nVidia GTX 580M clock speed stuck after clean system install

MitelusMorion

Estimable
Jul 30, 2015
4
0
4,510
Hello,

I've been doing a lot of searches around but can't seem to find an answer to my problem.
I have an Alienware M17x R3 with nVidia GTX 580M gpu. It worked really well until I had an HDD failure and had to replace it with a new one and perform a clean install of Windows 7.

Before: almost any game would run on max settings and FPS would be on the 40+.
After: almost any game running on minimal settings give FPS from 5-20 (on a good day I get 20).

Things I've done:

1. Clean installed the drivers (at least 2 times). Tried both original (old), intermediate and newest drivers.
2. Installed CPU-Z and other system monitoring tools. Temperatures never go beyond 65, either CPU or GPU.
3. Performed several GPU and CPU stress-tests: nVidia's, Dell's, FurMark. On Dell's and nVidia I get "PASS" on everything. On FurMark I get 5-7 FPS on 1600x900. Load goes all the way up to max, stays there for a while, then comes down for a bit, goes back up, and repeat. This oscillation in throttle does not alter FPS performance.
4. I have set power settings to high performance and selected the GPU to be on at all times (no switching between Intel and nVidia).
5. I've tried playing the usual games (BF4, DayZ, SotA, LOTRO, etc...) while having CPU-Z on, and here's the latest things I've found:

i. Voltage remains at 800 mV at all times.
ii. Clock speed all remain at idle values (74/162/147 MHz for Core/Memory/Shader, respectively). As if the GPU is not being used.
iii. GPU load jumps up to 99-100% during game and return to 0-1% after closing the game.

6. After the testing, I tried installing some nVidia software and manually adjusting the throttle to 620/1500/1420 MHz. Still behaves the same when I try to play a game.
7. Bios offers no option to turn ON/OFF GPU, but I believe this is not a problem since it is being recognized and has drivers installed.

Does anyone have any idea of what is going on? I have the feeling that I have forgotten to do something after reinstalling the system, but I've exhausted all options I had in hand.

Update: BIOS version is 70.24.2B.00.06.
 

MitelusMorion

Estimable
Jul 30, 2015
4
0
4,510


That was my one of my first assumptions. To test it I performed two graphics tests: one using only the iGPU (nVidia card disabled; didn't show on CPU-Z) and using the GPU (nVidia card enabled; showing on CPU-Z). Both tests returned the same FPS average on FurMark, but when the nVidia GPU was showing on CPU-Z, the load on the card went from 0% to 99-100% while performing the test. Oddly enough, the clock speed for the core/memory/shader remained on idle values.

Thoughts?
 

MitelusMorion

Estimable
Jul 30, 2015
4
0
4,510
UPDATE: Problem Solved.

Here's what has been causing the problem: a faulty charger.

The original charger that came with the AW m17x R3 started "failing" over 1 year ago. I've never noticed then that it caused a decrease in GPU performance, because most of the times when it wasn't recognized, I wasn't actually playing any games.

With time, the problem accentuated and, lately, it has gotten to a point that it wasn't charging the battery anymore. After reading a few posts on Dell's website, I drew the conclusion that my battery was dead and I would continue to run it on AC power. WRONG.

I've just performed these simple steps:

1. Powered Off.
2. Disconnected the charger from the wall for 30 seconds.
3. Removed the battery from the notebook.
4. Pressed Power button for 30 seconds to dump charge.
5. Reconnected the power cord.
6. Turned on the notebook and performed a Kombustion burn-in test

Results:

Clocks worked all the way to max performance.
Voltage went up to 870mV stock max.
FPS jumped from miserable 7 FPS to average 70 FPS.

---

I hope this helps anyone. Something very stupid was hindering performance.