Strange 1330 CPU throttlin going on...

Hi guys:

I've got a 1330 with a T7250 processor, 4gigs of ram and a 8400GS.

I run my notebook all day long undervolted (tested stability a lot using orthos) so temperatures are less.

I also use RMClock to max out the cpu frequency when connected to AC (so that it stays locked at 2 ghz whenever i am in AC mode)

Also, my 8400 gs is Overclocked to 600/825 from stock 400/600.

The thing is the following: when running COD4, every once in a while my frames drop noticeably. I then went to investigate this and using RMClock's monitoring tool i noticed that the frame drop was consistent to a throttling of the CPU.

However, whenever i test out my pc with ORTHOS or rthdribl i get higher temperatures than when running COD4 and the CPU doesn't ever thottle itself...

As you can see, this is a pretty strange thing because temperature is NOT being the "throttling factor" over here, as it is almost all times. I also have the gpu copper mod applied and haven't ever passed out 80 degrees on the gpu (which are pretty normal temps for an xps1330, specially overclocked)

What do you think could be causing this?? It is really bugging me a a lot. Could it be maybe unstable voltages or something?

Greetings

 

Do you have enough memory?

 

4 giga ram... oh, i do have only 14.1 GB free on my C: (windows installation) partition but i think that is plenty isn't it?

 

I think temperature is your throtteling factor and this is why I think so:

There is just one heatpipe that connects CPU, GPU and PM965:



If you test with ORTHOS only your CPU gets hot - nothing more. If you test with rthdribl only the GPU gets hot - nothing more. I didn't see you testing both. If you play COD4, you GPU will be somewhere around 80 - 100°C and will heat up the heat pipe to this value. The temp that the heat pipe can pull away from the CPU is proportional to the delta in temps between the heat pipe and the CPU core temp. Hence if the heat pipe is hotter, the CPU get's hotter.

Get HWMonitor:
http://www.cpuid.com/hwmonitor.php

It keeps track of the maximum temps reached for GPU as well as CPU. If throtteling kicks in, check back the max GPU + CPU temps. What are their values?

 

As above, use hwmonitor and keep an eye on the ACPI thermal zone (TZ) temps.

The ACPI is what will trigger the downclock on a single fan system.

If you dont know what ACPI is:

 

I really don't think temperatures are the issue... let met explain further on this:

I have run both ORTHOS AND rthdribl at the same time, and i forcedly did this in my bed where the vents get blocked. I reached more than 86 degrees on GPU and 75 in CPU (this is WAY higher than when playing COD4) .... there was NO throttling going on, constantly monitored it with RMClock....

Any further ideas?

 

Your implications are based on guessing, but not on measurements. I'm not saying you're wrong, but it is questionable how good your in-bed-vents-blocked-rthdribl-orthos test really represents COD4.

Since HWMonitor measures GPU temps as well, which RMClock doesn't and also stores the maximum temps, I suggest your repeat the test with HWMonitor. Especially since it looks you can reproduce the problem. Something is triggering the throttle and I can't think of something else than temperature.

BTW: Are you experienced enough to distinguish between TM1 and TM2 throttle? You may put up a RMClock screenshot, if you can't. I think it helps anyway.

I haven't tested this: But it may be that the latest A13 bios, throttles much earlier than older BIOS versions. In contrast to what flipfire said I can't confirm that Dell's throttle is a function merely of the PM965 ACPI temp. From my own measurements which I did with older BIOS version I was pretty convinced that the throttle trigger is a funtion of many variables such as: CPU temp, GPU temp, ACPI temp and maybe others (memory temp, ...).

 

dude... your clocks are wayyyy too high for a laptop that is prone to heat issues and bad cards...

your gpu is the problem, down clock it to stock

 

i dont think they are that high... i mean, i know they're pretty deviated from stock clocks but temperatures are very very good for such an OC

 

i still think that is the problem... and even if they are cooler for that oc, i still say its a bad idea

 

Btw you can only see the GPU Core temps NOT gpu memory temps. This is how people fry the gpu and ended up with crazy artifacts on screen.

Your memory clock is way to high. Put it down to 750mhz

You shouldnt be pushing the 8400gs that much.

 

He's correct, the GPU will downclock when something gets too hot to protect itself. I would be careful.