Threadripper Pro potential bug

[auer]

I work with various different systems where Memtest86 comes in handy. I noticed on the most recent release, when using a Threadripper (3xxx to 7xxx pre prod) the channel count will increase by 4 each start. I have gotten it as high as x24 channels. To recreate this, just quit the current test (press c then 2) then start the test again. I have not been able to recreate this on other AMD CPUs or any Intel CPUs. We have not tested Zeon processors.

MB used in TR 39xxx: Aorus TRX40 Pro wifi
MB used in TR Pro 59xxx: ASUS PRO WS WRX80E-SAGE WIFI-SI
RAM in TR 39xx system: KSM32ED8/32HC 32gb x8
RAM in TR PRO 59XX SYSTEM: M393AAG40M32-CAE 128gb x 8

I can't upload .zip files so here is a Goole Drive link with pictures of the issue in some .zip files and uploaded individually. The errors seen are un-related.

[David (PassMark)]

We'll look into it.

[keith]

Thanks for letting us know. We believe we have fixed the issue, can you give this build a try:

https://www.passmark.com/temp/memtes...-10.6.3005.zip

[auer]

Thanks for the quick response. It looks like the issue is still present on both the 3xxx and 5xxx series systems. I have not been able to test the 7xxx series.

[keith]

Thanks for the quick response. It looks like the issue is still present on both the 3xxx and 5xxx series systems. I have not been able to test the 7xxx series.

Thanks for the update. If possible, can you upload a copy of the MemTest86 logs for this new build.

[auer]

Logs are attached.

[keith]

Thanks for the logs.

Sorry, it seems that the previous download link didn't have the correct build (with the fix). Can you try this new build:

https://www.passmark.com/temp/memtes...-10.6.3006.zip

[auer]

That fixed the issue. The channels are now represented properly. Thank you!