Flashed 270X may only work in High Sierra or lower.
Mojave may shows black screen after driver loaded.
Original (unflashed) 270X can work in Mojave without any issue.
Same for 78xx.
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 1,1-5,1 GPU Compatibility List for cMP
- Thread starter Pentaxer
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
Flashed 270X may only work in High Sierra or lower.
Mojave may shows black screen after driver loaded.
Original (unflashed) 270X can work in Mojave without any issue.
Same for 78xx.
About two years ago I got hold of a "made in Japan" knock off R9 270X" definitely non-reference.
I have since sold it but it ran OK used the Netkas 7950efi. Showed bootscreens.
To be honest 7950's and in particular .. the HIS 7950 ( best cooling ever ! ) are great for low demand users .. . 7970's for more stressful stuff.
Sorry i didn't keep records for the R9 270X and the ROM I made is probably useless for reference cards.
The Gigabyte 280X variant doesn't block slot 2, I have a NVMe adaptor card there, fits easily. Had a Sapphire 7970 before that didn't block slot 2 either. When in slot 2 though both would block slot 3 of course. Also there are cards with more than 2 slots height, so maybe this point needs some clarification.
Last edited:
Need additional information.
1. It can't be flashed to have Boot Screen. Right?
2. What the earlist version of macOS does it support?
1. Can be flashed by MVC only...
2. Sierra would be the earliest supported OS, as first Nvidia web drivers for this chipset were released in April of 2017.
2. Sierra would be the earliest supported OS, as first Nvidia web drivers for this chipset were released in April of 2017.
Last edited:
26. NVIDIA Geforce GTX 1080 8GB Founders Edition (FE), Unflashed
Performance: 259% RPI.
Power consumption: needs additional power (8 pins).
Outs: 3 x DP, HDMI, DL-DVI.
Compatibility: DirectX 12.0, OpenGL 4.6, Metal GPUFamily1v4. macOS 10.12.x - 10.13.6.
Other: occupies 2 slots, no Boot Screen. NVIDIA chipset - Pascal. NVIDIA CUDA cores - 2560. User guide.
Unflashed (PC version): works without Boot Screen; can be flashed only by MVC.
#1 - As noted, needs to be flashed by MVC.
#2 - As noted above, Sierra at earliest.
Will note, this GPU does not need "additional power". The MP5,1/MP4,1 provides enough power without any modifications using both dual mini 6-pin cables. This is not "self powered" via PCIe, however. The stat, while it may be technically accurate, can be misleading and confusing. If this is to be a reference database for the future, would suggest rewording or relabeling cards powered this way.
Let's try that way - the variants:
1) from PCIE slot only;
2) 6 pins / dual 6 pins
3) 8 pins (cMP dual 6 pins -> 8 pins adapter is required);
4) 6 pins + 8 pins (needs some additional power from SATA);
5) dual 8 pins (needs much additional power - PIXLAS mod is required).
We have found the chipest ond coolest GPU with Mojave support - Sapphire Pulse Radeon RX 550. It's TDP is only 50 Watts (RX 560 has ~ 75 Watts TDP).
A man in the link above told, that 2 Gb card version doesn't work.
Need additional information - do another RX 550 4 Gb work in Mojave?
A man in the link above told, that 2 Gb card version doesn't work.
Need additional information - do another RX 550 4 Gb work in Mojave?
Very interesting document.
The opinion: each 6 pins connectors on MacPro's motherboard can provide 120W.
But i don't want to venture with my cMP
The opinion: each 6 pins connectors on MacPro's motherboard can provide 120W.
But i don't want to venture with my cMP
Today I was testing in Win10 Furmark my RX580 6 pin card. I tried a stress test then minimized the window then pressed again on the window in the taskbar and it restarted. Repeated the test 3-4 times with the same result. I measured the current with a Fluke 189 and a Fluke probe recording max current. Max current peak recorded was 8.1 Amps (around 96 W) from the boost connector A. I repeated the same test with dual 6 mini pin to 6 pin connector without any problems. What I noticed is that boost A and boost B were providing different amperage which means the impedance of the both branches is different. Boost A was 2.2 and Boost B 3.3 Amps. The moral of the story is that even a card with 75W input may need more power to work properly in extreme conditions.Very interesting document.
The opinion: each 6 pins connectors on MacPro's motherboard can provide 120W.
But i don't want to venture with my cMP
AMD Radeon RX 460 2 Gb
Performance:
Power consumption: from PCIE slot only. (75 W)
Outs: DL-DVI-D + DP + HDMI.
Compatibility: DirectX 12, OpenGL 4.5, Metal 2
Other: occupies 2 slots, no Boot Screen.
Unflashed (PC version): works out of the box
Not sure how to get the RPI so I'm leaving it blank. This was put into a cMP 4.1>5.1. Was able to install Mojave with no issues.
Performance:
Power consumption: from PCIE slot only. (75 W)
Outs: DL-DVI-D + DP + HDMI.
Compatibility: DirectX 12, OpenGL 4.5, Metal 2
Other: occupies 2 slots, no Boot Screen.
Unflashed (PC version): works out of the box
Not sure how to get the RPI so I'm leaving it blank. This was put into a cMP 4.1>5.1. Was able to install Mojave with no issues.
FYI
gtx titan black is working as well as the gtx 780ti, compared to eachother i see no different in them because they are based on the same kepler gpu. the benefit of titan black is the 6gb memory, while the 780ti is a little bit higher clocked.
i used a gtx780, gtx780ti and titan black in my cMP.
power consumption is too way high, under full load i always got system shut down.
i prevent this by using an evga powerlink adapter.
AFAIK, there is no custom rom to efi flash it by yourself.
it works OOB in mojave but cuda acceleration is not available because of no driver from nvidia.
gtx titan black is working as well as the gtx 780ti, compared to eachother i see no different in them because they are based on the same kepler gpu. the benefit of titan black is the 6gb memory, while the 780ti is a little bit higher clocked.
i used a gtx780, gtx780ti and titan black in my cMP.
power consumption is too way high, under full load i always got system shut down.
i prevent this by using an evga powerlink adapter.
AFAIK, there is no custom rom to efi flash it by yourself.
it works OOB in mojave but cuda acceleration is not available because of no driver from nvidia.
Can you please boot into windows and share with us a screenshot of techpowerup's GPU-Z output? This helps us identify which brand/model/GPU chip/memory speed is compatible without drivers.
That 8A is display limit. Your card most likely drawing more 100W at that moment.
The difference between A and B should be erroneous. If you install the card in slot 2, that difference should disappear.
For stress test, should not use window mode. But use Full screen mode, native (or even higher) resolution, no AA.
[doublepost=1554630938][/doublepost]
Kepler has Rev B card which was not natively supported in OSX. But Apple fixes that few years back, all Kepler GPU should be supported natively in MacOS.
dont have WIN on cMP but can take a shot later from pc...no problem.
but there arent different versions of the titan black, all are based on on kepler gpu with gk110-430-b1 chip and 2880 su's.
the 780 w/wo is using the gk110-300 or 425 gpu's, while the 425 variation also have 2880 shading units but the 300 just have 2304.
all of them have nativ supported in macOS since lion AFAIK.
but there arent different versions of the titan black, all are based on on kepler gpu with gk110-430-b1 chip and 2880 su's.
the 780 w/wo is using the gk110-300 or 425 gpu's, while the 425 variation also have 2880 shading units but the 300 just have 2304.
all of them have nativ supported in macOS since lion AFAIK.
Good to know. I am still confused as to why he would recently list it as requiring drivers and not supported though.
[doublepost=1554631388][/doublepost]
Thank you for sharing this information.