I promised a long story and here it is.
I just bought a new computer:
mobo: DFI Lanparty UT nf-4 Ultra-D
cpu: AMD 64 Venice 3200+
ram: 2x1gb OCZ DDR 400
gpu: Leadtek 6800GT
psu: Antec 450watt something
that should be all the important specs
Anywho, I started up the HL2 (everything at stock speeds), and it ran like a dream, for about 40 minutes, until it started hanging. Freezes for a few seconds every 5 minutes, then 10 seconds every 3 minutes, until it gets too bad to play, or it bluescreens. I thought 'ok, sure, overheating, no problem," and promptly pointed a table fan at my open case, while cackling wildly.
This solved my problem.
But, now I'm looking for a more permanent solution. I origionally blamed the video card for overheating, and looked at the temp. I was surprised by the idle temp of >50'C, and thought this was a big problem, possibly bad contact. But then I found that these temperatures are rather normal for this card. I did some testing.
With the FSB overclocked by 5mhz, the game reached the 'overheating' syndromes much faster. With the fan pointing at the case, it still experienced some of the syndromes, although much rarer.
All the information so far seems to be pointing towards overheating.
So i tested some more, and brought the clock back down to stock. From this point on, all speeds are stock.
I redid the contact to the chipset to the HSF. That did nothing. I started putting blame on the fact that the video card is directly over the chipset. In lieu of a picture, here is some crappy ascii
----o
(hyphens is the card, O is the chipset's HSF)
The end of the card does hover over the chipset, but the card has very little bulk at that point.
The blame on the chipset all but faded away the other night when I ran prime95 (with the fan off), and recived an error after a couple hours. This still MIGHT be the chipset, but a friend assured me that it is almost definetly the CPU or the ram, as the chipset does not handle the ram on the 64 bit motherboards.
I then ran memtest86, which ran for 10 hours with no errors, and is still running while I am at work.
After all this, I think I have it narrowed down to the CPU overheating. The temperature readout is not high, nor does the heatsink feel hot. I am not sure where the temp sensor on the 64 bit processors are, as there is no spot on the bottom without pins. I hope it is just bad contat between CPU and HSF.
Condensed version:
-HL2 @ stock w/o table fan = crash
-HL2 @ OC w/o table fan = faster crash
-HL2 @ stock w/ table fan = no crash
-HL2 @ OC w/ table fan = small stutter
-Prime95 @ stock w/o table fan = fail
-Memtest86 @ stock w/o table fan = stable
So, It seems the next thing to do woudl be one of the following
A) redo contact between CPU and HSF and pray
B) run prime95 with table fan
C) RMA everything and become a hermit
D) Run that wacky video test rthdribl
I will try to avoid C as my thatching skills are lacking. Thanks for any advice.
I just bought a new computer:
mobo: DFI Lanparty UT nf-4 Ultra-D
cpu: AMD 64 Venice 3200+
ram: 2x1gb OCZ DDR 400
gpu: Leadtek 6800GT
psu: Antec 450watt something
that should be all the important specs
Anywho, I started up the HL2 (everything at stock speeds), and it ran like a dream, for about 40 minutes, until it started hanging. Freezes for a few seconds every 5 minutes, then 10 seconds every 3 minutes, until it gets too bad to play, or it bluescreens. I thought 'ok, sure, overheating, no problem," and promptly pointed a table fan at my open case, while cackling wildly.
This solved my problem.
But, now I'm looking for a more permanent solution. I origionally blamed the video card for overheating, and looked at the temp. I was surprised by the idle temp of >50'C, and thought this was a big problem, possibly bad contact. But then I found that these temperatures are rather normal for this card. I did some testing.
With the FSB overclocked by 5mhz, the game reached the 'overheating' syndromes much faster. With the fan pointing at the case, it still experienced some of the syndromes, although much rarer.
All the information so far seems to be pointing towards overheating.
So i tested some more, and brought the clock back down to stock. From this point on, all speeds are stock.
I redid the contact to the chipset to the HSF. That did nothing. I started putting blame on the fact that the video card is directly over the chipset. In lieu of a picture, here is some crappy ascii
----o
(hyphens is the card, O is the chipset's HSF)
The end of the card does hover over the chipset, but the card has very little bulk at that point.
The blame on the chipset all but faded away the other night when I ran prime95 (with the fan off), and recived an error after a couple hours. This still MIGHT be the chipset, but a friend assured me that it is almost definetly the CPU or the ram, as the chipset does not handle the ram on the 64 bit motherboards.
I then ran memtest86, which ran for 10 hours with no errors, and is still running while I am at work.
After all this, I think I have it narrowed down to the CPU overheating. The temperature readout is not high, nor does the heatsink feel hot. I am not sure where the temp sensor on the 64 bit processors are, as there is no spot on the bottom without pins. I hope it is just bad contat between CPU and HSF.
Condensed version:
-HL2 @ stock w/o table fan = crash
-HL2 @ OC w/o table fan = faster crash
-HL2 @ stock w/ table fan = no crash
-HL2 @ OC w/ table fan = small stutter
-Prime95 @ stock w/o table fan = fail
-Memtest86 @ stock w/o table fan = stable
So, It seems the next thing to do woudl be one of the following
A) redo contact between CPU and HSF and pray
B) run prime95 with table fan
C) RMA everything and become a hermit
D) Run that wacky video test rthdribl
I will try to avoid C as my thatching skills are lacking. Thanks for any advice.
Facebook
Twitter