Acer Gemstone Blue downclocking GPU

Anyone else experienced this? How did you solve it?

 

Hi Kittie - this is pretty much the same with CPUs.

they clock down to save power.

you should find when you have the notebook plugged in and running a game the GPU should run at max.

 

I mean during a game, and when it's not overheating either.

 

Tried disabling PowerMizer..??

 

Yes. I've tried all the obvious stuff.

 

i have the same problem and unfortunately havent found a solution..

 

can i ask how you folks are finding where the GPU is down clocking?

are you using GPUZ?

 

Probably. Test it very simple: start GPU-Z and set it to monitor in background. Start the 3D screen saver of Windows Vista. It will go to 475/400 immediately. So, there's no problem. Unless it's not working.

 

TehSuigi and I have been discussing this in another thread...we have the same model laptop.

So here's what's going on, the CPU is overheating and gets down clocked; because of this less data is passed to the GPU, causing a lower framerate, which in turn causes the GPU to under clock because it doesn't need to work as hard.

I really haven't figured a way around it, if you figure something out PLEASE tell me. Remember, it's the CPU under clocking, not the GPU.

 

Bladezor, that's only the case with TF2 b/c it's so CPU-intensive.
Other games that stress the CPU may also cause it, but it's not the problem in every 3D gaming scenario.
Audiosurf works just fine, for example.

Come to think about it, perhaps it's b/c our T5550 is below the recommended system configuration?

 

I have a T7500 so I'm pretty sure it makes or passes the system requirements.

 

Ok, fixed the issue. Here's how:

I downloaded RightMark CPU Clock Utility.

Under Battery Info click Advanced CPU Settings. Click the Processor tab, and under the Automatic thermal protection section uncheck everything, that should keep it from underclocking your CPU, be warned though your CPU will get hot. I believe mine peaked at 45-50C which isn't TOO bad, especially for a mobile CPU.

 

RMClock does NOT read the temperature accurate - I found RealTemp was better.
And removing the thermal protections on your CPU sets off warning bells on my end. If that thing hits 100 degrees, you're outside of specified parameters.
Don't blame me if your processor melts into a glop of silicon.

 

ive got a t8300 penryn cpu and i experience the same problems when i play world in conflict... i used to think it was solely a gpu problem but the cpu downclocking makes more sense because i dont get the same problem when i play cod4 which i believe would be less cpu intensive than world in conflict...

just a quick question, what are safe temperatures for the cpu? can't we reset the automatic thermal protections so that it downclocks at a higher temperature?

 

The CPU is rated "safe" up to 100 degrees Celsius, but the going rule of thumb is to avoid prolonged periods of 90+.
Removing the Thermal Monitors 1 and 2 mean that your CPU won't downclock when it senses danger, nor will it shut down upon reaching 95 degrees.

I'm beginning to sense that the CPU cooling on the 6920G isn't adequate, or that Acer's thermal tables are so horribly cuffed up that it can't tell the fan when more cooling is needed.
Either way, I think it's just saved me from spending money on a CPU...but it will be costing me money to buy a notebook cooler.

 

I doesn't get higher than 60C under full load so I think I'll be fine.

 

Again, is that RMClock? RMClock reads the temps about 10-15 degrees lower than what they should be on my system - 25 celsius at idle just DOES NOT HAPPEN.

 

I get the same readings on both.

 

Both HWmonitor and RMclock report 15 to 20 C at idle and 40-45 C at load here. I've undervolted slightly but this still seems rather abnormal to me. It's a T7500.

 

I've got mine working fairly good, but I have to keep turning off throttling, IDA, and DFFS in RMClock because I believe the BIOS keeps setting it back.

 

Well, as long as you're happy Bladezor.
I'll still keep looking for a fix that doesn't involve throwing caution to the wind.

EDIT: After a little bit of monitoring, I've found that RMClock is exactly 15 degrees off from RealTemp. Now, instead of 25-55 idle-load, I'm seeing 40-70, which is a little more believable.
Gonna monitor my CPU during TF2 and see what its actual throttle switch temperature is.

 

So what have you got for us?

 

Well, it seemed that around 70 was the throttle trip on my CPU.
Which is kinda low, but may be related to ensuring my system stays in good shape longer.
Whatever - I'll try undervolting again at some point soon.