MemTest86 v7.5 (Build 1001) released - 21/Feb/2018

[streetwolf]

I notice that v7.5 runs slower than v.74. Also Patterns and State don't change as often as they used to. Is this true with v7.5?

[David (PassMark)]

There should be no significant speed difference between V7.4 and V7.5. Did it just feel different, or did you actually time it? The actual times should also be available in the log file.

[streetwolf]

There should be no significant speed difference between V7.4 and V7.5. Did it just feel different, or did you actually time it? The actual times should also be available in the log file.

It is definitely slower than 7.4 and as I mentioned the States animation is almost nonexistent. A possible reason might be that when I used your USB creator program to overwrite my 7.4 USB drive it failed on me. I manually formatted the device as NTFS and ran the USB creation program. It ran to completion. I'll try a new USB device and see what happens.

[streetwolf]

I built a new USB drive with 7.5 and it ran to completion. It still is slow. I recall that Test 2 ran pretty quick. Under 7.5 it runs slow. I'll have to try 7.4 again to compare.

[David (PassMark)]

A log file (MemTest86.log) is automatically created and updated while MemTest86 is running. This file is saved in the 'EFI/BOOT' directory in the USB drive's first partition.
Can you save the log for this machine for V7.4 and V7.5, then we can compare them.

[streetwolf]

A log file (MemTest86.log) is automatically created and updated while MemTest86 is running. This file is saved in the 'EFI/BOOT' directory in the USB drive's first partition.
Can you save the log for this machine for V7.4 and V7.5, then we can compare them.

First off, can you give me a link to v7.4? Secondly, what tests do you want me to run and how many passes? I don't want to tie up my machine longer than necessary.

[David (PassMark)]

One pass would be enough.
A possibility is that in V7.5 we added some additional motherboards to the blacklist. If your motherboard is on this blacklist, then it will just run single threaded. Which is slower than multi-threaded. The blacklist is to avoid crashing the machine on motherboards that have critical UEFI bugs in their firmware.

What motherboard and firmware version do you have?

Old V7.4 download is here
https://www.memtest86.com/downloads/...86-7.4-usb.zip

[streetwolf]

One pass would be enough.
A possibility is that in V7.5 we added some additional motherboards to the blacklist. If your motherboard is on this blacklist, then it will just run single threaded. Which is slower than multi-threaded. The blacklist is to avoid crashing the machine on motherboards that have critical UEFI bugs in their firmware.

What motherboard and firmware version do you have?

Old V7.4 download is here
https://www.memtest86.com/downloads/...86-7.4-usb.zip

My motherboard is GIGABYTE Z370 AORUS Gaming 7 BIOS F5m.

[streetwolf]

I ran 7.4 and 7.5. I only ran the first 3 tests because of time constraints. 7.4 ran 4 passes in 30 seconds. 7.5 took ~7.5 minutes

Desktop.zip

[keith]

Thanks for the logs.

Test 2 is the start of using all CPU cores in parallel. If possible, can you try running the tests in single CPU mode and see if you get a huge slowdown as well.

Also, are you using the same USB stick for both v7.4 and v7.5 runs?

[streetwolf]

Thanks for the logs.

Test 2 is the start of using all CPU cores in parallel. If possible, can you try running the tests in single CPU mode and see if you get a huge slowdown as well.

Also, are you using the same USB stick for both v7.4 and v7.5 runs?

I use 2 different sticks. If you think the stick is the problem I did install 7.5 on my 7.4 stick. What version of MT do you want me to run in single CPU mode?

[streetwolf]

I assumed you wanted me to use v7.4 and only use one CPU. I did this and Test 2 for example ran very fast. It showed that one CPU was being used and the States animation, the spinning '-' for CPU 0, was spinning. Under 7.5 this would spin very, very slowly.

[streetwolf]

I did some testing with v7.5. Used Test #2. Single CPU was fast. Round Robin was fast. Sequential and Parallel were slow. Also of note is that during and even after running a test I very often could not use Esc or c to exit out of the test. It appears my machine was hung. Even if the test ran successfully, often I could not enter any of the numbers on the resulting menu to get me back to the main menu.

[streetwolf]

I previously said that running with one CPU using v7.5 was fast like v7.4. I ran test #6 and 7.4 is twice as fast as 7.5.

[David (PassMark)]

We were not able to reproduce the problem here, V7.5 was in fact very marginally faster than V7.4.
Your problem might be a subtle version of the known multi-threaded UEFI BIOS bugs. Gigabyte did have a heap of bad boards. Hangs and restarts were some of the symptoms. Doesn't explain the V7.4 / V7.5 difference in behaviour however. Give us a couple of days, we should be able to send you a debug build to confirm if this is the issue.