR so it's the same as UD3H

It would be nice if we could access all of the memory settings though, might get those Sammy's going or something?!!

http://i.imgur.com/ssyJZ.jpg

[Leeghoofd]

Yeah it's not there with UD5, freaking renaming sigh , it's already complicated enough :p

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/biosud5H.png

edited the above post with the Giga bios name :p

[Lsdmeasap]

You had me going there for a minute, thought I was missing out on a setting again! :)

Here ya go!
http://www.mediafire.com/?pr8qa5zad6csc02

[Leeghoofd]

Thanks ! Now tWCL setting :p = tWL ?

[sin0822]

as far as the samsungs go, ED i reported this over and over. I also gave them a set of PCS of mine, so they could look into it too. Apart from the generic samsungs(b/c normal samsungs seems to OC very fine) and apart from PSC for the most part(as i am pretty sure I saw cookie OC with PCS off the charts man, he just put in all the timings manually) the other memory OC fine. So those two are on the list now for not fully compatible, shame too on the samsungs I bet it is something stupid. Are you using the most recent beta LSD? BTW what one thing that is EXTREMELY odd with the samsungs, try to just use 1 stick, i bet you will boot upwards of 2666. 2 sticks much lower, that shouldn't be. I also don't think it matters if you put the 1 sticks in DIMM1 or DIMM2.

There is some incompatibility with some kits, there are just so many DDR kits, however what i was trying to convery is always set XMP and then go from there with the GB boards.

[Leeghoofd]

Tried my BBSE kit on the UD5H, sort off stable at 2600mhz at 1.83Vdimm. any tighter on the tert timings and she will crash... for TRCD 11 I need to go 1.87Vdimm... I need Zen Effect's sticks looooooooool

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/BBSEUD5H2.png

C'mon Steve who cares about one stick versus two :p

[zeneffect]

Tried my BBSE kit on the UD5H, sort off stable at 2600mhz at 1.83Vdimm. any tighter on the tert timings and she will crash... for TRCD 11 I need to go 1.87Vdimm... I need Zen Effect's sticks looooooooool


you and me both!

ive got some psc im playing with right now... have some different bbse stashed away for a rainy day... cpu is totally NOT cooperating though lol.

[stasio]

Leeghoofd,
what you get with BBSE in XMP setting?
and what is your BBSE setting for daily use with Vdimm 1.65-1.66v?
Thanks.

My UD3H/D3H:
My XMP is 8-9-13-24-CR2 in Windows and BIOS MIT Current Status.
However,XMP is correct in BIOS Advanced Memory Setting/Ch.A
Pass Pi32 only with CL.9 (1.65V).

[stasio]

You had me going there for a minute, thought I was missing out on a setting again! :)

Here ya go!
http://www.mediafire.com/?pr8qa5zad6csc02

Lsd,
can you mod this 1.01.7 version:

http://www.mediafire.com/?zn69761upz4aem1

thanks.

[Leeghoofd]

Leeghoofd,
what you get with BBSE in XMP setting?
and what is your BBSE setting for daily use with Vdimm 1.65-1.66v?
Thanks.

Same as your settings trp 13, also read out in bios correctly.

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/BBSEXMP.png

Lowered trp to 8 seems 32M stable at 1.65Vdimm

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/BBSEXMPtrp8.png

[Leeghoofd]

daily it will be 2400c9-11-9 1.65Vdimm, for benching I tighten the timings to 8-10-8 at 1.75Vdimm. My bench CPU caps out at 2450ish speeds :p

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/BBSEXMPdaily.png

[stasio]

........Lowered trp to 8 seems 32M stable at 1.65Vdimm.....

Thanks,I'll try.

What about CR1?

[Leeghoofd]

Command rate 1 no problem either...

http://dl.dropbox.com/u/66364457/ASUS%20ZERO/BBSEXMPdaily1T.png

[stasio]

Thanks for your input.
Seems I must change my 2600K soon.

[Lsdmeasap]

as far as the samsungs go, ED i reported this over and over. I also gave them a set of PCS of mine, so they could look into it too. Apart from the generic samsungs(b/c normal samsungs seems to OC very fine) and apart from PSC for the most part(as i am pretty sure I saw cookie OC with PCS off the charts man, he just put in all the timings manually) the other memory OC fine. So those two are on the list now for not fully compatible, shame too on the samsungs I bet it is something stupid. Are you using the most recent beta LSD? BTW what one thing that is EXTREMELY odd with the samsungs, try to just use 1 stick, i bet you will boot upwards of 2666. 2 sticks much lower, that shouldn't be. I also don't think it matters if you put the 1 sticks in DIMM1 or DIMM2.

There is some incompatibility with some kits, there are just so many DDR kits, however what i was trying to convery is always set XMP and then go from there with the GB boards.

Ya, I sent in a report a long time ago too, about the Sammy's and BFR's, the BFR issues have been corrected shortly after my report (I think, haven't tested with them much since sending in the report so can't be 100% sure on that) but nothing done with Sammy's yet. Surely this is just one or two settings they aren't giving us access too? I posted an image here earlier showing all the settings currently in the BIOS, some of which we cannot set because they are hidden so it could be those, or could be a setting they've not added at all yet?

Seems like it should be an easy fix, not sure what the hold up is? I don't have any G.Skills, nor any contacts there to check about getting a set for testing, and I've been super short on cash lately so I can't afford to pickup a set just for testing. It's the same IC's (At least for those w/ double sided samsung based kits anyway), and the G.Skills work, so not sure what's going on? I know you say it's XMP, but that shouldn't matter really once you set things manually. In any event, if it is a XMP issue, then flashing a XMP into those cheaper ones should work and it doesn't from what others say, so that says to me right there XMP present or not isn't the issue.

Still waiting for someone to share a XMP with me so I can test!!

PSC has not been a problem for me at all, what issues are you seeing or hearing about with PSC? I can push mine to 2600, and can run good settings at 2400 all day long, with several different PSC kits. Is there a known issue with those I've missed out on hearing about? If so let me know, maybe I know the key and should be sharing it but didn't know others were having problems? :D

I'm mainly using F8g right now, since that's what I started my review testing with, but I have flashed in plenty of the more recent betas each time one comes out and then I play/test for a bit and see if anything's changed in regards to certain things. No luck with Sammy's on any of them though. I'll try the single stick sometime and see if I can see what's stopping dual from working at above 2133, might be something silly like you mentioned, but I'm leaning more towards something missing in the BIOS or something hidden we can't adjust, otherwise they'd probably have fixed it by now.

Dino, why you so quiet on this issue? Hahahah! :D

As for your last comment about XMP, I never set it and don't have problems (With kits that work that is). And yes, I've tested setting it on Z77 and don't really see any difference, except when at stock ram speeds XMP vs auto is obviously different, but at 2400Mhz on a 1600Mhz kit I don't see how a 1600MHz XMP profile is going to help anything anyway?


Lsd,
can you mod this 1.01.7 version:

http://www.mediafire.com/?zn69761upz4aem1

thanks.

Yes, sorry for the late reply!

Here's the updated MOD Version @ Blacker black, darker blue, and added back in tREFIX9 for Z77 (sorry Leeghoofd, looks like I missed that on the previous upload I posted for you, uploaded wrong MOD version)

http://i.imgur.com/s3T8F.png

http://www.mediafire.com/?vh56pp6z7l6zp77

[Leeghoofd]

Did the G.Skill review on the Gene V and found some extra stuff. Above 2400 they also set tWWSR to 7. Okay this decreases copy performance and might not be as efficient. But at least all of my sticks boot with just the main timings set... the user can fine tune, if he feels the need to...

Gigabyte Bios team must try to set some looser profiles to enhance ram compatibility. Allowing people to boot at high speeds with older/other sticks too. Think PJ mailed Tim about all of our findings. I hope they get it right on the P series, Dinos initial BBSE screenie on facebook at least looks promising.

Nice work on the tweakit Lsdmeasap :nana:

[sin0822]

Ya, I sent in a report a long time ago too, about the Sammy's and BFR's, the BFR issues have been corrected shortly after my report (I think, haven't tested with them much since sending in the report so can't be 100% sure on that) but nothing done with Sammy's yet. Surely this is just one or two settings they aren't giving us access too? I posted an image here earlier showing all the settings currently in the BIOS, some of which we cannot set because they are hidden so it could be those, or could be a setting they've not added at all yet?

Seems like it should be an easy fix, not sure what the hold up is? I don't have any G.Skills, nor any contacts there to check about getting a set for testing, and I've been super short on cash lately so I can't afford to pickup a set just for testing. It's the same IC's (At least for those w/ double sided samsung based kits anyway), and the G.Skills work, so not sure what's going on? I know you say it's XMP, but that shouldn't matter really once you set things manually. In any event, if it is a XMP issue, then flashing a XMP into those cheaper ones should work and it doesn't from what others say, so that says to me right there XMP present or not isn't the issue.

Still waiting for someone to share a XMP with me so I can test!!

PSC has not been a problem for me at all, what issues are you seeing or hearing about with PSC? I can push mine to 2600, and can run good settings at 2400 all day long, with several different PSC kits. Is there a known issue with those I've missed out on hearing about? If so let me know, maybe I know the key and should be sharing it but didn't know others were having problems? :D

I'm mainly using F8g right now, since that's what I started my review testing with, but I have flashed in plenty of the more recent betas each time one comes out and then I play/test for a bit and see if anything's changed in regards to certain things. No luck with Sammy's on any of them though. I'll try the single stick sometime and see if I can see what's stopping dual from working at above 2133, might be something silly like you mentioned, but I'm leaning more towards something missing in the BIOS or something hidden we can't adjust, otherwise they'd probably have fixed it by now.

Dino, why you so quiet on this issue? Hahahah! :D

As for your last comment about XMP, I never set it and don't have problems (With kits that work that is). And yes, I've tested setting it on Z77 and don't really see any difference, except when at stock ram speeds XMP vs auto is obviously different, but at 2400Mhz on a 1600Mhz kit I don't see how a 1600MHz XMP profile is going to help anything anyway?




Yes, sorry for the late reply!

Here's the updated MOD Version @ Blacker black, darker blue, and added back in tREFIX9 for Z77 (sorry Leeghoofd, looks like I missed that on the previous upload I posted for you, uploaded wrong MOD version)

http://i.imgur.com/s3T8F.png

http://www.mediafire.com/?vh56pp6z7l6zp77

You mean the TridentX kit is the same as the ULV sammy?

If you look at the most recent BETAs, one or two of them have some work done properly with the Sammy's. Before you couldn't set lower than 1.5v but only with the samsung ULV, any other kit was fine. Then with this latest round of BIOSes there was a fix in there to allow the Sammy's to default at 1.35v, you can see if after you flash some betas. But yea in terms of easy 2400 no go.

As far as the PSC go, i am betting you set all your 2nd and third timings yourself right? A lot of guys can't do this.

[Lsdmeasap]

Did the G.Skill review on the Gene V and found some extra stuff. Above 2400 they also set tWWSR to 7. Okay this decreases copy performance and might not be as efficient. But at least all of my sticks boot with just the main timings set... the user can fine tune, if he feels the need to...

Gigabyte Bios team must try to set some looser profiles to enhance ram compatibility. Allowing people to boot at high speeds with older/other sticks too. Think PJ mailed Tim about all of our findings. I hope they get it right on the P series, Dinos initial BBSE screenie on facebook at least looks promising.

Nice work on the tweakit Lsdmeasap :nana:

Yes, hopefully they get things ironed out soon!

Thanks, I wish I could figure out how to remove the green corners!

You mean the TridentX kit is the same as the ULV sammy?

If you look at the most recent BETAs, one or two of them have some work done properly with the Sammy's. Before you couldn't set lower than 1.5v but only with the samsung ULV, any other kit was fine. Then with this latest round of BIOSes there was a fix in there to allow the Sammy's to default at 1.35v, you can see if after you flash some betas. But yea in terms of easy 2400 no go.

As far as the PSC go, i am betting you set all your 2nd and third timings yourself right? A lot of guys can't do this.

Yes, that's what I meant, well at least for some of the early double sided samples anyway. Most are single sided now so it's not really a good comparison I suppose.

I did notice they've been working on the 1.35/1.5 issue, but that doesn't really help/affect the ability to use them above 2133MHz.

As for PSC, yes I do set 2nd/3rd timings myself, but auto works fine for me too most of the time (Just performance is not as good). What's the full IC model number of the ones you had issues with? XDY....A3G-A and XDZ....A3G-A work fine for me, I have a few other sets with similar IC's but I'm not 100% sure of hand what those model #'s are. But ya, in general you are right I do set those timings myself at higher speeds, but I start booting with auto and only adjust for better performance - not for stability or to make them bootable, ect.

Someone please share double-sided G.Skill Trident X 2400 XMP/SPD!! Thanks!

[sin0822]

Yea that would be great.

yea it is interesting that some of them are single sided. I am guessing it depends on their price? the $100 kit is single sided.

[Lsdmeasap]

Yea that would be great.

yea it is interesting that some of them are single sided. I am guessing it depends on their price? the $100 kit is single sided.

It's just a case of "Review samples (And some early retail) Vs. current and 95% of what's out there" - from what I see anyway. This applies to the $100 kit mainly, some of the more expensive kits have always been better (Better IC / or Better Binned, Dual sided, ect) and have not changed IC or dual to single like the $100 kit has.

Just tested a single stick tonight, but only for about 10 minutes as it was easy to see the outcome no matter what. Fine up to 2200 even at 1.55-1.6, no go at 2400 no matter how loose you set anything or how high you set the voltages (Tested up to 1.86). There is a hidden setting we can't adjust causing this, or a relevant/required setting missing from the BIOS, no way they could work at 2133-2200 so easily and then just stop dead in their tracks when trying 2400

[dinos22]

my kit is seen OK > TRRD =AUTO = 7

http://www.youtube.com/watch?v=wxcDC6zUAtU&t=5m47s

damn it, why do you guys think i see every thread on every forum and in a timely manner LOL!!!!

hey goofball, i got masbo an early board so you guys can mess around, any help is good help. it's not being ignored boys just lots going on. this is good feedback

[dinos22]

btw anyone have SPD flashing software for DDR3, i want to check something?

[Lsdmeasap]

Sorry Dino, didn't meant to poke on ya like that! :D And really, I was more wondering why you hadn't commented anywhere about the Samsung HYKO above 2133/2200 issue

You can use spdtool .63 on X58 boards, and P67 I think, not sure about Z68 - but I know it doesn't work on Z77, at least not for reading from the installed memory but you can open and read SPD's with it on Z77. Whenever I need to flash a SPD I do it on X58 just because it's easier that way for me since I know that chipset works, that way I'm not spending time trying with 5 other boards until I find a working one :D

As for the initial issue mentioned here, I don't see it as a board/BIOS issue from what I can tell, I mean it looks like tRRD is setting correctly in auto for many speeds I tested and posted about previously.

[zeneffect]

btw anyone have SPD flashing software for DDR3, i want to check something?

look into thaiphoon burner. apparently support sucks, but it works on z77 ud3 and ud5

[Leeghoofd]

my kit is seen OK > TRRD =AUTO = 7

Logic as you are using the XMP profile in the video ;)

See if your TridentX sticks boot at 2600c10-12-12-35 1T 1.65Vdimm and no other setting touched nor a profile loaded...

[dinos22]

That's what the profile is there for!

[Sam OCX]

UD3H F9, this (http://valid.canardpc.com/show_oc.php?id=2437518) CPU + TridentX 2600C10
load XMP, don't touch anything else = 15-03-powerdown-endless reboots ... bad job Gigabyte

[Lsdmeasap]

Thanks for your input.
Seems I must change my 2600K soon.

No problems here with BBSE and 2500k/Z77 at the timing/speed you mentioned, email in route with more info

http://i.imgur.com/1yDRo.jpg

Quick and dirty settings, and testing = no tweaks & voltage guesses :D


Thanks ! Now tWCL setting :p = tWL ?

Sorry, I missed this previously! Ya know, I thought that, so I changed it so I wouldn't get lost... but now I see this might not be the case!

Dino will have to confirm for us, or I'll try to ask the BIOS guys soon. I don't think tWL is the same as tWCL, related but not the same and we really need tWCL in the BIOS so hopefully they will add it soon if they've not just simply named tWCL to tWL (Unlikely, but ya never know?)

[Leeghoofd]

Sorry, I missed this previously!

No problem I'm used to being ignored :p

Hey goofball, i got masbo an early board so you guys can mess around, any help is good help. it's not being ignored boys just lots going on. this is good feedback

We are working on it boss ! you got mail soon !

[Lsdmeasap]

No problem I'm used to being ignored :p

Honest, I never ignore anyone!! I really missed it, sorry bud! :)

Anyway, I wanted to show you this, seems Franck thinks the same as I did on his new CPU-Tweaker

http://img232.imageshack.us/img232/13/twcltwl.jpg

http://www.tweakers.fr/download/CPU-Tweaker.rar

Hopefully Dino will see these concerns and let us know what's really what! Either tWL=tWCL or they are both currently locked to the same values, as both reflect changes in the BIOS to tWL. They should be adjustable separately I believe.

Anyway, if tWL does not = tWCL in GBT BIOS then they need to add tWCL into the BIOS for us, as it's very important for memory overclocking sometimes! That might even be the key issue or at least part of the issues with the cheap Samsungs!

Don't change in windows with either app, it'll crash ya out! At least it does me at my current settings, up or down causes crash even if I raise first in the BIOS and try to lower back to original value in windows.

I also just noticed a few days ago, you can change CAS on the fly with MemTweakit, very cool! Hopefully Franck can make CPU-Tweaker like that!

[Stelaras]

Here are some SPD files ...

The gskill2400c10 original is from the single side , but i modded it for dual side . Haven't tried it yet , because my thaiphoon version only works on P67 boards and i have to setup my old P67 ...

You can also use the 2400CL11 SPD from the ripjawsz (dual side) . It also have an XMP profile at 2400Mhz

None of these SPD's could make my low profile HYK0 boot higher than 2200 Mhz on UD5H board . I could run 2850Mhz with other rams like 2800Mhz kingston . Also no problems with BBSE & PSC at 2550+ Mhz .
My HYK0 on maximus z77 gene & formula can run spi32m at 2666Mhz+ with less than 1.7V .

[dinos22]

Ed

TWL in GIGABYTE bios is tWCL in that tweaker software yes

i dont think you can change TWL=tWCL on the fly though, it requires a reboot


Stelaras, do you have a specific part number or even photo of the ram you are using please?

[Lsdmeasap]

Here are some SPD files ...

The gskill2400c10 original is from the single side , but i modded it for dual side . Haven't tried it yet , because my thaiphoon version only works on P67 boards and i have to setup my old P67 ...

You can also use the 2400CL11 SPD from the ripjawsz (dual side) . It also have an XMP profile at 2400Mhz

None of these SPD's could make my low profile HYK0 boot higher than 2200 Mhz on UD5H board . I could run 2850Mhz with other rams like 2800Mhz kingston . Also no problems with BBSE & PSC at 2550+ Mhz .
My HYK0 on maximus z77 gene & formula can run spi32m at 2666Mhz+ with less than 1.7V .

Thanks Stelaras!!

I don't think modifying a single side SPD for dual side is going to work, since the dual side ones we need use different IC's. The single use Hynix and the dual use Samsung, that's why the dual side SPD is needed to try and attempt using it with the Sammy's

I'm not sure on the dual sided C11 Ripjawsz, do you know if those are Samsung IC? What about the Team SPD, do you know if that is double sided Samsung kit?

That's all I can do with the Sammy's too on Z77/X79, 2200 or so. Easy booting there, but dead stop at anything past that no matter what you do, even setting CAS 11-12, super loose timings for everything else, voltages 1.8-1.9, ect. Nothing works, and these work fine and proper at 2400-2600+ on other boards as you mention, and the IC's are used on many other kits that don't have issue going above 2200.


Ed

TWL in GIGABYTE bios is tWCL in that tweaker software yes

i dont think you can change TWL=tWCL on the fly though, it requires a reboot


Stelaras, do you have a specific part number or even photo of the ram you are using please?

Thanks for the confirmation on tWL!! The reason we are discussing it is because I changed the blue version of Mem Tweakit's lable from tWL like the BIOS to tWCL (Red version shows tWL), so that's how we got started talking about this.

These are Samsungs we have issues with, both links have good shots of the IC's and ram info
http://www.techpowerup.com/reviews/Samsung/MV-3V4G3/3.html
http://www.overclock3d.net/reviews/memory/samsung_green_8gb_vlp_1600mhz_kit_review/2

See this thread also, I'm sure you've been here a time or two :D
http://www.xtremesystems.org/forums/showthread.php?277635-Ultra-low-profile-1.35v-30nm-DDR3-capable-of-2400MHz.-Perfect-for-SFF-builds./page28

The BIOS team should already have a set for testing because Sin0822 sent them a set, and I know they've used them to already correct the 1.35/1.5 auto voltage issue.

[sin0822]

They actually got their own set now(matching batch numbers too), so i can have my sticks back to play with(mine are a mismatched pair of batch number even though it doesn't matter haha)! :D There is hope i think.

Are you going to leave it tWCL? or change it to tWL?

[dinos22]

i haven't seen that thread before actually but now i know about it and will email bios team with both this and XS threads to look at!

[Lsdmeasap]

They actually got their own set now(matching batch numbers too), so i can have my sticks back to play with(mine are a mismatched pair of batch number even though it doesn't matter haha)! :D There is hope i think.

Are you going to leave it tWCL? or change it to tWL?

Nice to know! I wondered why you had to send them your kit anyway? I thought surely Samsung would be willing to send them a kit, or since they are so cheap they could easily buy a set, but I guess maybe it's not that easy to find in HK/Taiwan?

In the mod version I will leave at tWCL, unless you guys all think I should change it back to defaults so it matches the BIOS? Or were you more asking Dino if they planned to change the BIOS to say tWCL since that is what it is? (Suggested!)

i haven't seen that thread before actually but now i know about it and will email bios team with both this and XS threads to look at!

Haha, sorry man! There I go again assuming you are everywhere at once, I'll try to cut back on that! :D I just thought since you frequent XS so often that you'd have ran across that thread more than a few times by now, but I guess not.

Anyway, ya that thread will show them what's possible with these kits, then hopefully they can figure out why GBT boards are holding them back / limiting them to the 2200 divider.

Hope they can figure it out, as that would make MANY people happy!! I personally only grabbed a set because of how cheap they were, so I could test things when helping others since so many users purchased these due to that XS thread and the huge one at OCN here (http://www.overclock.net/t/1210482/samsung-2x4gb-low-voltage-ddr3-1600-kit-ocing-to-2400-discussion).

Otherwise I'll normally be running BBSE or PSC at 2400+ anyway as that's no problem and they sticks look much better than the naked Sammy's :)

[sin0822]

Well this was a while ago, and I just sent them to they could look at them and see if they were getting the same results as i was. I think they had their own kit back then, but the more the merrier, some kits might not do as well as others. However many times it is best for manufacturers to go out and buy their own kits, to get the retail product which the end user will use. Also I don't know how popular the samsungs are over in Taiwan, I know in Europe they are hard to get so it might be the same in some other select regoins. The whole green thing is probably much bigger in the states.

Also the samsungs have no XMP like other kits that have to be evaluated and added to the QVL.

I think you should make it like the GB BIOS so it is easier for new users to use.

[stasio]

Seems I found sweet spot with my TeamX 2600 on UP5 tRRD=4 (PI32 pass).


http://i11.aijaa.com/b/00664/10650828.jpg

[Stelaras]

UP works way better than UD . :)

[sin0822]

those teams are based on samsung no? Nice one stasio!

[Leeghoofd]

Cool results Stasio, still on hold for me UP5

[stasio]

those teams are based on samsung no? Nice one stasio!

Yea,double sided Samsung HCH9

http://i458.photobucket.com/albums/qq306/owikh84/7970/DSC_0817.jpg

Thanks guys,slowly warming up......

http://i9.aijaa.com/b/00262/10652946.jpg

and 2666 (w/o tweaking)

http://i5.aijaa.com/b/00100/10653165.jpg

[sin0822]

awesome! So compared to your D3H the UP5 clocks better right?

[stasio]

As I just get TeamX last week,cann't say.
But D3H clock well with my BBSE.
Most important is to understand second and third subtiming and apply to specific memory kit.

Btw,

2400 .....9-11-11-21-1

http://i8.aijaa.com/b/00565/10656732.jpg

[sin0822]

Samsung ULV working correctly now with a solution:

First off the Z77X-UD5H running 2x4GB Samsung Green ULV@ 2600 Cas 10 T1
http://img545.imageshack.us/img545/7421/260032msam.jpg (http://imageshack.us/photo/my-images/545/260032msam.jpg/)

Second we have the beautiful Z77X-UP5 TH-CF running 2x4GB Samsung Green ULV@ 2600 Cas 10 T1
http://img832.imageshack.us/img832/1050/up52600sam.jpg (http://imageshack.us/photo/my-images/832/up52600sam.jpg/)

[dinos22]

Patrick got it figured out, what was the cause of the issue?

[Lsdmeasap]

Can't wait to try those BIOS out, shoot em my way - thanks!

My set runs great at the 2200 limit :D, even on older BIOSes! So I'm sure now that this is corrected they might even be able to show up my BFR's

9-10-10-24-1T @ 1.51V
http://i.imgur.com/Zjjb4.jpg (http://i.imgur.com/0oto4.jpg)

[Massman]

So what is the solution?

[sin0822]

Patrick got it figured out, what was the cause of the issue?

temperature of the ICes or something, so he said he had a few tricks up his sleeve and took care of it. I guess something that wasn't set for non XMP kits is my guess.

It was odd because at a cold boot he and I could both once in a while boot up at 2400mhz depending on the board, UD5H it worked for him, and UP5 it worked for me twice. Anyways this fix works all the time, if you set your memory and don't touch it, i can boot up on a cold boot, or i can restart the machine 50 times and have the same OC and stability, I tested those two conditions many times on both boards. Ran 32m many times both at 2400 and 2600. I didn't try 2666, i was already at 1.7v lol.

if anyone is wondering i used these timings:
Yea try this:
10
12
12
25

Auto
6
6
12
Auto
8
147
6
27
T1
Auto
Auto

7936
69
5
5
4
4
6
3
6
5
5

[Leeghoofd]

Guys I want the GB board to boot at 2600 without a XMP profile loaded. Yes indeed with subtimings set to AUTO. That was the intention why I started this thread :p If you start to set tRSSR and co to 6 ofcourse it will boot... We don't need magic bios fix for that :p

FYI I can't get the Plats working properly on the UD5H, without pumping wacky voltages... can you plz share the new bios asap to verify if something has changed in RAM OC behaviour...

[dinos22]

always a killjoy

[sin0822]

I confirmed I will now post it publicly: http://www.mediafire.com/download.php?c88l5bnj86gx37n

[sin0822]

also here is for UP5 Th http://www.mediafire.com/download.php?2a68g1hg5iv08o7

make cure to power cycle the board after flashing.

[dinos22]

I think every bios update should be power cycled but for f6 version it's a must

[Leeghoofd]

Reflashed F10X, first flash seemed bonkers. Board hang at bios screen and nothing, not even XMP profile worked anymore ( http://www.youtube.com/watch?v=43yL-GD5ThU ) . Reflashed, reset cmos, loaded optimised defaults, powered down and rebooted seemed to fix the hanging screen and previous XMP problem. But but but...

Nothing seems to have changed for my hardware setup or the topic of this thread... Some now known subtimings are still too tight if left on AUTO... sigh...

If I change TRRD to 6 and tRRSR and tWWSR to 5 she boots fine at 2666Mhz... otherwise 15-03 switch off... or I can load the XMP profile wOOOOOOOOOOOOOOOOOOOOOOOOt, but that's for pussies :p

[sin0822]

so you messed up a simple BIOS flash?

Zeneffect said that his flares kit now is much easier to work with with this BIOS. I would say things that mattered where made better, you obsession with wanting timings that are to tight to be loosened is just a tad odd. IDK anyone other than you who refuses to load modern XMP profiles.

[l0ud_sil3nc3]

so you messed up a simple BIOS flash?

Zeneffect said that his flares kit now is much easier to work with with this BIOS. I would say things that mattered where made better, you obsession with wanting timings that are to tight to be loosened is just a tad odd. IDK anyone other than you who refuses to load modern XMP profiles.

I never, ever load XMP profiles, I prefer the old fashioned way by manually setting primary/secondary timings.

Hopefully the UD3H gets the fix soon, it was a nightmare trying to get c7 Flares to boot at 2600c8 with tight subs, I had to use the 2400 divider and then raise bclk to get there because the 2600 divider was a no go. Plus profiles are beyond broken, at least once you save them they are.

Although Albrecht is correct you should be able to boot at 2600 on auto, just saying.

. . .but I agree that tighter timings are better than looser and the UP5 is pretty efficient with auto settings and seems to work a bit better than it's predecessors, so can I haz UD3H bios:D

[stasio]

^No big deal mod UD5H BIOS for UD3H.

[sin0822]

I never, ever load XMP profiles, I prefer the old fashioned way by manually setting primary/secondary timings.

Hopefully the UD3H gets the fix soon, it was a nightmare trying to get c7 Flares to boot at 2600c8 with tight subs, I had to use the 2400 divider and then raise bclk to get there because the 2600 divider was a no go. Plus profiles are beyond broken, at least once you save them they are.

Although Albrecht is correct you should be able to boot at 2600 on auto, just saying.

. . .but I agree that tighter timings are better than looser and the UP5 is pretty efficient with auto settings and seems to work a bit better than it's predecessors, so can I haz UD3H bios:D

How are the profiles broken? Their real use is to go from one board with the same BIOS to another board with the same BIOS.

Some things random things even in the non OC section might change or be added in separate BIOSes, they will cause issues with the profiles, like CPu PLL OV is one of them.

I have made and distributed profiles on the UD3H what BIOS or what were your issues with the profile sharing?

[dinos22]

Hopefully the UD3H gets the fix soon, it was a nightmare trying to get c7 Flares to boot at 2600c8 with tight subs, I had to use the 2400 divider and then raise bclk to get there because the 2600 divider was a no go. Plus profiles are beyond broken, at least once you save them they are.

didnt Zeneffect send you that board and RAM pretested at 2600? He had no issues with that specific hardware, how come you had a nightmare OCing them :D

[l0ud_sil3nc3]

didnt Zeneffect send you that board and RAM pretested at 2600? He had no issues with that specific hardware, how come you had a nightmare OCing them :D

yup and the profile he saved would not boot with the exact hardware he tested with.

1.4 vcore
1.215 vccio
1.125 vccsa
1.795 cpu pll

the only difference is that I used a different chip (weaker IMC)

I tried higher and lower vcore, vccio and vccsa along with cpu pll to no avail, to be honest this chip could just be that bad.

plus his profile for some reason had vdimm set at 1.97v and he told me he has never used that type of voltage, coupled with the fact that I saw 200 bclk (9ghz screen:D) in Windows under cpuz, but that is probably a cpuz bug.

this was on bios F9.

ps I also tried a few different kits of memory, not just the Flares and all kits can do 2600 no problems on other boards.

[Leeghoofd]

so you messed up a simple BIOS flash?

Zeneffect said that his flares kit now is much easier to work with with this BIOS. I would say things that mattered where made better, you obsession with wanting timings that are to tight to be loosened is just a tad odd. IDK anyone other than you who refuses to load modern XMP profiles.

Sometimes it will be better to post nothing then these types of comments due to your affiliation with a certain brand dude... sometimes posting these things will do more harm then what you think. Do you care ? Then plz don't treat us in this manner, as this will backfire in the long end...

I just adressed what to me feels as an issue and a lot of end users experienced the same thing with older ram on the GB boards (so plz don't talk about XMP profiles). We provided a workaround with the 3 looser timing settings. PJ even showed it to Hicookie in Paris. If you are too stubborn or too proud to let these settings be adjusted, well that's then darn good for you...

I just tell you a lot of end users ( not OCers hunting down records or knowing their way in the bios ) prefer other brands just due to the easy nature of their biosses... You want to increase sales ? Then make the biosses more forgiving at boot...

[dinos22]

hey loud it could be bad IMC but maybe it's contact issue also!!!

clean the memory gold fingers with electric contact cleaner spray as well as CPU pad and CPU socket. Clean DIMM slots too with it and try one stick at a time and try reinserting if it doesnt pick it up or let you OC to the max the first time

[l0ud_sil3nc3]

hey loud it could be bad IMC but maybe it's contact issue also!!!

clean the memory gold fingers with electric contact cleaner spray as well as CPU pad and CPU socket. Clean DIMM slots too with it and try one stick at a time and try reinserting if it doesnt pick it up or let you OC to the max the first time

I was thinking the same thing, I usually try to wipe any excess vaseline off the memory and definitely use the heat gun over the dimm slots to warm up any excess vaseline.

however the UD3H has no vaseline on it, or Zen did an incredible job cleaning it off:p

thanks again man, I will give everything a thorough cleaning, and give it another shot when I fill up on Monday.

[dinos22]

not just vas boards made, even just pure air

[l0ud_sil3nc3]

not just vas boards made, even just pure air

not quite sure I follow. . .

[dinos22]

even non insulated boards can sometimes have issues with RAM not doing the clocks but when you clean the dimm slots or reinsert it will sort itself out. if you're james, there could be a bread crumb in there!

[l0ud_sil3nc3]

^lol gotcha

pretty sure now after reading in my head it should have been "not just vas boards mate, even just pure air"

no bread crumbs in the dimms here:p

[Leeghoofd]

Already flashed back to F9 bios... too many issues on my daily rig with F10X... Worked okayish yesterday with the TridentX kit, this morning PC hangs again at startup screen. Hope the others have more success with it...

[zeneffect]

George,

I did an incredible job cleaning off the Vaseline. :)

The profiles will get corrupted (I think on UD3 profiles 1 and 2 are corrupted. The last one is ok iirc) when you fail the OC then spam clear cmos trying to recover.

@dino

James isnt the only one with bread crumbs. Sandwiches are popular world wide. Im guilty of using my motherboard as a plate sometimes. food > benching > sleep.

UD3 can be a PITA with PSC memory with a weak IMC. I had profiles saved on there for 2600 boots every time but they may have gotten corrupted (1.97v is crazy. I have no idea where that came from) When I dont have problems... it means everybody else does. When everybody else has no problems, that means that only I have issues... story of my life. Just be patient George, I'm fairly confident that the sammy fix bios will make things work well for you. If you want to, we can trade out that UD3 for a UD5. I can pre-test the bios for you as I still have alot of different PSC left over here.

*edit*

2666 is unstable with 32m F6X UP5
2600 is unstable with 32m F6X UP5

both tested on flares. Ive gotten 2600 stable with F5b and F5x. It I can boot higher now and alot easier, I lost some stability. checking gtl, mem tweakit, and cpu-tweaker, the settings look the same. Hopefully next bios revision will correct this :(

[sin0822]

Sometimes it will be better to post nothing then these types of comments due to your affiliation with a certain brand dude... sometimes posting these things will do more harm then what you think. Do you care ? Then plz don't treat us in this manner, as this will backfire in the long end...

I just adressed what to me feels as an issue and a lot of end users experienced the same thing with older ram on the GB boards (so plz don't talk about XMP profiles). We provided a workaround with the 3 looser timing settings. PJ even showed it to Hicookie in Paris. If you are too stubborn or too proud to let these settings be adjusted, well that's then darn good for you...

I just tell you a lot of end users ( not OCers hunting down records or knowing their way in the bios ) prefer other brands just due to the easy nature of their biosses... You want to increase sales ? Then make the biosses more forgiving at boot...

What the f**k is my affiliation with GB that is different than yours?
.

[l0ud_sil3nc3]

George,

I did an incredible job cleaning off the Vaseline. :)

The profiles will get corrupted (I think on UD3 profiles 1 and 2 are corrupted. The last one is ok iirc) when you fail the OC then spam clear cmos trying to recover.

@dino

James isnt the only one with bread crumbs. Sandwiches are popular world wide. Im guilty of using my motherboard as a plate sometimes. food > benching > sleep.

UD3 can be a PITA with PSC memory with a weak IMC. I had profiles saved on there for 2600 boots every time but they may have gotten corrupted (1.97v is crazy. I have no idea where that came from) When I dont have problems... it means everybody else does. When everybody else has no problems, that means that only I have issues... story of my life. Just be patient George, I'm fairly confident that the sammy fix bios will make things work well for you. If you want to, we can trade out that UD3 for a UD5. I can pre-test the bios for you as I still have alot of different PSC left over here.

*edit*

2666 is unstable with 32m F6X UP5
2600 is unstable with 32m F6X UP5

both tested on flares. Ive gotten 2600 stable with F5b and F5x. It I can boot higher now and alot easier, I lost some stability. checking gtl, mem tweakit, and cpu-tweaker, the settings look the same. Hopefully next bios revision will correct this :(

No worries I am sure they will resolve it in a bios update, however in the meantime I have other boards to play with.

so F6X lost the ability to run 32m at 2600?

so you would stick with F5B for the UD5H/UP5?

[Leeghoofd]

@Sin

Plz load the XMP 2600 profile for eg the Flare kit... good luck ! You want to help these boards to evolve or stand still ?

Secondly I never said you are paied by Gigabyte, people just link you to them as you are everywhere concerning help threads and co ( kuddos to you )... good advice: just be carefull what you preach and stop using the F word. We are all adults, no need for that temper mode at all...

[sin0822]

I just want to say this isn't a final release, it was just provided to me to test out to make sure samsung and some of the other things i complained about were fixed, i think they will just incorporate these fixes into future BIOSes. All i am trying to say is what i said in the first post of mine in this thread, I cannot recommend changes that will affect the performance of these boards because i don't work for gigabyte.

F6X is just to confirm a fix, it isn't a beta release. it was literally complied and sent to me like minutes after it was compiled. i was hesitant to put it online for this reason, but I had to go ahead to post it and I wanted to see if it fixed the samsung issue for many people.

I was also told this:
That this BIOS for UP5 requires a bit more tuning than before for 2-DIMM at 2600 than for 4-DIMM. At least with samsung, the same might be for the flares. I would try changing the stability level to like both 4s or both 5s, or even try 14s or 25s. Once hicookie told me that some kits with higher frequency require that change int he stability level, but it differs per kit and to find out you have ot try all levels, which i think is a pain.

Now leeghoofd, if you load the XMp and it wont work at all, then that is a problem i can report to be fixed, but from what i am seeing that isn't what you reported as a problem, you are reporting that without XMP it isn't working.

[Leeghoofd]

My report here is when you dial in old rams and just set manually the main timings : no boot, yet this boots on other brands. A too tight timing set is NOT an efficiency tweak, it's against any logic. It's not that the competitors put in tRSSR 7 or such...
Secondly why are the timing readouts borked in the bios ? Why even show them ? Confusing way of approaching things...

That's all this entire thread is about, it's not about loading or not loading XMP profiles ( as the BBSE and rams we prefer for benching don't have profiles for Ivy high clocks ). It's what works on other boards but doesn't work on GB boards... Why not fix it ? Beats me...

Next objective 16Gb 2666C10 as this is doing my head in ( again ) and this time with XMP profile :p

[l0ud_sil3nc3]

Now leeghoofd, if you load the XMp and it wont work at all, then that is a problem i can report to be fixed, but from what i am seeing that isn't what you reported as a problem, you are reporting that without XMP it isn't working.

You do realize that the G.Skill Flares XMP is 2000 7-9-7-24 right? We are talking about 2 dimms at 2600mhz correct?

just making sure we are on the same page.

[sin0822]

Okay. if you have issues with XMP you can email me and I will send the data so that it can be fixed. Otherwise if the Samsung issue is fixed then the F6X and F10X have done their jobs, your other issue might be fixed soon, but obviously this BIOS wasn't meant to deal with that.

[dinos22]

FFS sin and goofball take it easy, we are on a bloody online forum!

[dinos22]

Goofball so in your opinion what timings when set to auto should improve stability on nonXMP ram other than TRRD =7?

[Leeghoofd]

How about from 2600 divider on Trrd 7 - tRRSR & tWWSR 6, this should allow most dimms to boot if they can make it and the CPU IMC can handle it... tweakers can tweak later on...

It's weird at 2400 if I recall correct the bios sets 5, which boots also for me at 2600 if manually set. Yet if I leave it to auto the board must set 4 or such and hinders it from posting... 6 should be a safe setting. I honestly think the bios engineers made an error there...

Atm investigating why the 16GB 2666 doesn't work, comparing timings with the ROG board, can only boot at 2400mhz ( with XMP profile loaded wtf )

[dinos22]

so just 2600 divider for those timings?

[Leeghoofd]

yes 2400 always posted fine. Let's start with the 2600 divider with them 3 timings programmed a bit looser and see if it allows most of them old sticks to boot...

[dinos22]

tWWSR??

[Leeghoofd]

Yes increases copy bandwith alike tRRSR, if set too tight no post...

[zeneffect]

No worries I am sure they will resolve it in a bios update, however in the meantime I have other boards to play with.

so F6X lost the ability to run 32m at 2600?

so you would stick with F5B for the UD5H/UP5?

I went back to F5B for UP5. For UD5 I use F5 bios. its been the best bios for me so far, but some people dont like it. I dont know why, it seems the most "workable" to me.

[M.Beier]

What the f**k is my affiliation with GB that is different than yours?
.

You are FAE or 'technical' marketing, same as dino, only dino has the balls to admit it, and not manipulate.
To my knowledge, Leeghoofd is a hobby overclocker, and not a professional representing a company - I might be wrong, but I do think that was the difference he was refering to.

Alby; I admire that you keep on fighting for a better BIOS, I gave up after trying 9 of them(sorry, 12 was total amount across 2 boards), I simply just ended up branding the board as insuffecient, due BIOS.
Great hardware all the way through, it really is a pitty that the BIOS doesnt justify the hardware, Gigabyte used to have the best, but tides have changed - I am still waiting for another change, bringing my favorite motherboard vendor back on top of the game.

pro;
Problem for many(hint hint; ME, the GINGER), is that if you dont have a XMP, such as when you are binning cheap samsung ram - you are pretty much fucked.

EDIT; Fixed count, 3 BIOS tried on another board, 9 on UD5H... Pardon my ignorance of fast posting!!

[dinos22]

am i blind or what? i cant see that timing in bios?!?

[dinos22]

He is neither one of those. He's just an end user like you but he knows what he's doing/talking about so you confuse him with a paid professional


:DYou are FAE or 'technical' marketing, same as dino, only dino has the balls to admit it, and not manipulate.
To my knowledge, Leeghoofd is a hobby overclocker, and not a professional representing a company - I might be wrong, but I do think that was the difference he was refering to.

Alby; I admire that you keep on fighting for a better BIOS, I gave up after trying 12 of them, I simply just ended up branding the board as insuffecient, due BIOS.
Great hardware all the way through, it really is a pitty that the BIOS doesnt justify the hardware, Gigabyte used to have the best, but tides have changed - I am still waiting for another change, bringing my favorite motherboard vendor back on top of the game.

pro;
Problem for many(hint hint; ME, the GINGER), is that if you dont have a XMP, such as when you are binning cheap samsung ram - you are pretty much fucked.

[Leeghoofd]

tWWSR??

http://dl.dropbox.com/u/66364457/dinoblindman.jpg

[Christian Ney]

Dino just got a new nickname, BlindClocker

[Massman]

As for PSC/BBSE, it seems to come down to specific kit compatibility. I'm not surprised Zen's kit is working so well; Hicookie has one that works perfectly too. Pt1t had one that was doing 1270+ 8-11-7 with 1.72V at the GBT France office too (later it no longer worked). When I tested the BBSE in GBT France, couldn't get it to run anywhere near 1250 8-12, but when swapping out for the 2nd-gen Flare kit those speeds were easy.

Some kits just work and others really don't. It has very little to do with IMC quality or 'effort'.

[Leeghoofd]

Pt1t had one that was doing 1270+ 8-11-7 with 1.72V at the GBT France office too (later it no longer worked).

You gave them a little voltage bump :D ?

[zeneffect]

as far as PSC is concerned, Ive gotten all my PSC kits to work at 2600 with the right amount of "effort"

granted, its not as easy as it should be, but it is doable. The statement that some kits "simply do not work" does not reflect the behavior that I'm seeing. some kits are a royal pain in the ass but they do work. All timings are put in manually btw, I dont know if this is a key difference between your end user experience and mine in this matter.

[l0ud_sil3nc3]

As for PSC/BBSE, it seems to come down to specific kit compatibility. I'm not surprised Zen's kit is working so well; Hicookie has one that works perfectly too. Pt1t had one that was doing 1270+ 8-11-7 with 1.72V at the GBT France office too (later it no longer worked). When I tested the BBSE in GBT France, couldn't get it to run anywhere near 1250 8-12, but when swapping out for the 2nd-gen Flare kit those speeds were easy.

Some kits just work and others really don't. It has very little to do with IMC quality or 'effort'.

Was this with auto subs or tightened down secondary and tertiary timings?

[Massman]

as far as PSC is concerned, Ive gotten all my PSC kits to work at 2600 with the right amount of "effort"

granted, its not as easy as it should be, but it is doable. The statement that some kits "simply do not work" does not reflect the behavior that I'm seeing. some kits are a royal pain in the ass but they do work. All timings are put in manually btw, I dont know if this is a key difference between your end user experience and mine in this matter.

Maybe I should rephrase my statement. I never got them to work consistently. Every kit managed to do 2600C8 at some point, but hard reboots or CMOS clearing would break that capability.

Was this with auto subs or tightened down secondary and tertiary timings?

Tried any combination of auto/manual/loose/tight main/secondary/tertiary timings.

But I might just be missing something. If someone can give me a clue, I'll set up the rig right away :)

[l0ud_sil3nc3]

Maybe I should rephrase my statement. I never got them to work consistently. Every kit managed to do 2600C8 at some point, but hard reboots or CMOS clearing would break that capability.

So all 32M stable I assume, just resetting/clearing cmos and reentering timings to get it back up and running.


Tried any combination of auto/manual/loose/tight main/secondary/tertiary timings.

But I might just be missing something. If someone can give me a clue, I'll set up the rig right away :)

Hey don't forget me I need that clue as well:D

Well good to know that it needs finessing and does work, sometimes.

[Massman]

So all 32M stable I assume, just resetting/clearing cmos and reentering timings to get it back up and running.

Yes, 32M stable.

No, if it was just a matter of re-entering timings/voltages to get it back up and running, it would be consistent behavior :). The problem is that same settings result in different behavior.

[zeneffect]

Maybe I should rephrase my statement. I never got them to work consistently. Every kit managed to do 2600C8 at some point, but hard reboots or CMOS clearing would break that capability.



Tried any combination of auto/manual/loose/tight main/secondary/tertiary timings.

But I might just be missing something. If someone can give me a clue, I'll set up the rig right away :)

I'd have to agree with this statement. It is not consistent, though when I actually WANT to get it to run at desired speed, I can... it just takes a while and more effort than it should.

as far as "missing" something? yes you are... good bbse. The blue ripjaws (both kits) work without issue 100% of the time. :p

Do you think that when one kit works 100% and another kit doesnt, but sometimes does that there a variance in the spd? im willing to put up a thaiphoon burner dump of the blue ripjaws that i have if you want to experiment?

[sin0822]

What about with F10X or F6X on the UP5 TH is it more or less consistent? I ask because the test BIOSes use a different start up sequence which bypasses issues with thermals.

With this BIOS i noticed if one goes back in and changes a setting on the samsungs(like even go from 26x to 24x) then it might fail to boot, however booting with that memory timings set over and over isn't and issue either is a cold boot from the cold boots i tested. To make it boot again one must reset the CMOS and either load your profile or not.

IMO even if i set all the timings manually it is still a good idea to set the XMP profile, that way the timings int he background are consistent, some of those timings change also with dividers.

[zeneffect]

F6X is definately more consistant, but I cant get 32m to pass like I can on F5X and F5B. I cant even get to loop 2 with PSC.

[Massman]

I'd have to agree with this statement. It is not consistent, though when I actually WANT to get it to run at desired speed, I can... it just takes a while and more effort than it should.

as far as "missing" something? yes you are... good bbse. The blue ripjaws (both kits) work without issue 100% of the time. :p

Do you think that when one kit works 100% and another kit doesnt, but sometimes does that there a variance in the spd? im willing to put up a thaiphoon burner dump of the blue ripjaws that i have if you want to experiment?

Please do upload the SPD of your good kit - I'm willing to try everything (no, not in that way) to get 2600C8 running smoothly.

This kit is not strong enough?

http://hwbot.org/blog/wp-content//ripjawsx.jpg

[dumo]

Most blue Gskill 2133C8 can do 2600C8 air, but not all can run consistently

[Leeghoofd]

Did 2600 c8-11-8 1.85Vdimm on these PJ, one time lol for the 5ghz compo :p

I had major issues with F10X on UD5H regarding daily stability :
- erratic LAN port behaviour that stops working, never had it with F9, now daily 2-3 times I loose the port, lappie still works fine... so it's not the internet con, nor router...
- sometimes complete freeze at bios logo, if it passes the logo the rig is stable


Main UD5H issue atm : manual timings nor XMP profile working with 16Gb Corsair 2666C10 kit. 2400Mhz is stable, from 2600 it doesn't boot...

@Dumo you have the 16Gb tridentX kit can you check plz if it works fine with your kit. If so, can you plz pass timing details via GTL or memtweakit. I'm cross checking with the Gene board as the Corsairs work there at 2800mhzC11...

[zeneffect]

Green PCB > Black PCB (in most cases)

I will upload the SPD tonight.

[dumo]

@Dumo you have the 16Gb tridentX kit can you check plz if it works fine with your kit. If so, can you plz pass timing details via GTL or memtweakit...Works fine with G/F/E
http://img521.imageshack.us/img521/7789/screenshot011lp.png

Green PCB > Black PCB (in most cases)
Green and black pcbs
http://img845.imageshack.us/img845/1231/screenshot045y.png

[Leeghoofd]

Fun thing now with my UD5H is that it looses track of time... no idea what's going on since the F10X flash...

Thx Dumo for the info... stellar kit and IMC there man :)

[l0ud_sil3nc3]

Fun thing now with my UD5H is that it looses track of time... no idea what's going on since the F10X flash...

Thx Dumo for the info... stellar kit and IMC there man :)

Must be Gigabyte implementing their new pause feature:D

They still have some bugs to work out or so I have heard.

[zeneffect]

well... here is the SPD of one of the blue ripjaws I have that give me no headache at all. hope it helps, I will keep my fingers crossed.

[Massman]

well... here is the SPD of one of the blue ripjaws I have that give me no headache at all. hope it helps, I will keep my fingers crossed.

I've compared the SPD with the ones I have here and they seems 100% identical. Your kit has a green PCB?

If one type of PCB works better than the other, would that point to an electrical compatibility problem rather than SPD or weak mem IC? It would line up with the idea that you need to finetune the ref voltages?

[dinos22]

i have a second kit of green pcb based ram like zen's which OC the same as his black pcb ram. This other kit is better for air but not great subzero

[Massman]

Zen's are black PCB? Or green PCB?

[dinos22]

good ones are green but he has black pcb also. i have a green based PCB as well but it doesnt clock that well, just under 1300

[Sam OCX]

You can find gems with black PCB too, only a matter of binning. I saw more gems with green PCB, though.
After all, 2133C8 RipjawsX clock better than any other BBSE model out there, and they're so rare that you don't usually have choice.

[Massman]

good ones are green but he has black pcb also. i have a green based PCB as well but it doesnt clock that well, just under 1300

I assume you verified that 'just under 1300' is indeed the limit of the memory kit and not the board, right?

Just set up my rig again to see how high the blue BBSE kit can go with the new F6x BIOS. Test settings:

- DDR3-2400 divider
- C8-11-8
- 1.85Vdimm
- Vref all auto
- loose 2nd/3rd subs
- cpu around 4500

So, started testing:

- 100 BCLK BIOS to OS-> 32M stable
- 100 to 102 BCLK in OS -> fail after 1 loop
- 102 BCLK BIOS to OS -> 32M stable
- 102 to 104 BCLK in OS -> fail after 3 loops
- 102 to 104 BCLK BIOS to OS -> fail after 5 loops
- 104 to 100 to 104 BCLK BIOS to OS -> 32M stable
- 104 to 106 BCLK in OS -> fail after 5 loops
- 106 to 100 to 106 BCLK BIOS to OS -> can't boot/post

After that I had to clear CMOS and lost all capability for running 2400/104. It's now crashing at the 1st loop. Same settings in BIOS, same methodology, same everything. Given that fiddling with memory timings does not change the stability grade (loose/tight all result in early crashes) and that a certain method of power cycling does help, I think it's an electrical compatibility problem rather than BIOS at this point.

//edit: just a couple of minutes ago I was stuck at loop1 failure and after fiddling with auto/normal vref setting and power cycles, the memory is running 32M stable again :).
//edit2: soft reboot request in OS gave me a hard reboot instead; back in OS ... loop1 failure
//edit3: did 104 -> 100 -> 104 in BIOS (including power cycles) and it's running 32M stable again (http://hwbot.org/blog/wp-content//2400-104-boot-after-power-cycle-2.jpg). Did not change any voltage or timing.
//edit4: soft reboot request in OS turned into hard reboot instead (again); back in OS ... loop3 failure (http://hwbot.org/blog/wp-content//2400-104-boot-after-power-cycle-2-new-power-cycle.jpg).

[sin0822]

maybe i should have posted this or i might not have here, but F6X and F10X use a different bootup sequence which bypasses some things which limit the samsung's speeds, The BIOS isn't meant to be a beta BIOS in line with the others, just to fix 1 or two problems. The(person who made the BIOS) already knew it would have a negative affect on some other types of memory.
The inconsistency thing is a bit odd too, like you said, a different boot up sequence works. Like if I set 2600mhz with the samsung ULV and then goto 2400 then it will fail to boot, but if i go back in and change the multiplier to 26x no issue it works. I have to clear CMOS and load a saved profile for the 24x to work, and if it doesn't then apply the settings again and it works. It seems that it is fine with booting and such unless you go into the BIOS and change a memory setting. Of course with BBSE it prob is different, but this might be good for booting with higher multipliers for screenshots and such, but this boot sequence will just be engaged when the board recognizes samsung ULV. Otherwise it wont be like this in future BIOSes for your BBSE.

[Massman]

It's not a BIOS version problem.

[Leeghoofd]

C'mon Gigabyte Bios team get your act together plz... instead of hiding behind this automated reply:

http://dl.dropbox.com/u/66364457/meme.jpg

[sin0822]

all I am saying that it hold more weight if you didn't say it was on F6X, because F6X has known issues with some memory, and thus it will just be put into the issues that only exist on F6X, and basically be thought of as only occurring with F6X. That is what i am saying.

[zeneffect]

massman, try same settings but pull trfc to 95 or below. I dont think i have EVER ran my bbse that loose, but on UD5, trfc above 95 gave me headaches. during giga 4.5 32m comp, i used trfc 70.

also, are you sure you arent "overvolting" your bbse? my kit is very particular about voltage. 1.875 it will pass 1.880 fail on up5.

2nd

i found more 32m stability with bbse when i left imc voltage setting @ auto. i use vtt 1.2. I would use 1.15 imc thinking the .05 rule still applied and the voltages were tied, but when left at auto it actually worked better.

im not doubting your testing ability, im just relaying what i experienced in hopes it finds you a solution.

F6X was better for booting PSC. once i found i couldnt 32m the way F5B would (or at all), i flashed back.

settings i can boot easily with on all air:

1.2v VTT
auto IMC
1.875 vDIMM

2400 memory multiplier
107.15 bclk
memory timings top to bottom
8-11-7-21-33-4-4-16-19-7-75-4-16-1-2-2-A-A-1-1-1-1-4-1-4-4-4

these would be my sort of tight air baseline.

[dinos22]

I tried F6a bios, it was a headache to get ram stable even though it boots higher

stick to F5b i reckon

for my runs with PSC last night i had to get timings right, very tight or it would not even post LOL strange but it requires a lot of experimentation to get it going well

for 1300-1328MHz last night i did

2.1vdimm -50C (had trouble getting colder). in my first bench with RAM at -150C i could do 7-10-7 but last night i couldnt get 1300 to even post
DIMM slots 1&4 were clocking a bit higher and easier to have my ghetto pot installed
7-11-7-25(22)-10-4-4-8-12-6-74-4-16-1-A-A-15650(doesnt do shit)-122(doesnt do shit)-1-1-1-1-4-A(1 in OS)-A(4 in OS)-4-4

really bad wazza as i was just trying to get one run and keep at it but just didnt have the legs, kept crashing early. i think i will try again with -150C on RAM(gotta redo my ghetto RAM pot). Maybe 7-10-7 will work, that would improve efficiency considerably as well....particularly in closing loops

my kit is the rubbish bin that fester had out of 8 sticks, james has insane bin for those two 1400 sticks, fester has some nice bin also. I'm happy with a door stopper kit that goes as hard as these do even though fester says they are average LOL. It's hard work though, still trying to figure it out. subzero RAM is an interesting challenge

http://img442.imageshack.us/img442/3715/screen00b.jpg

zen's BBSE kit is very specific with voltage. 1.88 or 1.885 in bios depending on board was basically perfect voltage and it would behave best once i used that

if RAM doesnt run like it should it's usually contact issue 95% of fixes have had contact cleaner and hair dryer treatment with success

i also do not let the board boot loop if it fails OC, just power down, reset.
i also look at the bios LED lights and if i see a flight to backup bios i power cycle right away, no point wasting time giving myself a headache

[dinos22]

LOL at goofball talking about efficiency and he wants the bios to have a default TRRD7 and TRRSR at 6

[Leeghoofd]

Just wanted to lighten up the thread Dino... you know why I want them timings for the daily tweakers :p

[dinos22]

i know, was hoping someone would make a goofball meme :D

[Massman]

massman, try same settings but pull trfc to 95 or below. I dont think i have EVER ran my bbse that loose, but on UD5, trfc above 95 gave me headaches. during giga 4.5 32m comp, i used trfc 70.

also, are you sure you arent "overvolting" your bbse? my kit is very particular about voltage. 1.875 it will pass 1.880 fail on up5.

2nd

i found more 32m stability with bbse when i left imc voltage setting @ auto. i use vtt 1.2. I would use 1.15 imc thinking the .05 rule still applied and the voltages were tied, but when left at auto it actually worked better.

im not doubting your testing ability, im just relaying what i experienced in hopes it finds you a solution.

F6X was better for booting PSC. once i found i couldnt 32m the way F5B would (or at all), i flashed back.

settings i can boot easily with on all air:

1.2v VTT
auto IMC
1.875 vDIMM

2400 memory multiplier
107.15 bclk
memory timings top to bottom
8-11-7-21-33-4-4-16-19-7-75-4-16-1-2-2-A-A-1-1-1-1-4-1-4-4-4

these would be my sort of tight air baseline.

Thanks for the input, man! Much much appreciated! :).

I'll check for Vdimm and 'overvolting', although I have tried that before too. In the "'51' debug code - any fix or workaround? (http://hwbot.org/forum/showthread.php?t=50593)"-thread I also mentioned that I could work around the 51 problem by pumping 1.9V for POST and then lower it to 1.85V or 1.8V in OS for 32M stability. For now, I'm using what Leeghoofd says he used on the Gene to run 2600C8-11-8 32M. I do remember having to tighten up the Flares when looking for stability. Thanks for the reminder! Fyi, this is what I ran the Flare kit at when the UD5H would randomly boot up at 1300: http://i46.tinypic.com/24exjth.jpg

Just tried the settings you suggested on F5b and anything above the DDR3-2200 divider would just return an overclocking failed warning. I'll give it another go later today

[zeneffect]

Thanks for the input, man! Much much appreciated! :).

I'll check for Vdimm and 'overvolting', although I have tried that before too. In the "'51' debug code - any fix or workaround? (http://hwbot.org/forum/showthread.php?t=50593)"-thread I also mentioned that I could work around the 51 problem by pumping 1.9V for POST and then lower it to 1.85V or 1.8V in OS for 32M stability. For now, I'm using what Leeghoofd says he used on the Gene to run 2600C8-11-8 32M. I do remember having to tighten up the Flares when looking for stability. Thanks for the reminder! Fyi, this is what I ran the Flare kit at when the UD5H would randomly boot up at 1300: http://i46.tinypic.com/24exjth.jpg

Just tried the settings you suggested on F5b and anything above the DDR3-2200 divider would just return an overclocking failed warning. I'll give it another go later today

on UD5 + Flares I would start at 102bclk and walk it up .5 bclk at a time to avoid the hard shutoff with the 2400 multiplier + high bclk. for 2600 multi, I would have to start flares at 1.72 volts, let it post, re-enter bios then up voltage to 1.82v again in bios save and exit for 2600 32m stability.

Dino is right about my bbse ram kit though. UD5 = 1.885v UP5 = 1.875v to tame the sticks.

if f6 bios performs (is stable to pass 32m predictably and consistantly) and keeps booting like it does now.... man that would be sooooo killer.

[zeneffect]

so you guys got this figured out yet? im working on some stuff that might cause heads to spin... just gotta find a imc that will cooprate with me at 2800 c7. currently 2750 is the limit for me for flares in dual chan before the imc gives up.

dino and fester got "the next step forward" working so im not sure if its the board rather than the combination of components and approach to overclocking.

[Vivi]

my 2nd cpu clocks memory much better than previous. 1250 8-11-9 is now possible where previously it wasnt at all. So its like youngpro said, some cpus are just better at clocking memory than others.

"just gotta find a imc that will cooprate with me at 2800 c7" you talking ln2 clocks here right :O

[zeneffect]

my 2nd cpu clocks memory much better than previous. 1250 8-11-9 is now possible where previously it wasnt at all. So its like youngpro said, some cpus are just better at clocking memory than others.

"just gotta find a imc that will cooprate with me at 2800 c7" you talking ln2 clocks here right :O

yes, no other way to do it :)

[dumo]

Is UP7 out yet?

[l0ud_sil3nc3]

^ not quite yet, but very soon I assume.

[sin0822]

soon enough, but no not yet. It seems cookie has been going around the world overclocking it with some other overclockers, good move IMO.
I actually just got a set of this memory, the one in the title.

[zeneffect]

heh way to bring it back on topic sin :p

so massman, any luck with flares?

[Leeghoofd]

Nope Zen, we called it quits at 6 o'clock in the morning after trying several GB boards... To be honest I was ready to burn something Tapakah style !

[Massman]

heh way to bring it back on topic sin :p

so massman, any luck with flares?

I have some screenshots on an USB drive that kinda shows the problem. Booted up with the Gene at 2666C8-12 1.85V easily whereas any GBT failed at 2600Cx. I'll upload them later.

By the way, at Campus Party we tested some really crazy memory that could boot up at 2800C8-12-8 on the M5E, but also failed at 2600 on UP5.

Did you ever test your kit on a M5x?

[zeneffect]

not yet, im going to do that today. tested on formula... its a breeze... but on giga... dun dun dun.... not so easy but i can still pull wacky stuff.

http://i48.photobucket.com/albums/f210/icon22976/Capture-27008m.jpg

[Massman]

Damn, look at that. Magic sticks!

Rumours go that Cookie has a couple of sticks that are even better than those 2800C8-12's we tested. I actually wonder what his sticks will do on an M5x :p

[zeneffect]

Damn, look at that. Magic sticks!

Rumours go that Cookie has a couple of sticks that are even better than those 2800C8-12's we tested. I actually wonder what his sticks will do on an M5x :p

lol would that mean i have 2 kits of magic sticks? those were flares in the screen shot. psczzy breezy ;)

tell me more about the 2800 c8's. what sticks are they?

[Sam OCX]

If the screenshot above is PSC, then you'd need LN2 for such clocks which makes the result not impressive.

Gigabyte has a lot of problems running PSC and BBSE, but on ASUS it's not a breeze either. The same kit/settings that boots up 1400 8-12-8 will frequently crash in 32M at 1300 8-12-8 for no apparent reason.

If we're talking about the same kit, the "2800C8" stuff is a 2400 9-11-9 rated G.Skill PIS set produced on week 1127.
On air, that does 32M at 1200 7-10-7 at 1.59V (auto subs), 1200 6-11-6 at 1.82V (auto subs) and 1300 7-12-7 at 1.71V (auto subs), roughly 40-50MHz or 0.05-0.06V better than any retail PSC I have ever tested or seen.

[zeneffect]

If the screenshot above is PSC, then you'd need LN2 for such clocks which makes the result not impressive.

Gigabyte has a lot of problems running PSC and BBSE, but on ASUS it's not a breeze either. The same kit/settings that boots up 1400 8-12-8 will frequently crash in 32M at 1300 8-12-8 for no apparent reason.

If we're talking about the same kit, the "2800C8" stuff is a 2400 9-11-9 rated G.Skill PIS set produced on week 1127.
On air, that does 32M at 1200 7-10-7 at 1.59V (auto subs), 1200 6-11-6 at 1.82V (auto subs) and 1300 7-12-7 at 1.71V (auto subs), roughly 40-50MHz or 0.05-0.06V better than any retail PSC I have ever tested or seen.

have you tried c7 2700 with flares on gigabyte? 2600 is a chore @ c8... remember... no sponsorship, limited funds, no ram pot, not getting super sauce sticks flown in from taiwan, and im a less than average overclocker (am using ln2 + paper towls on ram.) the point isnt efficency, but to show that c7 2700+ can be done on gigabyte by stupid users like me and thats it... and ive had zero problems running bbse on ud3, ud5, or up5. I think thats just me though and my kit isnt "ordinary" even though its a retail kit from newegg. ive only bought 2 kits of the blue ripjawsx ever and neither give me problems.

im not saying that gigabyte doesnt have problems... im just saying that you guys have more problems than i do it seems... which is strange to me because im not doing anything out of the ordinary or with any "special" hardware.

[Leeghoofd]

you got perfectly matching hardware Zen. We don't, not even after 5 x CPU's, 3 x UD5H's, 3 x Flare kits, 1x BBSE kit, 1x TridentX, 1 x Kingston Predator and the last drop are the Corsairs Plats which are doing my head in on this board... not even talking about stability, just booting lol

[zeneffect]

have you tried shaking a stick at it? im out of ideas :(

[GoriLLakoS]

I'm tryin here with 45x100 just to finalize a review about UP4 with latest F3 Bios but it seems i cannot make it work!

CPU: 3770K
Mobo: UP4
Ram: Corsair Dominator Platinum 2666

Without XMP but with 1.65V on ram and 9-9-9-24 2T and all the other subtimmings on auto except tRRD that i have it on 7
i am getting not exact in round on 32M before even 1st loop!

Any ideas?

[Leeghoofd]

@ZEN so are we mate, so are we... I might take it to the pope for a blessing :p

And to top it all off... my board doesn't even keep track of time anymore... lol... so I swapped it... Dino can pick it up and investigate it :)

16GB Gorillakos ?

[GoriLLakoS]

Kit is 4x4 but i am only using 2x4 in 1+3 Slots (black color)

[Leeghoofd]

tomorrow I'll have PJ's UP5 and my rams back can quickly see if I have same issues....

[zeneffect]

I'm tryin here with 45x100 just to finalize a review about UP4 with latest F3 Bios but it seems i cannot make it work!

CPU: 3770K
Mobo: UP4
Ram: Corsair Dominator Platinum 2666

Without XMP but with 1.65V on ram and 9-9-9-24 2T and all the other subtimmings on auto except tRRD that i have it on 7
i am getting not exact in round on 32M before even 1st loop!

Any ideas?

isnt the memory rated for 2666 10-12-12-31? 9-9-9 might be too tight. I dont know 100% because i dont have this memory but it seems too tight to me given the spec. try 9-11-10?

[Leeghoofd]

Think he's running 1600Mhz or so Zen...

[l0ud_sil3nc3]

^oh wow that's bad, having issues at default is bad all around.

[zeneffect]

Think he's running 1600Mhz or so Zen...

wow ok... i thought he was running 2666. yea... not good.

[sin0822]

I will test that combo when I have a chance. I have that memory too.

[Vivi]

i think samsung just dont like 9-9-9 in general. even for me 9-12-11 is pushing it, and 9-11-11 is just not really possible. but 1400 9-12-12-28 boots just fine and runs 8m, 32m not yet ^^

[dinos22]

it's performance memory, you have to enable XMP for it to work properly, that's what it's there for to set timings and voltage for you for those that dont know what they are doing!

im running the Corsair Platinum Dominator 16GB 2133 kit at 2400MHz in my daily 24/7 rig now! :)

[dinos22]

nice clockage Zen

Fester is getting 2740MHz 32M with 7-11-7 and supertight subs 111144144 iirc on UP5 with his PSCs, gotta work on this gear to make it a bit more OC friendly though

[der8auer]

nice clockage Zen

Fester is getting 2740MHz 32M with 7-11-7 and supertight subs 111144144 iirc on UP5 with his PSCs, gotta work on this gear to make it a bit more OC friendly though

I don't know what causes this issue but no matter what I can't get my PSC kit properly stable on the GBT boards.

Sam tested the kit and it did 1200 6-9-6 32m stable on the M5E. I can't even get close to this. Always 15 or 51.

[Massman]

Is that on air or with the memory under LN2, Dino?

Maybe we should track what PSC/BBSE is actually working with what MB/BIOS/REV? Track it down to serial number?

[l0ud_sil3nc3]

^ I would assume cold with tertiary timings that tight

[GoriLLakoS]

It is the 1st time i'm playing with Giga so i am searching how it is working.
I changed to Profile1 of XMP and relax the timmings at 9-12-12-12-28 2T
and with a magic way, SUperPi 32M was done. Still ram is at 2133MHz as before.
As you said and Aristidis told me yesterday, for Samsung 9-9-9 is very tight and they don't like it.

Thanks :)

[Sam OCX]

Sam tested the kit and it did 1200 6-9-6 32m stable on the M5E. I can't even get close to this. Always 15 or 51.ehr, 6-11-6

[Vivi]

ehr, 6-11-6

6-9-6 would be pretty awesome tho ^^

[l0ud_sil3nc3]

6-9-6 would be pretty awesome tho ^^

6-9-6 under cold has been done before at 2500

http://www.xtremesystems.org/forums/showthread.php?280903-SuperPi32m-5Ghz-Ivy-All-Out-Challenge!&p=5101956&viewfull=1#post5101956

[GoriLLakoS]

I just got a UP5 and i am testing it.
1st Boot A6 Error, not even signal to monitor.
After many boots and CMOS the same.
I as using 3770K with the integrated VGA.

Then i used a 560 and i booted the 1st time.
Switch bios from F1 to F6X.
Everything ok with the 560.
Switch from 560 to integrated i am getting errors A6 or A0.

Anybody has the same problem?

[sin0822]

I think you mean "Ab", but Ab is not an error, it is just Ab and it means you are in the BIOS but video output is up to you.

If you have a GPU plugged in, the board always defaults to it. So you can pull the GPU out and then try integrated, if one of those outputs don't work then you can try another.

[GoriLLakoS]

Yeah...it is maybe Ab and i read in another forum that he thougths is was A6 too.
My problem is that a PCI-E VGA is giving me output but the integrated one is not working for me neither from VGA nor from DVI
plug.

I read that someone booted the PC and then plugged the VGA cable but this didn't worked for me.
I want to work with the integrated one as you understand :)

[zeneffect]

you can set the igp to default to the first display device in bios. alternatively, you can pull the discreet video card out completely, clear cmos, then boot with the onboard video connections.

[GoriLLakoS]

I have done all those things...nothing is working!
Seems like the plugs are dead or something...:(

[Massman]

I had a similar issue like Gorillakos on the UP4 before, but it got fixed magically. What I did was:

- boot up with pcie gpu
- set igp to default in bios
- save, exit shutdown
- remove pcie gpu
- boot with igp
- clear cmos (long hold)
- boot with igp always

After that, I never had issues with booting from the IGP again.

[sin0822]

yea you have to set INIT display first to IGFX, that is what i do and then remove the dGPU. I know it is kind of a pain.

[stasio]

Guys try new BIOS.

[GoriLLakoS]

@Massman i have done this more than 10 times...not working for me. In the end i am getting A0 Error and never see nothing on my screen.
@sin You say it's a kind of pain cause you are removing the VGA while you are in bios? :p
@stasio i got the F6X..which is the newest?

EDIT: Ok seen it the F9...going to put it :)

[GoriLLakoS]

F9 Installed again the same!

I switched to 3770S, same again!
I cleaned the CPU and installed everything from the start for 30th time and suddenly everything worked out :D :D :D

Thanks for all the support guys :T:

[sin0822]

which video output do u use?

Ab= you are in the BIOS
A0= you are in OS

[dinos22]

Is that on air or with the memory under LN2, Dino?

Maybe we should track what PSC/BBSE is actually working with what MB/BIOS/REV? Track it down to serial number?LN2 of course man, never seen people do that with air cooled RAM faints

6-9-6 under cold has been done before at 2500

http://www.xtremesystems.org/forums/showthread.php?280903-SuperPi32m-5Ghz-Ivy-All-Out-Challenge!&p=5101956&viewfull=1#post5101956 so did i on UD5H but that's not that impressive, 7-11-7 at 1300+ is faster
http://www.xtremesystems.org/forums/showthread.php?280903-SuperPi32m-5Ghz-Ivy-All-Out-Challenge!&p=5112064&viewfull=1#post5112064

[Leeghoofd]

With Massman in the Epicentre of GB knowledge, I'm eagerly awaiting a bios to try.... Heck I might even flash F14 tonite on the UD5H :)

[Gamer]

Seems like it failed? :p

[stasio]

Seems with proper BIOS (nickshih),guys easily reaching DRAM 2800+ on Asrock Z77 OC Formula.

http://www.xtremesystems.org/forums/showthread.php?282693-**-OFFICIAL-**-Asrock-Z77-OC-Formula-Overclocking-troubleshooting-Discussion/page10

[Leeghoofd]

Now that's what I call bios improvement

[stasio]

^ yea,you are right........and yesterday Nickshih with ASRock Z77 OC Formula Sets a New World Record............7.136 GHz!

http://www.techpowerup.com/172189/ASRock-Z77-OC-Formula-Sets-a-New-World-Record.html

[Leeghoofd]

reminds of the numerous FDI biosses per ram kit... could make some clockers swap sides...

[Alex@ro]

Can someone upload the SPD of the tridentx with sammy modules?Or any XMP of samsung based modules?I'm having great issues with a biostar TZ77x anything above 2133 will give me 55 code and not work regarding latency's and voltages of SA VCCIo etc.Have sammy 30nm YK0 they do 9-10-10-27-1 2133 with 1.5V but nothing more on this Biostar board,always 55 code.

Thanks,.

[Leeghoofd]

2666C11 plats on Gene V XMP profile :

http://dl.dropbox.com/u/66364457/Corsair%20Platinums/XMPtweakit.jpg

Try 11-13-13-30 2T 1.65Vdimm and see if 2400 boots with above sec and tertiary timings...

[K404]

If that's as Samsung as it looks, you could pull that tRFC time in by 80-100 cycles couldn't you? :)

[Leeghoofd]

Jups Kenny, TRFC 127 is smooth as butter... You can insanely lower tRAS too (18-20), tWWSR is too loose too on the ASUS, eg the MSI board sets 5 for the same XMP profile...:rolleyes:

[K404]

I mis-read the tRAS in your screeny, yea, 21 is good for me :) I can also pull tWR in to 10 without a problem. I've not got DEEP with timings though. First Ivy SPi session tonight though :D I hope this CPU is as good as I hope :)

[Leeghoofd]

On the UD5H ? fingers crossed you have at least a 6.7Ghz CPU...

[K404]

Thankyou for crossing your fingers :) Board stopped at 6.3GHz again and this is NOT a 6.3GHz CPU.

Board has had its chances. I'm moving to something else.

[Leeghoofd]

Vcore issues Kenny, shutting down ?

[K404]

Absolutely spot on, sir!

[Leeghoofd]

BIG sigh...

[K404]

UD5H = consistency problems?

[l0ud_sil3nc3]

ya I think they are great boards for 24/7 users / builds, but I think I will wait for the UD7 before wasting any more LN2 on another Giga (UD3/5)

I think it is time to try another vendors board Kenny:D

[K404]

I had a shot of a cold UD3H tonight..... PCI-E problems are gone, vCore problems are gone. :D

Different USB controller to UD5H though, which is a bit of a pain. on existing installs.

[Leeghoofd]

Also a tip : use one of the USB ports on the bottom of the board Kenny :)

[sin0822]

yea and if you use XP change it from smart auto to auto for xhci i think, but i think GB already did this recently. That changes intel USB 3.0 to USB 2.0, as XP drivers for intel USB 3.0 dont' exist.