MP 1,1-5,1 MVC EFI BIOS for AMD 580?

[MIKX]

h9826790

I upgraded to iStats Menus 6.31 ( 1110 )which now shows GPU power consumption.

Having read your comments re: IOreg in post #221.

I've been running X-Plane 10 at high graphics settings and I must admit that knowing the published TDP for RX 580 8gb GPUs . . . this version of iStats Menus seem to be pretty accurate.

My idle wattage shows between 50 ~ 74 w depending on what is running eg: lo-rez vids, browser etc.

But as you can see from the included pic. . while X-Plane is running it gets up to 192w

I'm powering my MSI Armor RX 580 : 2 X Mini 6 pin to 8 pin to EVGa Powerlink but I have also added an extra 2 X SATA to 6 in to the PowerLink as well just to be safe. I've seen many PC websites claiming that the RX 580's real TDP is higher than officially published by AMD.

Perhaps iStats Menus creator has included maths to compensate for IOReg's inaccuracies in this new version.

My intuition feels that the wattage use I'm seeing is pretty close to real consumption.

 

[macguru9999]

My fans are quite LOUD running say the heaven benchmark at 1920x1080. And they dont really slow down for a while unless you restart. Is this normal ? In day to day use the card works fine. I got a score of 2120 at a temp of 53degrees, its really just the fan noise i would find annoying if i was a gamer...

 

[zedex]

@macguru9999 -- when you test heaven at 1920x1080 what other settings do you apply...? Do you just set it to EXTREME..?

I've just installed a GTX Titan X (Maxwell) and I'm trying to run an apples for apples comparison

 

[macguru9999]

@macguru9999 -- when you test heaven at 1920x1080 what other settings do you apply...? Do you just set it to EXTREME..?

I've just installed a GTX Titan X (Maxwell) and I'm trying to run an apples for apples comparison

Actually I got this: (as you can see medium settings and i only used 1920x1080 because i think it would overload at 4k ?)
Unigine Heaven Benchmark 4.0

FPS:	84.2
Score:	2120
Min FPS:	12.0
Max FPS:	151.8

System

Platform:	Darwin 18.7.0 x86_64
CPU model:	Intel(R) Xeon(R) CPU W3690 @ 3.47GHz (3457MHz) x12
GPU model:	Radeon RX 580 (8192MB) x1

Settings

Render:	OpenGL
Mode:	1920x1080 fullscreen
Preset	Custom
Quality	Medium
Tessellation:	Disabled

Powered by UNIGINE Engine

Unigine Corp. © 2005-2013

 

[zedex]

By comparison..

( I would have expected better from the TITAN X (Maxwell Architecture )

 

[MarkC426]

Maybe the low result is related to it only showing as 256mb compared to the 8192mb....?
Does macOS not properly recognise a titan?

 

[zedex]

It's a good observation. The Titan X has 12GB VRAM - perhaps @h9826790 could shed some light on this anomaly.

I have NVIDIA Drivers [ NVIDIA WebDriver-387.10.10.15.15.108 ] installed and matching CUDA drivers BUT I'm not a GPU config expert so maybe I've not activated some kind of GPU acceleration feature..?

 

[MarkC426]

Mojave may also have better support.
Like yourself, I am still on HS.....

 

[macguru9999]

Mojave may also have better support.
Like yourself, I am still on HS.....

AFAIK mojave does not support nvidia cards at all so no joy there ! It would be interesting to compare the score you get in WINDOWS on your mac pro. Also, were the fans a bit noisy during the test ? (I know any answer to that is subjective)

 

[zedex]

UPDATE..

I have re-run the test to 'perfectly match' the parameters used by @macguru9999

I'm very surprised to see that in changing the one 'non-matching' parameter -- which upon closer inspection shows that the test run by macguru9999 was run in fullscreen whereas my 1920x1080 test was run in a true 1080P discrete window..

As you can see, by making this adjustment the score has changed quite dramatically

It seems unlikely that running the test in 'fullscreen' versus running it in 'window mode' explains the big boost in the test score

I know that when I ran the test the first time around I had Google Chrome Open with a YouTube video running in the background – maybe this undermined the true value of the original test score.

[ still no explanation for why the benchmark utility doesn't accurately report the video RAM available on Nvidia Graphics cards ]

 

[h9826790]

still no explanation for why the benchmark utility doesn't accurately report the video RAM available on Nvidia Graphics cards ]

That’s nothing special. The VRAM capacity often can’t be correctly reported in MacOS benchmarks. Won’t affect the result.

That “TitanX isn’t faster than RX580” result is because your setting is CPU limiting. You guys should set tessellation to max, and use higher resolution if possible.

Anyway, using Window mode (vs full screen) often cause big hit in 3D performance, that’s normal.

 

[macguru9999]

well i had other things open like chrome and eyetv but they were behind the fullscreen test ... anyways those scores are close enough to identical. The reason i did not use a higher resolution is because the fans go crazy. I have not tested enough cards to know what level of noise and power draw is ok, and dont need to use games anyways so i did not want to overwork the card.

 

[zedex]

SOME helpful insight from @h9826790 - I need to understand core GPU functionality better [ nature and impact of 3D performance ]. I'll resolve this question in another post.

For your benefit @macguru9999 - I did mean to say that I get no fan issues or acceleration of any kind. I use iStat and have fan profiles that I've optimised for cooling and noise.

I did have a 4,1 flashed to 5,1 originally and much has been documented about a bug with some models that triggered maximum fans / racing fans for 'no reason' (many theories). I experienced this - fan noise was intolerable. Like a jet engine. Plus - even if you use ear plugs - your instinct tells you that your cMP life expectancy must be severely limited when the fans are going crazy all the time. It was VERY distressing :/

 

[macguru9999]

I am talking about just the graphics card fan , not the mac fans. You cannot control that with software can you ? My machine fans are not spinning up much at all.I have macs fan control but have not needed to use it. But I can hear the graphics card fan - only when testing those benchmarks.

 

[zedex]

You are correct (to the best of my knowledge). Can't control GPU fan speeds with software. The NVIDIA cards are generally accepted to be cards that run much cooler

 

[h9826790]

If you guys want to have more control of the MVC RX580, you may search some Hackintosh forum, and get a PowerPlay table for the RX580.

Once you have that, you can build a PowerPlay kext to control the GPU clock, voltage, or fan curve, etc.

I am not 100% sure of this can work for a flashed card, but theoretically this can reduce the GPU operating temperature (when under stress) quite a bit, therefore, reduce fan noise.

 

[orph]

@zedex & @macguru9999 do not compare 'custom' settings the results can relay change a lot depending on what you set.
use the preset's only, like the extreme or extreme HD preset.

ps cMP life expectancy will not be effected badly by fans, if anything geting the heat out is a postive.
if the fans are to much do as h9826790 says, limit the card a tad and it will keep it cool/low fan speed.

 

[h9826790]

If you guys want to have more control of the MVC RX580, you may search some Hackintosh forum, and get a PowerPlay table for the RX580.

Once you have that, you can build a PowerPlay kext to control the GPU clock, voltage, or fan curve, etc.

I am not 100% sure if this can work for a flashed card, but theoretically, downvolt can reduce the GPU operating temperature (when under stress) quite a bit, therefore, reduce fan noise. At least, this is true for my own RX580 (I downvolt it by mod the ROM directly, however, this method is NOT feasible for MVC RX580).

 

[killawat]

I know it’s not the same setup but I am seeing bootscreens and FileVault prompts on stock rx 580 egpu Mac mini, 10.15

 

[MarkC426]

Yes, but that will have an efi gpu also.

 

[MIKX]

I know it’s not the same setup but I am seeing bootscreens and FileVault prompts on stock rx 580 egpu Mac mini, 10.15

Just to clarify . . has your RX580 been EFI modified by MVC ?

 

[tsialex]

I know it’s not the same setup but I am seeing bootscreens and FileVault prompts on stock rx 580 egpu Mac mini, 10.15

Mac mini 2018 supports GOP pre-boot configuration support natively and works out of the box with stock eGPUs or TB chassis, while MP5,1 needs UGA and won't work for pre-boot configuration support unless MVC flashed.

 

[tomsmods]

Looks like the MVC site is teasing a Vega56 card. Anyone know more?

 

[nicksilverstein]

Can someone please dump their MVC Rom of the RX 580? I'm sure I'm not the only one asking for this.

 

[Ludacrisvp]

yeah, thats not gonna happen.
it is read protected and on a chip that has more storage than what is shipped on that card.