Absolutely,but is so strange that this issue is present either on a cMP.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MP All Models AMD Polaris, Vega, & Navi GPU macOS Support
- Thread starter MisterAndrew
- Start date
- Sort by reaction score
I tested a few different BIOSs on my Pulse 56 today to see if the fan issue is still present in the latest beta. I tested the XFX Double Edition RX Vega 64 BIOS and the issue is still present. I also tested the PowerColor Red Devil RX Vega 56 and it does not have the issue. I tested the 64 Red Devil BIOS and the Mac would not boot, so my card wouldn't accept that one, but it probably doesn't have the issue since the 56 version doesn't. Now we have several Vega cards to choose from that don't have the fan issue. And those with a reference RX Vega 64 could try the Red Devil 64 BIOS if one of the Nitro+ 64 BIOSs doesn't work.
Edit: The Red Devil card is 316mm long, so it won’t fit in the cMP.
Edit: The Red Devil card is 316mm long, so it won’t fit in the cMP.
Last edited:
It looks like the 3 free games promotion isn’t limited to just Sapphire cards.
https://www.amd.com/en/gaming/raise-the-game
https://www.amd.com/en/gaming/raise-the-game
I'm trying to catch up on the latest...
Is seems that it's possible to run silent with a Vega card in Mojave (even if the odd 'hack' is necessary). Great—I'll test this when I get back from vacation with my Vega FE. I've waited with Mojave so far, but I think I'm ready.
Has HEVC encoding improved with Mojave? Are we able to tap into the Vega cards for accelerated HEVC performance? Mostly interested in encoding...
Is seems that it's possible to run silent with a Vega card in Mojave (even if the odd 'hack' is necessary). Great—I'll test this when I get back from vacation with my Vega FE. I've waited with Mojave so far, but I think I'm ready.
Has HEVC encoding improved with Mojave? Are we able to tap into the Vega cards for accelerated HEVC performance? Mostly interested in encoding...
Yes, first time some of these Vega cards are now completely silent when not under load. The fans on the Pulse 56 don't spin at all unless it's doing something fairly graphic intensive like a benchmark. Okay, it will be good to know if the fan issue on the FE has been fixed. Apple has at least been working on some cards, or something that some of these cards have in common. I don't know if HEVC encoding works.
The new fan issue is odd where it spins up under load and then doesn't stop spinning. The small H.265 video I tried playing that stopped the fans before doesn't seem to want to stop them now. But I found this video sample that is doing the trick: http://4ksamples.com/elysium-2013-2160p-1-minute-sample-footage/. You have to play it long enough for the fans to stop or they'll spin up again.
The new fan issue is odd where it spins up under load and then doesn't stop spinning. The small H.265 video I tried playing that stopped the fans before doesn't seem to want to stop them now. But I found this video sample that is doing the trick: http://4ksamples.com/elysium-2013-2160p-1-minute-sample-footage/. You have to play it long enough for the fans to stop or they'll spin up again.
Last edited:
OK. At least we're inching forward.
I had something similar with my RX 480 previously where I had to run the Heaven benchmark after reboots to get the fans into the correct mode. Never got fixed though, and I'm not sure we'll get a fix for the current cards either, except maybe 'by accident' as part of other optimisations.
I had something similar with my RX 480 previously where I had to run the Heaven benchmark after reboots to get the fans into the correct mode. Never got fixed though, and I'm not sure we'll get a fix for the current cards either, except maybe 'by accident' as part of other optimisations.
Considering they mention the following graphic cards in their eGPU documentation, we can only assume they work out the kinks at some point.
AMD Radeon RX 470, RX 480, RX 570, RX 580, Radeon Pro WX 7100, AMD Radeon RX VEGA 56, AMD Radeon RX VEGA 64, VEGA FRONTIER EDITION (AIR) and Radeon Pro WX 9100.
AMD Radeon RX 470, RX 480, RX 570, RX 580, Radeon Pro WX 7100, AMD Radeon RX VEGA 56, AMD Radeon RX VEGA 64, VEGA FRONTIER EDITION (AIR) and Radeon Pro WX 9100.
Great news! Apple responded to the bug report I sent and asked for a sysdiagnose file in order to further investigate the issue. So it appears Apple is officially supporting Vega graphics cards on the Mac Pro 5,1 for Mojave. I encourage others to submit a bug report if you are experiencing issues with your Vega graphics card (and also Polaris cards) running the latest Mojave beta on your genuine Mac Pro 5,1 (not 4,1 or hackintosh since those are unsupported.)
Submit the bug report here: http://bugreport.apple.com/
Select Mac hardware and describe the issue, including the specific model of your card. Attach a System Information file (Open System Information, click File, and Save...) and also a sysdiagnose file.
sysdiagnose Instructions:
https://developer.apple.com/service...S_X_Logs/sysdiagnose_Logging_Instructions.pdf
Submit the bug report here: http://bugreport.apple.com/
Select Mac hardware and describe the issue, including the specific model of your card. Attach a System Information file (Open System Information, click File, and Save...) and also a sysdiagnose file.
sysdiagnose Instructions:
https://developer.apple.com/service...S_X_Logs/sysdiagnose_Logging_Instructions.pdf
Cool beans MisterAndrew! For grins, maybe you could also file a bug report about the HDMI Audio driver not being updated with the device ID for the recommended AMD cards--maybe they'll fix it. Kinda sucks having a 4,1 for this because I doubt they're taking my bug reports seriously when they see my SN is for a 4,1. I hope everyone with genuine 5,1s are reporting the hell out of the various issues running the non-EFI Radeons that Apple recommended. Wondering if we should have a new thread here and in the Mojave forum outlining each of the "bugs" and encouraging real 5,1 users to report 'em while there's still time in the beta cycle.
They also replied to my bug report about no way to upgrade firmware with the RX580. Hopefully they will provide a decent solution later.
Woo!
this is going to be an interesting few days depending on how it all pans out
its worth mentioning that I dont have a Genuine 5,1 my MP is a flashed MP4,1 however during my bug reporting of the microcode BootROM issue, apple never asked for any sysdiagonose or system profiler dumps of my machine and I just said I had a MP5,1 I never said if it was a genuine or a flashed 5,1
so you MIGHT be able to get away with submitted bug reports even if your MacPro is a 4,1-5,1 flashed MacPro
[doublepost=1533896227][/doublepost]PS im curious if the (Sapphire pulse) RX 580 and (MSI) RX 560 are being Officially recommended for the MP5,1 now
do they now run in PCIe 2.0 mode properly now or? if they are still stuck in PCIe 1.1 mode it might be worth submitting a bug report on it now...
this is going to be an interesting few days depending on how it all pans out
its worth mentioning that I dont have a Genuine 5,1 my MP is a flashed MP4,1 however during my bug reporting of the microcode BootROM issue, apple never asked for any sysdiagonose or system profiler dumps of my machine and I just said I had a MP5,1 I never said if it was a genuine or a flashed 5,1
so you MIGHT be able to get away with submitted bug reports even if your MacPro is a 4,1-5,1 flashed MacPro
[doublepost=1533896227][/doublepost]PS im curious if the (Sapphire pulse) RX 580 and (MSI) RX 560 are being Officially recommended for the MP5,1 now
do they now run in PCIe 2.0 mode properly now or? if they are still stuck in PCIe 1.1 mode it might be worth submitting a bug report on it now...
Last edited:
To partially answer your question, no, they haven't unlocked the bandwidth connection on RX 560's mines a sapphire pulse version but they have the same device ID's so it doesn't make a difference, its still stuck at x8 lane width with 2.5gt/s bandwidth on Mojave.Woo!
this is going to be an interesting few days depending on how it all pans out
its worth mentioning that I dont have a Genuine 5,1 my MP is a flashed MP4,1 however during my bug reporting of the microcode BootROM issue, apple never asked for any sysdiagonose or system profiler dumps of my machine and I just said I had a MP5,1 I never said if it was a genuine or a flashed 5,1
so you MIGHT be able to get away with submitted bug reports even if your MacPro is a 4,1-5,1 flashed MacPro
[doublepost=1533896227][/doublepost]PS im curious if the (Sapphire pulse) RX 580 and (MSI) RX 560 are being Officially recommended for the MP5,1 now
do they now run in PCIe 2.0 mode properly now or? if they are still stuck in PCIe 1.1 mode it might be worth submitting a bug report on it now...
How would i go about using this bug report system to maybe get a solution for this?
Update: I just sent in a bug report I'll wait to see what happens.
Last edited:
For comparison sake.. Sharing results from a 470 4GB. Considering the Pulse RX580 is newer/faster, I was expecting to see more than an 8% difference.
It is actually a bit more than 8% in my case.
[doublepost=1533936532][/doublepost]
Please make sure you go to bugreport.apple.com to submit your bug report, but not just the build in bug reported in the beta OS.
I never receive any feed back from Apple except via bugreport.apple.com
When I ran the above Luxmark test, I realised that 10.14 has more aggressive on the temperature control of the PULSE RX580. It keep my card at around 70C with higher fan speed (up to around 2000RPM, still quite quiet).
Same card, same cMP, same condition, I boot back to 10.13.6 and run Luxmark again, the card's fan spin slower (around 1600RPM) and allow the card run up to ~75C during benchmarking.
[doublepost=1533936891][/doublepost]
PB5
[doublepost=1533937191][/doublepost]
Same card, same cMP, same condition, I boot back to 10.13.6 and run Luxmark again, the card's fan spin slower (around 1600RPM) and allow the card run up to ~75C during benchmarking.
[doublepost=1533936891][/doublepost]
PB5
[doublepost=1533937191][/doublepost]
Last edited:
Following up with a reference rx480 in Mojave with a 4,1. Again, I thought the difference would have been greater.
The rx580 is not that much better than the rx480 from a couple years ago with stock firmware.
Last edited:
I'm trying to figure out why h9826790's score is much higher. Same card and same macOS build? Is everything stock, unmodified? Perhaps Pixlas mod powered vs. logic board connectors?
Very clean 10.14 beta. Only for testing purpose, almost nothing installed and use for real.
iStat menu loaded at startup by default. Terminal is opened to monitor GPU parameters. No other apps / software opened when running the test. I ran the tests a few times and confirmed Terminal has virtually no penalty for this test.
Yes, I flashed the original firmware back in and do the test, same score. Before that, I ran a downvolted VBIOS to let the card run cooler but at the same clock, which should able to avoid thermal throttling even under very extreme condition. But my observation is that the original setting won’t be throttled in just Luxmark test.
cMP itself with unmodified 0089.B00 firmware.
My card is powered by Pixlas mod, but this shouldn’t cause any difference for the PULSE RX580.
PCIe / PSU fans are NOT at native setting, but controlled by my own fan control software. However, they should run slower than stock setting (assuming the GPU is connected to the mini 6pins in the native way) when a RX580 is under stress. So, this should has negative effect if it really affect the GPU performance.
I ran the test few more times in 10.14 and 10.13.6. 10.13.6 always produce higher score. But I think my last observation about fan / temperature is wrong. The difference most likely coming from the fuzzy fan logic of the Polaris VBIOS. Not from the different OS / driver version.