3D11 Physics - Ivy Bridge LCC

[Massman]

0. Introduction

Since Sandy Bridge-E, we now all know how the 3DMark11 Physics subtest does not only scale well with threads, but also scales very well with memory frequency. Ivy makes memory clocking very interesting again, so why no have a little memory tweaking entertainment using the Physics subtest?!

1. LCC rounds

Since 3D11 Physics is stressing all cores, it might not be so easy for everyone to run 5GHz low clock, I suggest we get three rankings:

4500 MHz LCC (join link: http://hwbot.org/competition/ivy_physics_lcc/stage/681_45gp_-_3d11_physics/)
6000 MHz LCC (join link: http://hwbot.org/competition/ivy_physics_lcc/stage/682_60gp_-_3d11_physics/)
Full Out Max MHz (join link: http://hwbot.org/competition/ivy_physics_lcc/stage/683_3d11_full_out/)


I'll make a hidden hwbot hce competition page so it'll be easier to track score and rank.

//edit: hidden hce page link => http://hwbot.org/competition/ivy_physics_lcc

2. Screenshot requirements


3DMark11 score
CPU-Z CPU
CPU-Z Memory
TweakIT / GIGABYTE GTL / ... (other subtiming detail tool)


3. Scores

3a - 4500 LCC


13095 (http://hwbot.org/submission/ 2296480/ ) by zeneffect // Core i7 3770K @ 4500.00 // GA-Z77X-UD5H // G.SKILL Flare @ 1256.1 8-11-8-27
13080 (http://hwbot.org/submission/ 2296236/ ) by Massman // Core i7 3770K @ 4499.98 // GA-Z77X-UD3H // G.SKILL Flare @ 1317 CL8-12-8-28 1T
13028 (http://hwbot.org/submission/ 2296239/ ) by Massman // Core i7 3770K @ 4498.02 // GA-Z77X-UD3H // G.SKILL RipjawsZ @ 1363.1 CL10-12-10-24
12823 (http://hwbot.org/submission/ 2296238/ ) by Massman // Core i7 3770K @ 4500.09 // GA-Z77X-UD3H // G.SKILL RipjawsZ @ 1428.7 CL11-13-13-32 1T



3b - 6000 LCC


15966 (http://hwbot.org/submission/ 2296237/ ) by Massman // Core i7 3770K @ 5998.01 // GA-Z77X-UD3H // G.SKILL Flare @ 1308.7 CL8-12-8-28 1T


3c - Full out


...


SELECT " ",ROUND(score)," ('http://hwbot.org/submission/",result_id,"/') by", user.name, "//", cpu_model.model, "@", cpu_oc, "//", mb_model.name, "//", manufacturer.name, memoryproductgroup.label, "@", result.memClock, result.memTimings
FROM result
JOIN USER USING (user_id)
JOIN cpu_model USING (cpu_id)
JOIN mb_model ON (result.mb_model_id = mb_model.model_id)
JOIN memoryproductgroup ON (result.mem_productgroup_id = memoryproductgroup.memproductgroup_id)
JOIN manufacturer ON (memoryproductgroup.manufacturer_id = manufacturer.manufacturer_id)
WHERE contest_id=369
ORDER BY contest_benchmark_id, score DESC;

[Massman]

Highest I got so far with RipjawZ Samsung kit

http://hwbot.org/blog/wp-content//12992.png

[dRweEz]

but if its hidden how will we track it :P
thanks dude this should be an interesting investigation
subscribed

[Massman]

I'll post the link here :).

It's hidden from mainpage, but it's still active.

[K404]

Any version/revision of 3D11? :)

Wish Futuremark added a version stamp to the GUI

[Massman]

Yup, any version (since we can't see from GUI).

[subaruwrc]

i like phsyx. im in.

[Massman]

Alright, here's the link to the hce page for score tracking: http://hwbot.org/competition/ivy_physics_lcc

Already uploaded some results :)

//edit: updated first post

[fuzz3l]

The links to the results are all broken. Anyway nice competition :)

[OBR]

its summer guys, go out sometimes :)

[sin0822]

pretty col if my power wasn't out i would participate.

[zeneffect]

subbed!

http://hwbot.org/submission/2296480

massman put a pretty good goal to reach for 4500mhz cap.

[zeneffect]

this thread is not as active as I would have hoped.

come on guys! 3dm11 is pretty relevant isnt it?

[zeneffect]

updated submission
http://hwbot.org/submission/2298517

dumo also put up a score :)
http://hwbot.org/submission/2298121

iduno if this is just lame-test or everybody else has something better to do with their non-existant lives... im gathering good data here.

[Massman]

13k for 4500 seems to be the target. I wonder what people with DDR3-3000 would be able to do ... :)

[zeneffect]

13k for 4500 seems to be the target. I wonder what people with DDR3-3000 would be able to do ... :)

do it. i have yet to go over 2680 on ivy. :(

fully tuned up to 2650ish i expect 13500... so 3000 is 14k+?

[Massman]

I tried DDR3-2800C11 without finetuning the subs and it was actually slower than the DDR3-2600+ C10.

As long as you can get tight subs and high clocks, I think you should be hitting 13.5K. Not sure if anyone is able to do 14k at 4500, though.

[Splave]

try 7-7-7 2133 might be surprised :)

[zeneffect]

@ splave

havent tried 7-7-7 2133 yet.... should I try GTX2 or just some Hypers?

@ massman

why is it so quiet/boring in here?

[Massman]

People might be busy with all the other competitions that are going on. Even I was spending all my benching time on the Team Cup last week :D

[Vivi]

i'll enter soon just chillax!

[Vivi]

nice score zen what made you jump 100 points?
my score is up too.

tWWSR from 5 to 4 gave me like 50.

And TWCL from 8 to 7 gave me quite a bit.
TWCL 6 lost me about 40 points.

[Leeghoofd]

tWWSR at 4 ?

[Vivi]

woops tWWSR yes..

5 was very stable
4 required some IMC and more VDim, 1.88v to 1.91v for it to be stable

[ARandomOWL]

Why is the leaderboard in no meaningful order? :/

[Massman]

The ranking is by score, the 'latest submissions' tab by submission date.

[zeneffect]

nice score zen what made you jump 100 points?
my score is up too.

tWWSR from 5 to 4 gave me like 50.

And TWCL from 8 to 7 gave me quite a bit.
TWCL 6 lost me about 40 points.

turn off core parking... more ram freq, and finding the "right" timings. some subs if you are too tight it hurts score. i see you are using UP5 :) time for me to switch boards and get with the program. next sub will be with up5

*edit*

ran UP5 last night... within 1 point of UD5 so efficency is the same. crash recovery was a disaster for me though... took me like 20 min to recover because clear cmos wasnt working. had to boot into 2nd bios, remove all hard drives and gpu... then reboot back into 1st bios only to find that all my settings were still the same that made me fail post constantly 66 reboot 66 reboot... since then though, no problems. probably was a 1 off thing.

[zeneffect]

UP5 is now rockin the way its supposed to. :D

vivi, f5a bios is pretty stable.

[zeneffect]

bored bored bored bored....

[dinos22]

would be a lot more interesting if it was 3D01 LCC lol :D

[Massman]

I'll put the competition up on frontpage after the Team Cup. Should attract more attention. Now it's kind of hidden in the forums.

[Vivi]

Yea make it live!

[Sam OCX]

why is this thread named after loadline calibration and not lowclock challenge?

[zeneffect]

why is this thread named after loadline calibration and not lowclock challenge?

SkIf9lNRY1A

[sin0822]

why is this thread named after loadline calibration and not lowclock challenge?

You forget lowest level cache

[zeneffect]

little leprachaun children (triple redundancy ftw)

[gnidaol]

http://cdn.memegenerator.net/instances/400x/26027492.jpg

=) nice kind of contest, should be held more often

[zeneffect]

it ran for 2 solid months

[Massman]

I opened it up for another month :).

[zeneffect]

it shows closed since sept 1st for all stages.

[George_o/c]

http://cdn.memegenerator.net/instances/400x/26027492.jpg

=) nice kind of contest, should be held more often

hahaha I loved a lot with this one :D

[T0lsty]

You are too late! This stage of the competition is no longer accepting submissions since 2012-09-01 12:00:00.0.

https://lh3.googleusercontent.com/-mtNBbfkdHfA/UEmjxqrPfmI/AAAAAAAAD70/64YPu5o-SXw/s1280/screen185.jpg

G.Skill F3-18400Cl8D-4Gbpis

[l0ud_sil3nc3]

^nice vdimm?

[T0lsty]

1.82V
can do some more mhz, but need to loose third timings.

[l0ud_sil3nc3]

BBSE


http://i.imgur.com/Vaus8.png (http://imgur.com/Vaus8)

[l0ud_sil3nc3]

good PSC never ceases to amaze. . .

http://i.imgur.com/J8fkR.jpg (http://imgur.com/J8fkR)

1.835vdimm on air:D

[JulioCesarSF]

What is my problem here? My score is low for a i7 3770k 4.8Ghz.

http://img831.imageshack.us/img831/6825/lowscore3dmark11physics.png
http://img831.imageshack.us/img831/6825/lowscore3dmark11physics.png

[Gunslinger]

RAM speed is too slow and timings are too loose for starters.

You'll probably want to run the original version of the benchmark as well, not the latest patched version.

[JulioCesarSF]

RAM speed is too slow and timings are too loose for starters.

You'll probably want to run the original version of the benchmark as well, not the latest patched version.

But I bought the 3dMark on steam, I can not run the old version.
My memory is by default 2133MHz 9-11-10-28 1.65v. = / Maybe 2133mhz 9-10-9-27 1.70v works?

http://www.gskill.com/products.php?index=403

[subaruwrc]

But I bought the 3dMark on steam, I can not run the old version.
My memory is by default 2133MHz 9-11-10-28 1.65v. = / Maybe 2133mhz 9-10-9-27 1.70v works?

http://www.gskill.com/products.php?index=403

yes maybe works but if you want real performance boost in this part of the bench, you should set 2400+mhz memory as a goal, with as tight timings as your memory can.
to tell the truth that memory is not the best for this.
you could try 2200 9 10 9 24 with it maybe

[GENiEBEN]

But I bought the 3dMark on steam, I can not run the old version.
My memory is by default 2133MHz 9-11-10-28 1.65v. = / Maybe 2133mhz 9-10-9-27 1.70v works?

http://www.gskill.com/products.php?index=403

Install STEAM version, go to Help and note down the cdkey. Uninstall then google the initial version, install, add the key.