ThrottleStop overwriting per-core clock configuration

[pantsaregood]

While using ThrottleStop to set Speed Shift EPP value, I observed that it appears to be overwriting the maximum turbo frequency of a number of cores (all except 17 and 18) with the multiplier of the highest core. This is on the X299 platform, so the difference in the multiplier that can be handled by the different cores can be fairly large.

For example:

If I set cores 0, 2, 5, 7, 9, 12, 13, and 14 to have a maximum multiplier of 50 and 1, 3, 4, 6, 8, 10, 11, 15, 16, and 17 to have a maximum multiplier of 48, loading all cores will reflect the cores operating at the expected multiplier.

If I use ThrottleStop 9.7.2 to modify the Speed Shift EPP value, cores 0-15 will use a 50 multiplier under load and leave 16 and 17 using a 48 multiplier.

This is a weird platform, but Windows doesn't appear to attempt to control EPP on its own - it just sits at 128 unless I use ThrottleStop.

 

[unclewebb]

X299 platform

The X299 platform is a platform that I have never owned or even used. It is almost impossible to write software like ThrottleStop without access to any hardware for testing purposes or access to the full Intel documentation.

Parts of ThrottleStop might work OK on X299 but other parts of ThrottleStop might not work at all or there might be bugs like you have found. I have no plans and no money to ever add full X299 support to ThrottleStop.

Perhaps you can use Quick CPU to adjust the Speed Shift values without any interference to the per core multiplier values.