BH-5 + DFI NF4 timing tests, results inside

[Evil_Spork]

for whatever reason i was driven to test TRAS settings with bh5 on my dfi board today.

i used everest read benchmark to test at 2-2-2-X at 9x265mhz
i recorded results over several tests and threw out the highest and lowest and any tests that just plain didnt fit in. unfitting numbers like a sudden 20mb/s drop for one test i threw out. i took all the numbers and got an average score.

i did this for all settings 0 through 12, and a few tests from 12 to 15 which arent worth reporting(very bad).

below is a chart of my findings. dont hold the exact values as rock solid as i would need a much larger number of tests to get better results, but what i have is pretty accurate.

i tested tras 2 many times to be sure the number was true. the value for tras 0 should be a bit higher, similar to tras 1.

accounting for margin of error i would say anything 0 through 5 will be similar performance, with a slight boost with tras 2 oddly.

ill post a64 tweaker screenshots tomorow, for now here is the chart of my findings:

http://www.evilspork.net/es/compstuff/bh5BW.jpg

[MaxxxRacer]

Interesting numbers. Thanks for doing more than one run on each timming... annoying when ppl arnt scientific about their testing.

[Evil_Spork]

Interesting numbers. Thanks for doing more than one run on each timming... annoying when ppl arnt scientific about their testing.
oh did i do more than one run..... lol, my computer nearly slapped me for running that test so many times! :p:

one note is i used an unmatched pair of KHX3200 bh5. they still are benchable at 280mhz 2-2-2-2.
they even have differant voltage ratings between the two sticks. :rolleyes:

one stick is 327WF, the other is 315WE. :cool: (the 315 is the 2.5v rated, the other is 2.6)

ill have to try my matched pair soon.

[comment]

for whatever reason i was driven to test TRAS settings with bh5 on my dfi board today.

i used everest read benchmark to test at 2-2-2-X at 9x265mhz
i recorded results over several tests and threw out the highest and lowest and any tests that just plain didnt fit in. unfitting numbers like a sudden 20mb/s drop for one test i threw out. i took all the numbers and got an average score.

i did this for all settings 0 through 12, and a few tests from 12 to 15 which arent worth reporting(very bad).

below is a chart of my findings. dont hold the exact values as rock solid as i would need a much larger number of tests to get better results, but what i have is pretty accurate.

i tested tras 2 many times to be sure the number was true. the value for tras 0 should be a bit higher, similar to tras 1.

accounting for margin of error i would say anything 0 through 5 will be similar performance, with a slight boost with tras 2 oddly.

ill post a64 tweaker screenshots tomorow, for now here is the chart of my findings:

http://www.evilspork.net/es/compstuff/bh5BW.jpgNice testing :toast:

What Evil Vdimm did you use Spork?

[Evil_Spork]

Nice testing :toast:

What Evil Vdimm did you use Spork?
3.6v

i had it at 3.8v for 280+, but i just tried and i only needed 3.5v lol! (at least for 265 which im booting at.

maybe a sticky for this thread?:)

[perry_78]

maybe a sticky for this thread

Nice try :D But then again, why not?

[btv94]

Can someone tell me what is the max vdimm on What is the max vdimm on DFI LANPARTY nF4 SLI-DR without mod?

Thanks,

[pershoot]

max vdimm by default is 3.2v. max vdimm after switching jumper to derive from +5 line is 4Vdimm. no mods necessary on this board for high vdimm.

[Ailleur]

Can someone tell me what is the max vdimm on What is the max vdimm on DFI LANPARTY nF4 SLI-DR without mod?

Thanks,

The default jumper settings will let you go up to 3.2V in the bios.
A simple jumper switch will give access to vdimm up to 4.0v (then draining from the 5v rail)

[S0nic]

Can someone tell me what is the max vdimm on What is the max vdimm on DFI LANPARTY nF4 SLI-DR without mod?

Thanks,

4V..

[btv94]

I guess I'll get this board over MSI K8N Neo4 Platinum SLI, for my BH5
Thanks for the info

[Evil_Spork]

bump for more views

[texuspete00]

I dont see a picture... is it me?

[SAE]

Yep, there was a pic, but it's gone... :D

P.S. E_S, I tried tras=2 instead of 0 and it was even better!!! Thanks, mate! :up:

[Evil_Spork]

Yep, there was a pic, but it's gone... :D

P.S. E_S, I tried tras=2 instead of 0 and it was even better!!! Thanks, mate! :up:
awesome! thats what i like to hear! :cool:

[blinky]

i tested superpi for a couple hours yeterday and found that 5 and 6 gave me the best scores

this is with a 3000 DTR claw 1mb

i think its different on different cpu's

[cantankerous]

5 is also best for me. Anything lower gives me worse super pi and 3dmark01 scores.

[SAE]

My scores do really get better... also try to test with hexus_pifast, guys :D

[Evil_Spork]

interesting..

SAE and i are both using winchesters and bh5, and we both see improvment.. ill have to test 3dmark

[Evil_Spork]

tested 3d01

my results are the same as SAE's, a good improvment!

Tras 5: 28,400
http://service.futuremark.com/compare?2k1=8481478

Tras 2: 28,626
http://service.futuremark.com/compare?2k1=8481463

[SAE]

SAE and i are both using winchesters and bh5

No, not exactly ;)
Mine is an FX-55 :hehe:

[HARDCORECLOCKER]

No, not exactly ;)
Mine is an FX-55 :hehe:

:D Reminds me to Your comment in another thread: People should buy new glasses.......... :toast:

[bachus_anonym]

tested 3d01

my results are the same as SAE's, a good improvment!

Tras 5: 28,400
http://service.futuremark.com/compare?2k1=8481478

Tras 2: 28,626
http://service.futuremark.com/compare?2k1=8481463

Well, If you're testing 3DMark2001 you should have really excluded Car Low test as it will give you random results due to physics involved (see barrel flying at some point and other time, it's not there ??? that's what I'm talking about). that's why you get such a huge difference in overall score - because of 21fps in Car Low ;)

[SAE]

:D Reminds me to Your comment in another thread: People should buy new glasses.......... :toast:
:lol:

No, I am sure he just did not check. :D

[Evil_Spork]

Well, If you're testing 3DMark2001 you should have really excluded Car Low test as it will give you random results due to physics involved (see barrel flying at some point and other time, it's not there ??? that's what I'm talking about). that's why you get such a huge difference in overall score - because of 21fps in Car Low ;)
i tested twice each, car low wasnt the deciding factor.

[bachus_anonym]

i tested twice each, car low wasnt the deciding factor.
sorry Spork, but I beg to differ ;) by comparing both runs you will see that the only significant difference is at Car Low (21fps)... and some test are got even lower frames in second run comparing to first one.

un-tick Car Low and you'll see what I am talking about ;)

EDIT: Just did little Photoshop action :D

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26674&stc=1

[Evil_Spork]

sorry Spork, but I beg to differ ;) by comparing both runs you will see that the only significant difference is at Car Low (21fps)... and some test are got even lower frames in second run comparing to first one.

un-tick Car Low and you'll see what I am talking about ;)

EDIT: Just did little Photoshop action :D

http://www.xtremesystems.org/forums/attachment.php?attachmentid=26674&stc=1
i tested twice with each and it wasnt a huge differance in car low. just those 2 were differant.

ill test more.

[STEvil]

always kill off car low when doing testing.

[Evil_Spork]

done..

ditched car low, heres what i got. looks like tras 2 still wins. :)

http://www.evilspork.net/es/compstuff/bh53dcompare.jpg

[SAE]

Hehe, good job my friend :up:

[Evil_Spork]

Hehe, good job my friend :up:
thank you :D

hopefully more people will see this and respond, and contribute results too. :)

[saaya]

very nice dominic! very nice! :toast:
can you do the same for cas 1.5 vs cas 2.0 and maybe some other timings as well? :D
dont know how long it took you to get those numbers :D

[STEvil]

I already bugged him about that, rained on his parade too as he says..

muahaha :D

[blinky]

you do realize that its different for different setups....

[Evil_Spork]

you do realize that its different for different setups....
yes, this is just for winchester + dfi nf4... but ALOT of people have the same setup so.... ;)

[Evil_Spork]

very nice dominic! very nice! :toast:
can you do the same for cas 1.5 vs cas 2.0 and maybe some other timings as well? :D
dont know how long it took you to get those numbers :D
forgot to do that! im on it!

[saaya]

sweeeeet :D

[SAE]

yes, this is just for winchester + dfi nf4... but ALOT of people have the same setup so.... ;)
Yep, also applies to FX and DFI :D

[Evil_Spork]

Yep, also applies to FX and DFI :D
we do have the same brand BH5.. might that have something to do with it?

maybe this only works with single sided bh5?

[Ailleur]

done..

ditched car low, heres what i got. looks like tras 2 still wins.

Looks well within the margin of error to me, i dont think a winner can be declared.
Cas 1.5 vs 2 should be more interesting

[Evil_Spork]

Looks well within the margin of error to me, i dont think a winner can be declared.
Cas 1.5 vs 2 should be more interesting
dunno when ill have time to test but ill run a full battery of tests and see how cas 1.5 affects my OC too. :)

[saaya]

well we know for sure that 2 or 3 are the best, dont argue whether its 2 or 3 :D

cant wait for the other results! *nudge* :D

[pershoot]

i can confirm that tRAS of 2 gives the most throughput on my system.

[Evil_Spork]

http://www.xtremesystems.org/forums/showthread.php?p=778090#post778090

in here i tested a ton of ultra tight timings, tras 2 was clearly faster in those cases.

[Evil_Spork]

bump, and expect more results from me soon. ill be testing more timings.

[$0m#0n#]

Gonna try the same with winnie @2.7Ghz & TCCD @300Mhz 2.5-3-3-7 :)

[Evil_Spork]

Gonna try the same with winnie @2.7Ghz & TCCD @300Mhz 2.5-3-3-7 :)
you tried 2.5-3-3-2?

[$0m#0n#]

you tried 2.5-3-3-2?
yea, but I dont remember the result but it was +/- like 6/7 I ll try now

[$0m#0n#]

k, here r first results I am testing it by 10 times @ each value

2x256MB A-DATA 566Mhz 437 @2.76V 2.5-3-3 1T

tras "0" 8187MB
tras "1" 8188MB
tras "2" 8190MB
tras "3" 8188.8MB
tras "4" 8188.7MB
tras "5" 8187MB

I am testing the other values now

I have only MSN/ICQ, I added u on MSN

[Evil_Spork]

k, here r first results I am testing it by 10 times @ each value

2x256MB A-DATA 566Mhz 437 @2.76V 2.5-3-3 1T

tras "0" 8187MB
tras "1" 8188MB
tras "2" 8190MB

I am testing the other values now
sweet!

ive been only testing 0, 2, 5 lately. they seem to be most popular and fastest. :)

edit: you have AIM? msg me at the name in my sig if so :) (my MSN is in my profile, oops msn wasnt connected... my cable went out earlier.)