univar

Well it seems like something is changing as the components heat up. I go from about 30 FPS to 1-2 FPS just by overclocking by another 50 MHz. Can anything on the motherboard switch something off or reduce something? Any ideas for other tests i can do to narow down the problem? What about checking the south bridge frequency, is there any way of doing that?

univar

Well i find it strange that a difference of 50 MHz would make a difference of 30 FPS to 2 FPS. It just seems that something is purposly limiting the bandwidth. Is there any other tests you can think of or anything else to look into? Also, it is strange that it happens only after everything has gotten a chance to warm up.

wrench97

What are you seeing for Voltages(12v, 5v, 3.3v) and temps(CPU, Chipset and Video card)
Check them in the Bios first on the PC health page(except the Video).
The in windows both before the issue and then again after it starts using a program like SensorView Pro> http://www.stvsoft.com/index.php?opt...d=17&Itemid=33

greenbrucelee

when you overclock your cpu ram frequency can up or down too much can cause problems and making it go below what the minum is for your board can cause problems.

When you overclock you need to stress test your cpu using an app like prime95 and you need to run it for 7 hours, you also need to have a temp moniotring program running so you can see how hot the cpu is getting.

You also need to test the ram as you are not just changing the speed of the cpu but the front side bus which affects the ram so you need to use an app like memtest 86 and run this for a few passes.

If you get no errors in any then your fine, but what your describing something like prime95 or memtest would probably flag it up.

A 50MHz higher setting could make a huge difference, when I overclock I always raise the FSB in 10s or 20s, raising it straight by 50MHz is not a good idea in my opinion. Remember not all cpus can be overclocked alot even 2 of the same make and model can be different to each other.

grimx133

Check cpu-z at the bottom of the mainboard tab, see what your pci express is running at when you get the low fps.
http://www.cpuid.com/cpuz.php

univar

Thank you very much for suggesting another software tool for measuring. I have been wondering about the GPU temp for some time but thought maybe PC Wiz was just wrong. I don't know how this can be accurate though withough melting the card into a piece of solid silicon!! Also, thanks for the feed back about the overclocking. Unfortunetly, overclocking is the only way i can repeat the problem and running pcmark 2005 seems to work all the way up to 2.8 GHz so i think the system is stable all except for the graphics card. In doing a little more research i found a "Slow Mode" issue long ago when the card was too cold. Is it possible the card is going into slow mode now for some other reason? How can you tell what mode it is in? I will check the PCI speed when i get a chance, is the number you are talking about the transfer rate of 8x?

Attached Images

SensorSnapShot.PNG (15.0 KB, 2 views)

univar

Alright, now i have some very interesting results. Tonight i have been running and playing a game for about 4 hours so the computer is good and hot. When the image in my last post was taken, it was just before i started playing with clock frequencies tonight. As you could see the CPU temp was WAY off the chart, which should trigger the "slow mode" so as to not over temp the card. Unfortunately i was playing all night with no problems. Then i OCed to 2.65, 2.7, 2.75, 2.8 and still nothing even with PCI OCed to 115 too, everything was running fine with a rate of 8x on the PCI. finally after checking the game i checked sensors view again and noticed the CPU was 45, much more reasonable. Then as i was watching it jumped up to 233. So then i tried the game again and TaDa!! It was going at about 1 FPS. So any ideas why the GPU would be reading incorrectly? Cause i am sure this is why it is triggering the over temp mode right?

Attached Images

SensorSnapShot.PNG (22.6 KB, 3 views)

wrench97

The reading is obviously incorrect, since at that temp it would have melted, but it may be high enough to slow the card down, do you have another card to test with?

greenbrucelee

yep if it really was 232 degrees you would have know about because your pc would have been on fire.

I was going to suggest downloading real temp but this only works for core 2 duo, quads and i7s

It also monitors GPU temps

if you have an nvidia card try this http://downloads.guru3d.com/download.php?det=1194 you can also download software from Nvidia itself

The ati catalyst should have it there anyway or can be downloaded from ATi

univar

I was thinking that should be incorrect too, do you know of any way i can determine if the card is in a slow mode? The new 190.xx gforce drivers just don't have all the readings the old ones used to have so i cant really get much from them. I do know that the "slow down" temp is something like 115C from other internet sites but i would like to know if it is in fact operating in slow mode if i could tell. Also, anyone know if this might still have some kind of warrenty? Got it from NewEgg it is an EVGA.

wrench97

A lot of EVGA cards are lifetime if registered when installed, get the numbers off the label and contact them> http://www.evga.com/support/warranty/

univar

Darn, i am sure i didn't register when i got it. I only recently started realizing that was a good idea. I think this temperature thing is promising but i would like to have conclusive proof that the card is being underclocked. Any thoughts?

univar

After looking at the warrenty site, i don't see that the -GS is under any warenty anyway?

wrench97

The GS is not the warranty code you have to get the long number(Part Number) off the card. I changed out and RMA'd a 7600GS last week, received a 8600GTS because they no longer have 7600 cards.
The only way to find out for sure is to swap in a different card.

univar

OK thank you, i will try and contact them today but i am almost sure i never registered. Bad luck for me i guess! Now i know better

univar

Well it turns out i did not register my product so no warrenty for me. I have since taken the graphics card out and air dustend the hole thing and re seated it. It did have quite a bit of dust in it. After i checked the temperatures i haven't had the mistake in some time. It wasn't until i opened PC Wiz that it came back. I have since uninstalled all but sensorsview and i think i have given up on the 4200+. I have not had a problem with 2.6 GHz for a few days now, even playing for hours. Thanks for all your help, i plan on getting all new hardward at christmas now.

wrench97

Let us know when your ready to pick out the new stuff, if need be you could start with the video card and PSU if necessary.