Question Temperature and performance issues with ASUS R7 250

[Architech88]

Greetings people,

I was using Windows 8.1 for a while and i had some performance issues in games but since my machine is a bit older, i got used to it.
My CPU is: AMD Athlon X4 860k, GPU is ASUS R7 250 2GB DDR5, SSD Kingston A300 series and Motherboard is MSI 7721.
With that hardware, i had ~50-60 FPS in a game called Warframe at minimum settings and everything turned off.
Last week, a friend on discord suggested me to swap to Windows 10. Which i did and i couldn't believe the improvement that OS did to my system and said game overall.
I had around 70-80 FPS at MEDIUM/HIGH settings with everything but shadows turned on!
To cut things short, few days later, i went to a store and purchased new, aftermarket cooler for my CPU, since i was using stock cooler. Said cooler is Antec A30, not the best quality heatsink but it keeps my CPU ~35-40C at full load, so i am happy with it. While i was replacing stock CPU cooler with it, i decided to re-apply thermal paste on my GPU as well, since i haven't changed it since i bought this GPU, 2 and a half years ago. After i set up everything, turned on my machine and that's where trouble began.
Performance dropped like they used to be on Windows 8.1... 40-50 FPS at minimum settings, and GPU temperature at around 70C at 60% load. I had around 53 before, on windows 8.1 at full load.
I thought i messed things up with thermal paste on my GPU, so i reapplied new thermal paste on GPU and things are the same.
So, does anyone have any idea what i did wrong? I'm losing my mind over here.

PS: Sorry for longer post, tried to explain the situation the best i could

 

[13thmonkey]

As a warning to everyone else, leave the thermal paste alone, it does not need changing on a whim.

@Architech88 it sounds to me like it has been poorly reassembled, or too much paste is on the GPU. The contact pressure between the heatsink and the chip is vital, too much paste and heat can't flow, the paste is meant to fill in microscopic gaps that's all. Or the cooler has not been tightened enough. Or the contact between the heat sink and the other hot components, memory and VRM's is no longer sufficient and they are overheating.

 

[Architech88]

As a warning to everyone else, leave the thermal paste alone, it does not need changing on a whim.

@Architech88 it sounds to me like it has been poorly reassembled, or too much paste is on the GPU. The contact pressure between the heatsink and the chip is vital, too much paste and heat can't flow, the paste is meant to fill in microscopic gaps that's all. Or the cooler has not been tightened enough. Or the contact between the heat sink and the other hot components, memory and VRM's is no longer sufficient and they are overheating.

I like to replace thermal paste at least once a year. Already burned one GPU before by not paying attention to paste, which somehow deteriorated due to heat. So i like to be sure everything is ok.
Going to replace it again. Will make sure to put only enough paste so it covers the chip and tighten the screws a bit tighter.
By the way, can the overheating be the cause of performance loss or something else got messed up with windows? Maybe some update?

 

[rolli59]

I have been building myself for over 20 years. I have never had to replace thermal paste even my oldest PCIe card (X1950Pro) still works. GPU maintenance for me is only regularly clean out the cooler.
I agree with 13thmonkey that the cooler install is bad.

 

[13thmonkey]

It could be thermally throttling, can't recall if 250's have an abnormally low throttle point, but it could be VRMs or Memory which may not be reported.

 

[Architech88]

Replace the paste again, tightened screws really tight and it's slightly better. Temperature is holding 65C at full load, which is ok i guess.
By the way, forgot to add, couple of days after swappin to Windows 10, my Radeon driver failed to load when i was turning on my PC. Was getting notification that Windows will use default video driver. Reinstalled it and was getting occasionally something about Radeon Wattman failing or something like that. Could something of that affect performance? Then only thing that comes to my mind right now, is to reinstall my windows.

 

[RememberThe5th]

I have 2x r7 250s, never had "that exact" issue ever, even when replacing paste, mine xfx's does not have vrm heatsinks.
Also check on gpuz does it run x8 or x16 mode.

 

[Architech88]

I have 2x r7 250s, never had "that exact" issue ever, even when replacing paste, mine xfx's does not have vrm heatsinks.
Also check on gpuz does it run x8 or x16 mode.

Not quite sure what you mean by x8 or x16 mode, so here's GPU-Z screenshot of this card.

 

[RememberThe5th]

Yes it is running x16, you see by Bus Interface where clearly says PCIe x16 3.0 (what it should be) @ x16 3.0 (what is currently running) (and question mark is to wake up the GPU from sleeping mode x8 for an example to the x16 by starting render).

What does in sensor say?
You can download furmark and run it windowed mode and see if sensor mode does show up PerfCAP reason (if its avalilable)

And by your mention UP

Reinstalled it and was getting occasionally something about Radeon Wattman failing or something like that. Could something of that affect performance? Then only thing that comes to my mind right now, is to reinstall my windows.

I would suggest you to try DDU CLICKABLE LINK!

If that fails, you can also install without amd software, just drivers.
By typing in start Device manager-> Display adapters -> Right click update-> Browse my...-> Then Browse -> find amd extraced installer (usually in windows patrition)-> tick include subfolders-> and press next.
It should install.

And to add I had a LOT of issues with amd drivers, especially the R7 250, currently using R9, I think the AMD drivers thinks this is an 7750, which can be true.

 

[Architech88]

Yes it is running x16, you see by Bus Interface where clearly says PCIe x16 3.0 (what it should be) @ x16 3.0 (what is currently running) (and question mark is to wake up the GPU from sleeping mode x8 for an example to the x16 by starting render).

What does in sensor say?
You can download furmark and run it windowed mode and see if sensor mode does show up PerfCAP reason (if its avalilable)

And by your mention UP

I would suggest you to try DDU CLICKABLE LINK!

If that fails, you can also install without amd software, just drivers.
By typing in start Device manager-> Display adapters -> Right click update-> Browse my...-> Then Browse -> find amd extraced installer (usually in windows patrition)-> tick include subfolders-> and press next.
It should install.

And to add I had a LOT of issues with amd drivers, especially the R7 250, currently using R9, I think the AMD drivers thinks this is an 7750, which can be true.

Ran the rendering test and buss interface changed numbers, during test it's v3.0 but when rendering is closed it says v1.1. but few moments after it goes back to v3.0.

Going to use the DDU to uninstall this driver. Any suggestions which driver should i use? AMD auto-detection or select one for R7?

 

[RememberThe5th]

Always select, I've never used the Auto Detect after I had troubles with my very old Gt 520.

I would watch the x16 mode rather than PCIE link mode. I think its normal to jump 3.0-1.1-3.0 after you close it.

 

[Architech88]

Always select, I've never used the Auto Detect after I had troubles with my very old Gt 520.

I would watch the x16 mode rather than PCIE link mode. I think its normal to jump 3.0-1.1-3.0 after you close it.

Tried both. Using DDU in safe mode then installing fresh new 19.1.1 driver for R7 - 200 - 250 series. Same performance issues.
Used DDU in safe mode again, went to device manager updated driver with path to C:\AMD\Radeon-Software-Adrenalin-2019-19.1.1-Win10-64Bit-Jan20 path. Same performance issues.

I guess i'll reinstall windows tomorrow and give it a try in hope i'll get the improvements i had when swapped to Windows 10.