Question about VRM cooling and mem temp reporting, and now the card is whining.

[viivo]

I finally got around to putting an aftermarket cooler on my 5870. While the core, shader and memory temps have decreased dramatically, VRM slowly increases the entire time I'm gaming, never hitting a plateau. I don't usually game for more than 30 minutes at a time, and while the highest I saw it go was 62c, I have a feeling longer play times would take it into dangerous temperatures. The way the HSF is designed, there is a fan blowing directly onto the VRM. Would it be worth it to MacGyver up a heatsink for them, or are those temps normal for the VRM?

Also, is the memory temperature reported the highest of the modules or an average of them all? The HSF covers one half of two memory modules. I put heatsinks on the other halves but I'm not sure how effective it is. I was just curious how the temps are reported so I can make sure I don't have hotspots or modules much higher in temperature than the others.

 

[Ben90]

62*C seems like a very good temp for a VRM. Most are rated for like 120*C. If you do think eventually they are getting a little too warm, a ghetto heat sink alternative is to glue screws to them

 

[Rifter]

67c is not even on the scale of being worried. my old 4890 would hot 100c, they are usually rated for 105-120C.

 

[badb0y]

How are you guys monitoring the VRM temperatures? I can't seem to figure it out on my 6950.

 

[Rifter]

I was using the XFX program that came with my card, cant remember the name of it but it told me GPU core temp, VRM temp as well as memory temp. I would assume any of the GPU temp sensing programs(i think even gpuz does it now) will be able to read the sensors and you should be able to tell pretty quick from the temps which reading is from which sensor.

It is also possible your particular card does not have the sensor.

 

[viivo]

I may not have been clear. I know how to monitor my temps, I was just curious how the memory temperature is calculated: by the highest module, an average of them all, or through some sort of main I/O? Or is it the temp of the actual memory controller? I know little about such things but I can't imagine every memory module containing their own probe.

And another question since I don't want to make a new thread: I noticed a whining sound coming from my card while testing stability today. Is this because of the overclock, or normal and I didn't hear it before because of the stock brick cooler? It's not the fan. More a buzzing "Weeezooooweeezooo."

 

[Ben90]

I may not have been clear. I know how to monitor my temps, I was just curious how the memory temperature is calculated: by the highest module, an average of them all, or through some sort of main I/O? Or is it the temp of the actual memory controller? I know little about such things but I can't imagine every memory module containing their own probe.

And another question since I don't want to make a new thread: I noticed a whining sound coming from my card while testing stability today. Is this because of the overclock, or normal and I didn't hear it before because of the stock brick cooler? It's not the fan. More a buzzing "Weeezooooweeezooo."

I don't know how ATi/AMD does things but I know Intel uses the highest reporting sensor when calculating core temps.

With regards to the whining, that generally stems from when the VRMs are loaded. Its totally annoying. It doesn't mean your card is defective, it just happens.