- Joined
- Oct 27, 2014
- Messages
- 25
- Motherboard
- Gigabyte GA-Z97X-UD7 TH
- CPU
- Core i7-4790
- Graphics
- GeForce GTX 980 (PNY)
- Mac
- Classic Mac
- Mobile Phone
Hi,
I have been experimenting with KGP's awesome guide for setting up X99-based systems (https://www.tonymacx86.com/threads/...ill-tridentz-aorus-gtx-1080-ti-xtreme.211621/ This guide really should be given top billing on TonyMac because it's the best high-performance HackPro you can build!).
Anyway, I have been running Sierra on X99 for about 6 months with good results using other recipes on tonymacx86.com, but following KGP's guide gave me a big boost in performance, and allowed my systems to work with 10-core (i7 Extreme) and 12-core (E5-2690 v3) CPUs, which I was never able to do when I was following other guides.
One thing that have have learned in my most recent experiments is that the FakeCPUID that I configure in Clover has a huge impact on performance. For instance, if I use the FakeCPUID suggested for Haswell-E processors (0x0306F0), Geekbench shows me a score of about 3000 for a single core of my 12-core 2.6 Ghz E5-2690. On the other hand, if I use the FakeCPUID suggested for Broadwell-E processors (0x040674), I get a score of about 4100 for a single core. Results for multicore are slightly better if I use the Broadwell-E code (score of 33,500) versus if I use the Haswell-E code (score of 32,000).
My question is, what is the relationship between the FakeCPUID and performance? Why do both FakeCPUID's work? Is there a good primer/guide that explains how to get the best performance out of a Hackintosh?
For my applications, I'm not very worried about having sleep work, and I don't mind disabling SpeedStep in the BIOS. I just want the best possible performance while the system is running, and then I shut it off when I'm not using it.
Would love to get some feedback here!
Regards,
AndrewL7332
I have been experimenting with KGP's awesome guide for setting up X99-based systems (https://www.tonymacx86.com/threads/...ill-tridentz-aorus-gtx-1080-ti-xtreme.211621/ This guide really should be given top billing on TonyMac because it's the best high-performance HackPro you can build!).
Anyway, I have been running Sierra on X99 for about 6 months with good results using other recipes on tonymacx86.com, but following KGP's guide gave me a big boost in performance, and allowed my systems to work with 10-core (i7 Extreme) and 12-core (E5-2690 v3) CPUs, which I was never able to do when I was following other guides.
One thing that have have learned in my most recent experiments is that the FakeCPUID that I configure in Clover has a huge impact on performance. For instance, if I use the FakeCPUID suggested for Haswell-E processors (0x0306F0), Geekbench shows me a score of about 3000 for a single core of my 12-core 2.6 Ghz E5-2690. On the other hand, if I use the FakeCPUID suggested for Broadwell-E processors (0x040674), I get a score of about 4100 for a single core. Results for multicore are slightly better if I use the Broadwell-E code (score of 33,500) versus if I use the Haswell-E code (score of 32,000).
My question is, what is the relationship between the FakeCPUID and performance? Why do both FakeCPUID's work? Is there a good primer/guide that explains how to get the best performance out of a Hackintosh?
For my applications, I'm not very worried about having sleep work, and I don't mind disabling SpeedStep in the BIOS. I just want the best possible performance while the system is running, and then I shut it off when I'm not using it.
Would love to get some feedback here!
Regards,
AndrewL7332