MP 1,1-5,1 Activate AMD hardware acceleration

[Synchro]

I upgraded from 0.6.7 to 0.6.8 in my 5,1 and my Bluetooth adapter stopped working. I have an upgraded wifi & bluetooth card:

Card Type: AirPort Extreme (0x14E4, 0x14A)
Firmware Version: Broadcom BCM43xx 1.0 (7.77.111.1 AirPortDriverBrcmNIC-1675.1)

But system info says "No information found" under Bluetooth. This worked fine in 0.6.7.

Weirdly, my Magic Keyboard works over BT when OC is displaying the boot picker (I can select the boot drive), but once I get to the macOS login screen, I can't type my password. Plugging the keyboard in via USB makes it work, just not over bluetooth.

 

[h9826790]

I upgraded from 0.6.7 to 0.6.8 in my 5,1 and my Bluetooth adapter stopped working. I have an upgraded wifi & bluetooth card:

Card Type: AirPort Extreme (0x14E4, 0x14A)
Firmware Version: Broadcom BCM43xx 1.0 (7.77.111.1 AirPortDriverBrcmNIC-1675.1)

But system info says "No information found" under Bluetooth. This worked fine in 0.6.7.

Weirdly, my Magic Keyboard works over BT when OC is displaying the boot picker (I can select the boot drive), but once I get to the macOS login screen, I can't type my password. Plugging the keyboard in via USB makes it work, just not over bluetooth.

Try a warm reboot.

You are not the only having this BT issue in Big Sur, and it's not OC version related.

 

[Ausdauersportler]

Try a warm reboot.

You are not the only having this BT issue in Big Sur, and it's not OC version related.

In the most recent OCLP this issue has been resolved. Just wondering if we - on the long run - really need more than one provider of OpenCore configs - a consolidation might be better from the users view?

 

[Synchro]

Try a warm reboot.

You are not the only having this BT issue in Big Sur, and it's not OC version related.

I tried that and it didn't help, still no BT. What is odd is that while BT works during OC boot device selection, I can't use my keyboard via a USB hub there, so I have to unplug to select boot device, then replug to enter my password.

 

[Dayo]

Just wondering if we - on the long run - really need more than one provider of OpenCore configs - a consolidation might be better from the users view?

The OCLP has a lot going for it and it may be that there will ultimately be no way around the "Full Spoofing" and "Rigid Patching" that it applies, particularly when the next iteration of Mac OS is released.

However, the minimal approach to implementing OpenCore on Real Macs that @cdf has put together, which the few OpenCore configs here follow, has served us very well in keeping our Macs safe as OpenCore itself has evolved over time.

Basically saying you ought to tone your recent "OCLP or Nothing" posts down a bit.

OpenCore on the Mac Pro

Has one of you tried OCLP instead of this particular version of OpenCore described and maintained here? If the problem exists there too you may open a GitHub issue - I am quite sure the developers might have an idea. But you have to test it on your system first.

forums.macrumors.com

As said, the OCLP might well be the only way at some point but it doesn't need these types of posts for this to be the case.

 

[Ausdauersportler]

The OCLP has a lot going for it and it may be that there will ultimately be no way around the "Full Spoofing" and "Rigid Patching" that it applies, particularly when the next iteration of Mac OS is released.

However, the minimal approach to implementing OpenCore on Real Macs that @cdf has put together, which the few OpenCore configs here follow, has served us very well in keeping our Macs safe as OpenCore itself has evolved over time.

Basically saying you ought to tone down your recent "OCLP or Nothing" posts a bit.

OpenCore on the Mac Pro

Has one of you tried OCLP instead of this particular version of OpenCore described and maintained here? If the problem exists there too you may open a GitHub issue - I am quite sure the developers might have an idea. But you have to test it on your system first.

forums.macrumors.com

As said, the OCLP might well be the only way at some point but it doesn't need these types of posts for this to be the case.

This maybe really new for you:

OLCP uses light spoofing by default. This is shadow boxing...you can choose to do full spoofing, although. You can do this using the config provided here, too.

The reason for that type of spoofing was a strong intervention from some very active users from this particular thread.

I am not saying "OCLP or Nothing" - I am asking to join the forces. Everything what has been developed or created in the past here and by other users published on other threads should and will go in the end into a single tool. You cannot stop this by sitting and waiting here.

 

[solaris8x86]

Try a warm reboot.

You are not the only having this BT issue in Big Sur, and it's not OC version related.

You may need a patched custom version for BCM4333. Download here.

GitHub - khronokernel/IO80211-Patches: Repo for patched IO80211Family Kexts

Repo for patched IO80211Family Kexts. Contribute to khronokernel/IO80211-Patches development by creating an account on GitHub.

github.com

Selected the Patched Catalina Kext.


I'd the same issue as yours but on BCM4322. I selected the Patched Mojave kext version from there. It works out for me.

 

[solaris8x86]

I think working on an old Mac Pro 2010 + Vega 56 with either Lilu method or Opencore method on hardware acceleration is really great.

My compute score of Geekbench is 49435 for both Metal and OpenCL.
Heaven benchmark score is 1790 (68 fps).

When it works on Final Cut Pro to export a 4k Prores422 project to h265/HEVC file at 70Mbps bitrate. The content of the clip is 5 min long. The export time is just 4 min....for a 4k HEVC encoding.... that is unimaginable fast... Later, on YouTube. I found out this performance is similar to a top config model of MacBook Pro 2019 16'.

I checked to YouTube. I saw a lot of users who have MacBook Pro 2018 + Thunderbolt 3 eGPU enclosure + 5700XT. They could get only around 39000 compute score from Geekbench. That thunderbolt cable thing is really a bottleneck for roundtrip data traffic and it is not as good as an internal PCI-e 2.0 on cMP. The PCI-e 2.0 wins thunderbolt 3 that making an older internal PCI-e Vega 56 big-winning an external more advanced 5700XT on thunderbolt 3. Impressive.

When a Vega 56 can easily win a 5700XT. Then we can tell there is a bottleneck problem on thunderbolt. So eGPU is a joke... It's functional. But not optimal. Poor thing to all Mac laptop users. They should go back order a cMP... if they really want to be the fastest in this hardware acceleration game. The old PCI-e 2.0 does its FXXKXXX job amazingly.

I got a Sapphire Vega 56 in last year at USD$260 on eBay and now it jumps to USD$500 on eBay due to EOL and lack of supply. Now, I did a calculation of what I spent. I spent that $260 only and in return of a MacBook Pro 2019 16' model performance. What a wise investment I did!!

10 year... just $260.... got a performance of a very late model of Mac... that's the point.

 

[solaris8x86]

My Mac Pro 2010 with Vega 56 is that fast now..... outperforms any RX5700 XT in eGPU enclosure for sure.

It's just that fast.... It is even running at over 50,000 score for Metal and OpenCL.

I'm still working on believing how come an old PCI-e 2.0 Mac Pro 5,1 is much faster than Thunderbolt 3 in the new generations of Mac and how come a Vega 56 (stock BIOS, no overclock) could kill RX5700 XT.

It's just turbo...

The conclusion. I don't have to buy a new Mac. I just don't see the point. 😀

 

[h9826790]

PCIe 2.0 x16 has much higher bandwidth than TB3 indeed.

 

[solaris8x86]

PCIe 2.0 x16 has much higher bandwidth than TB3 indeed.

If the old PCI is faster than Thunderbolt 3. Then why they design the thunderbolt to play themselves... It sounds like self-FXXK.

And Vega 56 is a 210W rated card. It just reaches to the ceiling of power consumption of the PSU that the Mac Pro 5,1 can handle without needing a PSU power Mod. It's just saving the Mod works for me and plug and play. That's why I didn't pick Vega 64. The 56 did its job to unlock the maximum performance ceiling in a stock configuration PSU in cMP.

 

[h9826790]

If the old PCI is faster than Thunderbolt 3. Then why they design the thunderbolt to play themselves... It sounds like self-FXXK.

And Vega 56 is a 210W rated card. It just reaches to the ceiling of power consumption of the PSU that the Mac Pro 5,1 can handle without needing a PSU power Mod. It's just saving the Mod works for me and plug and play. That's why I didn't pick Vega 64. The 56 did its job to unlock the maximum performance ceiling in a stock configuration PSU in cMP.

LAN is also faster (and older) than Wi-Fi. Then why design Wi-Fi?

Apparently, TB is more portable than PCIe.

If you willing to spend time to carefully calibrate the card's clock speed and voltage (by using PowerPlay table), even Vega64 / Radeon VII can be used on the cMP without PSU mod. And provide a bit more performance.

How to - downvolt a Radeon VII

I confirmed it's possible to run a radeon VII with the mini 6pin ONLY (no Pixlas mod, no extra PSU, no SATA power, but with power balancing) with my PowerPlay Table kext. I limit the card's power draw to 204W (170 x 120%). This is NOT a random number, but when I perform tests in Windows, I...

forums.macrumors.com

But Vega56 is quite a optimum choice indeed. Max performance for min input (from user).

 

[solaris8x86]

LAN is also faster (and older) than Wi-Fi. Then why design Wi-Fi?

Apparently, TB is more portable than PCIe.

If you willing to spend time to carefully calibrate the card's clock speed and voltage (by using PowerPlay table), even Vega64 / Radeon VII can be used on the cMP without PSU mod. And provide a bit more performance.

How to - downvolt a Radeon VII

I confirmed it's possible to run a radeon VII with the mini 6pin ONLY (no Pixlas mod, no extra PSU, no SATA power, but with power balancing) with my PowerPlay Table kext. I limit the card's power draw to 204W (170 x 120%). This is NOT a random number, but when I perform tests in Windows, I...

forums.macrumors.com

But Vega56 is quite a optimum choice indeed. Max performance for min input (from user).

I won't do such a silly thing to get a more expensive card and intentionally to degrade it thru downgrade procedures just because to suit the rating of PSU. It's illogical. I would rather just pick the right card to suit the maximum output of PSU can handle.

In your article. It is limited to 204W. But in my native installation. My Vega 56 would draw from:

PCIe Slot 1 = 2A max (12v x 2A = 24W)
PCIe Booster A = 8A max (12v x 8A = 96W)
PCIe Booster B = 8A max (12v x 8A = 96W)

Total drawn = 216W.

In AMD website, it spec is some kind of rated at 210W max. It hits 216W in my case without triggering any abnormal accidental reboot or power interruption. It is stable when working at this power consumption. The total PSU 12v output at max is around 460W when exporting a Final Cut Pro Prores422 project to h264 4k file in my system with Vega 56. It really nicely hitting the physical ceiling of what the Mac Pro PSU can supply the power.

 

[h9826790]

I won't do such a silly thing to get a more expensive card and intentionally to degrade it thru downgrade procedures just because to suit the rating of PSU. It's illogical. I would rather just pick the right card to suit the maximum output of PSU can handle.

In your article. It is limited to 204W. But in my native installation. My Vega 56 would draw from:

PCIe Slot 1 = 2A max (12v x 2A = 24W)
PCIe Booster A = 8A max (12v x 8A = 96W)
PCIe Booster B = 8A max (12v x 8A = 96W)

Total drawn = 216W.

In AMD website, it spec is some kind of rated at 210W max. It hits 216W in my case without triggering any abnormal accidental reboot or power interruption. It is stable when working at this power consumption. The total PSU 12v output at max is around 460W when exporting a Final Cut Pro Prores422 project to h264 4k file in my system.

8A is the display limit, your card actually draw more than that (but should be still quite a bit below the shutdown limit).

Anyway, there is practically no degrade in performance, my Radeon VII still perform at 50000 Luxmark range, always run at or above the base clock. I just make it draw less and run cooler.

That 204W is the card‘s max power draw measured (in Windows) when there is no artificial power draw limit (at the optimum voltage). So, set it to 204W shouldn’t cause any performance degradation. If I want to make sure 100% no limit on performance, I can’t make it 225W. I just make the card works better (cooler and quiter). There is nothing illogical.

That 204 isn’t for fitting the PSU limit, but the optimum setting for this specific card. I just want to show the others that can be done (no PSU mod to run a Radeon VII).

 

[solaris8x86]

8A is the display limit, your card actually draw more than that (but should be still quite a bit below the shutdown limit).

Anyway, there is practically no degrade in performance, my Radeon VII still perform at 50000 Luxmark range, always run at or above the base clock. I just make it draw less and run cooler.

That 204W is the card‘s max power draw measured (in Windows) when there is no artificial power draw limit (at the optimum voltage). So, set it to 204W shouldn’t cause any performance degradation. If I want to make sure 100% no limit on performance, I can’t make it 225W. I just make the card works better (cooler and quiter). There is nothing illogical.

That 204 isn’t for fitting the PSU limit, but the optimum setting for this specific card.

But my benchmark vs watt consumption matched out.

So I assume all of my 8A + 8A + 2A = 216W vs AMD claims = 210W are true values from iStat display due to mathematical matched out from the spec. Due to my card's physical performance can never beyond its factory spec.

I don't doubt the VII could use a less power wattage to produce the same performance as this is what all the new generation graphic cards do (less power consumption, higher performance). It has been a trend.

 

[h9826790]

But my benchmark vs watt consumption matched out.

So I assume all of my 8A + 8A + 2A = 216W vs AMD claims = 210W are true values from iStat display due to mathematical matched out from the spec. Due to my card's physical performance can never beyond its factory spec.

I don't doubt the VII could use a less power wattage to produce the same performance as this is what all the new generation graphic cards do (less power consumption, higher performance). It has been a trend.

I believe that's very very close to 8A, but just tiny bit over 8A. What I want to point out is that don't trust that 8A, when it shows 8A, it simply means "I can't show you the higher value". But in your case, that doesn't a big issue.

 

[solaris8x86]

I believe that's very very close to 8A, but just tiny bit over 8A. What I want to point out is that don't trust that 8A, when it shows 8A, it simply means "I can't show you the higher value". But in your case, that doesn't a big issue.

That's right. It is now already reaching to a performance of MacBook Pro 2019 16' model. It's a giant leap for earning. I have already earned 10 year time and expenditure on Apple.... with an expense of USD$250 only... who knows will it run for 5 more years... till Apple no longer support Intel binary and 100% moves to M1. By that time, it will become a 15 year old computer...

Jesus.. Apple really hard to earn my money.

 

[solaris8x86]

I believe that's very very close to 8A, but just tiny bit over 8A. What I want to point out is that don't trust that 8A, when it shows 8A, it simply means "I can't show you the higher value". But in your case, that doesn't a big issue.

Can you do a render test for me with your VII?

1. Create a 4K Final Cut Pro project with 4K h264 clip sources.
2. Make sure the clips are transcoded to Prores422 first.
3. Apply 2 x LUT filters on the clips. And export to HEVC 70Mbps bitrate and H264 70Mbps bitrate (single pass) respectively for a duration of exactly 5 min long each movie.

See how many second/min they go for each test. If you have time to do so, of course. Thanks.

 

[h9826790]

I don't mind, but

1) We need an identical sample video in order to make the test comparable

2) I can't access my cMP in the next few months. So, I won't be able to do that at least until the end of summer holiday 🥺

 

[solaris8x86]

I don't mind, but

1) We need an identical sample video in order to make the test comparable

2) I can't access my cMP in the next few months. So, I won't be able to do that at least until the end of summer holiday 🥺

If you have no Mac Pro. You are not even doing anything..

 

[KeesMacPro]

won't do such a silly thing to get a more expensive card and intentionally to degrade it thru downgrade procedures just because to suit the rating of PSU. It's illogical. I would rather just pick the right card to suit the maximum output of PSU can handle.

In your article. It is limited to 204W. But in my native installation. My Vega 56 would draw from:

PCIe Slot 1 = 2A max (12v x 2A = 24W)
PCIe Booster A = 8A max (12v x 8A = 96W)
PCIe Booster B = 8A max (12v x 8A = 96W)

Total drawn = 216W.

In AMD website, it spec is some kind of rated at 210W max. It hits 216W in my case without triggering any abnormal accidental reboot or power interruption. It is stable when working at this power consumption. The total PSU 12v output at max is around 460W when exporting a Final Cut Pro Prores422 project to h264 4k file in my system with Vega 56. It really nicely hitting the physical ceiling of what the Mac Pro PSU can supply the power.

Sounds to me like a couple of assumptions in your posts e.g. "If you have no Mac Pro. You are not even doing anything.."

Personally I would start with a Pixlas mod...

 

[solaris8x86]

Sounds to me like a couple of assumptions in your posts.

Personally I would start with a Pixlas mod...

There is no assumption. It is the spec from AMD and the fact data matches with the installation. That's all. It's just applied science proven with data. and the data also matched with the spec.

Pixel mod would just ruin the power cord unnecessary for Vega 56. May be required for Vega 64 or above. Pixel mod is a nasty solution by damaging the cords. I would rather to use an external power supply for safety if I really need it.

But I don't in my case as the spec of Vega 56 is within the PSU spec. Any mod is unnecessary. Just plug and play. The stability and performance will come and working at ceiling natively.

 

[KeesMacPro]

No time for this.....

 

[solaris8x86]

No time for this.....

Of course, mine is already working flawlessly.

 

[drunkmouth]

Had a skim through but can't see anything. Upgraded to 0.6.8 OpenCore but can't install Big Sur on my 5,1. Keep getting stuck in a boot loop. It installs then will restart the computer as expected, then launch the installer progress bar but restart at around 10% and repeat. Any suggestions or am I missing something painfully obvious? Thanks!