Is my OC actually working? Asus Rog Strix Z390-I Gaming, i7 9700k

[stasivanov]

So, here is the thing. I have slim case with i7 9700k, so I can't install massive CPU cooler, so I decided to OC default 3.6 GHZ to stable 4.5 Ghz on all Cores. I have setup this all in Bios, and temps are ok. At the same time I'm comparing results with the Intel Turbo-Boost, that gives much more heat, and even throttling.

So, in general I finished with 2 profiles in Bios, one for Turbo-Boost and one for 4.5Ghz on all Cores with manual voltage setting and other stuff.

First I checked everything in Windows, and apart from general stability tests (like OCCT, IntelBurnTest, or Aida) and I have some results in:
Windows 10 Cinebench (4.5 Ghz all Cores)
3630 pts
Windows 10 Geekbench 5 (4.5 Ghz all Cores)
1294 Single-Core, 7909 Multi-Core

Also results for Turbo-Boost mode with full power control by motherboard. I won't use this setup cause it's too hot (it actually reaches 100C and throttles to smth like 4.6-4.7 Ghz), but it's nice to note results:
Windows 10 Cinebench (Turbo-Boost)
3810 pts
Windows 10 Geekbench 5 (Turbo-Boost)
1354 Single-Core, 8300 Multi-Core

After I concluded for myself it's fine by me, I switched to setup Mac OS using this tutorial - https://www.tonymacx86.com/threads/...g-i7-8700k-sapphire-rx580-pulse-build.272572/. After setup kext for sensors, I realised that it seems that Mac OS is slower. And also, it actually feels like it's slower a bit.

First of all, while in Windows with stable 4.5 on all Cores its Task Manager shows that it's really stable 4.5 Ghz, instead in Mac OS it jumps from 1.3 Ghz to 4.5 Ghz. So it seems like the OC ratio set in Bios is just a max limit, and not a static value.
Then I made tests and here are results:
Mac OS 10.15.2 Cinebench (4.5 Ghz all Cores)
3590 pts
Mac OS 10.15.2 Geekbench 5 (4.5 Ghz all Cores)
1156 Single-Core, 7804 Multi-Core
Mac OS 10.15.2 Cinebench (Turbo-Boost)
3701 pts
Mac OS 10.15.2 Geekbench 5 (Turbo-Boost)
1203 Single-Core, 8138 Multi-Core

Also, it's strange but in Cinebench in Mac OS it says that I have 4 Cores and 8 Threads (just in case - i7 9700k has 8 Cores, 8 Threads).

For me it looks weird. I understand that Windows and Mac OS are different platforms, so software can give different results. But how can I make sure that my OC is actually working on the same level as in Windows? I think that the reason of such difference with results may be the point that the frequency is not really stable in Mac OS. Probably it would give better results if my frequency is always 4.5 Ghz.

P.S.: I'm newbie to OC in Mac OS and to OC in general, so please forgive me any stupid questions and comments in advance

 

[Shiitaki]

So, here is the thing. I have slim case with i7 9700k, so I can't install massive CPU cooler, so I decided to OC default 3.6 GHZ to stable 4.5 Ghz on all Cores. I have setup this all in Bios, and temps are ok. At the same time I'm comparing results with the Intel Turbo-Boost, that gives much more heat, and even throttling.

So, in general I finished with 2 profiles in Bios, one for Turbo-Boost and one for 4.5Ghz on all Cores with manual voltage setting and other stuff.

First I checked everything in Windows, and apart from general stability tests (like OCCT, IntelBurnTest, or Aida) and I have some results in:
Windows 10 Cinebench (4.5 Ghz all Cores)
3630 pts
Windows 10 Geekbench 5 (4.5 Ghz all Cores)
1294 Single-Core, 7909 Multi-Core

Also results for Turbo-Boost mode with full power control by motherboard. I won't use this setup cause it's too hot (it actually reaches 100C and throttles to smth like 4.6-4.7 Ghz), but it's nice to note results:
Windows 10 Cinebench (Turbo-Boost)
3810 pts
Windows 10 Geekbench 5 (Turbo-Boost)
1354 Single-Core, 8300 Multi-Core

After I concluded for myself it's fine by me, I switched to setup Mac OS using this tutorial - https://www.tonymacx86.com/threads/...g-i7-8700k-sapphire-rx580-pulse-build.272572/. After setup kext for sensors, I realised that it seems that Mac OS is slower. And also, it actually feels like it's slower a bit.

First of all, while in Windows with stable 4.5 on all Cores its Task Manager shows that it's really stable 4.5 Ghz, instead in Mac OS it jumps from 1.3 Ghz to 4.5 Ghz. So it seems like the OC ratio set in Bios is just a max limit, and not a static value.
Then I made tests and here are results:
Mac OS 10.15.2 Cinebench (4.5 Ghz all Cores)
3590 pts
Mac OS 10.15.2 Geekbench 5 (4.5 Ghz all Cores)
1156 Single-Core, 7804 Multi-Core
Mac OS 10.15.2 Cinebench (Turbo-Boost)
3701 pts
Mac OS 10.15.2 Geekbench 5 (Turbo-Boost)
1203 Single-Core, 8138 Multi-Core

Also, it's strange but in Cinebench in Mac OS it says that I have 4 Cores and 8 Threads (just in case - i7 9700k has 8 Cores, 8 Threads).

For me it looks weird. I understand that Windows and Mac OS are different platforms, so software can give different results. But how can I make sure that my OC is actually working on the same level as in Windows? I think that the reason of such difference with results may be the point that the frequency is not really stable in Mac OS. Probably it would give better results if my frequency is always 4.5 Ghz.

P.S.: I'm newbie to OC in Mac OS and to OC in general, so please forgive me any stupid questions and comments in advance

Overclocking in macOS is the same as Windows as long as you do it in the BIOS. MacOS is very power efficient and will use all the available power saving features. Windows should be ramping the CPU down to 800Mhz when there is no load, but likely Windows is doing ******** or you installed a program hogging resources. Ideally even Windowss should bring the cores down to 800mhz when they are not being used.

Another words, Windows is the one with the problem. Typically you set the turbo boots multiplier to the max speed and then macOS/Windows will hit that only when you need it. Often overclockers will disable the power saving features because it can cause instability, I don't know if you have done this or not. I don't, I prefer my CPU to last for years and see no reason to run the CPU at full throttle when most of the time it isn't doing anything away. The weird core/thread count happens but doesn't seem to cause any problems in macOS.

As long as the benchmarks are reasonable for your overclock then you know the overclock is working in macOS. For that matter that's the best way to verify an overclock is working inward Windows too, using benchmarks, even with stock settings you should run benchmarks to catch issues.

The real issue with overclocking is stability testing. The best test seems to her Ashes of the Singularity, and frustratingly even when overclock settings is stable with half a dozen benchmarks running simultaneously for an hour, Ashes can still crash an hour and a half in to a game. The issue with overclocking is that any crash makes you question the overclock, not sure it is worth it anymore.

 

[stasivanov]

Overclocking in macOS is the same as Windows as long as you do it in the BIOS. MacOS is very power efficient and will use all the available power saving features. Windows should be ramping the CPU down to 800Mhz when there is no load, but likely Windows is doing ******** or you installed a program hogging resources. Ideally even Windowss should bring the cores down to 800mhz when they are not being used.

Another words, Windows is the one with the problem. Typically you set the turbo boots multiplier to the max speed and then macOS/Windows will hit that only when you need it. Often overclockers will disable the power saving features because it can cause instability, I don't know if you have done this or not. I don't, I prefer my CPU to last for years and see no reason to run the CPU at full throttle when most of the time it isn't doing anything away. The weird core/thread count happens but doesn't seem to cause any problems in macOS.

As long as the benchmarks are reasonable for your overclock then you know the overclock is working in macOS. For that matter that's the best way to verify an overclock is working inward Windows too, using benchmarks, even with stock settings you should run benchmarks to catch issues.

The real issue with overclocking is stability testing. The best test seems to her Ashes of the Singularity, and frustratingly even when overclock settings is stable with half a dozen benchmarks running simultaneously for an hour, Ashes can still crash an hour and a half in to a game. The issue with overclocking is that any crash makes you question the overclock, not sure it is worth it anymore.

Thanks for the reply.

Obviously, everybody decides regarding stability testing on his own.

In my case I know that I can’t setup the best cooling needed to not have throttling with turbo boost. Cause with this technology it anyway reaches 100C to me. Even with the fact that it will be throttling only when I push on my cpu, 100C doesn’t sound good to me.
So that’s why I decided to have less frequency than can be provided by turbo boost, but have it stable and less than 80C under full load.

But does it make sense? I mean for this cpu turbo boost can give 4.9Ghz for 1 core and 4.6ghz for all cores. My cpu can’t handle 4.6 ghz on all cores, cause in automatic mode the motherboard gives too much voltage to cpu, so it becomes too hot and throttles to 4.6 first and then even to 4.4 or 4.3 or smth. So I thought actually that core ratio in bios defines a static frequency all the time, but instead it’s just a max value, right? That actually means that there is no sense to set core ratio less than turbo boost can give on all cores (4.6ghz for my cpu)

4.5ghz with 1.190V works pretty fine for me. But in general it’s less powerful than turbo boost with 90+C heat. With turbo boost the motherboard sometimes gives up to 1.4V for cpu, and that’s why it’s so hot. So I’m now wondering which configuration is a golden point for me. Turbo boost with overheat and throttling (assuming that such load will not be active for hours like in stress testings), or what else you would recommend to try/test?