MP All Models Tired of low Geekbench scores? Use RadeonBoost!

[borgo1971]

Go to sleep and repeat the test after sleep mode.

I don't understand if you're kidding me, anyway... I specify that I swapped the video card and the nvme drive (Crucial P1) some time ago, because the disk performance was unsatisfactory. Now I reverted the device, and while graphic performance won't be changed much, drive access seems a little poor (my fault, didn't test it before the new swap). Using Blackmagic Disk Speed Test, both read and write decrease from about not enthusiastic 1300 MB/s to less to 500 MB/s after some tests. But it's not this the question now.
But the surprise has come now: I waked remotely from sleep my MacPro, and launched Geekbench test again, and scored 45K in OpenCL and 50K in Metal test. So this raises my doubt about your sleep talk... ?
I will remain in doubt until tomorrow, if these results depend only on the fact that I'm connected remotely with Apple Remote Desktop, or if the short sleep was beneficial for the Mac and his gpu.

 

[Ausdauersportler]

I don't understand if you're kidding me, anyway... I specify that I swapped the video card and the nvme drive (Crucial P1) some time ago, because the disk performance was unsatisfactory. Now I reverted the device, and while graphic performance won't be changed much, drive access seems a little poor (my fault, didn't test it before the new swap). Using Blackmagic Disk Speed Test, both read and write decrease from about not enthusiastic 1300 MB/s to less to 500 MB/s after some tests. But it's not this the question now.
But the surprise has come now: I waked remotely from sleep my MacPro, and launched Geekbench test again, and scored 45K in OpenCL and 50K in Metal test. So this raises my doubt about your sleep talk... ?
I will remain in doubt until tomorrow, if these results depend only on the fact that I'm connected remotely with Apple Remote Desktop, or if the short sleep was beneficial for the Mac and his gpu.

Check #308 and the following few posts about this problem. In this time I thought it has to be solved through OC somehow but in the meantime I am sure a proper designed vBIOS is the key when trying to get the best values out of your Polaris based dGPU.

Unfortunately this kind of programming is not an easy task and beyond my abilities and knowledge.

GeekBench is not a real world benchmark. You will have to pick your own standard graphics application and run some real world tests longer than a few minutes (to get rid of data loading from and storing to disk times) before and after sleep. If you see some differences than you know sleeping is crucial for your MacPro, too.

EDIT:
You can even get higher GeekBench results when having an external display connected to the same dGPU. Doing this will drive the core clock up from the low power state. During the benchmark you see no up and down switching of the power state and you get better results (10% on my system, CL 36000 vs 40000 and Metal 41000 vs 44000). I guess it shows which poor benchmark design this tool really has.

 

[arw]

Sorry in advance, if this is the wrong place - I just didn't want to start a completely new thread but can do so, if it's too unrelated.
It is about my RX 570 performing as expected on day, bad on the other.
In Geekbench I get either around 43.000 or 25.000 Metal score - nothing in between. Some boots so, some boots the other.
GPU Core and Memory clock are the same in each scenario (according to iStat and ioreg).
I think I narrowed it down to having HiDPI forced on my single 4K-TV or not but it's not 100% consistent.
I had hoped enabling RadeonBoost in OpenCore would help but it makes no difference.
Any ideas what the reason for this behavior could be?
I have already successfully injected a custom fan curve via Open Core before, so I'd be happy to try something regarding PowerPlay etc., I just have no idea what I could do.
It uses the normal ATY,AMD,RadeonFramebuffer.
To get the HiDPI options in System Preferences > Display in the first place, I had to execute

sudo defaults write /Library/Preferences/com.apple.windowserver.plist DisplayResolutionEnabled -bool true

Thank you!

System: MacPro5,1 2010 Dual x5690 6-core; 48 GB 1333 MHz RAM; MSI RX 570 Armor OC 8GB; Big Sur 11.6; OpenCore 0.7.4

 

[startergo]

Sorry in advance, if this is the wrong place - I just didn't want to start a completely new thread but can do so, if it's too unrelated.
It is about my RX 570 performing as expected on day, bad on the other.
In Geekbench I get either around 43.000 or 25.000 Metal score - nothing in between. Some boots so, some boots the other.
GPU Core and Memory clock are the same in each scenario (according to iStat and ioreg).
I think I narrowed it down to having HiDPI forced on my single 4K-TV or not but it's not 100% consistent.
I had hoped enabling RadeonBoost in OpenCore would help but it makes no difference.
Any ideas what the reason for this behavior could be?
I have already successfully injected a custom fan curve via Open Core before, so I'd be happy to try something regarding PowerPlay etc., I just have no idea what I could do.
It uses the normal ATY,AMD,RadeonFramebuffer.
To get the HiDPI options in System Preferences > Display in the first place, I had to execute

sudo defaults write /Library/Preferences/com.apple.windowserver.plist DisplayResolutionEnabled -bool true

Thank you!

System: MacPro5,1 2010 Dual x5690 6-core; 48 GB 1333 MHz RAM; MSI RX 570 Armor OC 8GB; Big Sur 11.6; OpenCore 0.7.4

Always test it after sleep/wake cycle.

 

[mani girafe]

hello, I would like to know if radeonboost works without oc, I am under mojave on mcp 4.1.

 

[arw]

Always test it after sleep/wake cycle.

Yes, this "solves" the issue, but sometimes my Mac freezes/panics after wake.
However: At least in my case I seem to have found a workaround/solution. It is unrelated to RadeonBoost and I have no idea why macOS behaves like that - but it does:
- After a reboot, my scores are always fine
- The problem can be 100% triggerd by switching the resolution from (3840x2160) to (1920x1080 HiDPI) which unfortunately I need to do
- Only a reboot or sleep/wake helps
Solution: The high GPU scores remain, when I switch the resolution in 2 steps (3840x2160) > (1920x1080) and then (1920x1080) to (1920x1080 HiDPI)
I have found the cli tool displacer for automation, which needs no installation or SIP modification.

Next I have to get to the bottom of issue after wake, maybe injecting other information via OpenCore into my RX 570 may help, but that's a different problem.

System: MacPro5,1 2010 Dual x5690 6-core; 48 GB 1333 MHz RAM; MSI RX 570 Armor OC 8GB; Big Sur 11.6; MartinLo's OpenCore 0.7.4

 

[roobarb!]

Evening all, I've only discovered this recently, but my previously impressive OpenCL and Metal scores with Geekbench 5 seem to have dropped back to their previous non-RadeonBoost levels since moving to Monterey. I've not changed my RadeonBoost install through OpenCore and it still shows correctly as the VBIOS Version on my Radeon VII.

Back in September on Catalina, I was getting scores of 81917 (OpenCL) and 105346 (Metal). Running the same tests today I'm seeing 61775 (OpenCL) and 66344 (Metal). Have I missed some important configuration tweak that has come along since Monterey? Tested cold boot, warm boot, after sleep/wake; they're the latest tests on my Geekbench profile.

There've been no hardware changes between September 2021 2020* (!) and today, the big alteration being the necessity of moving from Catalina to Monterey and an associated change from the normal "Mac-F221BEC8" board ID to "Mac-27AD2F918AE68F61" of the new MacPro7,1. Everything is working perfectly, except for this weird drop.

*Wow, just realised those results are from a year earlier than I thought. Where the hell did that go?

 

[mikas]

Diidn't Apple intentionally cripple the VII performance. And the cure for that, Radeon Boost kext, is not working in Big Sur anymore.
Radeon Boost Big Sur
GeekBench Metal Scores (VII)
Something like that it seems, unfortunately.
And from this particular thread I found: CMMChris quote about Big Sur + RB

 

[roobarb!]

Diidn't Apple intentionally cripple the VII performance.

Hmm, I'm not sure. To be honest, I really only see the difference in my Geekbench scores. Luxmark, hardware compression, Cinebench - those results are all the same as they were in Catalina. I'm not sufficiently informed about the Radeon VII under macOS to really know what's going on. In everything I actually do the thing still flies, under Windows 11 for those rare gaming moments it's absolutely perfect. It's just these weird Geekbench scores.

Just tried the test under Windows 11 (no overclock, there's a slight overclock under macOS) and I got 89374 (OpenCL) straight away. Booted back into macOS now, will try here again...

...yeah, there we go. It's 61972 (OpenCL) today under macOS. Warm boots into both Windows 11 and macOS 12.2. ?‍♂️

Oof, I just disabled RadeonBoost through OpenCore and ran the test again (still a warm boot). I'm down to 50727 (OpenCL) and 60354 (Metal). Something's still going on.

Yeah, re-enable RadeonBoost and it jumps back up - 60252 (OpenCL) on this run. Bit down for some reason, maybe there was something in the background just after booting.

Okay, after a few minutes we're back to 62643 (OpenCL) and 64757 (Metal). RadeonBoost still seems to be doing something. Hope this lays some breadcrumbs for someone with more knowledge on this.

 

[mikas]

Yeah, I know there's something going on still it seems.

It's not like I dislike like my VIIs in any way, with or without the radeonboost kext. I love them. I just rendered a big bunch of arch visuals with a VII in what, 20-30 minutes or so (not including any of the modelling, material adjustement or camera prework of course). Comparing with one of my 6800XT, there's even no reason to compare, it's that close.

That's why I'm not worried about geekbench scores.

 

[roobarb!]

Yeah, I know there's something going on still it seems.

Nice.

It's not like I dislike like my VIIs in any way, with or without the radeonboost kext. I love them. I just rendered a big bunch of arch visuals with a VII in what, 20-30 minutes or so (not including any of the modelling, material adjustement or camera prework of course). Comparing with one of my 6800XT, there's even no reason to compare, it's that close.

That's why I'm not worried about geekbench scores.

Yeah, this is one of those curious problems. It's there, I remember about it sometimes, then I get on with my day and it doesn't bother me. The VII and NVMe (and OpenCore, of course) completely changed this machine for me. I'm frankly amazed I still just turn it on and use it as my main machine every day.

 

[roobarb!]

And from this particular thread I found: CMMChris quote about Big Sur + RB

Ha, I actually commented all over that post and I have no memory of having done so.

 

[Stex]

Oof, I just disabled RadeonBoost through OpenCore and ran the test again (still a warm boot). I'm down to 50727 (OpenCL) and 60354 (Metal). Something's still going on.

Yeah, re-enable RadeonBoost and it jumps back up - 60252 (OpenCL) on this run. Bit down for some reason, maybe there was something in the background just after booting.

Okay, after a few minutes we're back to 62643 (OpenCL) and 64757 (Metal). RadeonBoost still seems to be doing something. Hope this lays some breadcrumbs for someone with more knowledge on this.

I know your (both @roobarb! and @mikas) tests/comments related to the Radeon VII, but I was interested to see if with a Radeon RX 5700 there would be a similar benchmark scenario with both RadeonBoost disabled and enabled. I had actually tested this difference few months ago when I upgraded from a RX 580 to the RX 5700, but had not really focused on OpenCL results at the time. Here's what I got as benchmarks, they seem to somewhat mirror @roobarb! results (taking into consideration the difference in GPUs). You can find specs of my cMP and macOS/OC in my signature below.

Geekbench 5 Results
RadeonBoost disabled: OpenCL 52598; Metal 60525
RadeonBoost enabled: OpenCL 57341; Metal 62373

It would appear that yes (!), there's something going on! Hope this can (also) help someone with more knowledge on this.

EDIT: Just read the complete thread after posting the above, and based on the kext updates and subsequent comments/findings, I am now less sure my results above show performance impact from this kext. I could not find clear indications within this thread on whether navi cards benefit or not, although I did see navi cards crossed out in post #1.

EDIT 2, March 26th, 2022: I have disabled this kext based on the OP update below in post #440.

 

[Dewdman42]

RadeonBoost.kext appears to have broken my Mac after updating to MacOS 12.3. I get kernel panic during booting. Here is kernel panic report (see spoiler below). Removing RadeonBoost resolves the issue.

Spoiler: RadeonBoost Kernel Panic Log

panic(cpu 4 caller 0xffffff80191d38e3): Kernel trap at 0xffffff7f9a4462bf, type 14=page fault, registers:
CR0: 0x000000008001003b, CR2: 0x0000000003000000, CR3: 0x000000182915b03f, CR4: 0x00000000000226e0
RAX: 0xffffff7f9a446280, RBX: 0xffffff90cf36e014, RCX: 0x0000000002000000, RDX: 0x0000000003000000
RSP: 0xffffffd12eea39b0, RBP: 0xffffffd12eea39c0, RSI: 0x0000000000008200, RDI: 0xffffff90cf36e014
R8:  0xffffff8c0287d9d0, R9:  0x00000000ffffffff, R10: 0x000000000000001d, R11: 0xffffff9a6869d7c0
R12: 0xffffff7fb216bf21, R13: 0xffffff90ced3be80, R14: 0x0000000003000000, R15: 0x0000000000000006
RFL: 0x0000000000010206, RIP: 0xffffff7f9a4462bf, CS:  0x0000000000000008, SS:  0x0000000000000010
Fault CR2: 0x0000000003000000, Error code: 0x0000000000000002, Fault CPU: 0x4, PL: 0, VF: 1

Panicked task 0xffffff8c01d8b3b0: 10 threads: pid 167: WindowServer
Backtrace (CPU 4), panicked thread: 0xffffff959c368000, Frame : Return Address
0xffffffd12eea3360 : 0xffffff8019083e1d mach_kernel : _handle_debugger_trap + 0x41d
0xffffffd12eea33b0 : 0xffffff80191e3ca6 mach_kernel : _kdp_i386_trap + 0x116
0xffffffd12eea33f0 : 0xffffff80191d34fd mach_kernel : _kernel_trap + 0x51d
0xffffffd12eea3440 : 0xffffff8019023a60 mach_kernel : _return_from_trap + 0xe0
0xffffffd12eea3460 : 0xffffff80190841ed mach_kernel : _DebuggerTrapWithState + 0xad
0xffffffd12eea3580 : 0xffffff80190839a6 mach_kernel : _panic_trap_to_debugger + 0x2b6
0xffffffd12eea35e0 : 0xffffff80199164af mach_kernel : _panic + 0x84
0xffffffd12eea36d0 : 0xffffff80191d38e3 mach_kernel : _sync_iss_to_iks + 0x2c3
0xffffffd12eea3850 : 0xffffff80191d35d2 mach_kernel : _kernel_trap + 0x5f2
0xffffffd12eea38a0 : 0xffffff8019023a60 mach_kernel : _return_from_trap + 0xe0
0xffffffd12eea38c0 : 0xffffff7f9a4462bf com.apple.kext.AMDRadeonX4000HWLibs : __ZN34AMDRadeonX4000_AMDRadeonHWLibsCAIL20callPlatformFunctionEPKcbPvS2_S2_S2_ + 0x16434f
0xffffffd12eea39c0 : 0xffffff7f9a500c60 com.apple.kext.AMDRadeonX4000HWLibs : __ZN34AMDRadeonX4000_AMDRadeonHWLibsCAIL20callPlatformFunctionEPKcbPvS2_S2_S2_ + 0x21ecf0
0xffffffd12eea39d0 : 0xffffff7f9a4ffa80 com.apple.kext.AMDRadeonX4000HWLibs : __ZN34AMDRadeonX4000_AMDRadeonHWLibsCAIL20callPlatformFunctionEPKcbPvS2_S2_S2_ + 0x21db10
0xffffffd12eea3a00 : 0xffffff7f99dc8d58 com.apple.kext.AMD9500Controller : __ZN22BaffinPowerPlayManager24setupPowerTuneParametersEv + 0x1c8
0xffffffd12eea3a40 : 0xffffff7f99dc8583 com.apple.kext.AMD9500Controller : __ZN22BaffinPowerPlayManager10initializeEv + 0x2a3
0xffffffd12eea3a70 : 0xffffff7f99db7cdb com.apple.kext.AMD9500Controller : __ZN17AMD9500Controller19initializePowerPlayEv + 0x8b
0xffffffd12eea3aa0 : 0xffffff7f99e0b81a com.apple.kext.AMDFramebuffer : __ZN14AMDFramebuffer14enableHardwareEv + 0x1ea
0xffffffd12eea3b00 : 0xffffff7f99dfcc20 com.apple.kext.AMDFramebuffer : __ZN14AMDFramebuffer16enableControllerEv + 0x20
0xffffffd12eea3b50 : 0xffffff7fb215bcc1 com.apple.iokit.IOGraphicsFamily : __ZN13IOFramebuffer4openEv + 0x60f
0xffffffd12eea3be0 : 0xffffff7f99dfbda0 com.apple.kext.AMDFramebuffer : __ZN14AMDFramebuffer4openEv + 0x40
0xffffffd12eea3c10 : 0xffffff7fb2151f0e com.apple.iokit.IOGraphicsFamily : __ZN13IOFramebuffer13newUserClientEP4taskPvjPP12IOUserClient + 0x25e
0xffffffd12eea3ca0 : 0xffffff8019814aac mach_kernel : __ZN9IOService13newUserClientEP4taskPvjP12OSDictionaryPP12IOUserClient + 0x6c
0xffffffd12eea3d00 : 0xffffff80198830cb mach_kernel : _is_io_service_open_extended + 0x17b
0xffffffd12eea3d60 : 0xffffff8019189b87 mach_kernel : _iokit_server_routine + 0x4967
0xffffffd12eea3db0 : 0xffffff801908a6cb mach_kernel : _ipc_kobject_set_kobjidx + 0x39b
0xffffffd12eea3e10 : 0xffffff801905ecf3 mach_kernel : _ipc_kmsg_send + 0x243
0xffffffd12eea3e60 : 0xffffff8019075249 mach_kernel : _mach_msg_overwrite_trap + 0x2c9
0xffffffd12eea3ef0 : 0xffffff80191b6198 mach_kernel : _mach_call_munger64 + 0x1d8
0xffffffd12eea3fa0 : 0xffffff8019024246 mach_kernel : _hndl_mach_scall64 + 0x16
      Kernel Extensions in backtrace:
         com.apple.iokit.IOGraphicsFamily(594.0)[3ECB2385-311A-374E-9C0F-65639BA26A05]@0xffffff7fb2141000->0xffffff7fb216ffff
            dependency: com.apple.iokit.IOPCIFamily(2.9)[03565C8B-2960-3B32-92FD-E6E86917FE61]@0xffffff801bc03000->0xffffff801bc2efff
         com.apple.kext.AMD9500Controller(4.0.8)[73F58F88-8BFB-39FD-ADD4-B8D9169D4369]@0xffffff7f99d7f000->0xffffff7f99de3fff
            dependency: com.apple.AppleGraphicsDeviceControl(6.5.4)[3BBDC2AD-9889-3863-82EA-864DCE9D0DEF]@0xffffff7fb092c000->0xffffff7fb092ffff
            dependency: com.apple.iokit.IOACPIFamily(1.4)[CF250EA9-AE5C-3215-BCF5-763DB70A0228]@0xffffff801b760000->0xffffff801b761fff
            dependency: com.apple.iokit.IOGraphicsFamily(594)[3ECB2385-311A-374E-9C0F-65639BA26A05]@0xffffff7fb2141000->0xffffff7fb216ffff
            dependency: com.apple.iokit.IOPCIFamily(2.9)[03565C8B-2960-3B32-92FD-E6E86917FE61]@0xffffff801bc03000->0xffffff801bc2efff
            dependency: com.apple.kext.AMDSupport(4.0.8)[C9BA70CD-D5CA-32B9-B28F-45EE36A83E3F]@0xffffff7fb0628000->0xffffff7fb06edfff
         com.apple.kext.AMDRadeonX4000HWLibs(1.0)[731EE448-F9F2-34F7-B05A-A986391D762D]@0xffffff7f9a2e0000->0xffffff7f9a5ddfff
            dependency: com.apple.iokit.IOPCIFamily(2.9)[03565C8B-2960-3B32-92FD-E6E86917FE61]@0xffffff801bc03000->0xffffff801bc2efff
         com.apple.kext.AMDFramebuffer(4.0.8)[CCC9B796-5C11-38B9-807E-CAC9CE502386]@0xffffff7f99dfb000->0xffffff7f99e1ffff
            dependency: com.apple.iokit.IOACPIFamily(1.4)[CF250EA9-AE5C-3215-BCF5-763DB70A0228]@0xffffff801b760000->0xffffff801b761fff
            dependency: com.apple.iokit.IOGraphicsFamily(594)[3ECB2385-311A-374E-9C0F-65639BA26A05]@0xffffff7fb2141000->0xffffff7fb216ffff
            dependency: com.apple.iokit.IOPCIFamily(2.9)[03565C8B-2960-3B32-92FD-E6E86917FE61]@0xffffff801bc03000->0xffffff801bc2efff
            dependency: com.apple.kext.AMDSupport(4.0.8)[C9BA70CD-D5CA-32B9-B28F-45EE36A83E3F]@0xffffff7fb0628000->0xffffff7fb06edfff

Process name corresponding to current thread (0xffffff959c368000): WindowServer
Boot args: keepsyms=1 debug=0x100 chunklist-security-epoch=0 -chunklist-no-rev2-dev

Mac OS version:
21E230

Kernel version:
Darwin Kernel Version 21.4.0: Mon Feb 21 20:34:37 PST 2022; root:xnu-8020.101.4~2/RELEASE_X86_64
Kernel UUID: 9619904E-EA6D-3D08-89C4-B1C0451D39B8
KernelCache slide: 0x0000000018e00000
KernelCache base:  0xffffff8019000000
Kernel slide:      0x0000000018e10000
Kernel text base:  0xffffff8019010000
__HIB  text base: 0xffffff8018f00000
System model name: MacPro5,1 (Mac-27AD2F918AE68F61)
System shutdown begun: NO
Panic diags file available: YES (0x0)
Hibernation exit count: 0

System uptime in nanoseconds: 33537706168
Last Sleep:           absolute           base_tsc          base_nano
  Uptime  : 0x00000007cf00d264
  Sleep   : 0x0000000000000000 0x0000000000000000 0x0000000000000000
  Wake    : 0x0000000000000000 0x00000026972a14e0 0x0000000000000000
Compressor Info: 0% of compressed pages limit (OK) and 0% of segments limit (OK) with 0 swapfiles and OK swap space
Zone info:
  Foreign : 0xffffff801f6a3000 - 0xffffff801f6b1000
  Native  : 0xffffff80ce94f000 - 0xffffffa0ce94f000
  Readonly: 0xffffff859b61b000 - 0xffffff8734fb4000
  Metadata: 0xffffffe4040fe000 - 0xffffffe424bf1000
  Bitmaps : 0xffffffe424bf1000 - 0xffffffe444bf1000

last started kext at 32649545441: |IOAVB!F    1040.6 (addr 0xffffff7fb1f31000, size 77824)
loaded kexts:
com.khronokernel.FeatureUnlock    1.0.7
as.vit9696.WhateverGreen    1.5.8
as.vit9696.Lilu    1.6.0
>AGPM    127
>!AUpstreamUserClient    3.6.9
>!AMCCSControl    1.15
@kext.AMDFramebuffer    4.0.8
>AudioAUUC    1.70
>!AMikeyHIDDriver    131
>!AMikeyDriver    340.2
@kext.AMDRadeonX4000    4.0.8
>!AHDAHardwareConfigDriver    340.2
@filesystems.autofs    3.0
>!AHDA    340.2
@kext.AMDRadeonServiceManager    4.0.8
>!AGFXHDA    140.3
>!AGraphicsDevicePolicy    6.5.4
@AGDCPluginDisplayMetrics    6.5.4
>pmtelemetry    1
|IOUserEthernet    1.0.1
>!ALPC    3.1
>usb.!UUserHCI    1
>!AHV    1
>!ADiskImages2    126.100.13
>!AFWOHCI    5.7.5
@kext.AMD9500!C    4.0.8
>!AFIVRDriver    4.1.0
>!A!ISlowAdaptiveClocking    4.0.0
>!A!IMCEReporter    115
>ACPI_SMC_PlatformPlugin    1.0.0
>usb.!UHostBillboardDevice    1.0
>!AFileSystemDriver    3.0.1
@filesystems.tmpfs    1
@filesystems.lifs    1
@filesystems.hfs.kext    583.100.10
@BootCache    40
@!AFSCompression.!AFSCompressionTypeZlib    1.0.0
@!AFSCompression.!AFSCompressionTypeDataless    1.0.0d1
|SCSITaskUserClient    456.100.7
@filesystems.apfs    1934.101.3
>!I82574LEthernet    2.7.2
>AirPort.BrcmNIC    1400.1.1
>!AAHCIPort    351.100.4
@private.KextAudit    1.0
>!ARTC    2.0.1
>!AHPET    1.8
>!AACPIButtons    6.1
>!ASMBIOS    2.1
>!AACPIEC    6.1
>!AAPIC    1.7
>!A!ICPUPowerManagementClient    222.0.0
@!ASystemPolicy    2.0.0
@nke.applicationfirewall    402
|IOKitRegistryCompatibility    1
|EndpointSecurity    1
@Dont_Steal_Mac_OS_X    7.0.0
>!A!

 

[CMMChris]

Don't use this Kext on anything later than Catalina, starting from macOS Big Sur it doesn't work anymore for most GPUs due to Apple changing things. I also don't update it anymore because I am not building and using Hackintoshes anymore.

 

[Dayo]

Don't use this Kext on anything later than Catalina

Might be worth adding that to Post 1 in big bold text along with a warning that as from MacOS 12.3, usage results in the NVRAM being filled with KP dumps, which could cause a brick.

 

[CMMChris]

@Dayo Good idea, done!

 

[roobarb!]

it doesn't work anymore for most GPUs due to Apple changing things

@Dayo - Which GPU are you using? I'm using a VII here and RadeonBoost still seems to makes a difference and I've not seen any kernel panics. @CMMChris, is there a reason it would still make an improvement despite the Apple changes?

 

[CMMChris]

I think VII is one of the few GPU where it still works but I am not entirely sure. I am out of this for quite a while now.

 

[tommy chen]

I think VII is one of the few GPU where it still works but I am not entirely sure. I am out of this for quite a while now.

no, radeonboost has no effect on the VII since big sur

 

[roobarb!]

no, radeonboost has no effect on the VII since big sur

Huh, it seemed to do something for me, but I'm no expert in this stuff.

I've disabled it as advised now anyway. So is it correct that the improvements shown in Geekbench were entirely aesthetic? I've seen this bandied around but never fully understood how this could be the case.

 

[tommy chen]

you should consider that the radeonboost effect was only visible in geekbench,
and almost no increase in performance was noticeable in all other tests and also in the realtime tests.

geekbench can also not be believed at all, since different results come out with every measurement
and the highest values are also displayed after the sleep of a cMP.

 

[roobarb!]

almost no increase in performance was noticeable in all other tests and also in the realtime tests.

Well, I can certainly agree with that.

geekbench can also not be believed at all, since different results come out with every measurement
and the highest values are also displayed after the sleep of a cMP.

Weird. Fair enough, thank you for taking the time!

 

[h9826790]

It seems someone find a way to re-create the effect of RadeonBoost by using OCLP in Monterey

 

[tsialex]

It seems someone find a way to re-create the effect of RadeonBoost by using OCLP in Monterey

This is just disabling SMBIOS spoofing and rebooting. Dumb move if you need anything that requires SMBIOS spoofing like VideoToolbox hardware acceleration.

We discussed it last week on the main OC thread, seems entering sleep is still the best way to re-gain performance.