[SOLVED] How dependable is Passmark Performance Test results?

[clutchc]

In particular, the CPU test. I am getting results that show the old FX-8350 has nearly the same performance as the old i7-3770 across all tests. In fact, slightly faster in some. Back in the day that they were competing, that wouldn't seem to have been the case.

 

[CerianK]

They should be fairly similar in most tests, as AMD's 8 SMT hardware cores/4 modules will make up for ~10% lower single thread performance against the Intel's 8 virtual HT threads/4 cores.

As software/OSes have evolved to take advantage of more physical cores, the AMD would likely benefit more.

Newer versions of Passmark have rescaled slightly to take this into account, as the weighting of all of the individual tests toward calculating the final CPU score has changed:
(V8 and V9 calculations) https://www.passmark.com/forum/performancetest/4599-formula-cpu-mark-memory-mark-and-disk-mark
(older useless response) https://www.passmark.com/forum/performancetest/3312-cpu-mark-calculation

 

[stdragon]

I'm going to guess those results are current, meaning post Meltdown and Spectre which would have a significant impact on Ivy Bridge cores.

Only way to be sure is to disable Meltdown and Spectre mitigation in the registry, reboot (for changes to take effect), and then test again.

 

[prince_xaine]

I'm going to guess those results are current, meaning post Meltdown and Spectre which would have a significant impact on Ivy Bridge cores.

Only way to be sure is to disable Meltdown and Spectre mitigation in the registry, reboot (for changes to take effect), and then test again.

How do you do this? I would like to know for a couple of my older computers.

And I wouldn't look too deep into Passmark's performance results. It seems a larger number of people use Time Spy, meaning even some medium-high gaming computers can get rather high-end results. It's great for comparing performance between components though, although I prefer userbenchmark for this.

 

[clutchc]

I tend to agree with the theory that Passmark is weighted in the wrong way. And maybe the Spectre/Meltdown fixes are indeed causing issues. I just ran my older V7 of Passmark as Ceriank suggested and it's worse. The FX-8350 (CPU only) destroys the the i7-3370. That ain't right!

Maybe I'll run Time Spy and Firestrike using the same card and see how that compares. If anyone's interested, I'll post the Passmark V9 CPU test results too.

 

[stdragon]

I'm going to guess those results are current, meaning post Meltdown and Spectre which would have a significant impact on Ivy Bridge cores.

Only way to be sure is to disable Meltdown and Spectre mitigation in the registry, reboot (for changes to take effect), and then test again.

How do you do this? I would like to know for a couple of my older computers.

And I wouldn't look too deep into Passmark's performance results. It seems a larger number of people use Time Spy, meaning even some medium-high gaming computers can get rather high-end results. It's great for comparing performance between components though, although I prefer userbenchmark for this.

GRC provides a GUI utility 'InSpectre' that makes the registry changes for you. Once you click on the corresponding buttons, you'll have to reboot the computer for changes to take effect.

https://www.grc.com/inspectre.htm

And if you're brave, you can do it yourself. Again, reboot required after changes made.

https://support.microsoft.com/en-us/help/4072698/windows-server-speculative-execution-side-channel-vulnerabilities-prot

 

[prince_xaine]

I'm going to guess those results are current, meaning post Meltdown and Spectre which would have a significant impact on Ivy Bridge cores.

Only way to be sure is to disable Meltdown and Spectre mitigation in the registry, reboot (for changes to take effect), and then test again.

How do you do this? I would like to know for a couple of my older computers.

And I wouldn't look too deep into Passmark's performance results. It seems a larger number of people use Time Spy, meaning even some medium-high gaming computers can get rather high-end results. It's great for comparing performance between components though, although I prefer userbenchmark for this.

GRC provides a GUI utility 'InSpectre' that makes the registry changes for you. Once you click on the corresponding buttons, you'll have to reboot the computer for changes to take effect.

https://www.grc.com/inspectre.htm

And if you're brave, you can do it yourself. Again, reboot required after changes made.

https://support.microsoft.com/en-us/help/4072698/windows-server-speculative-execution-side-channel-vulnerabilities-prot

Thanks!