Just wanted to be sure. The contact gona be OK without the tape ( thickness ) ?Quote:
Originally Posted by gabe
Printable View
Just wanted to be sure. The contact gona be OK without the tape ( thickness ) ?Quote:
Originally Posted by gabe
almost forgot one more: don't smoke when you are spraying or :explode2: :zombie:Quote:
Originally Posted by EnJoY
yes, according to our tests it is much better.Quote:
Originally Posted by Boulard83
Thanks Gabe !
Thanks for the heads up. I was wondering how that one got missed :p:
Time to report Komodo's performance
Idle (rad fans throttling @ 1.1 to 1.2K)
GPU/MEM/Shader/VRM: 40c/40c/44c/44c
Furmark (rad fans @ 1.3 to 1.5K)
GPU/MEM/Shader/VRM: 52c/60c/65c/86c
I'm still using thermal pad for VRM because it's too troublesome for me to switch over to paste.
The high temp is not exactly a big deal since the VRM usually hovered around 60c when I play MW2 and Dirt 2.
Phil
I have similar VRM temp with the pads. But i barely brake ~41°c under furmark. Ambiant ~25°c and water ~31°c after 10 min of furmark at 950/1300 1.175c.
My ambient and water temps are 30+c and 42c, respectively. The cooling performance is also dependent on individual setup. Mine is far from optimal due to space and layout constraints.Quote:
Originally Posted by Boulard83
Phil
Got my Komodo in today... Man what a sexy block!!!
Hey Gabe, I have a question for you...what length of Bitspower direct connect do I need for connecting two of these in CrossFireX on EVGA Classified E759 (NF200 version)? I measured things and it looks like I need the 25mm connector (http://www.sidewindercomputers.com/bishg14simid.html), but I am not 100% sure. This also matches what you have printed here http://swiftnets.com/products/Komodo-gtx285.asp.
Thanks in advance.
Edit: Well, I just put in an order for the 25mm direct connect fittings. I read your page, measured, then measured again, then calculated, then calculated again and I always come out to ~25mm. I really hope I got this right.
I disagree. The fire is part of the fun. ;)Quote:
Originally Posted by gabe
Hmmm....well...Quote:
Originally Posted by gabe
I recently bought two Komodo blocks and installed one on one of my HD 5850 card to do some overclock testing while I am waiting for my second HD 5850.
I went straight for the Advanced Thermal Utilization step seen on page 5 here http://swiftnets.com/products/instal...ultilangue.pdf because I did not even want to bother with the thermal pads but I am getting awful VRM temps and I think I know why...the block does not seem to be making adequate contact with the VRMs at all, but there is no way to screw in the block any further without cutting the fasteners to shorten them.
At 1.225V and 1000/1250 overclock running Furmark the VRM temperatures go over 100C. At 1.187V and 975/1250 overclock the VRM temperatures hit high 90s. I did not dare let it run to see what the max temperatures would come to, but all I know is that they kept on rising when I stopped the test. The stock air cooler kept the VRM temperatures to ~80C at these voltages and clocks. GPU temps themselves are excellent topping out in mid 30s.
Any ideas what to do? As it is right now I have no choice but to back down the overclock to get the temperatures on the VRMs within working range (i.e., below 100C).
Something is definitely wrong, you might try a reseat or something...Quote:
Originally Posted by dejanh
My best I have got so far stable is 1081/1306 at 1.35v... My core maxed out at 35*C and my hottest VRM got to 65*C... I think I will stick with this overclock, I don't want to push my luck with BC2 coming out tomorrow hehe, its stable, and its a pretty respective overclock IMO.
It still isn't breaking a sweat yet, so i'm sure it will go further. :)
BTW, this is at 75.8*F ambient temps, and with my san ace 1011 fans at almost silent RPM's, but I have a pretty overkill watercooling setup.
I took down the loop, removed the block and I was right, the contact with the VRMs is terrible. Only 1 VRM seems to have made good contact and all of the ones on the back (rear) of the card (5 or so) did not make any sort of contact really. I re-did all of the block fasteners/bolts to tighten them as much as physically possible, re-applied the paste, put everything together again and did not get any improvement in VRM temperatures at all. Running Furmark 1.8.0 in windowed mode at 1920x1200 with 8x AA on loads the VRMs to 100C in ~2 minutes. This is brutally bad. Check out the screenshot below...Quote:
Originally Posted by Badfastbusa
@2 minutes hottest VRM is at 97C and climbing steadily...
http://img.techpowerup.org/100302/vr...er_remount.jpg
Gabe?
Did you clean the tape off really well off of the waterblock, along with the clear gooey leftover?
Yep all of the pad, all of the tape, all of it period. I've done it twice over, mounting problems are out of the question now. There is physically no way the block is gonna get any tighter to the graphics board.Quote:
Originally Posted by Badfastbusa
I really want to get Gabe's comment on this but most of all I want to see what happens with the incoming Asus EAH5850. I am currently running MSI R5850. I know that there has been issues in the past with HD5970 cards where certain components on the cards sat higher/lower than usual. These did not happen to be MSI cards but still, I am not excluding anything at this point. I cannot believe that Gabe would have missed this in design so I want to get his comment...then again, who knows. I will have to measure the thickness of both blocks now too, the one that is mounted and the one that is in the box still.
I wonder if anyone else with the Komodo and paste instead of the tape would be willing to do some Furmark 1.8.0 action at 1920x1200 with 8x AA. Just to point out too that running Kombustor instead of Furmark is not the same thing. The electrical current (amp) load on the VRMs is much higher with Furmark 1.8.0 and above settings than that of Kombustor. At least that's what I have observed.
Edit: I powered down the system and measured the thickness of the mounted block with the standoff vs. the thickness of the block with the standoff that is still in the box. There is a total of 0.35mm-0.4mm difference in thickness between the two blocks. This could be the reason for such awful VRM temps. The mounted block is ~15.55mm thick measured with a digital caliper from the top side of the block to the top of the standoff, and the one that is still in the box measures ~15.2mm from the top side of the block to the top of the standoff. Considering that we are dealing with very fine measurements here this could make all the difference between the VRMs cooling or not. Also note that I re-did the measurements several times and they always come out to within 0.05mm so the difference of >0.3mm is definitely there.
I am out of the country for 2 weeks travelling.. Will try to get back to you asap in between two meetings..Quote:
Originally Posted by dejanh
Thank you sir :up:Quote:
Originally Posted by gabe
Im also waiting for this Gabe ... im on thermal tape and the temp are so-so on the VRM ... i wish to switch to Paste but im afraid now ...
Don't do it. My card is now toast too. I'll show tomorrow that this does not work 100%(explanation + pics).Quote:
Originally Posted by Boulard83
Sorry to hear that, but the new 5850 and 5870 are supposed to have built in overheating protection against killer software such as Furmark. Plus honestly speaking you are pushing the card to its limit at such high resolution and AA settings.Quote:
Originally Posted by dejanh
One of the workarounds is to file the standoff to improve contact, but then it might warp the card in the long run as well. There's a reason why most full face blocks use thermal pad and that's to buffer for the different height tolerances.
Phil
What is this, http://www.moderatesystems.org?Quote:
Originally Posted by Philwong
I doubt the card has protection in that sense since I doubt VRM would be allowed to hit 110C+. Anyway, what killed the card is the mounting, and remounting, and remounting in hopes to get data for Gabe and to solve this problem. Besides, nothing that is in the installation manual is out of line in terms of proper installation. Unfortunately, I have two Komodo blocks and now a dead card for which I can show that paste method does not work. The uneaven thickness of the block, height of stanoffs, etc., all play a part and when you are dealing with measurements as small as fractions of a millimeter something being off by 0.1mm or 0.2mm or more is significantly off. I will have the pictures up very soon with some data. It's unfortunate I had to sacrifice a card to get this. :shakes:
Bummer about the card, but good information to know.
Dude, that freaking sucks to hear man, that is a ton of money wasted, and I am sure you are very unhappy, especially seeing as BFBC2 is out now and is epic and you can't even play it... However, I haven't had any troubles out of mine, and getting absolutely great temps... I followed the instructions to a T and everything has worked flawlessly... I ran furmark with the exact specs you wanted me to run it and my temps didn't ever get out of hand, so I don't know what to tell ya man...Quote:
Originally Posted by dejanh
It would be cool if gabe would throw his insight in on this and maybe give you a helping hand.
Unfortunately, the Komodo is not the most "extreme" block as far as VRM temp is concerned. It's all about expectation and you ought to be looking at EK for the lowest possible temp. Like I mentioned earlier, if Swiftech can bring down the temp by such a huge degree just by using paste, there's no reason why EK will not benefit from this tweak as well.Quote:
Originally Posted by dejanh
Phil