Is my GPU failing?
Is my GPU failing?
Recently, my GPU hits temperatures near 100°C and then drops to 60°C, yet maintains a FPS between 20 and 35. I’m not sure what’s going on anymore. I’m hoping someone can explain what might be happening.
I ran some benchmarks on Heaven Bench. Here are the results:
First attempt: temperature around 95-100°C
Unigine Heaven Benchmark 4.0
FPS: 49.1
Score: 1238
Min FPS: 8.6
Max FPS: 89.3
System: Windows NT 6.2 (build 9200) 64bit
CPU: Intel® Core™ i7-4770K @ 3.50GHz (3499MHz) x4
GPU: AMD Radeon R9 200 Series 16.150.2211.0 (3072MB) x1
Settings: Render → Direct3D11, Mode 1920x1080 8xAA fullscreen, Preset Custom, Quality Ultra, Tessellation disabled
Second attempt: temperature stays around 55-65°C
Unigine Heaven Benchmark 4.0
FPS: 29.2
Score: 735
Min FPS: 21.8
Max FPS: 46.3
System: Same as above
My GPU is an MSI R9 280x gaming model. Thanks for the assistance!
I can only express concern about the overheating. Most cases of overheating stem from a dirty heatsink or a malfunctioning fan. Does the fan speed increase when temperature rises? Does it achieve 100% speed when it gets that hot? If not, then it should.
First, ensure the heatsink is clean. Next, confirm the fan operates at an adequate rate. During load testing, check if the fan increases its speed. It must definitely reach 100% before reaching 100°C. If this doesn’t happen and you haven’t already adjusted to a custom fan profile, switch to one that reaches 100% between 70 and 80°C.
If all steps are followed and overheating persists, you may...
The lower score occurs because the GPU hasn't throttled yet. It's likely that once it reaches 100C, it starts throttling to manage temperature. Have you attempted removing the card and clearing dust from the heatsink? I also checked if cleaning before benchmarks resolved the issue.
Throttling occurs when a threshold is exceeded, usually temperature. It happens when that limit is surpassed, to protect the component—this can apply to both GPU and CPU—the clock is reduced to manage heat. For example, if your GPU reaches 100°C while running at 1.0GHz, the card will lower the clock speed to around 500MHz to prevent overheating and potential damage.
The card usually switches between speeds as it warms, cools, and heats up. This pattern can be tracked using tools like Afterburner. When benchmarking with Afterburner active, pause the test when the card warms up and check the Core Clock graph. A drop indicates throttling.
But in your case, it seems unusual because once the limit is reached, the speed stays at the lower setting rather than changing as expected.
Techgeek: Throttling occurs when a threshold is reached, usually temperature. It happens when that limit is exceeded, and the component—whether GPU or CPU—reduces its clock speed to manage heat. For example, if your GPU reaches 100°C while running at 1.0GHz, it will lower the clock to around 500MHz to prevent overheating and potential damage.
The card usually switches between speeds as it heats up, cools down, and warms again. This pattern can be tracked using tools like Afterburner. When you run benchmarks with Afterburner active and the card gets hot, press ALT-TAB and check the Core Clock graph. A drop indicates throttling.
But in your case it seems unusual because once the limit is hit, the speed stays at a lower setting rather than switching back and forth as expected.
I understand that sometimes if the driver fails or temperatures become too high, the system might not return to full speed until you restart.
I can only express concerns about overheating. Most cases of overheating stem from a dirty heatsink or a malfunctioning fan. Does the fan increase its speed when temperature rises? Does it achieve full rotation at higher temperatures? If not, it should. Begin by confirming the heatsink is clean. Next, ensure the fan operates at an adequate rate. During load testing, observe if the fan accelerates properly. It must reach 100% before reaching 100°C. If this doesn’t happen and you haven’t already implemented a custom fan profile, consider switching to one that activates between 70 and 80°C. If all else fails and overheating persists, removing the heatsink and fan might be necessary, along with replacing the TIM.