OC 0.7.6 radeon pro w5700 metal score was 64000, openCL was 54000 now OC 0.7.7 metal score 68000, openCL 57000What were your previous scores Gustav?
OC 0.7.6 radeon pro w5700 metal score was 64000, openCL was 54000 now OC 0.7.7 metal score 68000, openCL 57000What were your previous scores Gustav?
Can anyone comment or provide their opinion on the performance of their cMP across macOS Mohave, Catalina, Big Sur, and Monterey?
cMP4.1/5.1 dual xeon5680, evo plus 1TB, Crusial NVME P2 2TB, UCB3 pci card, RadeonPro W5700. I have been using OC for almost a year and a half, starting with version 0.5.8. my experience shows that the best CPU performance was on Mojave without open core, the weakest operating system was Catalina - and the CPU and GPU worked worse than on Mojave, although Catalina allowed to use Radeon Navi, but starting with version 0.7.0 and BigSur update The CPU began to work much better, almost like in Mojave, but the GPUs began to show significant performance gains. Gikbench5 Metal score on Catalina for the Radeon Pro W5700 was 60,000-61000, when switching to Big Sur Metal score became 63000, and with version 0.7.2 and BigSur 11.3, the drivers for Navi improved and the score became 64000, but with upgrade to Monterey with OC version 7.5 or 7.6 the score became 65000, today I update OC 7.7 and the geekbench5 metal score became 68000. BlackmagicRAW in BigSur 11.2 was 980, in BigSur 11.6 was 2000, for Monterey with OC 0.7.7 2032. But LuxMark BigSur and Monterey are about the same 2880 and 2910. Heaven benchmark with ultra settings for RadeonProW5700 -2205- for Monterey+OC 0.7.7, -2009 for OC 0.7.6 for BigSur score was about 1900, and i have comparison of Monterey and Catalina for radeon RX580 (sapphire dual bios)- high settings: for Mojave -1978, but for Catalina only -1740. As to real rendering with Twinmotion and UE4.7 or UE5 FPS performance increases with monterey significantly (about 15% compared to BigSur). some pictures attached.Can anyone comment or provide their opinion on the performance of their cMP across macOS Mohave, Catalina, Big Sur, and Monterey? Have people noticed if any specific macOS and OC combination performs significantly better than others? I'm considering refreshing my system but I'm not sure if there is a best option. I expect answers will of course be subjective and differ based on hardware specifics but I thought the feedback might be helpful to consider as a starting point.
Here's a snippet of the pertinent part of the config:Can you explain somthing more about this?
<key>UEFI</key>
<dict>
⋮
<key>Drivers</key>
<array>
⋮
<dict>
<key>Arguments</key>
<string>--gpio-setup</string>
<key>Comment</key>
<string>Audio driver</string>
<key>Enabled</key>
<true/>
<key>Path</key>
<string>AudioDxe.efi</string>
</dict>
⋮
</array>
Yes.could we hold only english and/or our language audio files?
It was never really needed. Spoofing BIOSVersion is a superfluous practice, the effectiveness of which is questionable. In fact, it does not prevent the staging of unwanted firmware updates (MultiUpdater.efi and related files will still appear on the ESP). Ultimately, it is BlacklistAppleUpdate that does the blocking.No more need to spoof biosversion number in OC 0.7.7?
It was never really needed. Spoofing BIOSVersion is a superfluous practice, the effectiveness of which is questionable. In fact, it does not prevent the staging of unwanted firmware updates (MultiUpdater.efi and related files will still appear on the ESP). Ultimately, it is BlacklistAppleUpdate that does the blocking.
More problematic with the classic Mac Pro is the data that inevitably gets written to the NVRAM during macOS installations. That's why it's important to maintain the health of the firmware chip by periodically reflashing a clean BootROM.
Nice improvement!OC 0.7.6 radeon pro w5700 metal score was 64000, openCL was 54000 now OC 0.7.7 metal score 68000, openCL 57000
Looking better inside Audio folder, there are two family of files: AXEFIAudio and OCEFIAudio, the latter with only english and russian. I would like to know if I'm right, guessing that the OCEFIAudio_en files are all we need.Yes.
I'll need tu study about this... any tip/link is welcome.More problematic with the classic Mac Pro is the data that inevitably gets written to the NVRAM during macOS installations. That's why it's important to maintain the health of the firmware chip by periodically reflashing a clean BootROM.
Even with ExposeSensitiveData set to 2, I won't get boot-path (used in the command that mounts EFI partition). What are the advantage of holding ExposeSensitiveData lower than 3?This is expected if you've set ExposeSensitiveData to 0. If you want to see the version number, you can set it to 2.
Yes, the OCEFIAudio files should be enough for navigating the OC boot menu. However, you may also want to keep the AXEFIAudio files (some of which are not localized) for audio in FileVault.Looking better inside Audio folder, there are two family of files: AXEFIAudio and OCEFIAudio, the latter with only english and russian. I would like to know if I'm right, guessing that the OCEFIAudio_en files are all we need.
This is a very popular topic. SeeI'll need tu study about this... any tip/link is welcome.
Both the APPLE folder and the Update container are normal. Regarding "dedicated container on ESP", I suppose you mean installing both Monterey and Big Sur in the same APFS container. This can be done, but it might be problematic because different versions of macOS can use different versions of APFS. Better to use separate containers.So I restarted the Mac with the power button, and found Apple folder inside EFI, a new "Update" container and about 16GB of files in destination container. For now I removed the Apple folder in EFI and the "Update" container, but now I wonder if it's safe to install another macOS in a dedicated container on ESP (I thought to install both, BigSur and Monterey to verify which troubles I'll get with my workflow)
In that case, use 3.Even with ExposeSensitiveData set to 2, I won't get boot-path (used in the command that mounts EFI partition).
Minimize the traces of OpenCore for security reasons.What are the advantage of holding ExposeSensitiveData lower than 3?
Hi @cdf, any updates on this? I have not been able to boot linux on my machine, only faced with a grey screen unfortunately.
I have EndeavourOS (Archlinux using .efi bootloader), which appears and boots in rEFInd, but does not appear in the OC bootpicker.How far did you get?
1. Use OpenCore boot picker to select Linux installer from USB [which variant of Linux?]
2. Linux installer starts okay
3. Install Linux using Linux installer [as recommended, just leave GRUB - or whatever your Linux's default bootloader is - do not unselect or disable it]
4. Reboot system
5. Newly installed Linux appears in OC picker
6. On selecting newly installed Linux, it starts and runs [I know it was not as far as this one]
I have EndeavourOS (Archlinux using .efi bootloader), which appears and boots in rEFInd, but does not appear in the OC bootpicker.
Thanks!It was never really needed. Spoofing BIOSVersion is a superfluous practice, the effectiveness of which is questionable. In fact, it does not prevent the staging of unwanted firmware updates (MultiUpdater.efi and related files will still appear on the ESP). Ultimately, it is BlacklistAppleUpdate that does the blocking.
More problematic with the classic Mac Pro is the data that inevitably gets written to the NVRAM during macOS installations. That's why it's important to maintain the health of the firmware chip by periodically reflashing a clean BootROM.
I haveReally stupid question, but you have installed driver OpenLinuxBoot.efi in OpenCore, right? If so, feel free to send a zipped debug log and I can have a look.
OpenLinuxBoot.efi
installed yes. But Linux showed up in the picker only after BlessOverride setup.Why would you expect opencore to be related to the video card performance? It is just a bootloader.Can any of the opencore gurus explain what could be causing this? After replacing my old RX580 8GB with radeon pro W5700, geekbench5 single core CPU performance dropped from 650 to 570 and multicore score from 6800 to 6400, which worsened the cpu rendering time in CinebenchR20-23 or corona render tests by almost 10%. But since now programs use GPU rendering, I'm not worried much. Moreover, Radeon Pro allows other tasks to be performed in parallel during rendering (RX580 8GB used all the resources and did not allow working on the cMP during rendering). That's why I was almost a year ago I replaced RX580 Sapphire with the RadeonPro W5500 - an equivalent replacement was obtained, but with more comfortable work in ArchViz, this replacement did not reduce the performance of the CPU, But when I almost immediately bought a more productive AMD RadeonPro W5700 to improve performance, the Geeckbench5 CPU tests showed a decrease in performance. I thought that I probably redistributed the processor resources due to the fact that the GPU is more modern and is not intended for the pci 2.0, although, strangely enough, radeon pro W5500 at the same time does not degrade the performance of the CPU.
what i noticed:
1. with RadeonPro W5700, there is always a 40% kernel task on one of the cores
2. temperature of process A in idle mode 50C, and B 37C(with W5500 or RX580 temperature indicators 40C and 32C)
3. graphic card path looks like PciRoot(0x0)/Pci(0x3,0x0)/Pci(0x0,0x0)/Pci(0x0,0x0)/Pci(0x0,0x0)
cMP4.1/5.1 dual xeon5680, evoPlus 1tb, usb3 pci card. OC 0.7.3- 0.7.4-0.7.5-0.7.6-0.7.7, Mac OS Monterey (optional BigSur, Mojave)
I read that if the hardware is not configured correctly, then since the cMP works in monterey like a hackintosh, then i ask: may be the hardware acceleration settings work incorrectly with kexts and with the system?Why would you expect opencore to be related to the video card performance? It is just a bootloader.
AMD hardware acceleration is the named "incorrectly" used here for acceleration of video conversion, compression, pixel transfer and color grading by the Apple VideoToolbox API.I read that if the hardware is not configured correctly, then since the cMP works in monterey like a hackintosh, then i ask: may be the hardware acceleration settings work incorrectly with kexts and with the system?
cMP 4.1/5.1 OC 0.7.7 mac os Monterey
Martin Lo answered me "if both computers show the same behaviour. It may be as simple as the GPU driver itself causing the issue (or one of the software that you installed on both of your computer. Not necessary the GPU's hardware problem.)" i try to understand, may be due to the big difference in the specifications amd radeon pro and mac edition radeon pro mac pro does not correctly allocate resources, because the specifications of the cards are different MAC edition has 40 units, 2560 processors, 1860MHz, 250W, and a simple AMD card has 36 units, 2304 processors, 1930MHz 205WAMD hardware acceleration is the named "incorrectly" used here for acceleration of video conversion, compression, pixel transfer and color grading by the Apple VideoToolbox API.
The ASICs that VideoToolbox API have direct access have absolutely nothing to do with 3D/OpenCL/Compute/shaders. Enabling or disabling VideoToolbox won't change your 3D benchmark results. The ASICs are a distinct part of the GPU.
This has nothing to do with OpenCore or even the GPU ASICs that provide VideoToolbox hardware acceleration for video operation, but with the Apple NAVI drivers for the 1x family. Also, the number of CUs is not important, drivers know how to scale dinamically/workaround it.Martin Lo answered me "if both computers show the same behaviour. It may be as simple as the GPU driver itself causing the issue (or one of the software that you installed on both of your computer. Not necessary the GPU's hardware problem.)" i try to understand, may be due to the big difference in the specifications amd radeon pro and mac edition radeon pro mac pro does not correctly allocate resources, because the specifications of the cards are different MAC edition has 40 units, 2560 processors, 1860MHz, 250W, and a simple AMD card has 36 units, 2304 processors, 1930MHz 205W
Thank you for replyThis has nothing to do with OpenCore or even the GPU ASICs that provide VideoToolbox hardware acceleration for video operation, but with the Apple NAVI drivers for the 1x family. Also, the number of CUs is not important, drivers know how to scale dinamically/workaround it.
Btw, just an anecdote - macOS Polaris drivers performance was dreadful from 10.2.6 to 10.14.4, only after almost two years since first supported Apple really unlashed the power of a RX 580 GPU with 10.14.5.
Anyway, this is off topic here since it's not related to OpenCore in any way, AMD hardware acceleration is for the video ASICs of the modern AMD GPUs, not the shaders/CUs.