The_Jager Posted January 6, 2005 Posted January 6, 2005 In an effort to trouble shoot my problem and alot of others with HL2 crashing with a "Memory could not be read" error which is documented here and a lot of other forums I reran Mem86 and Prime95 to see if everything was still in good shape. I reran Mem86 for 6 hours without any errors. In the past I have only ran the "Blend" in Prime95. I reran this overnight all ok. The next night I ran the "Small FFT" all ok. The next night the "In Place Large FFT's". It returned a "Rounding was .496 was expecting less than 0.4" error after 6 hours. I took the side off the case and reran it and also monitored the enviroment with SisSoftware. It failed after 2 hours with the same type of message. The temperature of the CPU, board & power just prior to the "hardware failure" was 55, 49, & 41. Right where it was running the whole test. What is different about this particular Prime 95 test and how would I go about isolating the problem? Is it a problem? Share this post Link to post Share on other sites More sharing options...
dmon Posted January 6, 2005 Posted January 6, 2005 I also fail that one. Share this post Link to post Share on other sites More sharing options...
ViperJohn Posted January 6, 2005 Posted January 6, 2005 Originally posted by The_Jager In an effort to trouble shoot my problem and alot of others with HL2 crashing with a "Memory could not be read" error which is documented here and a lot of other forums I reran Mem86 and Prime95 to see if everything was still in good shape. I reran Mem86 for 6 hours without any errors. In the past I have only ran the "Blend" in Prime95. I reran this overnight all ok. The next night I ran the "Small FFT" all ok. The next night the "In Place Large FFT's". It returned a "Rounding was .496 was expecting less than 0.4" error after 6 hours. I took the side off the case and reran it and also monitored the enviroment with SisSoftware. It failed after 2 hours with the same type of message. The temperature of the CPU, board & power just prior to the "hardware failure" was 55, 49, & 41. Right where it was running the whole test. What is different about this particular Prime 95 test and how would I go about isolating the problem? Is it a problem? Bump the Vcore .025 volts. The rounding error is coming from the FPU in the CPU. Viper Share this post Link to post Share on other sites More sharing options...
The_Jager Posted January 8, 2005 Posted January 8, 2005 Originally posted by ViperJohn Bump the Vcore .025 volts. The rounding error is coming from the FPU in the CPU. Viper Thanks for your reply.Ok I did this and reran Prime95 Inplace Large FFT's overnite. It was running when I got up. Went to put the coffee on and when I came back after 9 hrs 26 minutes it halted with an "Error: Illegal Sumout". Does this narrow things down any as to what the poblem might be? Would you venture a guess on where the most likely place to start? Share this post Link to post Share on other sites More sharing options...
ViperJohn Posted January 8, 2005 Posted January 8, 2005 Originally posted by The_Jager Thanks for your reply.Ok I did this and reran Prime95 Inplace Large FFT's overnite. It was running when I got up. Went to put the coffee on and when I came back after 9 hrs 26 minutes it halted with an "Error: Illegal Sumout". Does this narrow things down any as to what the poblem might be? Would you venture a guess on where the most likely place to start? That is again an FPU math error. At 9 hours and 26 minutes you are looking stable though. Viper Share this post Link to post Share on other sites More sharing options...
The_Jager Posted January 10, 2005 Posted January 10, 2005 I manually entered my memory timings in the BIOS and upped the memory voltage to 2.7. Reran Prime 95 and ran 10 hours w/o failure. Best of all I plalyed half life 2 three times each tiime in excess of an hour and no "cannot read memory" crash. If others are having this problem with HL2 they may try this it worked for me. Share this post Link to post Share on other sites More sharing options...
IvanAndreevich Posted January 10, 2005 Posted January 10, 2005 ViperJohn You sound pretty confident about the "FPU" etc. Got a link? Share this post Link to post Share on other sites More sharing options...
Covert_operatoR Posted January 10, 2005 Posted January 10, 2005 its the ONLY part that does any work. your cpu just merely becomes a big calculator crunching rather large sets of large numbers Share this post Link to post Share on other sites More sharing options...
pharmd24 Posted January 10, 2005 Posted January 10, 2005 What version of Prime95 are you using? As I understand it, versions prior to 23.8.1 have problems with 1gb of memory. Of course I have only had experience with using the blend test...... Share this post Link to post Share on other sites More sharing options...
Korrad Posted January 10, 2005 Posted January 10, 2005 Jager, I had similare case of prime instability at stock that cleared up with increasing Vcore (48 + hrs). However, once I started to overclock I lost prime stability again and upping vcore didn't help. Tried increasing vdimm to with no affect. I haven't really played further with getting prime stable on oc. I am guessing that I have to adjust my memory timing parameters. Just thought I would mention it as a headsup in case you run into the same thing. Share this post Link to post Share on other sites More sharing options...
IvanAndreevich Posted January 11, 2005 Posted January 11, 2005 >>its the ONLY part that does any work Oh really? What about the ALU heh. Share this post Link to post Share on other sites More sharing options...
psy^ Posted January 11, 2005 Posted January 11, 2005 Originally posted by IvanAndreevich >>its the ONLY part that does any work Oh really? What about the ALU heh. Prime95 really works the ALU in the test in question? Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now