More results with ASUS GTX580 DirectCU II

[SF3D]

Yeah, it's clearly 1st of april. I really fooled myself today.

We will have that another baby any second now, so I requested some short time to do bencing and got it. Then I did some HWbot heaven benching.



After that run I realized that score sucks and the reason was very old driver Ln2 is used.

I feel a bit dumb now

[G H Z]

lets see if this helps with no lid

Cool, awaiting your results. Read about how they prep this stuff before each mount when used on cryo equipment, but you wont have to do that yet, it's brand new.

[dinos22]

how much was it splave?

[G.Foyle]

Thanks Hipro and SF3D for suggestions

But something is still amiss, I'm having similar problems as Ronaldo.
Just blew another 15 l of LN2 on this card, using Arctic Cooling MX-2, with IHS still on, +400 mv VID mod, both OCP mods and CBB mod.
I get blue, black, gray and white screens when under -80 and above 1200 MHz. I never get any stars or black textures in Unigine - sometimes it passes 24 of 26 stages at 1350/1300, then suddenly a gray screen. It seems completely independent of temperature, voltage... (-80 to -160, 1,3 to 1,55 V, same random screens of death).
A reference card we got at MOA Poland did better

EDIT: just saw a new kind of screen of death: orange!

[matose]

Try different paste, seems like contact problem or really bad card

[SF3D]

Yeah. Colorful screens are a symptom of overheating. 1.55V and 1350MHz is huge load and bad paste will loose it's ability to transfer heat to pot. Then you will have these symptoms.
Your card might not be the best sample. That is one possibility of course.

E: I made expample pic of the situation. I pushed two Nvidia HS plates together with heavy force and then cooled them down to around -190C. In some point during the cooldown you can clearly hear when the plates loose contact and you can just lift the other part away. Other side was like that. (Paste is crappy ceramique, avoid it)

[matose]

If you have bad paste it's much better to bench at -80 or -100 the most. Under -120 the chances that the paste will be like Petri's picture is very high!

Antec Formula 7 stay ok until -190 C?

[Splave]

how much was it splave?

$180 cause it was a "sample" and they keep trying to make me buy more lol

[Calathea]

Antec Formula 7 stay ok until -190 C?

ME4ME tested Formula 7 on ln2, 980X with CB ~-135 in vantage cputest. Max cpu clock:

[Kabauterman]

@SF3D:

I have never seen something like this happening to our years long used AC MX-2 Paste.

But since the comparisons we also use the Freeze to be on the save side

[matose]

Weird because I saw many favorable reports for AF7

[K404]

I thought there Ceramique was beginning to lose its appeal because it has a higher viscosity and is fairly solid as "hot" as -115? I tried Titan Blue and even immediately after I removed the pot, it was still thin and quite fluid.

IIRC, my best 3D05 MHz have been with Titan Blue by ~75MHz, but i'm not gonna say 100% thats not because of degradation since or bad luck or whatever




Question... do you guys let the card (or CPU) warm up with the pot attached, before benching? I always do that to allow the thermal paste spread out as evenly as possible and also ensures no bad mounts......

[Hondacity]

I thought there Ceramique was beginning to lose its appeal because it has a higher viscosity and is fairly solid as "hot" as -115? I tried Titan Blue and even immediately after I removed the pot, it was still thin and quite fluid.

IIRC, my best 3D05 MHz have been with Titan Blue by ~75MHz, but i'm not gonna say 100% thats not because of degradation since or bad luck or whatever




Question... do you guys let the card (or CPU) warm up with the pot attached, before benching? I always do that to allow the thermal paste spread out as evenly as possible and also ensures no bad mounts......

bulls eye

[rbuass]

Did try Formula 6 and 7, IC7 Diamond and same issues.
Did remove heatspreader and lots of artifacts ()...then give up, and put again heat spreader...
I am almost sure that is the limit of mine card...
would like to buy OCZ Freeze, but look for Elmor and his friends bought all sock from OCZ...lol
This is a best result i got.


and at vantage max 1270/1200 (less than mine reference GTX 580 that reached 1275/1262)
Did try all and all range voltages and temperatures settings

Who can sell me OCZ freeze???.....lol

[Kabauterman]

It's also not easy to get the OCZ Freeze here in germany.

Just two Shops have them on thier Websites, just one Shop has it in Stock and yesterday I ordered 2 pcs so they no more have it in stock

[T_M]

http://teamau.net/509

[Kabauterman]

http://www.youtube.com/watch?v=OLJHG...e_gdata_player

so easy? How much was the clock gain?

[T_M]

Nothing, it was dead already :P

[Kabauterman]

did u tested with any alive card?

Or can someone say if the card is better clocking ofter the removal?

[matose]

It's supposed to be, but on some cards it's very risky to remove even with heat

[G.Foyle]

So it's not enough to cut the glue around the core, like on GTX280? The paste is a bit adhesive?

[Linuxfan]

Who can sell me OCZ freeze???.....lol

For the right price I will sell you anything!

[Hondacity]

So it's not enough to cut the glue around the core, like on GTX280? The paste is a bit adhesive?

video from T_M i think shows alot of info

http://teamau.net/509

[matose]

So it's not enough to cut the glue around the core, like on GTX280? The paste is a bit adhesive?

You must heat it, then use a sharp knife / cutter and insert it between the GPU's PCB and IHS. Some will pop very easy, some will not...

The new batches of GPU's are glued differently and it's more difficult to pop off. The DCU II I have was old batch but the Lightning is new batch

[Splave]

here is indium on Core mounted gave it acid wash first.

To mount it I wet a qtip with a very small amount of acetone, and slightly moistened the GPU core so sheet would stick. Put pot on and heat up to 100c and then cool down to room temp