Unofficial Opteron 939 Thread

[supershanks]

"In Place Large FFT's"

stresses ram whilst small FFTs stress CPU. So these can be used selectively. To Test overall use the blend option - this is the one used in the OC Database.

 

luck

[supershanks]

Yes! Yes! YES!!!! http://valid.x86-secret.com/show_oc.php?id=79038

Been nagging at me as an ambition since i got this puppy - 273x11 @1:1

not got my LA back from RMA yet but son's away so knicked his old G-Skill 2x512mb LC.

Was just playing with timings & it looked stable @ 265 so turned it up.

Think it may be a new record (low) for vcore 1.35x104% = 1.45v per multimeter.

 

1Mb Pi stable not checked much more yet.

[Sorrento]

IMO the Blend test is the to use... I prefer to do In-place large FFTs just to check the maximum cpu temperature, and to see the maximum power consumption. Also I can't load both cores with two Prime95, one on each, with the Blend test... only with the Large FFTs one.

 

I would stick to Blend for long runs, and use Large for small one or two hour runs. With blend you test everything, thats why its default.

[TheFlowerKing]

Yes! Yes! YES!!!! http://valid.x86-secret.com/show_oc.php?id=79038

Been nagging at me as an ambition since i got this puppy - 273x11 @1:1

not got my LA back from RMA yet but son's away so knicked his old G-Skill 2x512mb LC.

Was just playing with timings & it looked stable @ 265 so turned it up.

Think it may be a new record (low) for vcore 1.35x104% = 1.45v per multimeter.

 

1Mb Pi stable not checked much more yet.

 

Nice one SS, my 148 is 3GHz OCDB stable at 1.475v which I thought was good!

That's just tremendous Congrats.

[supershanks]

@TheFlowerKing thanks i've seen conflicting specs re 148 re vcore being 1.35v or 1.40v have a feeling from the performance of mine that it's 1.35v have checked the codes etc & can't specifically pin it down.

 

Note: Idle Voltages are calibrated to multimeter.

Idle

Load

 

PS yep saw your OCDB stable maybe i'll get there with a bit of tweaking.

[TheFlowerKing]

Yeah, was ocdb stable, but that was 1.475v in the bios, so who know what it is for real heh.

Kinda hits a brick wall at 3.1GHz @ 1.6v, but 2.8GHz is more than enough for me

[Clay]

Here's a bandwidth or 2

 

 

And as promised probably the worlds lowest 3dm01 score for and overclocked opteron.

 

 

It was better than I thought it would be!LOL If anyone want's to try to get lower than that. Be warned I have a 1meg card out in the trailer and it might just work!LOL

 

Collection of superpi's at this clocking.

 

[desire]

Update your CPU-z to v1.32. It will fill out all the missing info.

 

http://www.cpuid.com/download/cpu-z-132.zip

 

 

Really close to getting mine OCDB stable at 3GHz.

 

[Sorrento]

And as promised probably the worlds lowest 3dm01 score for and overclocked opteron.

 

 

It was better than I thought it would be!LOL If anyone want's to try to get lower than that. Be warned I have a 1meg card out in the trailer and it might just work!LOL

Aww, S3 pci graphic card w/8 megabytes, I wish I had one of those... you never know when you need a backup to flash video card's BIOS.

 

The 1mb card, now thats a scary thought :eek2:

[before]

IMO the Blend test is the to use... I prefer to do In-place large FFTs just to check the maximum cpu temperature, and to see the maximum power consumption. Also I can't load both cores with two Prime95, one on each, with the Blend test... only with the Large FFTs one.

 

I would stick to Blend for long runs, and use Large for small one or two hour runs. With blend you test everything, thats why its default.

In personally prefer StressCPU to assess for overheating issues. This is a small program to torture-test your CPU in order to make sure that you don't have overheating problems. It will only run on SSE-equipped x86 CPUs, and it is executing a special version of the Gromacs innerloops that mixes SSE and normal assembly instructions to heat your CPU as much as possible.

 

On the other hand, it can be used as a benchie; and seconding me, Fouge @OverClocking-Masters.com has modded the program that gives now computation time as well as CPU clock speed. There are 3 choices: 10K iterations which is like Super Pi 1M, 100K iterations which is like Super PI 8M, 1M iterations which is a real stability test during about 1 hour.

 

http://olivier.fouge.free.fr/Bench/StressCPUBench.zip

 

I've started a database for this bench, some of you may wish to participate here.

 

http://www.beforeoverclock.com/b/showthread.php?t=33

 

:cat:

[supershanks]

@TheFlowerKing Yep i run 24/7 2900~2950 264~268x11 tlike you saty it's a good clip. but it's nice to know it can

 

luck :nod:

[Sw4y]

Update your CPU-z to v1.32. It will fill out all the missing info.

 

http://www.cpuid.com/download/cpu-z-132.zip

 

 

Really close to getting mine OCDB stable at 3GHz.

 

 

Hey desire, your running the 133 divider arent you? how long is your rig prime stable because i just bought the CACJE 0551 and the same board. Im hoping to get 300x10 w/ my 2x512 OCZ EL Plat's. I have yet to order my board tho, waiting on the tax return.