Yes. But considering the Sapphire Rx 580 pulse’s TDP is 145W, I believe Vega 56 reference’s 150w TDP should be safe.
Last edited:
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
But it's not. Reference VEGA 56 total board power is 210W, pulling up to ~360W under load. VEGA 64 goes up to 410W.
Last edited:
Update after 24 hours.... no unusual fan noise, I did run Luxball and the fan did spin up, but not too loud and shortly after it spun down again.
Performance in FCPX seems improved which is what I wanted.
[doublepost=1556481334][/doublepost]
where do you find these numbers? i looked on the Sapphire website and found very little info.
Vega 56 reference has two bioses which have two different TDP (150w & 165w). You better have a source of your number. Please note, some people flash their vega 56 with a vega 64 bios to increase TDP. We are here to talk about vega 56 reference original eco bios.
AMD Radeon RX VEGA 56: Typical Board Power (Desktop) 210 W.
All reviews mention this. All reviews show ~360 W under load for VEGA 56 and ~410 W for VEGA 64 under load.
I have no idea where you see those reviews. Here is review.
https://www.tomshardware.com/reviews/radeon-rx-vega-56,5202-21.html
The typical bios 2 power is about 202w.
Another test says Vega 56 power is 237W under Furmark(most likely bios 1 result).
https://www.techpowerup.com/reviews/AMD/Radeon_RX_Vega_56/29.html
But, here is Sapphire pulse RX580 which recommended by Apple. Its power comsuption is 234W under load.
https://www.pcworld.com/article/319...ew-solid-gaming-on-a-tight-budget.html?page=8
Last edited:
Not sure where to put this, but I figured this crowd might be interested.
AMD GDDR6 GPU PCB leaked, baseless speculation video below:
AMD GDDR6 GPU PCB leaked, baseless speculation video below:
Don't compare the TDP of Polaris and Vega cards to make a determination. Polaris cards take a good portion of power from the PCIe slot. Vega cards take very little power from the PCIe slot. So little that you should disregard the PCIe slot from the equation. When a Vega card wants juice it will pull it from the boost connections. It's a good thing. Just make sure it can get enough juice when it needs it. It's too dangerous to use the logic board connections IMO. Connect it directly to a power supply.
According to Alchemistics post and Sytem reports it seems that the latest beta drivers send more power through the pci slot than previously.
AMD Polaris & Vega GPU macOS Support
maybe if playing games or running endless Furmark tests then Pixlas mod would be recommended, for my use in FCPX im happy with my set up. VEGa 56 runs quietly and is faster than the RX480 that i had installed before.
I missed that. That's really interesting. It looks like it was drawing about 1.5A from the slot in 10.14.4 and about 2.5A in 10.14.5.
Can you tell me what diagnostic is used for those reports? I would gladly add reports to help the community.
Yes, this is what I see, too. I was holding a test report for Vega 56 reference on 10.14.5.
Last edited by a moderator:
H264 de and encode work with a little bit Lilu
Did you perform the modifications in the other thread, or is this native?
Sapphire Pulse Vega 56 in a Mac Pro 2010 (5,1). Xeon 3.46GHz 12 core, 64GB RAM. Mojave 10.14.4.
Under LuxMark OpenCL Stress Test Full Load. Watt consumption is.
Here is the metal score by Geekbench 4.
Here is the OpenCL score by Geekbench 4.
I'm a Final Cut Pro user. I have some findings. When export a movie using h264 coding. It can fully draw the GPU power at full loading. The Xeon CPU loading (usr) would become lower. I can tell because it draws 4A on each PCI-e AUX-A and B port. The export speed is fairly fast. However, when export a movie using h265 (HEVC) coding. Final Cut Pro wouldn't make use the GPU performance but from the Xeon CPU. I can tell because it draws less than 1A on each PCI-e AUX-A and B port. It seems that this card doesn't process h265 hardware acceleration or encoding feature by Final Cut Pro. The export time of a movie using h265 is 5 times longer than h264 in Final Cut Pro X when exporting the same length of movie clip.
To all Final Cut Pro users, be aware.
Here is a power consumption when exporting a movie by Final Cut Pro.
- No fan top-speed issue. It turns on when it needs. Most of the time both fans are off automatically if the environment is cool (my room).
- Just direct plug-and-play. No need any power-related modification. It just works out.
- The card is 28cm long. It can fit into the PCI bay flawlessly. No need to remove the PCIe fans in the front.
Under LuxMark OpenCL Stress Test Full Load. Watt consumption is.
- PCIe AUX-A = 12.3V x 6.79A = 83.5W
- PCIe AUX-B = 12.3V x 7.99A = 98.2W
- PCI slot = 12.3V x 1.65A = 20.2W
Here is the metal score by Geekbench 4.
Here is the OpenCL score by Geekbench 4.
I'm a Final Cut Pro user. I have some findings. When export a movie using h264 coding. It can fully draw the GPU power at full loading. The Xeon CPU loading (usr) would become lower. I can tell because it draws 4A on each PCI-e AUX-A and B port. The export speed is fairly fast. However, when export a movie using h265 (HEVC) coding. Final Cut Pro wouldn't make use the GPU performance but from the Xeon CPU. I can tell because it draws less than 1A on each PCI-e AUX-A and B port. It seems that this card doesn't process h265 hardware acceleration or encoding feature by Final Cut Pro. The export time of a movie using h265 is 5 times longer than h264 in Final Cut Pro X when exporting the same length of movie clip.
To all Final Cut Pro users, be aware.
Here is a power consumption when exporting a movie by Final Cut Pro.
- PCIe AUX-A = 12.3V x 3.68A = 45W
- PCIe AUX-B = 12.3V x 4.6A = 57W
- PCI slot = 12.3V x 1.65A = 20.2W
Last edited:
The Radeon RX VEGA 56 has a TDP of 210 W. It will exceed that in certain games or applications.
May be in games. But my routine is mainly on FCPX jobs. It has never drawn higher than 122W. And under the LuxMark stress test. It was max at 201W only and I've never experienced any abnormal power cut issue. So, in my applications. It works out without the need of any power-related mods. Besides, its BIOS-1 has a power limitation at 225W max. The Mac Pro 2010 would just fine to handle this power consumption.
I feel like we need to test at the wall instead of using software.
I run a VEGA FE using only the dual 6-pin motherboard connectors.
Actually, I did some research before buying the card. I know Vega 64 (300W) and Frontier (300W) would go beyond the Watt capacity the Mac Pro 2010 can provide. So when using 64 and Frontier they need to do a mod is understandable.
Each PCIe AUX-A = 75W + PCIe AUX-B = 75W + PCI slot = 75W. Total max load it can provide is = 225W where AUX-A and B can provide 75W max is a theoretical technical specification. In my real case. Each AUX port can actually support up to 8A. Which is actually 98W each.
Therefore, in a practical world (tested from my case by LuxMark Stress Test). The Mac Pro 2010 can provide up to 100W + 100W + 75W = Total 275W max for the graphic card. This is really the ceiling that it can handle.
My objective is I don't want to do a power-mod to my Mac but demands a similar performance from Vega 64. So I got a Vega 56 which its rating is 180W max on AMD website (your 210W doesn't matter). In my case, it is 201W max when using a stress test software. Practical application on FCPX on max load is actually just 122W (It won't go higher anymore in ANY cases that I have tried. It just won't draw more power for the card. 122W that is.). It's far less than the theoretical ceiling at 225W and practical ceiling at 275W max. So it is pretty safe now. And when comparing to the OpenCL/Metal scores. The Vega 56 and 64 are very close. So, I've achieved my objective now (no need to do a mod but it comes with a pretty close to a Vega 64 performance). The Vega 56 is the best pick for an unmodified Mac Pro. Win-win.
Last edited:
It will never pull 75W from the PCIe slot though, highest I've seen is around 2.5A (30W).
But good thing your setup is stable
There is the point of view that each cMP's microFIT 6 pins connector can provide 120W+. Look at the spoiler of p. 3.3.