Swiftech Releases Komodo HD5800 F/C Waterblock For ATI

[dejanh]

Phil, listen, I do not know how to put it to you more simply...

I found that paste/grease/whatever-you-want-to-call-it method does not work "as advertised". With some blocks/cards/combination there of it does not work period. I do not know how else to put this to you so that you will understand what I am trying to show Gabe.

If you still feel like pushing whatever indeterminate point you are pushing then please stop, I am not interested in hearing your opinion any more. At this point unless you want to do the same tests as me and show that the paste method works perfectly for you with stress test screenshots and block pics and all, or not, please refrain from posting back. Why did you bother commenting on the shorting issue then? Or are you suggesting we should just keep quiet when a problem is found. The problem with the paste method is crystal clear, from pictures, from measurements, from explanation.

I'll leave it up to Gabe now to comment.

As far as my card goes, remounting was done about 5 times total. Did it have something to do with the card dying, probably. Am I blaming the block for "killing" the card directly, no. At no point in time did I come into this thread whining how "omfg, my Komodo killed my card". I would almost 100% certainly say it is a combination of the stress the card underwent while diagnosing this problem and mounting and remounting. Either way at the end of the day I found a problem with the Komodo block (possibly in combination with at least some cards) and if I can help one person think twice about using the paste or understand that using paste may not get them the desired effect then I have done my part for the community. After all, you yourself said instructions are not foolproof.

[Badfastbusa]

Phil, listen, I do not know how to put it to you more simply...

I found that paste/grease/whatever-you-want-to-call-it method does not work "as advertised". With some blocks/cards/combination there of it does not work period. I do not know how else to put this to you so that you will understand what I am trying to show Gabe.

If you still feel like pushing whatever indeterminate point you are pushing then please stop, I am not interested in hearing your opinion any more. At this point unless you want to do the same tests as me and show that the paste method works perfectly for you with stress test screenshots and block pics and all, or not, please refrain from posting back. Why did you bother commenting on the shorting issue then? Or are you suggesting we should just keep quiet when a problem is found. The problem with the paste method is crystal clear, from pictures, from measurements, from explanation.

I'll leave it up to Gabe now to comment.

As far as my card goes, remounting was done about 5 times total. Did it have something to do with the card dying, probably. Am I blaming the block for "killing" the card directly, no. At no point in time did I come into this thread whining how "omfg, my Komodo killed my card". I would almost 100% certainly say it is a combination of the stress the card underwent while diagnosing this problem and mounting and remounting. Either way at the end of the day I found a problem with the Komodo block (possibly in combination with at least some cards) and if I can help one person think twice about using the paste or understand that using paste may not get them the desired effect then I have done my part for the community. After all, you yourself said instructions are not foolproof.

Unfortunately, some people here, no matter how many times or different ways you say something, just dont "get it" for some reason... In my worklog thread, I had to say about 6 different times that my overclocks were just for fun and not to be meant for 24/7 stable clocks, and one after another, they each wanted a ton of stability tests LMAO

In conclusion, there are some really smart people here, along with some real idiots.

[Boulard83]

Unfortunately, some people here, no matter how many times or different ways you say something, just dont "get it" for some reason... In my worklog thread, I had to say about 6 different times that my overclocks were just for fun and not to be meant for 24/7 stable clocks, and one after another, they each wanted a ton of stability tests LMAO

In conclusion, there are some really smart people here, along with some real idiots.

[Philwong]

Phil, listen, I do not know how to put it to you more simply...

I found that paste/grease/whatever-you-want-to-call-it method does not work "as advertised". With some blocks/cards/combination there of it does not work period. I do not know how else to put this to you so that you will understand what I am trying to show Gabe.

If you still feel like pushing whatever indeterminate point you are pushing then please stop, I am not interested in hearing your opinion any more. At this point unless you want to do the same tests as me and show that the paste method works perfectly for you with stress test screenshots and block pics and all, or not, please refrain from posting back. Why did you bother commenting on the shorting issue then? Or are you suggesting we should just keep quiet when a problem is found. The problem with the paste method is crystal clear, from pictures, from measurements, from explanation.

I'll leave it up to Gabe now to comment.

As far as my card goes, remounting was done about 5 times total. Did it have something to do with the card dying, probably. Am I blaming the block for "killing" the card directly, no. At no point in time did I come into this thread whining how "omfg, my Komodo killed my card". I would almost 100% certainly say it is a combination of the stress the card underwent while diagnosing this problem and mounting and remounting. Either way at the end of the day I found a problem with the Komodo block (possibly in combination with at least some cards) and if I can help one person think twice about using the paste or understand that using paste may not get them the desired effect then I have done my part for the community. After all, you yourself said instructions are not foolproof.

Since when have I contest the mounting problem or have you been reading between lines? Like it or not this is still an open forum and you're most welcome to ignore my posts if you desires. For those who prefer to leave sarcastic remarks, pls go ahead and stress test your Komodo at high reso and AA, then whine that temps are bad for all you want . S-h-i-t happens when you play with fire. Remember to do your homework and buy the right block next time. It's not as if there's no obvious hint that the Komodo does not stack up well in VRM cooling. Period.

Phil

[gabe]

It's not as if there's no obvious hint that the Komodo does not stack up well in VRM cooling. Period.

Phil

Your final statement either shows a poor understanding of the thermal and mechanical requirements to cool these vrm's or an unabated support for your favorite brand, both of which are perfectly understandable and acceptable I should say. I would hope nonetheless that my response above will shed some more light for you on the problem at hands.

As vendors, we make choices as to what we want to accomplish with respect to our users. I already said and demonstrated in our lab that our solution can cool the VRM's just as well as any other extreme solution. The reality is, the heat loads for these components are such that just about ANY full-cover solution providing a decent TIM will cool them enough to be substantially below operating specs. The difference as I pointed above, is that we try really hard to do it safely.

[Philwong]

Your final statement either shows a poor understanding of the thermal and mechanical requirements to cool these vrm's or an unabated support for your favorite brand, both of which are perfectly understandable and acceptable I should say. I would hope nonetheless that my response above will shed some more light for you on the problem at hands.

As vendors, we make choices as to what we want to accomplish with respect to our users. I already said and demonstrated in our lab that our solution can cool the VRM's just as well as any other extreme solution. The reality is, the heat loads for these components are such that just about ANY full-cover solution providing a decent TIM will cool them enough to be substantially below operating specs. The difference as I pointed above, is that we try really hard to do it safely.

Gabe, the EK is superb in VRM cooling but is quite a hassle to install. Whereas the Komodo is quite the opposite with the added advantage of better GPU cooling. The difference here is EK is channeling coolant over the VRM, while Komodo uses micro pins for the GPU. Like I mentioned earlier, the thermal paste tweak should work for EK and its already low VRM temp as well. To me, it is a matter of personal priorities until a near-perfect block appears. For the record, I'm not affiliated with EK and neither do I own any of their products.

May I suggest you contact the German reviewer to carry out third party testing between your block and EK's using thermal paste for the VRM? Could you also share with us the resolution and AA settings used during your in-house testing and whether the heat output increases appropriately with higher Furmark settings?

In addition, it may be advisable to warm your customers that there may be contact issue when using paste instead of pads regardless of whether they are seasoned or advanced users. At least could bench or evaluate the card cautiously after installation.

Lastly, if you could revise the block and force the coolant to flow closer to the VRM, that could be the safest solution to enhance Komodo's performance.

Phil

[gabe]

Gabe, the EK is superb in VRM cooling but is quite a hassle to install. Whereas the Komodo is quite the opposite with the added advantage of better GPU cooling. The difference here is EK is channeling coolant over the VRM, while Komodo uses micro pins for the GPU. Like I mentioned earlier, the thermal paste tweak should work for EK and its already low VRM temp as well. To me, it is a matter of personal priorities until a near-perfect block appears.

May I suggest you contact the German reviewer to carry out third party testing between your block and EK's using thermal paste for the VRM?

In addition, it may be advisable to warm your customers that there may be contact issue when using paste instead of pads regardless of whether they are seasoned or advanced users. At least could bench or evaluate the card cautiously after installation.

Lastly, if you could revise the block and force the coolant to flow closer to the VRM, that could be the safest solution to enhance Komodo's performance.

Phil

Thank you for your suggestions, but your comments confirm an incorrect thermal and mechancial assessment. Again I do not blame you for such, as you lack some basic information, sot I will try to it provide below.

1/If you take the time to compare the distance between the heat sources and the coolant between said competing solution and Komodo's, you might be surprised to find that the coolant might actually flow closer to the hot spots in the Komodo.



2/ Nonetheless, irrespective of the preceding statement, it really doesn't matter because the heat load doesn't warrant immediate coolant proximity to do the job well above and beyond what is necessary (which I would point out completely redeems said competing solution from my previous statement)

3/ The key element in the whole affair is not thermal but mechanical: the TIM joint is the key element here, not the thermal design.

So now you should understand why a solution favoring TIM at the detriment of safety presents an advantage. Again, as cooling solution designers we all have the freedom to make a choice in our design considerations, and as a matter of design philosophy, I choose safety first, AND performance second.

[Philwong]

Thanks for clarifying.


Here's a closeup of the EK block. Check out the "step" visible from this particular angle.

Phil

[gabe]

Here's a closeup of the EK block. Check out the "step" visible from this particular angle.

Phil

It is difficult to tell without a transparent view in overlay on the board, but the step I see there appears to be for clearance of the inductors, which would place the vrm's East of that, thus at the very edge of the flow path. But really it doesn't matter one way or the other. The flow path in these, and other blocks is close enough to provide more than ample cooling to the VRM's.

The difference in performance is directly linked to the TIM joint, and the quality of the TIM joint is directly linked to the gaps that are been allowed by design. I will go one step further, if you reduce tolerance gaps with the VRM's, you take the risk to kill the GPU TIM.