GTX580: 1.025VID: 950/1900/2300: stock cooler: guaranteed?!?

[cole2109]

I hate this stupid 3DM11
I7 930@4000MHz 7470 points, GPU 1005/2290
I7 930@4650MHz 7600 points, GPU 1006/2335

Only 130 points difference

[MTP04]

I hate this stupid 3DM11
I7 930@4000MHz 7470 points, GPU 1005/2290
I7 930@4650MHz 7600 points, GPU 1006/2335

Only 130 points difference

I haven't run the benchmark yet myself so I can't say for sure, but it probably responds more to increased gpu clocks rather than increased cpu clocks.

[OC Nub]

I hate this stupid 3DM11
I7 930@4000MHz 7470 points, GPU 1005/2290
I7 930@4650MHz 7600 points, GPU 1006/2335

Only 130 points difference

Thats a good score, looks like 3dm11 likes X58/Bloomfield.

[cole2109]

I haven't run the benchmark yet myself so I can't say for sure, but it probably responds more to increased gpu clocks rather than increased cpu clocks.

Yes i know 3DM11 is more GPU friendly, but 650MHz more on CPU and just 100+ points

Cant figure out why I score higher with a 470 in 3dm05 vs this 580. Was able to hit 49,400's with a GTX470 at 890ish mhz, same cpu clocked at 5400mhz.
This is what I get with the 580. Maybe I need a fresh install.

I tried 05 to and got low score. I think that 580 or latest drivers are not working properly on 05.

[aussie-revhead]

My 580 doesnt like 05/03/01/AM3 either, its not too bad in 11. It only really shines in Vantage and 06.

[Dumo]

Vantage run

[OC Nub]

Vantage run

Very nice!

Please check your PM. Think I screwed up

[Asmola]

Dumo, is that 2600K chilled-water-cooled? High voltage with HT, no decreasing visible?

[Dumo]

Dumo, is that 2600K chilled-water-cooled? High voltage with HT, no decreasing visible?

It cooled with single stage. For vantage I use 1.62Vcore bios LLC2 HT on and 1.575Vcore bios LLC2 for other 3D marks.

So far so good, of course high Vcore used only for a few rounds of benching and never run it with air or water

[Asmola]

Damn, i need to buy single stage for my self too! LN2 is worthless with SB and it also cost's way too much..

[OC Nub]

Had a bad scare for a couple hours. I flashed the card and bricked it, during the process I erased the eeprom to try and get it to flash using ..nvflash -Protectoff...nvflash -Eraseeprom, because it wouldnt allow me to override the protectoin. Then it wouldnt allow me to reflash the card, tried everything, booted back to a black screen. Nothing seemed to work. Finally got it running again, guess I can cancell the RMA lol. Finally tried this... nvflash --index=2(adapter number) biosname.rom

I was getting real nervous for about an hour.

Thanks Dumo for helping.

[Dumo]

No problem
I put a note for that just in case it happened to me

To remove protected bios use nvflash -r

Playing with '01 now...Trying to pass 130K sli run

[Demo]

My Crysis OC results are odd

with 1.088v @ 900Mhz core min FPS is a bit lower at 19fps while I get a higher and avg fps and with stock speeds I get a higher min with 24ps while avg and max is at bit lower.. also the tests really have HUGE variances in the min fps per batch run may it be stock or overclocked, warheard on the other hand is pretty consistent with min fps and gives me a boost of about 3-4fps minimum and 3-4max from 797Mhz (stock) to 900Mhz

Seems like my Q6600 @ 3.4Ghz finally met a GPU that bottlenecks it!

[p3gaz_001]

Vantage run

is your 580 on water?

[Asmola]

is your 580 on water?

My card does the same so i believe not.

[p3gaz_001]

My card does the same so i believe not.

mine i think does not... actually i'm with stock cooler..and i can't push more than 935mhz 1.1v.....

[NapalmV5]

mine i think does not... actually i'm with stock cooler..and i can't push more than 935mhz 1.1v.....

not as bad as you may think its pretty good bump up the voltage for higher clocks.. good luck!

-----

guys please read the 1st post.. the 1.025V in the thread title is the vid not the voltage 950 runs @

[p3gaz_001]

not as bad as you may think its pretty good bump up the voltage for higher clocks.. good luck!

-----

guys please read the 1st post.. the 1.025V in the thread title is the vid not the voltage 950 runs @

nice to hear that btw, i've got 1.000 vid as default, and 1.150 max, wich voltage you suggest me to hold on daily / oc use?

[NapalmV5]

nice to hear that btw, i've got 1.000 vid as default, and 1.150 max, wich voltage you suggest me to hold on daily / oc use?

depends on the temps/fan setting/your preference.. you can run it @ 1.1v nonstop no problem

[NapalmV5]

title changed thanx Gautam!

[p3gaz_001]

depends on the temps/fan setting/your preference.. you can run it @ 1.1v nonstop no problem

do you think that if a get under liquid cooling my 580 i can push more event with less voltage?



this was a run of 3dm06 made just yesterday , as i said b4, 580 stock cooling... let me share a pic if you guys don't mind ...
------------------------------


and a koolance ready to go ..

[Demo]

I have a stability question,

With several loops on crysis using the ambush and GPU maps, maxing everything out I dont have any crashes / artifacts, same with the warhead benchmark and 3DM demo looped several times.

I do get a hardlock with Dragon Age 2, but im NOT sure if its caused by the GPU OC ( 1.088v , 900Mhz Core ) Dragon Age 2 stays at about 50-60% GPU usage.. The music in the background or sound continues but the system hard locks.. its NOT that often though and very very hard to reproduce!

[CryptiK]

Comparing the GTX480 PCB and GTX580 reveals differences and similarities. One thing I noticed immediately on the 580 was the lack of filter caps after the chokes, despite there being spaces for them. On the 480, the caps are present. You will also see on the Direct CUII there are filter caps post choke. Anyone with some tantalums lying around want to fill in the blanks on the 580 and see if it makes any difference in clocks?

Pics (first 2 pics from Hardware Canucks review, 3rd pic from techpowerup review):

[Demo]

Since GDDR5 has some sort of ECC, whats the best way to test for stability on this ?

[CryptiK]

Diminishing returns vs clockspeed increase probably. Plot a graph and see when it starts to become non-linear.

Since these cards mem chokes seem borderline under-specced when OC'ing and like to fail, how far is even safe to push the mem 24/7?