More results with ASUS GTX580 DirectCU II

[Patch]

Black screen reset over 1.44v in Vantage GT2. Vid and/or smartdoctor
1300 mhz does fine at 1.4v. Same issue at -70 to -130.
HCP 1200
OCP mods (overall and phase)
PLL 1.32

Only have a tpq 1200 OC to try, so can't evaluate single rail.

[SF3D]

Black screen reset over 1.44v in Vantage GT2. Vid and/or smartdoctor
1300 mhz does fine at 1.4v. Same issue at -70 to -130.
HCP 1200
OCP mods (overall and phase)
PLL 1.32

Only have a tpq 1200 OC to try, so can't evaluate single rail.

Change your PSU to Antec HCP or Corsair AX1200. Those solved my similar issues. I have been using Thermaltake TP 1200W PSU's few years and this was first monster which triggered the OCP in it.

Your card seems to be good, so fix that and you will see a lot better results.

[runmc]

I think my problem may be with thermal paste, not so good contact, and getting the pot too cold before the card warms up and losing my contact with pot and gpu core.

Did you just finger tight the thumbscrews on the pot?

[hipro5]

May I add something here which I believe NONE of you have ever thought about?.......

You ALL say nowadays that you use other thermal paste than Arctic Ceramic and you're right.....

You ALL say that after some time you can't bench at the frequencies you did when start puring LN2 in pot of the card.....

You ALL say that at VERY LOW temps you hear a "crack noise" and the card can't bench any more at the frequencies it did.......

You ALL put OCZ freeze thermal paste or any other kind to keep the temps as low as possible on the core.


Well..........HAVE ANY of you though IF the thermal paste that it's BETWEEN the core and the heatspreader CAN HANDLE the VERY LOW temps?....
I believe NOT.....

START from IN THERE then, and AFTERWARDS apply whatever thermal paste you wish......


My suggestion is: REMOVE THE HEATSPREADER......


.

[rbuass]

Mine was very dificult to remove.
I just tryed to remove GTX 580 headspreader it can t get out.Any idea to remove?

[steponz]

need to heat it up.... put a pot on it and heat it up like 60 +.. remove and it should be alot easier to remove.. sometimes fall off...

Reference cards would just fall off most of the time..

[hipro5]

Like steponz said....You NEED to heat it to remove it.....

Take a look how "poor" this heatspreader attached the core....NOT even the half of it....

[rbuass]

I jusr tryed about 100 degrees (GTX 580 reference)...and here is impossible...
Will try with Direct CU2
hahahaha

[steponz]

Really on a reference?? you can take pliers after heating it to pop it off.. you just have to be very careful not to pull the chip off the board.

[Gautam]

My 480 lightning's spreader just decided to come off with the pot without me doing anything...:S

[hipro5]

Try with a turbo torch to heat the heatspreader AROUND the core (NOT in the middle - to it's edges - ) and lift it up with a razor or something with CARE....

[Splave]

hipro do you recommend leaving it ihs less or just changing the ihs TIM thanks

[steponz]

I just popped the IHS on my 580 with pliers.. just have to do it gently.. after heating up...

They do fall off quite a bit....

Taking off the IHS seems to be the best.. your taking off the middle layer....

[rbuass]

Thanks a lot for all you guys...
First thing i will try when coming home you remove this ¨%$#
I remember i heat a lot heatspreader from my reference GTX 580 bit it didn t fall ...then i gave up afraid to kill my only card...lol
Now will try like you told...heat in the edge and put a hot pot about 5 minutes...then pull of gently (lol)..

Best regards for all

[hipro5]

hipro do you recommend leaving it ihs less or just changing the ihs TIM thanks

You could remove it OR if you could find some Artctic Aloumina (not with epoxy - the OLD one - ) you could put AROUND the core a lot of it to cool down the core's PCB traces that carries all this current.....

[Splave]

Now your talkin

[SF3D]

Nothing new on removing the HS on these GPU's : http://www.youtube.com/watch?v=99VMW6Zfz1Q
Of course it did not help anything with refernce design, cause pcb could not deliver enough amps.

I did not gain anything without HS on DC II really, but when the HS was removed and OCZ freeze was applied to core / hs / pot combination it worked pretty well.

Too bad time run out for more testing. It was looking very promising.

[SoF]

well I think I forgot to show until now: stummerwinter and kovsk heating up a GT during Cebit this year - looks like some kinda pretesting before testing it on the GPU

[steponz]

Nothing new on removing the HS on these GPU's : http://www.youtube.com/watch?v=99VMW6Zfz1Q
Of course it did not help anything with refernce design, cause pcb could not deliver enough amps.

I did not gain anything without HS on DC II really, but when the HS was removed and OCZ freeze was applied to core / hs / pot combination it worked pretty well.

Too bad time run out for more testing. It was looking very promising.

Actually I had experiences some differences in the reference card..

With, the card would only do 1220's in 03, with it removed 1300....

I think every card could be different of course.. Just my experience..

[SF3D]

Actually I had experiences some differences in the reference card..

With, the card would only do 1220's in 03, with it removed 1300....

I think every card could be different of course.. Just my experience..

Yeah..you are propably right. That was my first 580 and it was not the winner for sure

[oanvoanc]

tested my card yesterday and here are is a first result under ln2:

card: asus directcu II
pot: old otterauge pot with thin copper base
thermal compound: noctua nt-h1
shaminos f-bios
smartdoctor (posted by shamino on kpcooling-forum)
-80-90 degree celsius load
1,38 volt directly set with smartdoctor
no mods at all

1300 core in 3dmark03
cbb -60 degree celsius, cb not tested yet.


could you guys explain a little bit the mods posted by sf3d?
my first mod on the card will be cbb mod which should be self-explaining...

but how about vgpu mod which adds 0,4volts? is it necessary? how about just using smarddoctor to increase voltages? any experiences?

overall ocp- and per phase ocp mod:
when will the ocp ususally start to kick in? i seemed to have no issues till 1,38v and 1300mhz 03 stable.

what will change after the vpll mod?

thanks for your help guys

[matose]

With my DCU II it wasn't that hard: heat gun on GPU to about 70-80 degrees, Swiss Army knife between PCB and IHS and POP .

Actually the TIM was applied pretty good. LN2 test tommorow so keep you posted !

[ME4ME]

Everyone now popping of those IHS'es in order to find the holy grail in GPU clocks

[Splave]

this is the 1627 03 GPU, TIM is dry and like powder it actually flaked off in chunks lol 110c on kingpin fattie plus butter knife

[rbuass]

Use some different thermal paste. I have not tested antec formula 7, so can not say anything how that works. But what I would suggest is, that go to -50C first and set the voltages, which will be used during 3D runs. Then cool down to -100C and try 1100MHz for example. If it can run, try 1200MHz and then 1300MHz etc.

If you go directly to -130C and then start benchmark + set voltages, your thremal paste have propably lost contact and you will get lock up or black screen.

Grey screen is usually too hot gpu for the voltage +MHz combination. Blue screen is something weird and I have never seen it with these GPU's.

So try again with easier style first.
My rought approach have been giving a lot of issues since GTX 480 Lightning. I saw these issues first time back then and finally I have learned from my mistakes. Well, almost

hahaha
Thanks SF3D,,,,you always a very friendly guy and it will helps for sure...
I hope this card no sucks and it is a good card...or will need to wait about 2 months to come another one to me...
regards