I'm not sure if it's a GPU model ex Pascal or Web Driver issue, but I've personally never had any issues my GTX 960 using the Web Drivers
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.
AMD Radeon RX Vega Plug-and-Play in 10.13
- Thread starter theitsage
- Start date
- Sort by reaction score
I hear of less issues with Maxwell compared to Pascal as well.
Makes me think that perhaps the GTX 980 may be the BEST option....
Can you actually screen share using an iPad for the web drivers? I heard the GPU has to be installed for the web drivers to even initialize. IS this true?
Makes me think that perhaps the GTX 980 may be the BEST option....
Can you actually screen share using an iPad for the web drivers? I heard the GPU has to be installed for the web drivers to even initialize. IS this true?
I don't know, but I'm sure someone will know.
[doublepost=1505282972][/doublepost]
I have also heard of this happening with 9.. series cards.
No, I have tested ASUS RX 580 8GB, since we ordered s big batch, and the pulse was not available at that time.
The results are mixed though, so I am trying to find a pattern and then post my findings
Yea, me too. The problem with buying a GTX 980 today is that you'll spend around the same amount used that you would for a brand new GTX 1060 8GB. And 980 = 1060 = RX 480/580. So buying a GTX 980 seems a bit silly today, unless you could score one for a superb deal.
Alright, results are in.
After multiple tests, these are my observations for the RX 580 model we have in stock (Asus RX 580 dual fan):
• The gpu power draw somehow caps at 180W, and about 355W for two
• The gpu often draws most power from the Pcie Slot, and not from the aux 1 and 2 mini-6 pins
• The gpu power draw from the Pcie slot comes close to the max allowed (75W) but never exceed 70W
• Luxmark of about 14700 points (luxBall) and Geekbench OpenCL of about 139000 points for this model
.. so all in all a great GPU for these cMP.
Dual Rx 580 definitely shine.
ps. the inconsistencies were just me looking at the wrong numbers during late-office hours
After multiple tests, these are my observations for the RX 580 model we have in stock (Asus RX 580 dual fan):
• The gpu power draw somehow caps at 180W, and about 355W for two
• The gpu often draws most power from the Pcie Slot, and not from the aux 1 and 2 mini-6 pins
• The gpu power draw from the Pcie slot comes close to the max allowed (75W) but never exceed 70W
• Luxmark of about 14700 points (luxBall) and Geekbench OpenCL of about 139000 points for this model
.. so all in all a great GPU for these cMP.
Dual Rx 580 definitely shine.
ps. the inconsistencies were just me looking at the wrong numbers during late-office hours
Alright, results are in.
After multiple tests, these are my observations for the RX 580 model we have in stock (Asus RX 580 dual fan):
• The gpu power draw somehow caps at 180W, and about 355W for two
• The gpu often draws most power from the Pcie Slot, and not from the aux 1 and 2 mini-6 pins
• The gpu power draw from the Pcie slot comes close to the max allowed (75W) but never exceed 70W
• Luxmark of about 14700 points (luxBall) and Geekbench OpenCL of about 139000 points for this model
.. so all in all a great GPU for these cMP.
Dual Rx 580 definitely shine.
ps. the inconsistencies were just me looking at the wrong numbers during late-office hours
Thanks for the detail report. So, even Furmark won't draw >75W from slot?
When running FurMark, make sure to test with a high resolution (I've used 2560x1440) and no MSAA to maximize power draw.
This is awesome. Thank you very much for doing this!Alright, results are in.
After multiple tests, these are my observations for the RX 580 model we have in stock (Asus RX 580 dual fan):
• The gpu power draw somehow caps at 180W, and about 355W for two
• The gpu often draws most power from the Pcie Slot, and not from the aux 1 and 2 mini-6 pins
• The gpu power draw from the Pcie slot comes close to the max allowed (75W) but never exceed 70W
• Luxmark of about 14700 points (luxBall) and Geekbench OpenCL of about 139000 points for this model
.. so all in all a great GPU for these cMP.
Dual Rx 580 definitely shine.
ps. the inconsistencies were just me looking at the wrong numbers during late-office hours
I'd also like to see a table of supported RX 580s. Looking for just the 8GB Pulse model is futile.
Glad to hear that Asus 8GB RX 580 seems to be good....
Keep the testing going guys!
I'd help with the testing, but the RX 480/580's are $200-250 cards to me. They are not $300-500 performers, so I feel kinda ripped off at the current inflated prices.
Glad to hear that Asus 8GB RX 580 seems to be good....
Keep the testing going guys!
I'd help with the testing, but the RX 480/580's are $200-250 cards to me. They are not $300-500 performers, so I feel kinda ripped off at the current inflated prices.
Last edited:
Which GPU are you looking for? All the Vega cards currently available are reference cards.
Here are a couple links for a reference RX 480 and RX 580:
XFX reference RX 480 @ Newegg: https://www.newegg.com/Product/Prod...6&cm_re=xfx_rx_480-_-9SIABVR6475376-_-Product
MSI reference RX 580 @ Newegg: https://www.newegg.com/Product/ComboBundleDetails.aspx?ItemList=Combo.3629238
Last edited:
I had a Sapphire Pulse RX580 8GB delivered earlier this week. Tests were run on a MacPro 5,1 w/ X5690 on macOS 10.12.6.
Assuming W = V * A (I don't know if this is correct)
Volts (12.15) and Amps are from iStat Menu readings
LuxMark 13334
LuxBall HDR Benchmark (OpenGL GPU)
Individual Max for PCIe Boost A/B - 52W (4.25A) and 55W (4.5A), Max for PCI Slot 1 - 55W (4.5A)
FurMark 950 points (FPS: 94)
open 'GpuTest.app' --args '/test=fur /width=2560 /height=1440 /msaa=0 /benchmark /benchmark_duration_ms=10000'
Individual Max for PCIe Boost A/B - 79W (6.5A) and 85W (7.0A), Max for PCI Slot 1 - 80W (6.6A)
Geekbench 4
OpenGL Score: 137129
Assuming W = V * A (I don't know if this is correct)
Volts (12.15) and Amps are from iStat Menu readings
LuxMark 13334
LuxBall HDR Benchmark (OpenGL GPU)
Individual Max for PCIe Boost A/B - 52W (4.25A) and 55W (4.5A), Max for PCI Slot 1 - 55W (4.5A)
FurMark 950 points (FPS: 94)
open 'GpuTest.app' --args '/test=fur /width=2560 /height=1440 /msaa=0 /benchmark /benchmark_duration_ms=10000'
Individual Max for PCIe Boost A/B - 79W (6.5A) and 85W (7.0A), Max for PCI Slot 1 - 80W (6.6A)
Geekbench 4
OpenGL Score: 137129
Seems like thus far, the RX 580 is not burning out the PCI-E slot...
Now, to wait for them to be to $200-250 where they belong.
Now, to wait for them to be to $200-250 where they belong.
Are these values really within tolerances then? Is the cMP safe while this amount of power is being drawn for a prolonged period of time?
PCIe slots are rated for 75w, Mac Pro or not.
If RX580's had a chance to burn slots, you'd see it happening with PCs, not just Mac Pros.
This has been said before, but it's a good read: https://videocardz.com/61667/what-r...80-exceeding-pci-express-power-specifications
If RX580's had a chance to burn slots, you'd see it happening with PCs, not just Mac Pros.
This has been said before, but it's a good read: https://videocardz.com/61667/what-r...80-exceeding-pci-express-power-specifications
Strightly speaking, No. However, the Luxmark value looks very good to me. That's a very good normal high demand reference.
Unigine Heaven / Valley should be more demanding as well (peak value), but with that power draw. Seem no real world usage can exceed the limit.
By considering Furmark can draw 15-20% more than almost any other demanding operation. I expect the card will only draw up to 65W from the slot and the 6pins in any real world operation. Even if we lower the buffer to just 10% only, it should still able to just stay within 75W.
Last edited:
Just ordered a brand new reference MSI RX 580 8GB. I'll let you guys know how it is. It's discontinued from MSI and I guess some stores are selling off their batches.
Amazon has a few sellers that have them in stock. $306 shipped, no tax. It's $50 too much, but it's finally time to upgrade and prepare for VR while keeping MacOS.
Amazon has a few sellers that have them in stock. $306 shipped, no tax. It's $50 too much, but it's finally time to upgrade and prepare for VR while keeping MacOS.
True. I also forgot to say I actually also tested Valley and Heaven, and the power draw was in line with LuxMark (just about 5-10 W more, spread across the connections, but never reaching the limit), and a little lower than GeekBench (OpenCL or Metal)
Does the Sapphire 580 Pulse card offer any functional advantage over other brand cards other than showing correct name in 'About this mac'? I'll appreciate if someone can help me understand the difference between various brands before I jump to some non Sapphire Pulse card. Also, is there any difference in the Sapphire cards with Part#: 113-4E3531U or 113-4E353BUpart (https://forums.macrumors.com/thread...os-10-12-6-10-13.2050945/page-3#post-24819720) and the one with Mfg. Part: 11265-05-20G?
The difference at this moment should be limited to correct identification. If don't want to wait, or prefer better / cheaper RX580, other brands / models should work as well. However, you better make sure they are the reference card. Some brand like XFX love to produce custom PCB with custom ROM, those cards have high chance not working in cMP.