9800 Pro + Call of Duty crashing constantly

[Jay]

A friend is having problems running COD on his newly upgraded system:
-> Epox 8RDA3+, XP 3200+, 512M Crucial, FIC Radeon 9800 Pro, SB Live 5.1, Antec
400W PS

He just upgraded his mobo, proc and memory. Old components were:
-> Epox 8KHA+, XP 1900+, 2x256M Crucal

He's running Windows XP, latest drivers (SP1, mobo, video, sound), COD 1.2 and
about every 2-5 minutes COD will lock solid, get a VPU Recover, or get graphic
bars on the screen and sometimes show a partial desktop and crash.

After searching the web, he's tried many things suggested such as AGP set to 4X,
bilinear instead of trilinear, Catalyst 3.10, ATI hotfixes, latest Omega
drivers, disable Fastwrite. No changes.

He's been playing BF1942, UT 2004, Halo and AA for hours with almost no problems
(almost because BF1942 crashed twice today, once to the desktop, once with
graphic bars).

At standstill CPU temp is 55c, system temp 39c, +12 at 11.85, +5 at 5.05.
Nothing is overclocked! System at stock settings.

The kicker is that his old 8KHA+ system ran COD with zero problems. The only
thing he doesn't know is what video drivers were used at the time. The new
system is a complete format and reinstall of XP, so there is no old baggage
there.

So it's gotta be related to the new mobo-- speed, or pushing the 9800 Pro to
it's limitis or something. But when his other games play with no problems, it
kind of contradicts that.

Can anybody offer any suggestions?? Thanks in advance!

Jay

 

[dino]

first off- Disable VPU Recover..I am running a 9600XT with 2504 Omega's....I
had some minor issues until I did a ground up format. Did he reformat with
the new board?And what are his new components???

 

[dino]

sorry missed the 8KHA..

 

[Wayne Youngman]

A friend is having problems running COD on his newly upgraded system:
-> Epox 8RDA3+, XP 3200+, 512M Crucial, FIC Radeon 9800 Pro, SB Live 5.1, Antec
400W PS

He just upgraded his mobo, proc and memory. Old components were:
-> Epox 8KHA+, XP 1900+, 2x256M Crucal

He's running Windows XP, latest drivers (SP1, mobo, video, sound), COD 1.2 and
about every 2-5 minutes COD will lock solid, get a VPU Recover, or get graphic
bars on the screen and sometimes show a partial desktop and crash.

Hi,
I set-up up two Athlon machines that run COD, one uses a 9800 with CAT v3.7
and the other uses a 9600XT with CAT v3.10, works fine, never crashed once!.
Has your friend done some proper stability testing like Prime95 and
Memtest?. Also I'm not convinced that some motherboards give the right
voltage to the graphics card, but don't know if that would make COD crash.

Do some stability testing of the CPU and Memory, then post back. May not be
a graphics problem!

 

[Jay]

Hey Dino,
The 8RDA3+ is the latest board. He did disable VPU recover but it still gets VPU
recover errors. Go figure.

The installation is a fresh install.

 

[Jay]

Hi Wayne,
What motherboards/procs are you using on your working systems?

FWIW my friend and I swapped video cards today. Same exact board, FIC 9800 Pro.
My board seems to work better-- he was able to play COD for a couple hours this
afterboon and maybe a half hour this evening before it VPU recovered.

I wondered about upping voltages but I wouldn't know what to mess with.

Re: Prim95, memtest.... thought of that but was learning towards video as the
problem, not proc/memory. Though you may be right; I'll have him try that....

Jay

 

[DaveW]

The CPU temp is actually rather high at 55C!

 

[NTH]

The CPU temp is actually rather high at 55C!

That's what I was thinking as well.

NTH

 

[Wayne Youngman]

Hi Wayne,
What motherboards/procs are you using on your working systems?

Hi,
here is the tech spec of both working COD systems:
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

*Mercury-1*
Barton (AQXEA) XP2500+ @ 2.2GHz (10x220) - 1.775vCore
CoolerMaster Aero 7 Lite - 3,200rpm
ABIT NF7-S v2.0 (BIOS d20)
512MB Dual TwiSTER PC3500 @ DDR440 1:1 (2.0,3,3,9 - 2.8v)
Sapphire Atlantis 9800 - 3.3ns Samsung (325/290 Default)
WD-SE 240GB (2x120GB) SATA RAID-0 (NTFS - 16k Stripe)
Antec SX630II Mini-Tower Case Inc 300w PSU
2 x CoolerMaster 80mm Blue Neon Fans
WinXP PRO inc. SP1
nVidia Unified v3.13
Cat 3.7 - DX9.0b

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

*SOUL-RIPPER*
AN7 rev1.0 (BIOS 13)
T-Bred XP2400+ (AIUHB) 2.2GHz - 1.85vCore
512MB (Single Stick) Crucial PC3200 (CT6464Z40B)
Sapphire Radeon 9600XT (FireBlade Edition)
Western Digital S.E 120GB SATA (7,200rpm - 8MB)
LITE-ON 52x32x52 Burner
LITE-ON 16x DVD
ANTEC SX835 II
ANTEC 350w Smart-Power (Dual Fan)
4 x 80mm Case fans
WinXP SP/1
DX 9.0b
nVidia Drivers (CD)
Cat 3.10 (CD)

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

I'm currently looking into an issue regarding AGP 8x and AGP 4x voltages set
by the motherboards. Even though I couldn't crash COD, if I left
*SOUL-RIPPER* (9600XT) running Prime95 *and* 3DMark looping for more than
24hours it would reboot the system!!! (lol yes I know this is *extreme*
circumstances, but my other builds have been alright)

 

[Wayne Youngman]

"DaveW" wrote

The CPU temp is actually rather high at 55C!

Hi,

55°C is fine, and shouldn't crash a system. Yes it could be better (lower),
but should still work fine. . .

 

[Jay]

The CPU is a retail AMD w/ their heatsink. Suspecting their heatsink
tape/wax/crap might be the problem, we yanked the sink and put down some Arctic
Alumina.

The sink could use some lapping, it's not as smooth as I'd like.....

 

[Jay]

Hey Wayne, tried running Memtest and Prime95.

Memtest ran overlight (about 15 minutes short of 12 hours) with zero problems.

However....

Prime95 runs about 3 minutes on Torture/mixed before aborting. Longest run has
been 7 minutes.

We opened the case and put a small desktop fan (10") set on high, blasting over
the memory and proc. No change, Prime still dies in 3 minutes.

Whaddya think -- bum proc or mobo? Is 12 hours of Memtest sufficient enough to
rule the memory as innocent?

 

[Wayne Youngman]

Hey Wayne, tried running Memtest and Prime95.

Memtest ran overlight (about 15 minutes short of 12 hours) with zero problems.

However....

Prime95 runs about 3 minutes on Torture/mixed before aborting. Longest run has
been 7 minutes.

Hi,
if Prime95 is failing in that short a time that system is *UNSTABLE*. The
fact that you ran Memtest86 overnight is good. Tell me more about the
system specs again please, it looks like your CPU isn't receiving enough
juice (vCore). In these circumstances you may be able to run WindowsXP, do
some surfing etc without knowing there was a problem. Its when you run
something that's really CPU intensive the problem becomes apparent.

 

[John Lewis]

The CPU is a retail AMD w/ their heatsink. Suspecting their heatsink
tape/wax/crap might be the problem, we yanked the sink and put down some Arctic
Alumina.

The sink could use some lapping, it's not as smooth as I'd like.....

Ati has issued a special patch for COD to fix crash problems with Ati
boards 9500 and above, Cat3.8 and higher.

See:-

http://www.avault.com/pcrl/patches_temp.asp?patch=callduty

John Lewis

 

[Gordon Scott]

The CPU is a retail AMD w/ their heatsink. Suspecting their heatsink
tape/wax/crap might be the problem, we yanked the sink and put down some Arctic
Alumina.

The sink could use some lapping, it's not as smooth as I'd like.....

actually I have seen discussions and tests where a smoother sink might
not be the advantage as previously thought. The pits and grooves in the
rougher finish actually fill with thermal compound giving better
disipation than a scenario where perfectly flat finishes tend to squeeze
out all the compound. Makes sense to me, otherwise why wouldn't AMD spec
a smoother finish for the stock sinks that fit its specifications.

Gordon

 

[Wayne Youngman]

actually I have seen discussions and tests where a smoother sink might
not be the advantage as previously thought. The pits and grooves in the
rougher finish actually fill with thermal compound giving better
disipation than a scenario where perfectly flat finishes tend to squeeze
out all the compound. Makes sense to me, otherwise why wouldn't AMD spec
a smoother finish for the stock sinks that fit its specifications.

Hi,
that can't be right?. I haven't seen any information that leads me too
believe that a roughly finished heatsink would perform better thank a mega
smooth.ultra lapped mirror finished one. The thermal goop is just there to
fill a *micro* chasms in the finished, because as we know, air doesn't
transmit heat very well. I suspect if the finish of the CPU slug and HS
were smooth enough you wouldn't need any heat-sink compound at all. If the
compound is getting *squeezed* out then its being applied too thickly!. . .

 

[Mike P]

That patch is actually beta 3.10 cat's. I used them and the official 3.10,
both were fine in CoD.

Mike

 

[Jay]

Taken from USDM.... know you wanted Vcore but thought I'd dump them all.

I did try bumping Vcore to 1.7 but that didn't help, so went back to defaults.

Vdd 1.632
Vcore 1.664
Vdimm 2.752
+5 5.053
+12 11.856
Vagp 1.520
Vbat 2.96
6VSB 4.969
CPU 54c
Sys 40c

AMD's stock HS is gone, replaced by a CAK II-38.

 

[Jay]

Interesting page at overclockers.com comparing heatsinks and their abilities to
dissipate heat. AMD's stock sink was pretty feeble.

http://www.overclockers.com/articles373/socketA.asp