[un]official GTX580 Overclocking Thread

[CryptiK]

Ok i find out that my friend has the same thing with his GTX570 MSI brand, during ATItool error test he has the same high pitched peeeeeeeeeeeeeeeeeeeee and during a game he has the wooooooooo or pooooooooooh so you both are right, it's just how the made the PCB's, nothing to worry about.

Oh and the reason why we can hear it clearly on his GTX570 because he has a different cooler also so nothing that can block the sound (naked PCB).

Oh btw tip for overclockers, difference ingame from 900MHz to 950MHz result in 2-3FPS more but you need a bit more voltage to obtain stable clock and you will get higher temps in my situation it will result in 7 to 9 dregees hotter so it wasn't worth it.

People that use watercooling or any other cooling solution that won't be needing the power pins on the PCB of the GPU you can best set fan min and max to 0%, you need less voltage for your clock.

I'm not seeing how this could be the case, with no load (due to fan not being connected) there is no current draw from the fan circuitry, and setting it to 0% would not be giving the card any 'extra power' as it were.

[Demo]

Thats probably because all the CPU's out there are currently a bottleneck to the GTX580 hence minimal performance gain from 900 to 950Mhz.

[donmarkoni]

oh yes.. i'm in!!

wow donmarkoni 1ghz on the gpu and it makes lower scores ...

http://3dmark.com/3dm11/959068

2mhz less on the memory.

Those were quick runs, before I made a post. Probably not too optimized.

Basically it has like 1-2fps performance boost?


@Topic

Is there a GPU block that can fit on the 580 while the stock VRM/RAM sinks unibody is still attached?

[Demo]

So what GPU block is that?

[CryptiK]

Something from this page perhaps: http://www.ekwaterblocks.com/shop/bl...ga-blocks.html

[p3gaz_001]

page not found... anyway is an universal EK gpu waterblock .

[CryptiK]

That's what I linked....works for me?

[donmarkoni]

Yeah, that's it. Universal WB, EK VGA Supreme, older model, restricted flow, as my CPU WB. I like them like that.

[X4er0]

I'm not seeing how this could be the case, with no load (due to fan not being connected) there is no current draw from the fan circuitry, and setting it to 0% would not be giving the card any 'extra power' as it were.

Than it was a fluke for me and the guys at guru3d, my theory same as yours first, nothing connected no load or current needed, first i needed +1.13 for 900MHz now i set it on 1.10000 and it's 100% stable, nor artifacts or errors.

[CryptiK]

MSI Afterburner wont allow it, even if in user define screen I set 0% for all set point in the graph.

Do you have to BIOS mod the fan values to 0%?

[Paradoxxx]

This is what I've got so far at MAX voltage 1.1475, stock cooling with fan speed at full. Temps rose to 63C during tests.



Interestingly enough, I got a lower overall and graphics score as compared to the guy in post #8 even though my clocks are higher. CPU/Driver related?

[Demo]

I get about 66-67c at full fan speed with 1.150v

[cowie]

modded 580 lightning bios for 1.21v

http://download1024.mediafire.com/1c...hting+bios.rom

[cowie]

580 lightnig stock 3d11


580 lightning stock volt oc .997vid(dmm)



was able to loop 944/1200 more then a few times no issues at all i'm sure there is more for hwbot type benching was just setting the paste i had done a few minutes earlier

+200vm+1.027v 1.19v


max i could set with software voltage wise 1.36v(1.35ish dmm)have not tried to move jumpers yets.
just set i did not let card go in 3d with this for obvious reasons being on stock cooler
with a modded bios add .06v thuogh software np.

[X4er0]

MSI Afterburner wont allow it, even if in user define screen I set 0% for all set point in the graph.

Do you have to BIOS mod the fan values to 0%?

I always overclock via bios, first test in MSI afterburner to see how much volt is needed for 100% stability on a clockspeed and i use ATItool because it's very fragile.

Then i go for temps check and i use furmark and OCCT for bizarre temps.

If everything is allright i extract the bios than i alter bios with de speeds and set fan min and max on 0% so if i use GPU-Z now it says 0%, then i upload the bios and go back in MSI afterburner and see if i can lower volt while maintaining stability, that's how i found out i can achieve 900MHz and 4200MHz with 1.1 Volt instead of 1.13.

Hope this helps you with the right steps and order.

Now my GTX580 gets 58 degrees celsius max on load during DOW 2 Retribution and Crysis 2, cooler is Shaman BUT i bought 2 bags of small heatsinks to cover up the VRM's, i used a GTX280 large heatsink from thermalright for it and it gets warm to the touch, not so hot that you'll burn your finger btw.

edit: if you need help holla at me...

[Paradoxxx]

This is what I've got so far at MAX voltage 1.1475, stock cooling with fan speed at full. Temps rose to 63C during tests.



Interestingly enough, I got a lower overall and graphics score as compared to the guy in post #8 even though my clocks are higher. CPU/Driver related?

Just to answer my own question here, the difference lies with the driver used. The 267.60 Driver produces a higher score.

[Demo]

Could it also be the core difference from 985 to 990 ?

[Paradoxxx]

Could it also be the core difference from 985 to 990 ?

Yes the GPU change would have had some effect on the score, Certainly nothing to warrant a 300+ score increase.

I redid the test with the same GPU clocks used before from proper comparison.

[p3gaz_001]

266.58 driver i don't think were well optimized for this our GTX580, 260.70 & 270.51 i think are the best drivers we can have for our GTX's, btw i do not really care about that bla bla driver not approved on the 3dmark11 page...... :|

i've made two runs, 1ghz gpu & 992mhz gpu with 4.5ghz on cpu

[Demo]

Just want to ask for some input, this is a rail diagram on my PSU.. Silverstone 750ZF

http://www.jonnyguru.com/modules.php...=Story&reid=25



The 8Pin is on the 12V4 rail so that what Im using on my GTX580 for the 6pin Im also getting it from the 12V4, would it make any difference on the PSU loading if I use the 12V3 6pin instead?

I think I have all my fans drawing from the 12V3, 3x 140MM LED fans, 2x 120MM LED fan.

[p3gaz_001]

this is my crysis2 gameplay ...

http://img217.imageshack.us/g/crysis...515142231.jpg/

i'm really enjoying it, hope to see it in DX11 ....

[p3gaz_001]

watch out

[CryptiK]

580 lightnig stock 3d11
[IMG]http://img405.imageshack.us/img405/1881/580stock.jpg

580 lightning stock volt oc .997vid(dmm)

[IMG]http://img808.imageshack.us/img808/4151/580stockvoltoc.jpg

was able to loop 944/1200 more then a few times no issues at all i'm sure there is more for hwbot type benching was just setting the paste i had done a few minutes earlier

+200vm+1.027v 1.19v
[IMG]http://img816.imageshack.us/img816/3699/580100mvram119vsoftware.jpg

max i could set with software voltage wise 1.36v(1.35ish dmm)have not tried to move jumpers yets.
just set i did not let card go in 3d with this for obvious reasons being on stock cooler
with a modded bios add .06v thuogh software np.

[IMG]http://img203.imageshack.us/img203/9590/maxvoltsoftwarenobiosfl.jpg

Is that a 900MHz core speed on 0.993v ??!!!

watch out

[img]http://www.pctunerup.com/up/results/_201104/20110418030733_3dm06htoffOK.jpg

Nice score. HT on would make that better again.

[p3gaz_001]

@Cryptic i know that, but at this speed HT on makes me bluescreen ....

unigine 2.1 hwbot version



unigine 2.1 goliath mode (all maxxed out)

[vern]

@Cryptic i know that, but at this speed HT on makes me bluescreen ....

unigine 2.1 hwbot version



unigine 2.1 goliath mode (all maxxed out)

Can run 1000Core@3DMV?
Thanks