What's the likelihood of your NB Chipset affecting your GPU Temp?

[Emphacy]

Okay here's the story.
My first post on this forum was due to my Radeon HD 4870 reaching a rather insane temperature of 148 degrees Celsius.

Kinda strange since thermal trip is at around 125 degrees, tested the temps on GPU-Z and EVEREST Ultimate, definitely the Memory overheating.
The memory sits around 15 degrees higher than the core. It's not overclocked, I can't overclock it with the temps I get off playing 10mins of COD WAW.

Question is then, what's the likelihood of my northbridge which under load usually reaches about 70 degrees, affecting my GPU memory Temp?
The card uses an Akasa Vortexx Neo cooler, suppose to be one of the best. It's either faulty, or the NB is causing problems for it...

I'm seriously unsure whether to try another cooler on it, although the one i'm using gets the best reviews.
OR... Go and buy another HD 4870 off the bay for a nice £50 then maybe crossfire it when I get a new mobo...

I really am not gonna spend over £100 on a 5850 or something when my 4870 still does the job.
Opinions much appreciated guys, I really am stuck on options here.

 

[MichaelM]

What thermal paste did you use when putting it on, did you do it correctly and how long ago was it?

edit - I assume separate heatsinks where put on the memory chips?

 

[Emphacy]

I used AC Silver MX2 Thermal Paste, been on there for about 4 months.
I used the thermal pads that came with the cooler on the memory.

I checked contact of them all like recommended in the last thread, no bad contact what so ever.

 

[MichaelM]

NB does seem to be pretty high, nearly double what I would expect someone to be getting although I can't say I have taken notice of ATX motherboard temperatures. Not that I imagine they would be prone to rise over a normal idle of maybe 25-30C.

I assume the temperatures are in Celsius and not Fahrenheit.
I assume you are seeing the temperatures with high idles and scaling realistically so its not dodgy sensors (fluctuating wildly or at a constant max).
I assume someone else on the other thread asked you to remove the side panel to see if that changed the readings by much.
I assume if the chipset temperature readings are displayed in the bios screen they match the 3d party software.
I assume you are using the latest versions of the 3rd party software to monitor temperatures.

Long story short, assume something simple is being overlooked or there is a physical issue with the hardware. Easier to troubleshoot this stuff in person. As its on a forum I feel i've reached the maximum of my useful input, sorry.

 

[^Andy^]

Ok 148 degrees would burn your hand so my question is...

...does it?

If it doesn't the sensor may just be giving a duff reading. You didn't say if the card is still working correctly at that ridiculous temperature?

 

[Emphacy]

Yes it burns my hand... Even at idle, which sits around 70 degrees celcius.
Seems rather high to me... The northbridge as said sits around 55 degrees idle, reaches 70 under load, overclocked FSB is most likely the reason.

If I take the side panel of theres a change of about 2/3 degrees celcius, although I have 3 120mm and an 80mm intake fan and a 120mm and 80mm exhaust.

Got all the latest software, all read the same thing. Temperatures stay constant, only rise to insane temps during load, (i.e. playing games)
Thanks for your help guys but it really does look like a faulty card, may buy another one from the bay...

 

[Greebo]

You sure it's Celsius? Only reason is that the 4870 should thermal throttle well before it reaches 120C and at 120C the card will blow:

http://www.guru3d.com/article/radeon-hd-4870-review--asus/5

Googling reveals many people claiming 140C+ but on closer inspection the temp reading was in Fereinheit.

quotes:
"During testing we felt the need to point two case fans at the card, blowing across the rear and the large cooler. The metal plates on the rear of the HD 4870 X2 were painfully hot to touch even after relatively short periods of gaming, so don't think about cramming the HD 4870 X2 into a small, poorly cooled case." http://www.bit-tech.net/custompc/reviews…

"card is fast don't get me wrong, but it runs EXTREMELY EXTREMELY HOT be sure to enable manual fan control and set the fans to atleast 60% or else you'll get random BSOD's (blue screen of death)"

"The black plastic thermal duct covers a pair of skived copper heatsinks and several subsidiary aluminum heatsinks for the memory and VRMs. A copper heat pipe helps balance the thermal load. All components are actively cooled by the fan, with the exception of the GDDR5 memory chips which get their own passive heatsink. During testing, we noticed that the back of this video card got quite hot, we'd estimate in the 60-70 degree Celsius range (140-158 Fahrenheit)."

 

[JTrickle]

As stated above, your gpu is not hitting 148.

Have you got a link to the other thread you mention where it discusses what problems you are having with the card?

 

[Emphacy]

Link to Thread

Yes mate, it's exactly what I thought, would have been black and crispy by that temperature, which leads me to believe it's even more faulty.

Also system immediately shutdown. I left it for 5mins booted up again, EVEREST displayed 148 degrees celcius. (showed this in sidebar before shutdown, sidebar stays like it was before the PC shut down on start up, until everest is restarted)
Loaded up GPU-Z and it was just just throttling down from 98 degrees.

 

[drunkenmaster]

When asking for advice like this, essentially temps and what could cause it, more information is needed.

System spec, what mobo, which hd 4870, what case, take a screen shot of highest temps in gpu-z and post it, take a pic of your case and post it.

Do you have a 3rd party hd 4870 and a cooler that only fits properly to a reference design pcb? What mobo do you have, what kind of chipset heatsink, does the mobo have an alternate pci-e slot, have you tried the gpu in that slot further from the northbridge, etc, etc. Are you using a "proper" case or a tiny box with one 80mm fan exhaust, if its a decent case, is the front/rear fans working properly, is the gpu fan running properly in use, if it is running, does the fan have speed control either via its own controller or because its connected to the normal header on the graphics card that lets you change it like normal in CCC, if so, have you whacked the speed up.

Most of those things would be answered with system spec and a pic of gpu-z/case.

 

[Emphacy]

When asking for advice like this, essentially temps and what could cause it, more information is needed.

System spec, what mobo, which hd 4870, what case, take a screen shot of highest temps in gpu-z and post it, take a pic of your case and post it.

Do you have a 3rd party hd 4870 and a cooler that only fits properly to a reference design pcb? What mobo do you have, what kind of chipset heatsink, does the mobo have an alternate pci-e slot, have you tried the gpu in that slot further from the northbridge, etc, etc. Are you using a "proper" case or a tiny box with one 80mm fan exhaust, if its a decent case, is the front/rear fans working properly, is the gpu fan running properly in use, if it is running, does the fan have speed control either via its own controller or because its connected to the normal header on the graphics card that lets you change it like normal in CCC, if so, have you whacked the speed up.

Most of those things would be answered with system spec and a pic of gpu-z/case.

All information is in this thread and the link given to my first thread regarding this problem of mine. I did have a signature displaying setup but changed it. MSI P43 Neo-F. 1 PCI-E slot, NB chipset sits around 3/4 inch above GPU.

 

[Emphacy]

Think i've sussed it guys...
Using aftermarket coolers on HD 4870's cause huge heat problems as i've read from many reviews and other forums.
Generally aftermarket coolers do well at cooling the core, while they are more poor than the stock cooler at cooling the cards VRM's.

I ran Furmark on 1280 by 1024 resolution with XTreme Burning mode on, and well the GPU never went above 80 degrees celcius, and the memory reached about 104/5 degrees celcius max! (Also used GPU-Z).

VDDC Temps where around 130 degrees celcius! The vortexx neo doesn't even cool the VRM's at all, doesn't even touch them.
I think it's time to revert back to a stock cooler. Although, still doesn't quite explain the 148 degrees celcius MEMIO Temp.

 

[^Andy^]

Think i've sussed it guys...
Using aftermarket coolers on HD 4870's cause huge heat problems as i've read from many reviews and other forums.
Generally aftermarket coolers do well at cooling the core, while they are more poor than the stock cooler at cooling the cards VRM's.

I ran Furmark on 1280 by 1024 resolution with XTreme Burning mode on, and well the GPU never went above 80 degrees celcius, and the memory reached about 104/5 degrees celcius max! (Also used GPU-Z).

VDDC Temps where around 130 degrees celcius! The vortexx neo doesn't even cool the VRM's at all, doesn't even touch them.
I think it's time to revert back to a stock cooler. Although, still doesn't quite explain the 148 degrees celcius MEMIO Temp.

I used the zalman vrm coolers on all of my old 4870's along with the zalman vf1000 - worked perfectly.

 

[Jakus]

Question is then, what's the likelihood of my northbridge which under load usually reaches about 70 degrees, affecting my GPU memory Temp?
The card uses an Akasa Vortexx Neo cooler, suppose to be one of the best. It's either faulty, or the NB is causing problems for it...

More likely a GFX card cooking the motherboard and surrounding chipset... lots of the coolers fitted to GFX cards spill the hot exhaust onto the motherboard.
Quality reference coolers like Nvidia high end cards expel as much heat as possible away out the back of the case for good reason at the expense of a few extra degrees to the GPU.
ATI who currently do not have a clue how to make a quiet efficient cooler block half that exhaust with an unnecessary Dvi.
From memory Your vortex cooler was based on the old artic cooling 4 cooler and was very good.
The reference 4890 had a stonking cooler with extra heatpipes if You could find a SH one from the members market it will cool your memory and vrms perfectly