thanks for the help hazzan :up:
Wish me luck, rigged up my hairdryer :rofl:
Printable View
alright the ghetto mod worked great. 2.5hrs benching cold half of which was full pot and no frost on mem on either side and or on the backside at all :clap:
mem ic's measured 20-30c on either side. PCB backside near power area stayed at 35-39c backside by switches 28-33c and backside near bracket 30-34c. :up:
Bad news is the card still wouldnt clock at all like it should
here is my 5 pages of notes condensed
Start settings +250c +250m +300pll
1005/1100 @ -48c pass [no issues]
1055/1100 @ -70c freeze [5s into bench]
Start settings +300c +300m +300pll
1055/1100 @ -75c fail [1m20s *crate drop causes volt spike grey screen]
Start settings +200c +300m +300pll
1055/1100 @ -65c fail [grey screen]
Start settings +150c +300, +300pll
1055/1100 @ -65c fail [grey screen]
1055/1100 @ -75c pass
1100/1100 @ -80c fail [checkerboard screen]
1100/1100 @ -70c fail [grey screen]
Start settings +100c +300m +300pll
1100/1100 @ -75c freeze [screen frozen in bench]
1100/1100 @ -97c pass
1150/1100 @ -97c fail [grey screen]
Start settings +125c +300m +300pll
1150/1100 @ -100c fail [display driver 5s in]
1150/1100 @ -120c fail [display driver 2s in]
Start settings +075c +300m +300pll
1150/1100 @ -100c fail [display driver 2s in]
Start settings +125c +325m +300pll
1150/1100 @ -120c fail [display drive 5s in]
Start settings +200c +325m +300pll
1150/1100 @ -135c fail [display driver 15s in]
Start settings +250c +300m +300pll
1150/1100 @ -140c fail [checkerboard screen]
Start settings +225c +300m +300pll
1150/1100 @ -145c fail [screen freeze 42s in]
Start settings +212c +300m +100pll
1150/1100 @ -165c fail [display driver 5s in]
Start settings +175c +300m +300pll
1150/1100 @ -165c fail [display driver 5s in]
Start settings +237c +300m +300pll
1150/1100 @ -165c fail [freeze screen 58s in]
Start settings +250c +300m +300pll
1150/1100 @ -165c fail [freeze screen 58s in]
Start settings + 231c +300m +300pll
1150/1100 @ -165c fail [grey screen 1m25s in]
Start settings +224c +300m +300pll
1150/1100 @ -165c fail [display driver 1m25s in]
Start settings +325c +300m +300pll
1150/1100 @ -165c fail [grey screen 20s in]
Start settings +237 +300m +300pll
1150/1000 @ -165c [freeze screen during last 2 seconds of bench :clap:]
Start settings +237 +325m +300pll
1150/100 @ -165c [grey screen 20s in]
::Tidbits::
Core Voltage is too unstable?@#?
+231 = 1.383 with upper 1.398 @ (crate drop) lower 1.379
+224 = 1.377 with upper 1.389 @ (crate drop) lower 1.361
+293 = 1.452 with upper 1.467 @ (crate drop) lower 1.436
When the crate hits the water in vantage the voltage spikes which usually causes the grey screen to appear. If you pass the crate dropping it grey screens when the jetski jumps over the camer which spikes as well. :down:
Is it worth it to do the hard mod and forget the software? Im guessing not because you guys are using it no problem. What a waste!
Pics are no frost on board after 2.5 hours :D too bad it didnt matter
I wouldn't push PLL like that until you see scaling problems and the other voltages have been explored. It almost sounds like your card has bad IHS contact, maybe check that out.
maybe it will fall off when I take the pot off rofl
Mine did, no joke. :p: Came right off along with the pot.
So much testing and still nothing huh? :( If you did something wrong there should've been some improvement by now.
Good to see more and more people getting up to 1400 at least. Congrats on a good card/score Hazzan :)
I tried mine for the first time last night and also hit a limit too low. I kept the PCB warm by mounting a waterblock on the back of the PCB on top of 2 layers of liquid electrical tape and a thermal pad for better heat/cold transfer. Then I put my rads in front of the kerosene heater and the water temps got up to around 70C. Only had to hit it with the hair dryer ever 5 minutes or so....full pot for about 2 hours.
But, it didn't help my clocks :( Similar to Allens results, had stability troubles past 1150 (only tried +250 to +350 on the core). memory would not clock past 2100....Just my first go though, I'll try again later ;)
(I'll post up some pics of the water block later)
I must see pics of your heater! thats an awesome Idea miah :D
People with good clockers does your volts fluctuate so much?
+293 = 1.452 with upper 1.467 @ (crate drop) lower 1.436 Swing of .031v
A lot of input in this thread, but unfortunately only few have got the card to 1400mhz.
I did not have any more luck with this than splave and others with problems.
This issue does not have that much to do with the PCB temperature and
I am quite sure, that it has nothing to do with pll temperature either (too cold pll crystal will give you sharp triangles)
because I removed PLL parts and took them away from pcb, without any effect.
Ram temperature does not change the core overclock.
I removed the HS and it could take a bit more voltage,
but the card can not do even 1mhz more on the core clocks.
Voltages I have tested are from 1.35V up to 1.6V. 1.55V is best voltage for my cards and if there is any less, they will just fail.
I have got all tested cards up to 1250MHz on core freq, but then I fail.
So, I am out of options here and I decided to leave this design for more skilled people :(
P.S People should write only real voltages from voltage meter like splave just did, cause for example +100 is totally different in all cards. It depends on P0 VID, temperature of the core, and the current load (mhz + V) :up:
:rofl: @ club
Here are the pictures....I was pretty impressed, but I'm thining of adapting a lager aluminum block to cover a wider area that I'll mount in-between the water block and the PCB....Should work a bit better.
http://i161.photobucket.com/albums/t...llen/GPUWB.jpg
http://i161.photobucket.com/albums/t...n/Kerosene.jpg
Love seeing stuff like that.
Just careful you dont crush any surface components with your mounting pressure
lol that's awesome miah. Too bad you didn't get big clocks though :-/
That's a darn good idea.
Hope im the first person to ever suggest this:
Water heating loop with a constantly running kettle as the reservoir :D
I thought about this too....but the kitchen is upstairs and the loop would have been about 50ft long :eek:
My other idea was to run a hose straight from the tap, it uses a propane instant water heater and can get the water temp up to 75C :up:.....But I was worried about the lake I'd form in the ally outside my window :rofl:
I could have done it with a camping stove (propane), but the other thing that worried me was the pump seeing such high temps.
I was actually REALLY suprized how hot the water got, this worked much better than expected. I love my FFF&FM, it's so flexible is application :p: The Kerosene is really cheap for me too (standard heating for Japanese homes), so it's a pretty cost effective solution too (where electricity for the hair dryer is outrageous) :D
You don't own an electric kettle?!?!?
No....electricity is expensive here....we use gas ;)
i got a little interesting insight to these issues,
dino and i took a reference 480 last night that struggled to hit 1075mhz stable at any temps, voltage, continual grey screens..
when we heated the PCB so the card was not too cold, all of a sudden it will bench at 1175 and was still scaling... when we though okay lets let it get cold again, issues again...
Miah you've got a very bright solution there, but maybe not very effective.. By using that waterblock your also warming up the GPU..
The 480 core really needs all the cooling it can get. Warming up just the places which i think elmor marked at his pictures should do the job just fine :)