RX590 should work
For 10.14.4, don't know any difference. But for older OS, without Orinoco framebuffer may cause sleep / wake issue.
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
RX590 should work
For 10.14.4, don't know any difference. But for older OS, without Orinoco framebuffer may cause sleep / wake issue.
but as i know rx590 will be identified not like rx590 by default? it will be like radeon polaris 8gb something like that...
Hello everyone, super late to this thread but just thought I’d share my finding with regards to Vega in the cMP! My Mac Pro is a 2010 server model, can get into specs but don’t know if that’s super relevant. I do have a SYBA IO crest installed and due to my local market having a surplus of $350 AirBoost Vega 56s I decided to pick one up (for reference, a RX 580 is nearly $300 after taxes new, life’s hard with Canadian Ruples). The blower style slightly attracted my attention, as it means the fan is only partially blocked off by the IO crest installed in the slot above. And so, after living with this card for about a month and a half, here are my findings: the air boost (from my experience) can be powered purely and benchmarked (heaven and valley specifically) through dual SATA to 8 pin off the optical bays and a dual mini 6 pin to 8 pin. I did encounter some issues very recently though where the Mac Pro would just shut off during a heaven benchmark. This problem only persisted in heaven so I’ll do more investigation. I am fairly new to this, but thought I’d share! All testing was done under Mojave 10.14.4. I will continue testing soon enough. I’d like to reiterate this only happens under heaven, valley worked flawlessly in all my testing. So maybe heaven is just being weird. As well, don’t take my use of the dual 6 pin to 8 pin and sata connectors as something smart to do - pixalas mod or separate power supply if you can. I’m just testing this for the time being until I have the time to do the mod properly.
Edit - it seems to be a problem with heaven, it shuts down sometimes sometimes doesn’t. I have yet to replicate this in valley. Can’t seem to pinpoint the problem - likely has to due with an uneven draw from heaven putting too much pressure on the SATA connectors. The applications I use this for mean I will likely never even come close to the power ceiling, but even then it’ll still be doing the pixlas mod and testing other things out to see if I can come across a solution. The card theoretically gets 150W from the dual 6 pins and should get 75+W from dual sata and theoretically 75W from PCIe but it likely uses 40W. MSI says the card should be more than satisfied with 210W. Uneven draw is likely the culprit. Even when pushing with highest 1440P settings in Valley I can’t get the system to restart - and I haven’t had this problem before. Potentially a problem with heaven, potentially with my way of powering the card (likely the latter).
Edit - it seems to be a problem with heaven, it shuts down sometimes sometimes doesn’t. I have yet to replicate this in valley. Can’t seem to pinpoint the problem - likely has to due with an uneven draw from heaven putting too much pressure on the SATA connectors. The applications I use this for mean I will likely never even come close to the power ceiling, but even then it’ll still be doing the pixlas mod and testing other things out to see if I can come across a solution. The card theoretically gets 150W from the dual 6 pins and should get 75+W from dual sata and theoretically 75W from PCIe but it likely uses 40W. MSI says the card should be more than satisfied with 210W. Uneven draw is likely the culprit. Even when pushing with highest 1440P settings in Valley I can’t get the system to restart - and I haven’t had this problem before. Potentially a problem with heaven, potentially with my way of powering the card (likely the latter).
Last edited:
10.14.5 beta 2 (build 18F108f) contains 4 new Vega frame-buffer personalties.
-Guanta
-Hullinco
-Ichabod
-Itata
Itata looks like it may be Vega 10. The others may be Vega 20. It would be interesting if one of them is for Radeon VII.
Guanta has 4 outputs according to the frame-buffer-lister script. Hullinco has 2. Ichabod has 1. And Itata has 6.
Update: Radeon VII is using the default frame-buffer personality. Thanks to user eksu. Link
-Guanta
-Hullinco
-Ichabod
-Itata
Itata looks like it may be Vega 10. The others may be Vega 20. It would be interesting if one of them is for Radeon VII.
Guanta has 4 outputs according to the frame-buffer-lister script. Hullinco has 2. Ichabod has 1. And Itata has 6.
Update: Radeon VII is using the default frame-buffer personality. Thanks to user eksu. Link
Last edited:
I received a replacement Pulse Vega 56 from Sapphire's warranty service provider Althon Micro. I installed the card and I ran LuxMark. So far so good! Score 26411 in 10.14.4. My experience with Althon Micro was pretty good. It was kind of weird dealing with Sapphire though. When I opened a ticket on their website and sent them a message they responded that I had to contact my "local point office," whatever that is, to request warranty service. I dug around their site and found the email for Althon Micro so I emailed them and they immediately responded that I fill out an RMA request form. I submitted that and then a couple days later I heard back from Sapphire again who told me to contact Althon Micro and to send them a copy of my receipt. So I did that and a little while later that day I got an RMA number with instructions on sending them the GPU. About a week later I have the replacement. If I need to request warranty service again I think I'll bypass Sapphire and just email Althon Micro an RMA form and my receipt.
Point often means lead or "head", or tip of the spear/arrow/blade; probably a translation error. Local authorized distributor sounds about right.
Thanks Andrew for the heads up. Since the 10.14.5 beta somehow influences the fan behavior of Vega cards, it could be connected to the new iMac Pro Vega64X card from the 2019 facelift. Maybe there is one Vega64X frame buffer reserved for this card.
It basically means a Vega card will become the very best choice for 5.1 owners. I sincerely hope this is the Scenario. I will test it as soon as the 10.14.5 is official.
Even on a Gigabyte 56 ref the fan speeds are normal on stock ROM!
That's great! This was a long time coming!! But better late than never.
[doublepost=1554906518][/doublepost]
Possibly. I think the Vega 64X is just a higher clocked Vega 64, so it should be using the same framebuffer as the 64.
Same here.
I tried every possible rom with less TPD without luck so far.
Slighty underclocking could potentially remove the need for pixlas mod.
Doing this with wattman enables me to game under w10 without power failures.
But there‘s no way to edit the roms permanently cause only certified roms are accepted by the card.
Under 10.14.5 beta I no longer have fan issues on stock ROM(reference vega 56). Furthermore, I can run Furmark at 1080p with no system shutdown so the power supplied to the card with this ROM seems sufficient, not needing Pixlas
You may try Furmark, full screen, native resolution, no AA.
IMO, this is the real correct setting to test if the card is properly power.
I flashed the XFX Double Edition 64 bios to the Pulse 56 and the fan behaves the same in macOS as it does in Windows. The downside is that the fan runs about 957 rpm at idle (as opposed to zero RPM) with this bios, but it is still rather quiet. That must be due to the increased clocks & voltage. LuxMark score is about 2000 higher. Haven't noticed any issues. Bios is here in case anyone wants to use it: RX-VEGATDFF6. It's the only nano PCB 64 bios I know of.
No, some fan issues persist, e.g. fan doesn't spin up past the acoustic limit, but even by setting the limit to max rpm, fan rpm never reaches max rpm. I can only hit about 2200rpm running furmark, max rpm is set to 3300.
Then again, reading early reviews from when the cards were released complain about the fans a lot, Windows solves this with Wattman, we can only edit the ppt in kext or flash another firmware...
You can add the powerplaytable to the kext, look for VGtab.
Fyi It's included in GPUtest, not as a stand-alone app on OSX.
Mine also defaults to ~900rpm, which is a value I can't find in the powerplaytable, although TargetGfxClk is close, haven't tried editing this yet.
Invisible ink?!?
Interesting, the dark mode extension (Brave browser) only change that post’s front colour, but not other posts. I have absolutely no idea about this until I read this message again on my phone.
[doublepost=1554975192][/doublepost]
Obviously he know where to find Furmark, he used that already. I just suggest a better setting for him to do power draw test.
But yeah, I often seen other members said there is no Furmark for MacOS.
Thanks! I tried it today.
Installing 10.4.5 beta on my MacPro 5,1 fried my LG5K's backlight. I do use the monitor in conjunction with the TB3 TitanRidge Card. It started flickering during the installation, so I've disconnected the monitor. Afterwards even connecting the LG to a MacBook Pro gave me flickery backlight on higher brightness.
Might be a coincidence. Apple already exchanged the monitor though.
Regarding the GPU. The Sapphire Vega 56 (Reference cooler) on BIOS switch nearer to the fan shows me the 150w TPD in GPUz.
After the update the fan issue is gone! It does idle at ~700 instead of turning off completely now, which I prefer! It did a high pitch noise before, every time the fans started spinning. It does run cooler ~40° (before ~50°) at idle now as well - obviously. (Screenshot added)
I did stress test the gpu with furmark on highest resolution. What I observed:
The power draw never goes near 7.99A on each PCIe Boost Port! It used to go above in the past, with the 180w TPD. It therefore stays within an acceptable range and only goes up to around 7.5A.
(see Screenshots of istat showing furmark)
Compared to heaven the PCIe Boost draws are very similar compared to furmark. But with furmark the only difference was that the PCI Slot power draw went up to ~2.4A compared to ~1.4 in heaven.
This is phenomenal! This does allow me to run the Vega56 in my MP5,1 without a power mod, on 10.4.5 and the 150w TPD BIOS.
All this without frying the two 6 pin to 8 pin ports. They seem to max out at around 90watts now. Compared to probably >100w before.
There is one downside to mention. With these setting the GPU seems to only hit a core clock of ~1.35GHz compared to ~1.45GHz before
I did a geekbench test before and after the beta update (both on the 150w BIOS) and got around 20% less score!
This seems weird, because I didn't change the settings on the card..
Anyways I do like the lower power usage and feel like this is a more stable way of running the card (with a bit less performance) productively without a mod. On FCPX etc. the power draw if even less hungry.
I add some screenshots which help to interpret the results with before & after.
So far, this is it. Hope you enjoy the numbers ;-).
Attachments
Last edited:
Is your card actually overheating? I don’t think it’s realistic for most people to be regularly putting Furmark level loads on the card.
Last edited:
hey
how do you power the red dragon: dual mini 6 pin (on MB) to 8 pin on VGA + sata (on MB) to 6 pin on VGA? I have ordered the same GPU... is it ok?
[doublepost=1555073403][/doublepost]
so, finally what did you buy?
lol I ended up just keeping my Nvidia GTX 1080 Ti Founders Edition for now. It runs all the Windows games on Bootcamp and 1440p Ultra (anthem, battlefield V, etc). It also has bootscreen by Macvidcards. The only drawback is that there are no Mojave drivers for it. I also like that the GPU is a rear blower so it blows the hot air out of my MacPro’s case.
I have a hard time paying to downgrade to an AMD card with somewhat less performance just to get Mojave on my Mac. I would also be losing the boot screen support which I find very useful. But I am interested on the Radeon VII or maybe Navi as eventually I will have to upgrade to Mojave.
Ps: I have done a Pixlas mod to power this GPU.
-(((
I was thinking to buy 1080ti but then i heard about some problems with nv web drivers and gave up.... ordered vega 56 red dragon.
Are you using a 6 pins and 8 pins cable connected to the Mac Pro motherboard directly as usual? No additional PSU or cable mod?
Last edited: