Errors during Torture Test
 

I need some help. I am running a P4 2.26G with 512 of 1066 on a Gigabyte GA-8IHXP. I ahve been getting 2 different errors. 1st of them is Fatal Error: Reading from Temp File. The 2nd error "Fatal Error: Rounding was .4989318848, expected less then .4, Hardware failure detected." I have already replaced the RAM and I am getting the same thing. I have flashed the BIOS to F7, the lastest. I have formated a drive and installed win2k on it and tried to Run it and the errors are still there. I have ver 22.12 of Prime. I used to get alot of BSOD but after the flash they went away. And I jusst ran superPI and I got one error "error in round" on 1 million digits. Anyone have a clue which way to go?

There are a variety of possible causes:

1) A bad CPU. If you can try underclocking the chip and see if things get better.
2) Overheating. This is more likely if the problem only occurs after prime95 has run for a while and the temps are more than 60C. Check your temps using one of the freely available programs.
3) Inadequate power supply. Check you voltages using a free utility program.
4) A bad motherboard. I've no idea how you'll prove this is the cause.
5) Bad memory. Try underclocking memory - that is running it as PC800 memory - look for a 3/4 memory setting in BIOS.

It will not be easy to find the cause. If this is a new computer and under warranty, you might consider returning the whole unit. If you've assembled the unit from parts - well... good luck.

Most likely it is a faulty cpu.

I had the same problem with one of my AMD. I plugged in a new one. Never saw it again.

Get memtest98 and run it overnight just to rule out bad memory.

It is Winter time so it is unlikely the heat. Even if it is, it would be either the cpu or the cpu cooling. Intel cpu is supposed to run cool so that part is less likely. do you overclock by any chance?

I ran into a "fatal error" on the temp file couple days ago - turns out my disk drive on that machine had managed to fill up (I *hate* useless logger programs, and programs that don't let *you* specify where the logs go).

I think it was a temp file error on write, though - not read.

I RMA'd my motherboard. In the mean time I have put my old 500Mhz P3 back together and it runs fine. I hope I don't have to RMA the CPU as well. Thanks for the help everyone.

Drivers, drivers, drivers.

A person just had this problem on our team and all he had to do was upgrade his video card drivers.

It was an ATI video card. For some reason, ATI has problems with their drivers, that if it is not the perfect driver for that vid card, it will produce errors.

Well, I have replaced everything but the CPU at this point. Has anyone ever sent a CPU back to Intel? What do they want to hear?

[quote="sjhanson"]Has anyone ever sent a CPU back to Intel? What do they want to hear?[/quote]

I sent one to AMD. Their TSC line is polite and capable. I described my problem well and they took it well. They assigned me an RMA number right away and emailed instructions on RMA process within seconds. I dont know if they have done any testing on the chip but replace is shipped in about 3 days after they received. They gave me a 1900+ for my 1800+ although I am much prefer getting a TB-A core 1800+.

I told them FPU failed torture testing. I mentioned prime95 but they may have their ways.

Key is to buy the warranty. Retail package in other words.

I ran the torture test again last night and got an error I have not had before, does anyone now what this means? "Fatal Error: Final result was E83AD81D, expected: 3EB29644". The test only ran for 10 min and got to test 27 of the 1024k FFT.

Well, looks like you have a problem. You should consider returning the CPU.

When the torture test happens Prime95 runs some tests whose results are known. In your case, in case 27 the result was not what is expected. Definitely a CPU problem.

I had this occur recently on a rock solid machine.

It occurred in the 1792K FFT range. It does not occur in any other range.

So far I haven't been able to fix the problem. This machine does have an ATI vid card though. I am afraid this may be causing the problem.