Please post 7900GT / GTX overclock results

[pinto]

Asus 7900GT standard clocks
stock volts stock cooling:




1.55v and zalman VF900 no vmem mod:

[GOESA]

Asus 7900GT standard clocks
stock volts stock cooling:




1.55v and zalman VF900 no vmem mod:

How hight temps do you get with the VF900 and the 1.55v mod?

/Grau

[Mysterfix]

Just got my card! Gonna through it in and test it on stock cooler before I give it some water.

[pinto]

How hight temps do you get with the VF900 and the 1.55v mod?

/Grau

approx 55° load and 42° idle.

[ncc-1709]

2x galaxy 7900 gt. stock 450/1320 oc stock cooling+volts to 550/1820

[alpha0ne]

2x galaxy 7900 gt. stock 450/1320 oc stock cooling+volts to 550/1820

Very nice, I just got my MSI NX7900GT 450/1320 and if I can get the same as you I'll be a very happy man

[Raphie]

Which one do you have? The 675 cored version? If so, you may want to edit the BIOS so that you can restore the delta back to 50 MHz from 25 MHz. The ROP/Shader is often the limiting item, not the vertex. Mine is now at 60 MHz and it has improved things. I have the regular EVGA 7800 GTX.

at 729/910

3D05 12605
3D06 7006

mine is factory OCed @ 695/1750

[pinto]

vmem mod done, at least head room of 15mhz more to come:

[killxswitch]

XFX Extreme 550/1630 stock clocks, $329 at Monarch.

I haven't had time to mess with it yet, but I did run 3dmark05/06 at 600/1630 yesterday at stock volts and cooling.

Edit: core maxed at 615 at stock cooling. Still finding upper range of memory, right now it's at 1800.

[Nilson]

***Deimos***,
Check this out:

713/913 on GT:
http://www.xtremesystems.org/forums/...&postcount=660
vs
http://img81.imageshack.us/my.php?im...k0663998mh.jpg

713/1826 vs 740/1978

6512 vs 6399. Yeah the CPU in the first case +125MHz.

mark06 doesn't so sense to freq in case one core CPU.

But the core and memory! pinto should get much more in HDR/SM3.0 tests!!! Anyway - we can add 113 marks due to CPU compesation but... +27MHz on GPU and +162MHz - it does make sense.

Also, what do you think about 11.1k in mark05 on 740/1850 GT and one core 2.8GHz AMD? I think 11.9-12.0k should be here. Since 11.769 was done by Shark-357 on 716/1000 GT with the same one core AMD CPU.

Also, if you compare Shark-357 with pinto - you will see that Shark-357 can hit 12.0k in mark05 on 716/1000 with dual core Opteron at 2.8GHz.

[SF3D]

***Deimos***,
Check this out:

713/913 on GT:
http://www.xtremesystems.org/forums/...&postcount=660
vs
http://img81.imageshack.us/my.php?im...k0663998mh.jpg

713/1826 vs 740/1978

6512 vs 6399. Yeah the CPU in the first case +125MHz.

But the core and memory! pinto should get much more in HDR/SM3.0 tests!!! Anyway - we can add 101 mark due to CPU compesation but... +27MHz on GPU and +162MHz - it does make sense.

If you look accurately, pintos HDR/SM3.0 score is driven with 2D clocks not 740MHz. That's why score is so low.

This is normal HDR/SM 3.0 score:
http://img320.imageshack.us/img320/6803/06markit2ni.jpg

All results I have seen here, are just what they should be! You can leave this topic now

[Nilson]

You can leave this topic now

If you look accurately, pintos HDR/SM3.0 score is driven with 2D clocks not 740MHz. That's why score is so low.

Yeah? Who says that? Could you please point to me the reason for 2D clocks in HDR/SM3.0? Overheat or not enough volts?

This is normal HDR/SM 3.0 score:
http://img320.imageshack.us/img320/6803/06markit2ni.jpg

All results I have seen here, are just what they should be!

You post 6610 marks on dual core 2.8GHz Opti with 705/900. It is correct points for these clocks and CPU! Thats what I'm talking about.

You didn't get me. As I said earlier in my post for you - "you should be able to get on your system 6700 mark06 only with 720/1800." Now you prove it.

Shamino get 800/2126 and 7013 marks. His points in HDR/SM3.0 = 2744. Try to prove that it's correct points for 800/2126 - and I'll leave this topic.

I'm talking about issues on GT in case of 730-800MHz on GPU. Why we can't get stable clocks in tests? I mean 3D clocks in one subtests and 2D clocks in over subtests at the same test run? I saw many results on 730, 740, 750 MHz for 7900GT - no one can pass mark06 only on 3D freq. Aslo, does anybody can prove that on 740MHz or 750MHz he can pass mark05 without freq drops or skiping GPU cycles? No one I suppose. 12.2k on GT with 750/1820 with delta +20? I can believe if it's one core GPU but not two cores.

Can you try to flash 740 for all modes and run mark05 and mark06 tests? but it doesn't helps if we have power circuit issues.

[Mysterfix]

Doing 705/1800 on water so far with 1.55v

[Nilson]

Mysterfix, please post your 05 and 06 results on 2.8GHz....

[Mysterfix]

OK Dude here ya go.

<a href="http://img339.imageshack.us/my.php?image=screenshot0016uh.jpg" target="_blank"><img src="http://img339.imageshack.us/img339/4741/screenshot0016uh.th.jpg" border="0" alt="Free Image Hosting at www.ImageShack.us" /></a>
<a href="http://img114.imageshack.us/my.php?image=screenshot0027zg.jpg" target="_blank"><img src="http://img114.imageshack.us/img114/2760/screenshot0027zg.th.jpg" border="0" alt="Free Image Hosting at www.ImageShack.us" /></a>

[RommelD]

@Mysterfix: What version of the 7900 GT EE do you have?

Does anybody know that the absolute maximum overclock of the Extreme versions is higher then the normal 7900 gt's? I think a lot of people do not want to spend another extra 10% for a factory higher clocked card if the max. overclock is the same for both cards. Does anybody know? Do the card manufacturers put their better samples on the Extreme cards or do all the chips cope with the higher speeds? So this goes for both the gpu and the mem.

Penny for your thoughts

[Mysterfix]

I have the 520/1500 version.

[Max Tractor]

Has anybody fitted an Accelero X1 to their 7900gt, or does anybody know if it will fit, web search found conflicting results

[Ebola]

***Deimos***,
Check this out:

713/913 on GT:
http://www.xtremesystems.org/forums/...&postcount=660
vs
http://img81.imageshack.us/my.php?im...k0663998mh.jpg

713/1826 vs 740/1978

6512 vs 6399. Yeah the CPU in the first case +125MHz.

mark06 doesn't so sense to freq in case one core CPU.

But the core and memory! pinto should get much more in HDR/SM3.0 tests!!! Anyway - we can add 113 marks due to CPU compesation but... +27MHz on GPU and +162MHz - it does make sense.

Also, what do you think about 11.1k in mark05 on 740/1850 GT and one core 2.8GHz AMD? I think 11.9-12.0k should be here. Since 11.769 was done by Shark-357 on 716/1000 GT with the same one core AMD CPU.

Also, if you compare Shark-357 with pinto - you will see that Shark-357 can hit 12.0k in mark05 on 716/1000 with dual core Opteron at 2.8GHz.

To my mind pinto's board has some idling issue, i got similar

Even if frequency raise very high, the score doesn't improve from a certain point. For me over 910 MHz on the VRAM doesn't help, and moreover lower the performance even if the RAM can raise up to 1GHz

Same for the GPU but hard to sense, should be around 720-740 MHz

Am thinking of the issue mentionned in the Vmod section where either the OVP is in cause or the switching frequency of teh regulator, both can be patched with some modding that i will try soon

Waiting for a swithc like SF3D used, easy way for switching from 1.5 to 1.7 without removing the board.

[vocch]

Has anybody fitted an Accelero X1 to their 7900gt, or does anybody know if it will fit, web search found conflicting results

Go and get yourself a Zalman VF-900.

[Nilson]

OK Dude here ya go.

<a href="http://img339.imageshack.us/my.php?image=screenshot0016uh.jpg" target="_blank"><img src="http://img339.imageshack.us/img339/4741/screenshot0016uh.th.jpg" border="0" alt="Free Image Hosting at www.ImageShack.us" /></a>
<a href="http://img114.imageshack.us/my.php?image=screenshot0027zg.jpg" target="_blank"><img src="http://img114.imageshack.us/img114/2760/screenshot0027zg.th.jpg" border="0" alt="Free Image Hosting at www.ImageShack.us" /></a>

Thanks a lot

These shots are very stable and all scores are fine. I mean without any issues with GPU or Memory.

It is prove that 720/1800 for 7900GT is around: 6.7k+ mark06 and 12k+ mark05.

[ArcticOC]

Just did the 1,4V mod. I choosed to use a micro switch for the mod, this make the mod really easy to enable/disable. Did this because I often work with noisless solutions, and I have plans to make a non fan cooler for the card!

The card is a XFX GeForce 7900GT eXtreme. Running default @ 520/1500

Could run @ 541/1600 before the mod..
..after the mod the nForce drivers reports 636/1764 as the highest clock.

[Clint]

I got my eVGA 256-P2-N563-AX today from nn_step

Initial clocks a bit mediocre @ 541MHz / 1680MHz with stock cooler.

Gonna do some work on it and report later..

I did mount my NV Silencer 6 without any modifications whatsoever, everything fits like it's meant for it....precise length and clearance
Temps I see with the NV on slowest fan is 42 idle and 52 load....and dead silent

[Nilson]

ArcticOC, nice work man so neat vmod

[Mysterfix]

I got my eVGA 256-P2-N563-AX today from nn_step

Initial clocks a bit mediocre @ 541MHz / 1680MHz with stock cooler.

Gonna do some work on it and report later..

I did mount my NV Silencer 6 without any modifications whatsoever, everything fits like it's meant for it....precise length and clearance
Temps I see with the NV on slowest fan is 42 idle and 52 load....and dead silent

Mine topped out at about 560/1660 on stock everything.