What is my computer doing?

Factual Questions
1

Okay, so today I finally got fed up with Windows 98SE after the 14,003rd crash this week. I ran out to Best Buy and picked up a full copy of XP (and a pretty Samsung 17" CRT to help me calm down).

Here’s the problem. I’ve been playing Madden NFL 2004 and GTA: Vice City on my (fairly) new computer. Both games freeze and/or cause a crash every ten minutes or so.

The system is based around an AOpen AK79G-VN motherboard (AMD Athlon 2200+), with onboard everything kindly provided by nVidia (GeForce 4, not FX). Plenty of HD space (7 gig on the “System” (read: Windows) partition, 30+ on the other half), lots of room for air (it sits on top of my desk), and updated drivers. I’m not overclocking it (well, actually I am, but only up to the standard processor speed, which for some reason isn’t the default)

Still, the damn thing crashes more than AirTran. My surge protector is old (6 years) but I don’t think that’s causing the crashes. Madden either simply freezes, and keeps me from doing anything except hitting reset, or it does something odd. It doesn’t crash, exactly… it just, well, disappears. It will just poof and the desktop reappears. Nothing in the taskbar (occasionally the madden tab will be there until I click on it, when it disappears) and the system will be completely stable, but the game is gone.

GTA crashes more traditionally… the screen just stops refreshing and the sound begins to loop. Hitting escape allows me to end task, and bring Windows back up.

Here’s the really scary one though. Sometimes (during either game) the system will just die. Everything goes black and the HD winds down in less than a second, and a very quiet alarm starts going off inside the case. I checked the manual and the only alarm the board is supposed to have is for chassis intrusion, and I don’t have a sensor hooked up for that. The system is still sort of running… I have to manually power it down for the power light to go out… but I certainly can’t operate it.

Any ideas what’s causing all this… and how can I fix it?

2

Have you updated the video driver to the latest and greatest, or are you still using what came on the xp cd? That’s the most likely culprit. I’d do 2 things immediately, if you have not done so already.

  1. Get the latest Win XP service packs from Microsoft’s web site.
  2. Get the latest XP video drivers from Nvidia’s website.
3

I agree. Most likely a software rather than hardware issue here. Chances are, you might need the newest version of DirectX 9.0 or 9.0b(A Microsoft 2d/3d Graphics driver system) to run these newer game titles effectively.

I would run “Windows Update” (from the Start menu) while online and download all the “Critical Updates” and all the “Driver Updates” for your machine. (http://windowsupdate.microsoft.com) After that is complete, look for the video drivers from NVidia for your video card (http://www.nvidia.com) and after that is complete, look for the latest DirectX from Microsoft (http://www.microsoft.com/directx/)

4

If it crashed madly under 98 then crashes madly under XP, I’m inclined to blame it on the hardware rather than the software. I’m wondering if the alarm you are hearing is an overtemp alarm that is shutting the CPU down.

After it crashes, if you can reboot it at that point go into the BIOS setup and look for something that will tell you the temperatures. Often it’s listed under system information or environment or system monitoring or some such. If the cpu is up above 45 deg C or the overall case temp is much above room temp then you have a problem.

5

Assuming you’ve checked and discounted the CPU temperature according to engineer_comp_geek’s directions, I’m betting on the power supply. Is it a 250W or 300W model? You might also check the +5V and +12V voltages in the BIOS. If either of those are significantly higher or lower than they should be, replace the power supply. If the power supply is fine, it’s probably the motherboard itself that’s dying.

6

What are the video settings for the game. If you have all the video options maximized then it might be that the onboard video cannot handle it. That happened to my son’s computer with Halo and the computer would crash. We had to lower the video settings. On board video usually is not as good as a separate video card.

7

engineer_comp_geek, this is the second time in the past week I’ve seen you quote the “over 45 deg. is bad” number. I’m wondering, where are you getting this number from?

dutchboy208, check your MB manual again, or the web for error beep codes. Those beeps are telling you something important.

8

I got some reliabiilty training as part of my engineering degree at WVU and a bit more training when I did (among other things) reliability analysis for airborn RADAR and FLIR systems while working for a defense contractor. They didn’t let me keep the books (for some reason they are picky about you taking things out of a classified building, go figure…) so I don’t have a good cite for you at the moment.

9

Here’s one web site for motherboard beep codes.

10

General computer convention knowledge states that most CPU’s should be below 50C to maintain a stable envirnoment. AMD and Pentium chips start to melt down at a whopping 85-90C.

Here is a guide that may be helpful to you: http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/27375.pdf

Generally, running 45C or less is great for any processor running a .13 die process (such as earlier Athlons, Athlon XP Palomino, Pentium III, Early Pentium 4s) 50C or less is great for processors that run faster than that.

.09 die size chips should be in the 40’s for optimal performance and long life of the processor.

11

Manufacturers usually rate their electronic parts (like CPUs) to 85 or 90 degrees C, but if you hang around the PC hardware and CPU overclocking bulletin boards, you’ll see that most of these enthusiasts try to keep the CPU temperature under 50. Run at spec in a well-ventilated case, most CPUs stay even cooler, even down in the 30s. Higher temperatures are often blamed for stability problems. Anecdotal, I know. A valid result from empirical experimentation, maybe.

Also, at least some of the Athlon chips use a temperature sensor that’s not actually inside the chip, but mounted in the socket beneath it, and they may yield lower-than-actual readings.

12

You are overclocking it. If I’m understanding the OP correctly, you’ve overclocked the chip to 2.2 GHz, right? If that’s correct, you’ve overclocked the chip by 22% which, in my experience, is enough to cause the kind of thing you’re experiencing.

The 2200+ is a 1.8 GHz chip. AMD’s numbers don’t have anything to do with the actual speed of the processor; IIRC, it’s supposed to be an “equivalent to Intel processor X” kind of thing.

FWIW, I’ve been running a 2200+ for the last year or so, and I’ve found it to be extremely sensitive to overclocking. I finally just left it alone.

13

Here’s a little more…from SharkyExtreme.com

"If you’re thinking of buying the Athlon XP 2200+ for overclocking purposes, then stop right where you are. The highest reliable core speed we could get was 1931 MHz using a 143 MHz FSB and a core voltage of 1.775V. You may be looking at the core voltage and wondering why we didn’t use 1.8V or higher, and that would illustrate the main problem we had when overclocking. "

14

I’m betting on TheInterruptingCow here. That “I’m not overclocking it (well, actually I am,…” bit scares me. AMD (and most people in computing), realize that Mhz is not the be-all and end-all of computer performance, but lots of other people don’t. To make sure people bought their stuff, AMD tacked a number and a “+” on the end of their chips. Technically, that 2200 is part of the name, as it has nothing to do with the chip speed, and is meant to be roughly comparable to the speed of whatever Intel’s comparable product is.

Personally I think it’s kinda dumb, but I also think HD manufacturers selling 1,000,000,000 byte Gigs is kinda dumb too.

15

I don’t think you understood correctly. In the BIOS on my MB, there are 2 options for clock speed - one that is the advertised clock speed, and another one that is actually underclocked. This underclocked one is the default, for some reason.

Please correct me if I’m wrong. Also, thanks to those above who answered my CPU temp question.

16

Well, what were the “advertised” and “default” clock speeds in the BIOS screen? What megaherz are you running the chip at now?

17

Keep in mind that I’m not the OP.

My AMD 1800+ runs at a clock speed of 1.53 GHz. This is the higher of the two clock speeds supported automatically by my BIOS. The other was either 1.1 or 1.3 GHz, I can’t recall exactly at the moment. When I first installed the system and every time I flashed the BIOS, it would revert to the lower clock speed as the default.

I don’t think there is an overclocked clock speed option automatically supported. For that, you have to choose “Manual” (the third option, after 1.1 and 1.5 GHz) and then play around with Vcore.

18

Just a little clarification on CPU temps. The Oaximum Operating Temperature for most CPUs is around 70C. On AMD processors it ranges between 69-73C depending on model, for Intels it varies on the individual CPU level, topping out at around 79C. This is the maximum temperature that the CPU is claimed to operate properly, producing correct results and not crashing. The RECOMMENDED Maximum, which is the hottest they suggest the CPU should run at for optimal lifespan, is between 45-55C. Intel P4s are generally around 45C, AMD CPUs around 50C. The Absolute Maximum Operating Temperature is when the CPU is expected to be permanently damaged. AMD CPUs are rated at 85C, 90C, or 95C depending on model, with Intel CPUs generally within this range as well, or perhaps a bit higher. Reducing the operating temperature by 10C will double the lifespan of the CPU.

19

This is just a quick and dirty approximation. However, it is good enough that they actually teach it to engineers in college (or at least they did in my school). The general rule is every 10 deg C above room temp cuts the expected life span in half (which is basically saying exactly the same thing you did just the other way around), except this rule is notoriously inaccurate at either end of the range. The reliability numbers really don’t start to head south until about 45 deg C and once you get up above 70 deg C they go down a lot faster than the 10 deg C rule would predict.

Coincidentally, the human pain threshold is right around 40 to 45 deg C for most people (I once built a machine for a nuerobiologist to test this, so I can be pretty sure about the range). This leads to another rule when designing electronic stuff. If you can’t leave your finger on it then it needs more cooling. I personally refer to this as the highly scientific “ouch” test. I should probably mention that it’s not a good idea to go poking around inside your PC unless you know what you are doing. It’s a good way to damage your PC and possibly hurt yourself.

FYI - it’s not uncommon for PCs to run a little hot, especially some of the big name desktops that aim for very low cost. They are banking that you’ll upgrade your PC long before the reduced long term reliability comes into play. Laptops also tend to run hot. Cooling fans take up space and power, two things that laptops have a very short supply of.

20

Okay, I f’ed this one up a bit (oddly enough, when I saw Jake4’s posts, I thought they were mine and had magically been posted already).

I forgot to mention that switching to XP had stopped the crashing problems. I was asking the question to find out why the system was so unstable before , really.

The CPU is actually the Athlon XP 2600+, not the 2200+. What I said before about the speed being jacked down as the default was correct, though.

Here are a few settings from my BIOS:
CPU Frequency: 166 x 12.5 = 2.08 GHz
CPU Bus Freq. : 166 x 12.5 = 2.07 (?) GHz
AGP Bus Freq. : 166/2.52 = 66.67 MHz
CPU voltage is 1.65, AGP voltage is 1.53

Now, the CPU frequency is the one I increased, because as I understood it, the 2600 is supposed to run at 2 GHz.
I included the AGP frequency because I don’t know what it is, and if it might be important to answering my questions.

While we’re on the subject, what exactly do the AGP aperture size and on-chip frame buffer size do? I have 512 MB of RAM, of which 64 megs are reserved for video, leaving 448 megs for everything else. The on-chip frame buffer size just happens to be 64 megs… is that the 64 megs reserved for video?

Oh, and the chip temperature immediately after a crash is comfortably under 35 degrees.