4870 Waterblock Options

[jonny_ftm]

I had mosfet sinks on the voltage chips and a 120mm fan sitting on top and it still crashed after 2-3 minutes in furmark on my 4870.

What you say is described in many many other forums if you google it. Early in this thread, there's the area getting too hot. It is back of the card that needs airflow

[SNiiPE_DoGG]

I dont get why you all are so anal about not chopping the stock heatsink for the PWMS this is what I get with my voltmodded 4870s running in a water loop together (obviously when I load one card the other one gets hotter)

[gabe]

I had mosfet sinks on the voltage chips and a 120mm fan sitting on top and it still crashed after 2-3 minutes in furmark on my 4870.

We did some further investigation about cooling the Vitec inductors and confirm that operating temp range is -40 to 125C.

These do get hot, and it is perfectly normal. The amount of ventilation provided by the stock cooling solution (the component is located under the fan inlet) is equivalent or less than that described by several posters in this thread.

According to our informed sources, there is simply no need to sink these components, and required cooling is none other than a mild air circulation around the card.

the crashes reported in this thread are occuring for other reasons. software? overclock ?

[dnottis]

We did some further investigation about cooling the Vitec inductors and confirm that operating temp range is -40 to 125C.

These do get hot, and it is perfectly normal. The amount of ventilation provided by the stock cooling solution (the component is located under the fan inlet) is equivalent or less than that described by several posters in this thread.

According to our informed sources, there is simply no need to sink these components, and required cooling is none other than a mild air circulation around the card.

the crashes reported in this thread are occuring for other reasons. software? overclock ?

Thats not exactly true. I have been working with people at ATI and have gotten alot of information - albeit I haven't "fixed" the issue yet. There is absolutely a need to have sinks on the VRM hardware and in fact large heat sinks. I used the MCR21s from the GTX from the VRMs and was told that they are too small to cool them. Even with active airflow it may not be enough, they suggest blocking the entire area. The Vitec and Pulse chips may not need additional cooling but will need airflow.

Another piece of information that was critical is that the 4870 without the stock cooler will allow the GPU to draw more current. Apparently this is tied in with the 3 pin header of the stock fan.

Im not sure how much more I can share at this time, as I do have more particulars about the circuitry of the HD4870 - there is also a VRM regulator that I want to try and sink and see what effect this has (VT1165MF). You can find it's location in the link below.

http://en.expreview.com/2008/06/24/f...and-hd-4850/4/


The latest beta build of Everest 1455 will show diode and VRM temps. get it here - http://www.9down.com/EVEREST-Ultimat...55-Beta-29037/

With the VRMs, Vitec and pulse chips sinked AND a 120mm fan blowing over the area when I run Furmark the VRM temp in Everest 1455 beta shows the VRM spiking to 125C and then I get a black screen - thermal shutdown.

I'm awaiting replies on a few emails I have out with ATI and have an RMA number just in case. I have no issues in games but I really want to know where the VRM temp in Everest is coming from so I can sink that chip and see if this is in fact the problem. Others claim it's a power issue which I fail to believe in my case as I'm using a quad rail 1000 watt PSU.

[babalouj]

my aluminum unisink copy is going well. I am almost finished milling it out. I will report back with my VRM temp results when I get the thing strapped on.

[Kaane]

ok the chips that are getting too hot on my card are the small 5 chips next to the Vitec chip, not the vitec chip itself. So I think we are talking about 2 different things.

[dnottis]

I've changed the sinks around, sinked some other parts that I thought might be part of the issue.



I actually set this 80mm fan right on top of the VRM area. This was pushing about 30 CFM right over the card.



VRM hit 125c and shut down. Clearly it's not a temperature issue because every heatsink was cool to the touch even at the point it went to a black screen.

[dnottis]

Here is what I believe and what we need to test. The latest build of Everest shows the GPU VRM amperage draw (last reading) for the HD4870s. I need some others to run Furmark and watch the amperage draw. I dont think it's a temperature issue anymore, nothing is really hitting 125C / 250F as there is nothing on the card that hot. However, I see the GPU VRM amperage draw hit 72amps before I shutdown and the D601 light is a critical power trip on the 4870.

So I need you guys to run Furmark with latest Everest (build 1455) and see if your amperage draw is maxing your PSU rails. I have a TT1000 with quad rails. Rail 3 and 4 are dedicated to the PCI-e connectors. Each rail is 36a. My PC is going to a black screen when the HD4870 hit 72a draw, the maximum amperage I have available.


Links for latest Everest are here -
http://www.9down.com/EVEREST-Ultimat...55-Beta-29037/
http://middown.com/news_3945_Everest...1455-Beta.html

GPU VRM is the bottom reading in Everest -

[Kaane]

that is way too high of a draw, you must have a defective card.

[dnottis]

that is way too high of a draw, you must have a defective card.

Dunno, thats why I wanted to see someone elses results. Nothing else pulls this much, RTHDRIBL pulls 51a max. So is it the card or the application?

[Uberbob102000]

That has to be the amperage going to the core not drawn from the PSU because otherwise your card would be drawing 864 watts from your PSU.

[dnottis]

Gentlemen - here is the fix.

[Uberbob102000]

I just tested with mine and it maxed out at 80.01 A of draw.

[gabe]

Thats not exactly true. I have been working with people at ATI and have gotten alot of information - albeit I haven't "fixed" the issue yet. There is absolutely a need to have sinks on the VRM hardware and in fact large heat sinks. I used the MCR21s from the GTX from the VRMs and was told that they are too small to cool them. Even with active airflow it may not be enough, they suggest blocking the entire area. The Vitec and Pulse chips may not need additional cooling but will need airflow.

I am only referring to the Vitec inductor, not the VRM.. you are in fact confirming what I said: "The Vitec and Pulse chips may not need additional cooling but will need airflow"

It's interesting to read that the MC21's might not be sufficient to cool the VRM's.. something to further investigate.

[dnottis]

I am only referring to the Vitec inductor, not the VRM.. you are in fact confirming what I said: "The Vitec and Pulse chips may not need additional cooling but will need airflow"

It's interesting to read that the MC21's might not be sufficient to cool the VRM's.. something to further investigate.

Control chip of VRM part, VT1165MF, U601 on the PCB - needs a heatsink too as per ATI.



covering all VDDC/MVDDC/MVDDQ regulators chips & inductors in addition to GPU including following parts :

U603-5 & U701 , U901 , ML601 , L701 , L901

[Eastcoasthandle]

Unless I am unaware of something, why not use the OEM faceplate?

[gabe]

Control chip of VRM part, VT1165MF, U601 on the PCB - needs a heatsink too as per ATI.

if "ATI" says VT1165MF needs sinking, then why not cooling it with the stock heatsink? Further contradicting this information, the VT1165MF receives no air flow whatsoever from the stock fan because the step that reaches the VRM's completely blocks any airflow across the entire width of the heatsink.

Either your contact at ATI is not in the thermal/mechanical group, or he is and he disagrees with the reference design, or he is simply mistaken, or you might have misunderstood.

[evilrad]

Unless I am unaware of something, why not use the OEM faceplate?

I don't know about other brands, but once I removed the thick thermal puddy from the faceplate on my Asus 4870 it became useless. It doesn't even touch the ram chips. Even when the puddy was on it, the ram was barely touching the puddy. Is every company using the same design? I haven't seen anyone else complain about this problem (unless they don't realize it). So I'm wondering if this is only related to cards from Asus? Does anyone else have this same issue?

[dnottis]

if "ATI" says VT1165MF needs sinking, then why not cooling it with the stock heatsink? Further contradicting this information, the VT1165MF receives no air flow whatsoever from the stock fan because the step that reaches the VRM's completely blocks any airflow across the entire width of the heatsink.

Either your contact at ATI is not in the thermal/mechanical group, or he is and he disagrees with the reference design, or he is simply mistaken, or you might have misunderstood.

Engineer @ ATI - direct quote -
"U601 is the critical regulator here. It looks to me like he does have a sink on there, but it really should be blocked."

I have to check and see if the backplate does in fact contact that chip. There is no thermal pad on it however.


Look just above that chip in the Pic silkscreened on the PCB - its labelled U601.

[dnottis]

I don't know about other brands, but once I removed the thick thermal puddy from the faceplate on my Asus 4870 it became useless. It doesn't even touch the ram chips. Even when the puddy was on it, the ram was barely touching the puddy. Is every company using the same design? I haven't seen anyone else complain about this problem (unless they don't realize it). So I'm wondering if this is only related to cards from Asus? Does anyone else have this same issue?

They are all the same so far, none with third part cooling are out that I've seen.

[dnottis]

Unless I am unaware of something, why not use the OEM faceplate?

Can you run Furmark and run the latest build of Everest 1455 to see what temps and VRM amperage draw is?

Furmark - http://www.ozone3d.net/benchmarks/fur/

Everest - http://www.lavalys.com/beta/everestu...q0we6kylnx.zip

[gabe]

Engineer @ ATI - direct quote -
"U601 is the critical regulator here. It looks to me like he does have a sink on there, but it really should be blocked."

I have to check and see if the backplate does in fact contact that chip. There is no thermal pad on it however.


Look just above that chip in the Pic silkscreened on the PCB - its labelled U601.

I do understand that U601 is the VT1165, but it isn't sinked, the back plate doesn't touch it, and there is no airflow whatsoever in that area, none, nada, zip

[dnottis]

I do understand that U601 is the VT1165, but it isn't sinked, the back plate doesn't touch it, and there is no airflow whatsoever in that area, none, nada, zip

Yea, but Furmark crashed on me with the stock cooler too. So even though there isn't, that looks to be the cause of the crashes with the HD4870.

Its sad that ATI didn't address the thermal concerns with U601 now in the wild users are having stability issues from this.

[Eastcoasthandle]

Can you run Furmark and run the latest build of Everest 1455 to see what temps and VRM amperage draw is?

Furmark - http://www.ozone3d.net/benchmarks/fur/

Everest - http://www.lavalys.com/beta/everestu...q0we6kylnx.zip

79C full load using FurMark.

[dnottis]

Run the stability not benchmark - let it go longer than 1 minute lol

Even when I was crashing I'd get about 1 minute.