Which manufacturer is going to be ready with 5850/5870 blocks?

[Bojamijams]

Try adding a pair of nylon washers between the block and the backplate

[millertime359]

Maybe sand down the nylon spacer tubes?

[Mech0z]

Gabe - I tried my MCW60 on my 5870 last week, and found it made v.poor contact with the GPU. I fiddled around with it a bit, but it would only contact one small part of the GPU. Not sure if the GPU is a mm lower vs. the shim or something, or do we need a new mounting bracket for this card? BTW... glad you are doing a full heatsink solution for this card, hopefully it doesnt take too long to land over here in the UK!

Is this a general problem? Just thinking if I should buy fullcover or MCW60 when I get my 5870

[gabe]

Gabe - I tried my MCW60 on my 5870 last week, and found it made v.poor contact with the GPU. I fiddled around with it a bit, but it would only contact one small part of the GPU. Not sure if the GPU is a mm lower vs. the shim or something, or do we need a new mounting bracket for this card? BTW... glad you are doing a full heatsink solution for this card, hopefully it doesnt take too long to land over here in the UK!

I checked the specifications (from AMD reference 3D Models)

The normal distance from silicon surface to pcb is:

4870X2 model for reference: 2.26mm
5850 and 5870 models: 2.41mm

From our MCW60 specs, the normal distance from surface of the copper base platform to bottom of the legs is 1.85mm; on specs, that leaves a 0.56mm gap between feet and PCB, which is standard.

There could be several reasons for a poor contact between this particular MCW60 and the silicon, listed below are the most plausible explanations:

1/ The copper base plaftorm was over-lapped and is out of specs (possible) - Nominal distance should be comprised between 0.76 and 0.63mm
2/ The chrome hold-down bracket is bent downwards at the corners (from earlier mounts), lifting the entire block upwards (likely)
3/ Combination of both of the above


Diagnostic & remedies:

1/ Carefully measure the height of the copper base platform with a caliper or dial indicator to verify compliance with above specs - If below 0.63mm, remedies are to RMA block, or implement #3 below.
2/ Place a straight edge on the hold-down bracket, and verify bow. If excessive bow is observed, remedies are to request bracket replacement from customer service (michelle@swiftech.com) or implement #3 below.
3/DIY immediate remedy: place the block feet down on 120~200 grit dry/wet sand paper (tape edges of paper preferably on glass table). Holding the block flat, rub the feet in a circular motion for a few minutes (2~5mn).

[MikeMK]

Thanks for the comprehensive response Gabe! Ive not actually used this block on an ATI card since Ive had it... its been on a nvidia GTX285, so this is the first time Ive used the ATI hold-down bracket. I'll check it and see if its bent first, and then measure the copper base to see if either/both are the problem. Hopefully its easily fixable.

[punkrockpolak]

Do you know how well the 5870 will perform with the heatsink? It looks plenty thick and I suppose there's no reason it shouldn't do well, but someone testing it would be reassuring.

[millertime359]

Do you know how well the 5870 will perform with the heatsink? It looks plenty thick and I suppose there's no reason it shouldn't do well, but someone testing it would be reassuring.

Most of the full cover blocks only ran water over the ram and GPU. The VRMs were passively cooled.

Alot of people use them here on GTX260s, 280s, 4870s, 4870sx2 and haven't had any issues.

[zbogorgon]

[Locke]

yeah i am liking that "hybrid" solution since I already have an MCW60. Now I just need to find an HD5850... Anyone got a spare I could borrow for a few years?

[zalbard]

1120 Mhz GPU clock on water, omg!

[Bojamijams]

Nice core clock! What voltage is that?

[fornowagain]

Nice core clock! What voltage is that?

1.36v

Dude I need that block

[dnottis]

What did it take to achieve that?

Flashed to ASUS bios and MSI after burner?

[zbogorgon]

What did it take to achieve that?

Flashed to ASUS bios and MSI after burner?

not really, I use a software tool "in development" for setting clocks and voltages

[Razrback16]

Received a picture of our block today, expect them available Oct 8-12 depending on the air carrier. They'll run our usual $99.99. We will also have a block to fit the HD 5850 around Oct. 19th for the same price.

The fact the acrylic/stainless steel top covers the whole block would indicate that water flows over the vregs directly.

Thanks Dean, the block looks great. Question -- will it provide active water cooling to the VRM area or will the VRM cooling be similar to the 4850/4870 blocks where the heat was just sort of pumped into the memory cooled area via other means?

[Razrback16]

Swiftech will be supporting the 5850 and 5870 with both hybrid and full cover solutions.

The MCW60 is already fully compatible with both cards.

Individual heatsink solutions will be offered respectively for 5850 and 5870.
MSRP $29.95 - ETA End of october.

Rendering:

I really like the sink setup over the VRMs. That may be the route I go. Gabe, will you guys release something like that for the 5850X2?

[dnottis]

not really, I use a software tool "in development" for setting clocks and voltages

Alright, at least give me the name!!! Rivatuner? 1.3 volts? so you still have your stock bios on there??

[Mech0z]

I dont dare to buy a 5850 before I have seen how they clock on water compared to 5870
Want more results ^^

[Razrback16]

Most of the full cover blocks only ran water over the ram and GPU. The VRMs were passively cooled.

Alot of people use them here on GTX260s, 280s, 4870s, 4870sx2 and haven't had any issues.

There aren't issues unless you overvolt. At least with mine -- overvolting would result in 130-140*C VRM temperatures. Verified good mounting on both cards. Stock volts will load around 100*C. I'm very interested to find out if the new full cover blocks will have a water path to actively cool the VRMs.

[eternal_fantasy]

Thanks Dean, the block looks great. Question -- will it provide active water cooling to the VRM area or will the VRM cooling be similar to the 4850/4870 blocks where the heat was just sort of pumped into the memory cooled area via other means?

You have quoted the the answer to your question.

[DeanB]

Thanks Dean, the block looks great. Question -- will it provide active water cooling to the VRM area or will the VRM cooling be similar to the 4850/4870 blocks where the heat was just sort of pumped into the memory cooled area via other means?

I haven't seen the internals yet, but the outer design leads me to believe that water does flow over the vregs.

[millertime359]

There aren't issues unless you overvolt. At least with mine -- overvolting would result in 130-140*C VRM temperatures. Verified good mounting on both cards. Stock volts will load around 100*C. I'm very interested to find out if the new full cover blocks will have a water path to actively cool the VRMs.

Eddy said they they did or got alot closer to the VRMs than before.

[Razrback16]

I haven't seen the internals yet, but the outer design leads me to believe that water does flow over the vregs.

Very encouraging, thank you.

[MikeMK]

I see EK now have the 5870 blocks up for order on his site!

[fornowagain]

Ordered