Advertisement

Dual Xeon problem

Started by October 29, 2005 04:29 PM
8 comments, last by fastlane69 19 years, 3 months ago
I'm developing a MP application using conitec's 3D Game Studio. I've had this problem for over two years now, ever since I bought my Dual Xeon box; here's the root problem: This application runs fine on pentium4 box (1.4 Ghz, All-in-Wonder Radeon) , I start the server and the client on the same box with no problems. On my Xeon box though (Hyper Threaded Dual 2.66 Ghz, AGPx8 128MB ATI xtasy 9600) however, the server starts fine, but anywhere from immediatly to 10 seconds after I start my client the entire system locks up. I have NEVER seen XP lock up for any reason and thus this is worrisome. Here are some observations: - I realize the 3DGS is not formally supported on Xeon processors, but my understanding is the Xeons are P4 with SMP support. Furthermore, this is a DX application and for both reasons, I don't know why this would be the problem. I've tried running this by taking out a processor and run into the same problems. - Another weirdness is that everything works fine under some installations of XP and then when I reinstall (as I did recently to switch from non-raid IDE to raid SATA), the problem comes back (BTW, this problem also exsisted on the non-raid IDE, so it's not the new raid sata). - By applying an affinity mask to the applications and run the server and client on separate processors, I can usually stabilize the first client for a while. If I start a second client however, one again XP will lock within seconds or minutes. - I have tried taking out one processor and I get the same behaivior. - I've run World of Warcraft, Star Wars Galaxy with no problems. Although once in a while, SWG did lock up on me, adding only fuel to this mystery. Here are my questions: - Could it be that the video card somehow doesn't support SMP or is causing the lockup? Can I change my AGP settings in bios to solve this and if so how? - Is there some fundamental difference between Xeons and P4 that makes working on a Xeon as a workstation fundamentally flawed? - I can further stabilize by diabling HT from my box...has anyone heard of problems with HT and DX or any other application. - My motherboard is a Tyan S2668 AN...could this be the problem? - Above all, how in the heck do I go about narrowing down WHERE the problem is originating? I've tried running Task Manager and looking for unusual behaiviour in thread assignment or page faults but so far haven't seen a pattern? My gut feeling is that it's video related (since it only does it when the graphical client runs...I can leave multiple servers running with no lockup) - Lastly, can anyone recommend any more hardware/software technical sites to visit and post this problem on? I would really like to keep this box since it's so sweet to develop on...it compiles my levels faster and is so fast. But while I've found some workarounds over the last two years that allow my some illusion of stability, I would really like to get to the bottom of this so that if it is a problem with 3DGS I can go to them with data...and if worse comes to worse, I can sell this box and get a more conventional single processor P4 to develop on. I thank you in advance for any and all advice you might have on this matter.
This sounds like a hardware problem, rather than a networking problem.

When you say "locks up" does it mean "blue screen"? Or does it mean "doesn't respond to keyboard and mouse"? You can find out by installing WinDbg on a second machine, and targeting the first machine (boot it with /debug -- instructions are found in the WinDbg help and install information).

WinDbg, aka "debugging tools for windows," can be downloaded from the Windows DDK pages on MSDN.
enum Bool { True, False, FileNotFound };
Advertisement
Thanks for the quick and informative reply, HP.

Lock up is latter, "failure to respond"....Hell I haven't seen the blue screen of death for many many many years (1998 actually).

That's awesome...WinDbg is exactly what I needed and I concur, it's very much indicative of a Hardware problem...now to find out where.

As soon as I come back from Washington and the Serious Game Summit, I'm going to by a new video card and see if that fixes it...it's the only piece of hardware I haven't replaced in my quest to nail this down (even the mobo is new). If it still acts up, it's WinDbg-o-Mania!!

Thanks again...I really appreciate the help. :)
Hmmm, a dual Xeon setup. Hope it isn't the new dual core version.

To avoid wasting time, you could make sure there aren't any warning messages or labels in the manuals that came with the system or on the case itself (might be on the inside of the side/top panel, the one you would normally remove to get at the internals - with good packages you can often find a motherboard diagram here too).

Having a list of common warnings helps:

Only warning I see is against sarcastic staff members... :)

Seriously though, it's pre-dual-core and in every other regard, this system is amazing...of course it would be the ONE application that I bought it for that would cause me problems. :(
Well on a serious note, I can't see the CPU being the problem (even though the hyperthreading crutch can lead to some odd/should not happen cache situations). One possible cause is the AGP slot itself - at least with older Ati cards (though I've always used All in Wonders which carry their own driver issues) they can be sensitive to what order and what revision of GART driver is installed. I believe the current Radeon drivers lets you tweak/turn off advanced GART functionality, which would allow you to determine if it is a driver issue.
Advertisement
Yes, I too strongly believe the AGP slot is the culprit since all my past troubles seem to point to Video being the issue; the mobo and the card are both 8X agp and it is an older card.
All other non-video intensive software screams on this system as does the server application of this same software.

Thanks for the GART advice...more fodder to test out when I come back.

Really appreciate it. :)
Much ado about nothing I'm afraid! :(

The driver I was using was 3 months old. The NEW driver, 2 months old, fixes it. Man-o-man, more fool me for not checking up on drivers, but since this had plagued me for two years, I had no confidence that that was the problem.

So now I'm happy, nay, exstatic to say that my system is up and running. Let me tell you, developing on a dual xeon, raid-0, 2 gig ram, 256 Video Ram system is F'n Sweet!!!

I want to give a shout out to HP for tuning me into WinDbg...it helped me isolate that the issue was the video and ultimetely led me back to the begining...to updated drivers. Thank a million!

Ok, now it's back to work and a sigh of relief over not having to spend many hours tracking this problem down!!
I've heard WinDbg referred to as "teh r0x0r".

Glad you're up and running. Make sure to make a game that is fun to play :-)
enum Bool { True, False, FileNotFound };
Since the market is innundated with games that are fun to play, I'm actually trying to appease a niche market and make the worlds most boring game. I think this is an untapped market with much potential for growth!

:D

This topic is closed to new replies.

Advertisement