The setting that stresses memory is a joke if you are testing your proc. It only uses around 3-5% of your proc but most of your memory. If you used the small files test it would have errored out with a quickness.
They have S&M 1.51 out now and I believe that no other program can heat up a CPU like S&M 1.51 on the FPU setting. It beats Prime95, CPU Burn-in and even CPU Burn. It takes my CPU about 3C higher than CPU burn which is the second best. However, S&M 1.51 cannot crash my system when Prime95 on the small files (setting #1) can do it in a matter of minutes. I kind of use all of these programs to test for different things. One program cannot do it all IMHO.
So far it is OK as a temp producer
I set everything check..test is long with 100% cpu...max temp is 53.5c but seems to stay around 49's....holding 100% cpu utilization....I will wait untl whole test is done but I dont think I will see nay issues cause heat wasn't an issue for me...
sidenote...my prime95 temp at this speed is 50c with a occasional flash of 51c so I was probably like a 50.2 or someting..that is using speedfan which is also what S&M uses...
FH gets my cpu to 49-50c so basically I am not see what is so grand about this...Sounds like a bunch of morons in the limk running the blende test of prime and then running this...blended test usually only gets me to 48c max at this temp....either waY I am seeing no great swings as they have stated....
Update: test completed fine JPEYTON!!! NO fast crash here!!!
Max temp I saw was 53.9c which is about 3c higher then my highest p95 temp....
I may use it in the future but seems not long enough perhaps....How can you make this thing run like 24 hours??? Was there a settting I didn't see???
- Aug 23, 2003
- 25,375
- 142
- 116
Edited to denote version 1.5; and S&M is a good way to do extreme load testing of your power and cooling systems. A weak power supply will definitely make it error out; as will poor CPU, memory, or motherboard cooling (caps get burning hot too).
i dont think s&m is a good stability tester....why? cuz it gets "unusually" hot...hotter than a system would EVER get from any tasks i can even think of
what other program or game would get your system that hot?
since you cant really replicate it (at least i cant think of a way, unless you just have poor caseflow in the first place), it doesn't really necessitate a place in my stability testers toolbox
i can guarantee you that if you slap an extra five to ten degrees celcius to 95 percent of peoples' overclock here that it would cause otherwise "prime stable" overclocks to be worthless.
if you can pass memtest for 24 hours and then prime for 24 hours and say 3dmark or any game looping for 24 hours....thats pretty darn stable, even if its "only" at the temps that those programs are able to muster up...which is pretty much "normal" use
sure maybe during the summertime the temps would be equal to what s&m can cause your system to become but i'll just wait till summertime to drop my overclock
edit: its a damn good cpu burner tho!
what other program or game would get your system that hot?
since you cant really replicate it (at least i cant think of a way, unless you just have poor caseflow in the first place), it doesn't really necessitate a place in my stability testers toolbox
i can guarantee you that if you slap an extra five to ten degrees celcius to 95 percent of peoples' overclock here that it would cause otherwise "prime stable" overclocks to be worthless.
if you can pass memtest for 24 hours and then prime for 24 hours and say 3dmark or any game looping for 24 hours....thats pretty darn stable, even if its "only" at the temps that those programs are able to muster up...which is pretty much "normal" use
sure maybe during the summertime the temps would be equal to what s&m can cause your system to become but i'll just wait till summertime to drop my overclock
edit: its a damn good cpu burner tho!
VirtualLarry
No Lifer
- Aug 25, 2001
- 56,571
- 10,206
- 126
My personal guess is that he was referring to the original "Memtest86", and not the newer "Memtest86+". I've had one occasion thus far where the original did not pick up any memory issues, whereas the newer version, and other diagnostic programs did, so I don't really trust the original Memtest86 anymore myself. I'm guessing that by "Windows Memtest" they were referring to MS's Windows-based memory-testing tool, which I think actually tests memory via execute-based fetches.
Doesn't matter...the fact is the windows based one only test unused ram and since about 1/4 of moine is tied up at bootup then it is not testing all the ram...NOt effective enough....
VirtualLarry
No Lifer
- Aug 25, 2001
- 56,571
- 10,206
- 126
Ok, I guess what I was trying to say is that:
1) I agree with you that the Windows-based memory-tester offers only incomplete coverage of the DRAM address space, however the testing that it does perform on that address space is generally effective.
2) The original Memtest86 may have complete coverage, but based on my experiences, at least for a KT400-based system, the tests are useless and utterly ineffective.
3) The newer Memtest86+, which has specific support for newer systems/chipsets, has both complete coverage, and effective testing.
So while 1) may not be effective enough, it's still marginally better than 2), but really, use of 3), in conjunction with some additional testing tools is probably best. Which is what you suggested as well.
1) I agree with you that the Windows-based memory-tester offers only incomplete coverage of the DRAM address space, however the testing that it does perform on that address space is generally effective.
2) The original Memtest86 may have complete coverage, but based on my experiences, at least for a KT400-based system, the tests are useless and utterly ineffective.
3) The newer Memtest86+, which has specific support for newer systems/chipsets, has both complete coverage, and effective testing.
So while 1) may not be effective enough, it's still marginally better than 2), but really, use of 3), in conjunction with some additional testing tools is probably best. Which is what you suggested as well.
Prime95 is not a good memory stress test. I use it for testing the CPU.
BTW, prime keeps my CPU at 100% even when using 90% of my memory. The only way it will use less than that is on a SMP system, or if you use more memory than your system has available. In the latter case it will be too busy paging to and from the drive to do much of anything.
BTW, prime keeps my CPU at 100% even when using 90% of my memory. The only way it will use less than that is on a SMP system, or if you use more memory than your system has available. In the latter case it will be too busy paging to and from the drive to do much of anything.
i ahve been having problems with blend and posted a thread about it
Post
blend keeps failing no matter what i do, could it be that it is trying to use too much memory, i have had that issue where cpu drop to liek 20% usage and prime never goes to the next step. should i lower the max mem to use and by how much or to what, how would i determine that?
Post
blend keeps failing no matter what i do, could it be that it is trying to use too much memory, i have had that issue where cpu drop to liek 20% usage and prime never goes to the next step. should i lower the max mem to use and by how much or to what, how would i determine that?
In my experience, Prime has been a great stability test. If I had stability problems, Prime would always bring them out. I use it to test overclocks, and I haven't had a crash yet.
i was able to solve a prime blend error, dont know it this helps anyone but i will post it anyway. and this is a legit find, confirmed at the prime forum by the man who wrote prime95.
my post with solution
my post with solution
flexy
Diamond Member
- Sep 28, 2001
- 8,464
- 155
- 106
thats why i prefer OCCT....OCCT is more sensitive than prime...i had situations where prime needs an hour and OCCT complained after 3 mins already.
regarding memory testing:
YES...prime/occt (assuming you use the option to test plenty of your ram) is a GOOD memory test too..actually better than memtest.
memtest can run fine here for hours.....but under the same config i could fai OCCT/prime in a few seconds....memtest is only good for testing whether your mem is defective and whether it gets enough voltage.
(Eg, test #5 is good for testing voltages for your mem..if it passes 15 runs #5 your mem is ok and memory voltages are ok too)
STILL - you might have to tweak timings afterwards to make it run stable in windows - thats when you use OCCT or prime.
regarding memory testing:
YES...prime/occt (assuming you use the option to test plenty of your ram) is a GOOD memory test too..actually better than memtest.
memtest can run fine here for hours.....but under the same config i could fai OCCT/prime in a few seconds....memtest is only good for testing whether your mem is defective and whether it gets enough voltage.
(Eg, test #5 is good for testing voltages for your mem..if it passes 15 runs #5 your mem is ok and memory voltages are ok too)
STILL - you might have to tweak timings afterwards to make it run stable in windows - thats when you use OCCT or prime.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter