PDA

View Full Version : System Troubleshooting Help Needed (lots of data, very few answers)


OWA
07-21-05, 08:41 PM
I'm having a really hard time getting the SLI system in my sig stable. I've tried very loose and stock settings and when doing so have hard locks in about 5 or so games before the games even get to the menu. Switching memory helped some (tried 3 different kinds) but didn't solve it. I've tried everything I can think of and nothing completely resolves the issue. Last night I stumbled upon something that seems to have solved the issue but I'm not sure what this is telling me.

Okay, what seems to have solved it is underclocking my CPU quite a bit. Instead of the stock 2.4ghz, I set the multiplier to 10, HTT is at 4x, etc. so now I'm running at 2ghz. Doing this, I haven't had any lockups. I still thought it might be a heat issue so I added a fan to the top of the fanless heatplate covering the chipset (didn't really help), opened the case and put a huge fan blowing into the case. This lowered the system temps (chipset) and CPU some although the video cards pretty much stayed the same. I then got rid of the fan, closed the case, the temps went up but I was still stable as long as I'm underclocked.

Here are some sample temps.

Big fan blowing in an open case, underclocked CPU, stock video card settings
CPU 37C
System 31C
GPU 1 75C
GPU 2 69C
STABLE

No fan, case closed, underclocked CPU, stock video card settings
CPU 46C
System 45C
GPU 1 77C
GPU 2 79C
STABLE

At stock settings for the CPU, video cards, etc. my temps are largely the same as the last case. The CPU usually stays below 50C or right at 50C, System hovers around 45-50C and the GPUs are in the upper 70s but I have major stability issues.

I've also tried several different voltages for everything that I can change and none of that seems to matter. I didn't experiment with the CPU voltages very much though since it seemed stable in 2D tests. I did bump it up a little but maybe I should experiment more with that.

So, the bottom-line is, what is this (underclocked CPU making things stable) telling me? Maybe some sort of power issue? I'm using an Enermax 600W PS that is SLI approved but could there be something wrong with it that is causing instability? Anyway, I have a ton of data but until the underclocking didn't seem to be getting anywhere. The underclocking seems like it should be a pretty good hint at what the problem is but I'm not sure what it's telling me. So, I'm hoping some of y'all will have some ideas on the issue so I can finally start doing something other than trying to get things stable. Oh, I seem to be completely stable in 2D also and have even tried underclocking the 7800s to things like 300/1000, etc. (didn't help a bit).

OWA
07-21-05, 08:57 PM
Another data point, just tried changing the multiplier to 11x instead of 10x which raised the CPU to 2.21. Locks although temps were basically the same as 10x. That was the only change.

Edit: What made me think it's heat is that I can usually run the games if I run them first (right after booting). If I play any other games or whatever first, then they'll lock. But, since my temps don't seem that bad or all that different from when running underclocked, I don't really know what that is saying.

Zerocool
07-22-05, 06:45 PM
update mobo BIOS maybe?

-=DVS=-
07-22-05, 06:59 PM
Just for kicks try setting ram to 333 , and up multy back to 11.
Alternatively up CPU voltage to 1.5 .

OWA
07-22-05, 09:39 PM
Just for kicks try setting ram to 333 , and up multy back to 11.
Alternatively up CPU voltage to 1.5 .
Thanks for the suggestion but unfortunately, that didn't work either. I played with it a while last night and only seem to be able to get things reasonably stable when using x10.5 for the multiplier or less. I have to play with the voltages a little to get it stable at x10.5. At x10, I can pretty much leave things at stock settings. At x11, if I crank the voltage on almost everything to the max, things are almost stable but I still get locks. I even tried cranking the voltage to the max on the CPU.

macatak
07-22-05, 09:57 PM
As stated earlier, could be BIOS related, latest version is 1005. Also a few beta ones available

http://support.asus.com/download/download.aspx?SLanguage=en-us

OWA
07-23-05, 07:05 AM
It came with 1004 but I quickly went with the latest beta 1006-004. Once I talked to a few people with stable systems that were using 1005, I flashed back to it.

I just don't understand it. It seems like it's a heat issue but my temps all seem pretty good. For example, I can play almost every game right after booting. Run rthdribl just for a couple of minutes and I can usually still run them. Run it for around 5-10 minutes, enough to get everything heated up, and I get hard locks in several games even though reported temps don't seem all that bad. I thought there were a few games where I was completely stable, like Doom 3. But, launched Doom 3 last night and just let it sit there for a while on the beginning level and eventually it locked as well. It just takes a lot longer than some of the others.

I've also found that I get lockups just using a single 7800GTX as well although it's less frequent and takes a little longer.

I think I'll grab the A64 3500+ from my other system and try that next.

OWA
07-23-05, 01:41 PM
More info on why I don't think it's heat. It seems heat related b/c I won't be able to play games after running rthdribl for very long. I also found out today that I can use the windows version of memtest to have the same effect. Run it a little while and then the games will lock up almost immediately. Start games right after a reboot and they usually work for a little while.

Today I tried letting it sit for 4 hours after being heated up so my temps were very low and games still wouldn't start without locking in a few seconds. So it appears to be more the act of doing anything intensive on the computer leaves it in such a state that games won't work. Almost like something is being left in a bad state. In 2D I'm always fine and seem to be fine even overclocking the CPU to around 2.64ghz.

jAkUp
07-23-05, 04:02 PM
Set everything to stock, run CPUz and see what your voltage to your CPU is reading... is it lower than 1.35v?

Also, try putting a little extra vcore to the CPU. Set everything to stock, and try for a 1.45v setting. See if that cures any issues you are having.

It may be that your PSU is severely underloading the voltage supplied to your CPU under load. If you have a spare PSU you could try that as well.

OWA
07-23-05, 05:18 PM
Set everything to stock, run CPUz and see what your voltage to your CPU is reading... is it lower than 1.35v?

Also, try putting a little extra vcore to the CPU. Set everything to stock, and try for a 1.45v setting. See if that cures any issues you are having.

It may be that your PSU is severely underloading the voltage supplied to your CPU under load. If you have a spare PSU you could try that as well.
Thanks for the help. I meant to run CPUz to check that before I disassembled everything once again but forgot. Nvmonitor didn't correctly read it always showing 1.550. Everest Home edition showed around 1.34 to 1.35 when not under load. Unfortunately, I didn't see the voltages under load since I was using nvmonitor to watch temps and stuff when running rthdribl. I did try various voltages with the CPU though, undervolting and overvolting. The specs are kind of weird in that it says the voltage is supposed 1.35 - 1.45 or something like that instead of just a single number. Anyway, I had to go to about 1.50-1.55 to even get close to being stable with just a x11 multiplier (mine is supposed to be x12).

I did purchase an Antec 480W (neopower) PS to test it in non-SLI mode and had the same issue. So, I just took the AMD 64 3500+ CPU from my other system and now things seem to be stable. I've been completely stable with one 7800GTX so I'll try SLI soon. I really don't understand what could be wrong though with the X2 4800+. I seem to be completely stable in 2D. I ran memtest for around a couple of hours, looped 3DMark05 for around an hour or so, tried overclocking the CPU to around 2.6 and everything still seemed stable in 2D (and 3DM05). So, I don't understand why the CPU would affect things only when gaming in 3D and why 3DMark05, for example, would seem to run fine even though games wouldn't.

I'm now debating whether to install the X2 one more time just hoping it'll work this time around. What do you think? Wishful thinking? Should I just cut my losses and just RMA it or what? Actually, I guess I should test SLI first just to see if the 3500+ / system is still stable with that. I guess it could still be some weird motherboard or other issue.

OWA
07-23-05, 05:48 PM
Dang, seems to be stable with SLI...actually, I'm really glad about that but still can't figure out what the deal is with the X2 4800+.

-=DVS=-
07-23-05, 05:57 PM
Could be bad CPU , it happens.

jAkUp
07-23-05, 06:06 PM
Yup, probably a bad CPU.

floydenheimer
07-24-05, 12:34 AM
Do you have Cool n Quiet enabled in your bios?
Have you installed the AMD Athlon drivers?

2 things you can try are:

Try disabling CnQ.

Another is adding a /usepmtimer in your boot.ini file right after /fastdetect so the line would look something like this:

multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Microsoft Windows XP Professional" /noexecute=optin /fastdetect /usepmtimer

(don't copy and paste that line, it might not work for where you have your disk! it's for example only)

Also some 3D games/apps do not handle dual cores well and you need to assign affinity to a single core before you load any levels and such (at the initial menu selection screen ctrl+alt+del out and assign affinity to one of the cpus)

Hope this helps!

floydenheimer
07-24-05, 12:39 AM
Also, slight chance here but it's possible- do you have your mb mounted upside-down by chance? Like say in a Lian-Li V1000? The heatpipe on the nf4 chip doesn't work if the board is mounted upside-down and overheats.

floydenheimer
07-24-05, 01:04 AM
This is a handy tool to assign an app/game to your first cpu upon launch.
http://www.activeplus.com/us/freeware/runfirst/

wollyka
07-24-05, 09:02 AM
Hard luck it seems a bad cpu
but if u have spare rams try changing them, it once happened to me, playing games locked my system, no overheating, no PSU problems,even running memtest didn't show anything but fortunely i had spare rams so i changed them and problem solved.. maybe u can try that b4 sending the cpu back..

Rampant CL
07-24-05, 11:20 AM
What happens with only one gpu on stock settings? does that work fine?

OWA
07-24-05, 02:12 PM
Do you have Cool n Quiet enabled in your bios?
Have you installed the AMD Athlon drivers?

2 things you can try are:

Try disabling CnQ.

Another is adding a /usepmtimer in your boot.ini file right after /fastdetect so the line would look something like this:

multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Microsoft Windows XP Professional" /noexecute=optin /fastdetect /usepmtimer

(don't copy and paste that line, it might not work for where you have your disk! it's for example only)

Also some 3D games/apps do not handle dual cores well and you need to assign affinity to a single core before you load any levels and such (at the initial menu selection screen ctrl+alt+del out and assign affinity to one of the cpus)

Hope this helps!
I'm not using CnQ. I'm pretty sure I installed the AMD drivers but I'll double check that. Thanks for the other suggestions. I'll give them a shot. Right now I'm having to just run it at around 2.10 ghz (x10 multiplier and 210 for the FSB). I screwed up my other CPU (the A64 3500+) the last time I swapped things so I'm down to one marginally stable, underclocked system. :(

OWA
07-24-05, 02:13 PM
Also, slight chance here but it's possible- do you have your mb mounted upside-down by chance? Like say in a Lian-Li V1000? The heatpipe on the nf4 chip doesn't work if the board is mounted upside-down and overheats.
It's mounted in a Antec P160 case so it's a normal mounting.

This is a handy tool to assign an app/game to your first cpu upon launch.
http://www.activeplus.com/us/freeware/runfirst/
Thanks. I'll give that a shot.

What happens with only one gpu on stock settings? does that work fine?
I thought so initially but it's still unstable. It just happens a little less frequently. I also tried underclock the single card to like 300/1000 and that didn't seem to matter.

Hard luck it seems a bad cpu
but if u have spare rams try changing them, it once happened to me, playing games locked my system, no overheating, no PSU problems,even running memtest didn't show anything but fortunely i had spare rams so i changed them and problem solved.. maybe u can try that b4 sending the cpu back..
I thought it was the RAM at first. Changing RAM did improve things somewhat but didn't completely solve it. I've tried 3 different types of RAM (OCZ, PNY and Kingston) but I'm not sure any of them are on the approved list. I was going to get some on the list but I've PM'ed a few people that are using the same motherboard, memory, etc. and the RAM is fine for them. Also, the OCZ RAM was fine when I swapped out the X2 for the A64 3500+ so I don't think it's that. I may still try to get some on the approved list so if I talk to tech support or whatever I can say I'm using approved parts, etc.

floydenheimer
07-24-05, 02:39 PM
If you installed the AMD drivers you need to have the /usepmtimer in your boot.ini.

I had a lot of weird issues with certain games if I didn't assign them to a single cpu before loading a game. Some games just don't handle the dual cores well. GTA:SA and EQ are 2 examples of games that will only run stably on 1 cpu. Doom 3 and MVP Baseball 2005 are fine without assigning affinity.

OWA
07-24-05, 02:55 PM
Actually, after installing the AMD drivers, things seem to be working so far. If that's it, somebody please shoot me!! I broke my A64 3500+ CPU during all the switching CPUs testing yesterday and have spent at least 2 weeks trying to get the system stable. Initially, I had a bad SATA cable that was causing overall system instability. It must have been during all that that I had installed the AMD drivers. I'm sure I did it initially since I even copied the drivers off the CD to my install directory (a place I keep all the drivers and stuff that I install right after WinXP). I've since reinstalled 7 or 8 times and during those installs must have skipped installing the AMD drivers. I probably skipped them thinking they didn't matter since Windows seemed to correctly identify the CPU. Man ol' man, if this really fixed the issue, I can't thank you enough. It has been driving me crazy.

OWA
07-24-05, 03:06 PM
Okay, it doesn't appear to have completely solved my issues but it's now something I can work with. Painkiller has been my test case since it would usually hard-lock on the opening logo (about 2 seconds after trying to launch the game). It worked a few times but eventually, my 4th attempt to run it, it hung right before the main menu. It's a different type of hang though. I was usually getting hard-locks that required a reset to recover from or it would lock and I could still break-out of Painkiller to kill it but it would not be responding when I'd do that. This time, the game was still running it just wouldn't let me get to the menu. The first two times it required me to hit ESC a few more times than usual so maybe I can solve this with the utility you recommended. Anyway, looks like I'm finally getting some where. I checked the boot.ini and the AMD driver install had added the /usepmtimer already.

Edit: Wow, it really does seem like that was the issue. I've been able to run other things that were hard-locking also. Man do I feel stupid. Thanks again for making me double-check the AMD drivers.

floydenheimer
07-25-05, 01:00 AM
Hope you're running well!

Glad I could lend you a hand. :)