Skinnee Labs - GTX480 Block Round-Up (working thread)

[naokaji]

Click
Although the conclusion is favourable for the product, to me personally it's pointless...
Such a small improvement for that kind of money

Looking at how they tested it seems cool it told them to test it all wrong so they would have some excuses if anyone runs into problems

I remember a few people not being big fans of CoolIt, mostly due to leaking systems I believe.

There was a nice little fall out between bit-tech and cool it a while ago over the cpu version where the review samples kept failing on bit-tech, i'll see if I can dig up the link.
nvm, found it: http://www.bit-tech.net/hardware/coo...t-domino-alc/5

[cstkl1]

Skinnee is this you??
http://hwbot.org/article/news/remova...m_loc_incident
herefore, we issue a 1-year ban to all the people involved: AndreYang, Hiwa, eXtremetweaker.de, KJ and Skinnee.

Check post one, he's banned already. Good thing he has his own site so we still can read the review when it's done.

lol..
u think all that free stuff he got is to post on his site??

lol
this review aint gonna happen
and when it does i think we will have a new gc already.

[sirsiddius]

So this was why our GTX 480 waterblock results were taking so long. While we were begging you for results and while you were assuring us you were working you ass off, you and your cronies were actually messing around with LN2 and conspiring to cheat your way to a win when you figured out you could'nt achieve one legitmatly.

However look at bright side, now that your banned you can devote more time to review testing stuff; I still have enough faith in you to not automatically assume that your test results are 'tweaked' like your 3Dmark score.

[the_dope_chaud]

How can you have faith in any information he hands out after what just happened ? Everything he post/review now, I will take with a HUGE grain of salt....

[Camaroz06]

Well color me disappoint...

[p2501]

Screwing up a competition and his hardware reviews don't really need to be connected. I'm inclined to trust the review data he puts out, Vapors and his testing were profound and I trust that will stay the same.

I'd be really interested in hearing his side of this story. No reason in building up a name in the community and then smashing it to pieces with a pulling a stunt like that. No need to burn him at the stakes IMO.

[sirsiddius]

Here's his story: http://www.xtremesystems.org/forums/...15#post4479815 post number 114

[Waterlogged]

Cam, I'm only gonna talk about this "incident" once. If we were standing together in a room right now, I would you, and ask "why weren't you thinking?".

Pull yourself together, recommit yourself to getting reviews done and hopefully when you come back, we can move forward like nothing ever happened.

[p2501]

Here's his story: http://www.xtremesystems.org/forums/...15#post4479815 post number 114

Must have missed that, thank you.

[JoeChuo]

This is bad! Broke my heart!

[Alexandr0s]

Pull yourself together, recommit yourself to getting reviews done and hopefully when you come back, we can move forward like nothing ever happened.

+1. Hope to see you back soon .

[Boogerlad]

I hope skinnee doesn't get pissed off and join rrr/cause more drama.

[Vapor]

lol..
u think all that free stuff he got is to post on his site??

lol
this review aint gonna happen
and when it does i think we will have a new gc already.

Review IS going to happen (the data has been done for awhile, just the writing is what's holding it up)....and yes, the review sample blocks were sent for the review to be posted on skinneelabs.

I hope skinnee doesn't get pissed off and join rrr/cause more drama.

Never happening



Anyway, he's asked me to post part of the final review data so that you guys know he isn't gone or leaving or any of that. The time away from the forums (and even longer away from benching) does mean more time for watercooling

Below is the chart for flowrate, stock GPU temperatures, and stock VRM temperatures.

Compared to the graph he posted a few hundred posts back, VRM temps differ slightly on some blocks because in the initial runs, the temperature probes used to measure VRM temperatures were making contact with the block, artificially lowering VRM temperature readouts. Not all blocks had thermal tape on the chokes so not all blocks were effected--those that were effected were rerun with individual pieces of thermal tape just on the chokes, rather than one long strip over the entire area.

Flowrate:


Stock clock/voltage GPU Temps:


Stock clock/voltage VRM Temps:

[Hondacity]

crap evga is good - gpu

crap evga is bad - vrm lol stupid...but 66c isn't bad...

EVGA FTW!!!

thanks for posting

[Church]

Finally!
To me it looks that overall desirability (temps vs flow) Koolance #1, EK #2. EVGA #3 because of top temps (but what's with that monstrous restriction? Now sum that restriction up in fermi-sli setups .. )

[Vapor]

For a clarification: "All TIM" means that TIM was used on the VRMs/chokes rather than thermal tape. Not all blocks can be run like this due to clearance issues....but if it can, it definitely shows an improvement in VRM temps

[Alexandr0s]

The Danger Den looks like a great block .

[Vapor]

Finally!
To me it looks that overall desirability (temps vs flow) Koolance #1, EK #2. EVGA #3 because of top temps (but what's with that monstrous restriction? Now sum that restriction up in fermi-sli setups .. )

Ah, in SLI you can (and should, at least with the eVGA and other restrictive blocks) run the GPUs in parallel, which lowers restriction a HUGE amount.

[Vapor]

Also, the VRM temps reported are from Dallas One Wire probes hot-glued and TIM'd between the chokes. They're not software VRM temperatures, but temperatures of the nearest area the temperature probes would fit. The readouts are indicative of VRM temperatures but are not a perfect representation.

Also the importance of the readings is much more difficult to quantify than just straight GPU temperatures. Basically, they need to be held within thermal spec (100+C) to maintain stability. When overclocked and overvolted, that becomes increasingly difficult to do and when it fails to control the temperature, you get runaway power draw and temperature and ultimately failure. So they ARE important, but their stability is a nearly-binary entity. At all test settings, all blocks passed. But at higher settings (above what the Overclocked test settings were), not all blocks could handle the load and that did negatively effect overclockability.

[Wryknow]

Can you share how high you went on the overclocks?

[Systemlord]

Also, the VRM temps reported are from Dallas One Wire probes hot-glued and TIM'd between the chokes. They're not software VRM temperatures, but temperatures of the nearest area the temperature probes would fit. The readouts are indicative of VRM temperatures but are not a perfect representation.

Also the importance of the readings is much more difficult to quantify than just straight GPU temperatures. Basically, they need to be held within thermal spec (100+C) to maintain stability. When overclocked and overvolted, that becomes increasingly difficult to do and when it fails to control the temperature, you get runaway power draw and temperature and ultimately failure. So they ARE important, but their stability is a nearly-binary entity. At all test settings, all blocks passed. But at higher settings (above what the Overclocked test settings were), not all blocks could handle the load and that did negatively effect overclockability.

I'm just dying to see which blocks coundn't handle higher overclocks! I think the Danger Den and EK blocks came out on top because of higher flow rates, GPU and VRM temps with flow rate being very important! The highly restrictive blocks in my opinion are not desirable!

[Vapor]

877 core @ 1138mV (set....drooped to 1070-1075mV or so under load)

I'm just dying to see which blocks coundn't handle higher overclocks! I think the Danger Den and EK blocks came out on top because of higher flow rates, GPU and VRM temps with flow rate being very important! The highly restrictive blocks in my opinion are not desirable!

All the pretesting to decide final settings was done on the Bitspower and at voltages above like 1180mV (set, not actual), it was having runaway issues (i.e., failure).

All blocks passed the overclocked settings for this test (877 core at 1138mV), but better blocks can definitely take more voltage. I don't know where the eVGA block would begin to fail, but it was really close to 100C for VRM temps at the standard overclocked settings, so possibly not much more than that. Putting a fan on it helps significantly though

[Wryknow]

Awesome. Thanks for posting this and please pass the appreciation on to Skinnee!

[Systemlord]

877 core @ 1138mV (set....drooped to 1070-1075mV or so under load)

All the pretesting to decide final settings was done on the Bitspower and at voltages above like 1180mV (set, not actual), it was having runaway issues (i.e., failure).

All blocks passed the overclocked settings for this test (877 core at 1138mV), but better blocks can definitely take more voltage. I don't know where the eVGA block would begin to fail, but it was really close to 100C for VRM temps at the standard overclocked settings, so possibly not much more than that. Putting a fan on it helps significantly though

So if someone wants the highest overclock on their water cooled GTX 480 should we just flash the BIOS with at setting of 1.2v and hope for the best? I've never flashed a graphics card BIOS before, but would like to learn. I expected by now a guide with pictures showing beginners how to set voltages and flash a graphics cards BIOS.

[Vapor]

Use MSI Afterburner....much easier than flashing BIOSes