5720 3D slowdowns during games

This is an old problem that has plagued this system since it came out. You are playing a 3D game and frame rates are great until suddenly they drop to less than 1/5 normal speed. The game is unplayable. You sit and wait and after a minute or two the game speed picks back up to normal and you can go back to playing.

Most people react and say things like try turning down the graphics detail to improve performance, etc. This is not the issue and you have to experience it to understand. It's as if the video card is downclocking itself after prolonged gaming periods and then spools back up to normal speeds after it cools down for a minute or so. The process then often will repeat itself every five to ten minutes.

I've run third party software to log CPU speed and GPU speed while this happens and as best I can tell there is no indication that anything is downclocking during gameplay. I even try keeping the GPU fan manually set to maximum in case it's a cooling issue. Temperatures are reasonable and fan speed doesn't seem to affect the problem one way or another.

Sager lists the current and latest graphics drivers for the 5720 as 79.31 available for download from their website. These are from last winter and NVidia's latest drivers for the mobile 7800GTX are 84.63 and they claim to be compatible with the Clevo M570A but they don't specifically list the Sager 5720.

I have tried both drivers going back and forth using driver cleaner in safe mode between installs and while the latest drivers offer greater overall performance there is no difference in the occurence of this problem.

I did have this problem way back last Winter the first time I tried to upgrade to NVidia reference drivers from 79.31 and solved the problem after reverting back to 79.31 but lately the problem has been coming back.

I was wondering if there has been anyone who knows exactly what's causing this to happen on the 5720 and how to prevent it.

Thanks!

I figured cooling was the issue and since I travel with my laptop an underdesk fan or even a bulky portable cooler are somewhat impracticle but it could just be time to take it apart and clean the excess dust out again.

What I was wondering is what is actually causing the slowdowns to occur? If cooling is the solution than the graphics card must be auto downclocking to save itself from potentially damaging heat. From the temperatures I've monitored it doesn't seem to be dangerously hot so the safety threshold must be set too low. The system should be able to run intensive 3D games without overheating in it's default configuration at normal room temperatures.

Maybe there is something in the drivers that can be changed to increase the threshold slightly to prevent this from happenning since I doupt the graphics card is getting so hot that it's actually in danger of damaging itself. On the other hand I suppose it's possible the slim and sleek design of the M570 is just a little bit too tight for the 7800GTX to breath properly. The 7900GTX should run a little cooler with it's smaller manufacturing process. Other mobile PCs sporting the 7800GTX are much larger.

I'm not sure if anyone still follows this problem but it seems after more than a year of frustration and assumed heat/throttling issues the real culprit has been found.....

http://news.com.com/8301-10784_3-9740960-7.html

I had tried everything from every CPU/GPU monitoring/clocking app I could find to tearing the thing apart and re-seating the CPI/GPU heatsinks with arctic silver. All with no luck. As it turns out the culprit is my Logitech MX518 gaming mouse!

Here is a possible solution that may or may not work for you....

http://forums.vugames.com/thread.jsp...art=0&tstart=0

Bump.

Since someone else seems to be having this problem still with a newer machine and an 8700M I figured I should bump this thread. I never did manage to solve this problem and it was hitting me again when I played portal. I've changed pretty much everything at the software level including switching to Vista 32bit and all new drivers (of course).

I'm wondering if anyone else is having these problems on the newer Sagers.

The guy in the other thread (next line down) said Sager replaced one of his RAM modules and the problem went away. Try testing your RAM because that's what I'm going to do. I'm not optimistic though because I suspect the RAM must need to really get warm for this to happen. I don't see how RAM errors can cause these slowdowns.

Like you said the problem appears to be a throttling issue but I've tested that possibility thoroughly by monitoring temperatures and all aspects of system performance such as CPU, FSB, RAM, and GPU speed while playing FEAR. The slowdowns with sudden and consistent frame rate drops to about 20% normal occurred but none of the parameters were out of normal. I've even tried disabling the CPUs thermal protection feature (not a wise thing btw) and replacing the thermal paste on all the heatsinks.

The problem occurs in some games but not others. I can run BF2 at 1920x1200 all day with no problem but when playing Portal I had to turn the settings way way down just to be able to get through the slowdowns without tearing my hair out.

Anyway, I'm going to run some tests on memory....I'll post later.

OK, ran some tests just now and had some interesting results.

I wanted to stress the system in a way that was similar to 3D games while in a controlled environment. I used Orthos (just Prime95 in a package) to maximize the stress on the CPU and RAM.

I used ATi Tool to maximize the stress on the graphics card (7800GTX) and control clock speeds. I used NVidia NTune for redundancy monitoring of the GPU.

I used Rightmark CPU clock utility and Windows task manager to monitor all aspects of the CPU and bus speed.

Using the 3D View in ATi Tool I was able to get a consistent load on the GPU while monitoring the frame rates of the 3D render in a window. Under default GPU settings (400/1100MHz) it averaged around 300fps. I've attached a picture (1920x1200 resolution) to show the setup and what happened to the frame rates under load.

screenshot

What happenned is a consistent drop of framerates as low as 20fps from 300fps for about one to two seconds occuring every eight seconds. The slowdowns didn't occur immediately but after a couple minutes the fan speeds began to spool up automatically then the problems began to show shortly after that. Increasing the CPU fan speed to maximum (fn-F2) increased the cycle time to about every 11 seconds. Lifting the back of the computer approximately 1.5 inches above the desk to allow better airflow underneath increased the cycle time to approximately 40 seconds. I ran the computer in this configuration for at least ten cycles each time to ensure the results were consistent. You will notice from the screenshot that there is no indication of any drop in CPU or GPU output but that doesn't mean the graphics card isn't throttling, it just doesn't show up on the monitors.

Another test I ran was to underclock the graphics card down to 200/600MHz or approximately half speed. Running the same tests gave me a consistent frame rate of about 100fps and the slowdowns didn't occur at all. The internal fans also never spooled up to maximum.

So clearly I was wrong about this in some respects. It is definitely graphics card temperature related. I'm not sure exactly what's happening to the graphics card to cause this to happen but I may forward this to NVidia and see if they respond.

Something else...I had this same problem on my new 640MB 8800GTS desktop graphics card last year when I bought it but only while playing the game FEAR. I've since bought a new case and built a new system with this same card and the problem has never come back. My new case has much better airflow and stays much cooler but I never tried playing that specific game again and the problem never occurred with any other games.

Since this problem seems to occur on Clevo 570 style notebooks and various NVidia mobile graphics cards it seems to me this is a combination of inadequate cooling on this particular notebook and a fundamental flaw (or a feature designed to save the chip) in the design of Nvidia graphics cards. The reason why it affects some people more than others is simply that all chips don't come off the line created equal, some heat up more than others. But this problem affects enough people that it needs to be addressed.

In the meantime, I would urge caution buying a high end NVidia graphics card in a smaller or slimmer notebook like the 570 without thoroughly testing it before the warranty or return for refund date expires. If you're forced to deal with this issue elevate the notebook, blow a fan across the bottom, or just downclock the GPU a little until the problem goes away.