You can try to do that to see if that is the problem.
remember clean up with kext utility
then you start again
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.
Dual RX 480s in Mac Pro 5,1 running MacOS (w/ How-to)
- Thread starter theitsage
- Start date
- Sort by reaction score
You can try to do that to see if that is the problem.
remember clean up with kext utility
then you start again
I am not at the office atm, but the MSI RX480 Gaming X 8G (8GB GDDR5) has a single 8 pin instead if the reference 6 pin
FileVault is disabled.
So that SIP enabled thing.. was definitely a fluke somehow. The next reboot after I made that post the kext did not load so SIP was enabled. I had to go back and disable it again in recovery mode.. after that thing were fine again.
Sorry for not responding earlier. I don't have notifications enabled for this thread (will change that) and I hadn't been to this site for a while
UPDATE:
with macOS Sierra Beta 6 I see a significant improvement in the RX480 performance.
To be sure, I run every test 3 times.
Heaven is exactly the same, but Luxmark scored 600 extra points (all the benchmarks had more or less the same improvement)
Btw, I know this might be old news for those on the developer beta - only posted to have a reference to the public beta as well
with macOS Sierra Beta 6 I see a significant improvement in the RX480 performance.
To be sure, I run every test 3 times.
Heaven is exactly the same, but Luxmark scored 600 extra points (all the benchmarks had more or less the same improvement)
Btw, I know this might be old news for those on the developer beta - only posted to have a reference to the public beta as well
Attachments
Last edited:
I stil having trouble with installing RX 480 even It was successful for few weeks but suddenly the acceleration stopped without any reasons. I tried several times with that method but not working AT ALL. Are there any addition process for installing RX 480?
No but try re editing the kext file?
Also, you can try to put back the original, use kext utility, then re-do the whole kext hack
Stupid question for you sage……You didn't mention that SIP does/does not need to be switched back on after the kext mods?
I downloaded and installed it. Just needed to modify the kext and disable SIP again to support the RX 480 and everything seems to work as before so far. I've done luxmark, valley, and geekbench 4 benchmarks and I have not noticed any significant performance difference. Scores in luxmark were a few 100 points lower, valley was the same, but the geekbench 4 compute score went up about 800 points. CPU multi-core also went up about 400 points in geekbench 4.
macOS (in luxmark for instance) now shows the proper clock speed for me. 10.12.1 reported 130mhz while 10.12.2 correctly reports 1330mhz.
macOS (in luxmark for instance) now shows the proper clock speed for me. 10.12.1 reported 130mhz while 10.12.2 correctly reports 1330mhz.
Last edited:
Are you talking about the backplate? mine is metal, and it is generally really not recommended to remove it
Last edited:
The update removed acceleration from the RX480. Repeated the steps from theitsage and everything worked again.
Last edited:
Oh, sorry. No, I ment it turned of the acceleration of the GPU. I'll go ahead and edit my post.
Oh, I see, that sounds much better. Thanks for the confirmation.
Glad to see people are getting their card back up and running in 10.12.2.
I know this thread is focused on RX 480 in Sierra, but has anyone tested their RX 480 in Windows 10 (with the Anniversary Update) to see if they're getting proper PCIe 16x 2.0 support? I've heard that there should be no problems with getting the full PCIe 2.0 bandwidth in windows, but when I use GPU-Z and run the render test to stress the card, the bus interface still shows the card running at x16 1.1.
Could someone do me a huge favor if they have this setup and check to see what GPU-Z is reporting when running the render test?
Thanks!
I know this thread is focused on RX 480 in Sierra, but has anyone tested their RX 480 in Windows 10 (with the Anniversary Update) to see if they're getting proper PCIe 16x 2.0 support? I've heard that there should be no problems with getting the full PCIe 2.0 bandwidth in windows, but when I use GPU-Z and run the render test to stress the card, the bus interface still shows the card running at x16 1.1.
Could someone do me a huge favor if they have this setup and check to see what GPU-Z is reporting when running the render test?
Thanks!
Under Windows, with a non-EFI card, you're limited to PCIe 1.1.
But in my experience of several days playing Rise of the Tomb and Doom on one RX480 in a 3,1 with the setting almost all the way up in Full HD resolution, it still gets you the performance you need.
Thanks for letting me know. It's good to have a definitive answer so I don't need to keep searching around everywhere as to why it isn't being reported correctly. It is being reported correctly
I've played Rise of the Tomb Raider for about 45 minutes with no artifacts/shutdowns/freezes on my 5,1. I've been watching the wattage use and it's concerning of course. I'm using an 8 pin card which doesn't help. With the new Radeon Crimson ReLive Edition drivers you now can use the Chill function in WattMan which should help lower the power usage in non-DX12/Vulkan games. I'm running the card off the internal PSU so I'm taking my chances here. I'm a bit scared to run any benchmarks since I've seen usage spike over 200w then jump back down to varying around 150w or lower.
Re: ROTTR performance...
I think I can attest to this as well. I'm seeing framerates within what seems like relatively average range with settings maxed out at 1440p/DX12/FreeSync. It looks like there's generally a 5-15% performance loss depending on the game and resolution when running PCIe 1.1 vs 3.0. At least these cards are running at 16x and not 4x for instance. 8x is not really significant. Here's an article about PCIe scaling on the Fury X: https://www.techpowerup.com/reviews/AMD/R9_Fury_X_PCI-Express_Scaling
Unless there's a way of making Mac EFI roms in the future like Mac Video Cards has done in the past for the GTX 700 series, I guess we'll just have to settle for a PCIe 1.1 / 2.5 GT/s link which was news to me in Windows at least.
I haven't created my own rom before or know how if this is even a possibility, but with future Polaris 10 (and maybe Polaris 11 for lower configurations) and Vega support coming to the Retina iMacs and Mac Pro respectively, is there a chance someone can extract the EFI roms for use on a desktop card? Rumors are that the Vega (or possibly a revised Fiji under the moniker Dracarys) chip in the Mac Pro will be a custom chip for Apple. And I will assume the new Retina iMacs will still have a mobile based variant.
I wonder if there is any hope out there for EFI support in the future.
Last edited:
Thanks for letting me know. It's good to have a definitive answer so I don't need to keep searching around everywhere as to why it isn't being reported correctly. It is being reported correctly
I've played Rise of the Tomb Raider for about 45 minutes with no artifacts/shutdowns/freezes on my 5,1. I've been watching the wattage use and it's concerning of course. I'm using an 8 pin card which doesn't help. With the new Radeon Crimson ReLive Edition drivers you now can use the Chill function in WattMan which should help lower the power usage in non-DX12/Vulkan games. I'm running the card off the internal PSU so I'm taking my chances here. I'm a bit scared to run any benchmarks since I've seen usage spike over 200w then jump back down to varying around 150w or lower.
Re: ROTTR performance...
I think I can attest to this as well. I'm seeing framerates within what seems like relatively average range with settings maxed out at 1440p/DX12/FreeSync. It looks like there's generally a 5-15% performance loss depending on the game and resolution when running PCIe 1.1 vs 3.0. At least these cards are running at 16x and not 4x for instance. 8x is not really significant. Here's an article about PCIe scaling on the Fury X: https://www.techpowerup.com/reviews/AMD/R9_Fury_X_PCI-Express_Scaling
Unless there's a way of making Mac EFI roms in the future like Mac Video Cards has done in the past for the GTX 700 series, I guess we'll just have to settle for a PCIe 1.1 / 2.5 GT/s link which was news to me in Windows at least.
I haven't created my own rom before or know how if this is even a possibility, but with future Polaris 10 (and maybe Polaris 11 for lower configurations) and Vega support coming to the Retina iMacs and Mac Pro respectively, is there a chance someone can extract the EFI roms for use on a desktop card? Rumors are that the Vega (or possibly a revised Fiji under the moniker Dracarys) chip in the Mac Pro will be a custom chip for Apple. And I will assume the new Retina iMacs will still have a mobile based variant.
I wonder if there is any hope out there for EFI support in the future.
For AMD card, EFI may not be required for PCIe 2.0, but a resistor mod may be required. However, it's very hard to find out which resistor has to be removed, and hard to know if there is any side effect. For HD7950 / 7970, that's the resistor R17. Once remove that resistor, PCIe 2.0 will be avail in both OSX and Windows regardless if the card has Mac EFI or not. And actually, purely flash the card with Mac EFI won't enable PCIe 2.0 in both OSX and Windows.