Tweet
no errors with memtest86+ / plenty of errors with memtest86
I am experiencing no errors with the latest memtest86+ but when I use the latest memtest86 version I get plenty of errors. Which one should I trust ? This should not happen. I have ran both programs several times and always get the same difference in error reporting which I consider surprising. Looking for your feedback.
-
-
Can't really comment without knowing more details.
Maybe you can E-Mail or post us the details of the errors you are seeing, plus the details of the hardware.
What I can say is that the original MemTest86 (without the plus) is being actively supported with new features like ECC and UEFI. But there has been nearly no development done on MemTest86+ recently. -
hardware: Acer Aspire 5050 laptop / CPU AMD Turion 64 Mobile MK-36 (2 GHz) / RAM 1 GByte (this is a 2005 year computer). Memtest-86 v4.1.0 always crashes at Time 0:04:55 with 101 errors (errors are split between tests 3 and 8 with most errors 260 happening in test
.
Comment
-
Would have been interesting to see the full details of the errors. Especially the memory address ranges.
If the memory addresses for the errors are > 1MB, then I would think they are real errors. If all the errors are under the 1MB range it isn't so clear.
Was the machine completely stable in normal use?Comment
-
I see you have posted the same question elsewhere on other forums, with additional information that you didn't include here.
You initially said you were using the latest MemTest release. But in fact you were using the older MemTest86 V4.1 release and didn't to mention here that when you used MemTest86 V4.2 you didn't get any errors.
Based on the V4.2 results I would think the RAM is OK.Comment
-
memory address ranges that are giving errors with Memtest86 v4.1.0: Lowest Error Address 0.0 MB, Highest Error Address: 0.0MB. The machine is stable and I have no reasons to think it is not. I just wanted to make sure that I can use it for important work. I ran Prime95 without problems.Comment
-
I was using v4.1.0 that was the most recent until some days ago when v.4.2.0 was released (of which I was not aware). Still think that v.4.1.0 released January 2013 should be reliable to test on a 2005 platform, don't you agree ?Comment
-
I would like to add that I have initially tested this computer with Memtest86 v1.70 and got several errors also on Failing Address 0.0MB. So it seems that for some reason Memtest86 has not been able to work correctly for several years on 2005 Acer/AMD hardware, which I find a bit odd.Comment
-
More information that I find interesting/odd: this AMD Turion 64 Mobile Mk-36 CPU which has L1 Cache of 64K is shown by all versions of Memtest86 to have 128K L1 Cache which is incorrect.Comment
-
We only started maintaining MemTest86 recently (with the V4.2 release).Still think that v.4.1.0 released January 2013 should be reliable to test on a 2005 platform, don't you agree ?
I think there are a bunch of edge cases (rare hardware, Intel based Macs, some UEFI machines, etc..) that MemTest86 (and the Plus fork) doesn't deal with very well.
We are trying to fix these up as we become aware of them.
The specs seem to say that the CPU have,AMD Turion 64 Mobile Mk-36 CPU which has L1 Cache of 64K is shown by all versions of Memtest86 to have 128K L1 Cache which is incorrect.
64 KB 2-way associative instruction cache
plus
64 KB 2-way associative data cache
Which makes 128 KB in total. I haven't checked but I assume the total of instruction and data caches are used for all CPU types.
Comment
Comment