pastrychef's Asus ROG Strix Z370-G Gaming (WI-FI AC) build w/ i9-9900K + AMD 6600 XT

[pastrychef]

Hi @pastrychef ,
these days i am using VideoProc and I saw that in the options the 'graphics' is N/A.
For any video to convert, the software never uses the GPU.
I have been using the AMD Vega Power Injection for a long time.
VideoProc is fine for showing whether hardware acceleration can work in supported software. It just doesn't use the GPU for its own tasks.
Is it possible to fix this?
In the BIOS settings, no IGPU, only PCIE (MSI Radeon RX Vega 56 Air Boost).
View attachment 543533View attachment 543534View attachment 543535

I don't know how VideoProc actually processes videos. I use Handbrake for video conversions and that also only uses CPU...

PowerTable injection should be fine.

Your BIOS settings are the same as mine.

 

[dclive]

I don't know how VideoProc actually processes videos. I use Handbrake for video conversions and that also only uses CPU...

PowerTable injection should be fine.

Your BIOS settings are the same as mine.

What's the indication, in Handbrake, of what method (QSV, CPU, GPU) you're using?

In a generic 1920x1080 h264 file that I converted into h265, "x265" preset (assumedly CPU only) gets me about 40 fps, at mostly 80% or so CPU utilization. That same file source, going to "VideoToolbox" preset (assumedly able to use AMD GPU) gets me about 200 fps, at mostly 60-70% CPU utilization.

Reminder: This is a B660M-HDV with an i5-12400F and AMD 5700, so no iGPU for VideoToolbox to choose. It's either CPU or GPU here.

 

[pastrychef]

What's the indication, in Handbrake, of what method (QSV, CPU, GPU) you're using?

In a generic 1920x1080 h264 file that I converted into h265, "x265" preset (assumedly CPU only) gets me about 40 fps, at mostly 80% or so CPU utilization. That same file source, going to "VideoToolbox" preset (assumedly able to use AMD GPU) gets me about 200 fps, at mostly 60-70% CPU utilization.

Reminder: This is a B660M-HDV with an i5-12400F and AMD 5700, so no iGPU for VideoToolbox to choose. It's either CPU or GPU here.

Hmm... Even when selecting the "VideoToolbox" options in Handbrake, it's only using CPU on my system.

Please try playing this video in VLC and see if your GPU is being used to decode it. TIA.

 

[dclive]

Hmm... Even when selecting the "VideoToolbox" options in Handbrake, it's only using CPU on my system.

Please try playing this video in VLC and see if your GPU is being used to decode it. TIA.

That link went to a very confusing page. I then went to here : https://joumxyzptlk.de/3dmark_gallery_portroyal.html and then selected the

3DMark Port Royal Demo without DoF blur in 8K 120 fps.​

...7k 60fps file on the lower left of that one upper section. That downloaded a file 799.4mb in size.

I played it in VLC. I don't see how to get into fullscreen, but in windowed mode, there was noticeable stuttering, and cmd-I showed several hundred dropped frames. CPU was at 80% or thereabouts.

Overall, not terribly impressive.

 

[dclive]

I'm just waiting to see how difficult it is to open up the Mac Studio to clean out dust. But, I'm really close to getting one...

As much as I wan the M1 Ultra, I think it's overkill for me... I'd be perfectly happy with an M1 Max model. I'd still choose it over a laptop because I know very well that batteries degrade and die over time. I'd prefer not to have to unglue a battery from a MacBook. Also, with the larger heatsink and fans, in the Studio, there would never be any concerns of throttling.

My issue with "just" the M1 Max is performance - it's just nothing special anymore. Sure, 3 years ago it would be incredible, but right now, it's about as fast as an i5-12400F ($300 upgrade for most in this thread; keep your existing RAM) + AMD 5700: Cinebench for me is about 11700 range, and for the M1 Max about 12300 range; about the same. If I jumped to the Max, i believe my GPU performance would take a hit in most day to day things, as I'm currently using an AMD 5700 (Metal: around 64,000), and the M1 Max only gives about 66,000 or so; about the same, in Metal. That's not compelling - that's just a sidegrade, plus my performance in Intel apps will be negatively impacted.

The Ultra, OTOH, would be an interesting upgrade - about double the performance of the i5-12400F (Cinebench of around 23,500) with an OpenCL score of 77100, compared to the AMD 5700's 58,700. (I've not seen Metal scores for the Ultra yet.). But at $4k+TT&L.

Alas, after reviewing these scores, maybe it's not as impressive after all. Hackintosh still has legs for day-to-day use, although for specialized ProRes, 8K, etc. work, obviously, the higher M1s walk all over it. I just don't do that.

 

[pastrychef]

That link went to a very confusing page. I then went to here : https://joumxyzptlk.de/3dmark_gallery_portroyal.html and then selected the

3DMark Port Royal Demo without DoF blur in 8K 120 fps.​

...7k 60fps file on the lower left of that one upper section. That downloaded a file 799.4mb in size.

I played it in VLC. I don't see how to get into fullscreen, but in windowed mode, there was noticeable stuttering, and cmd-I showed several hundred dropped frames. CPU was at 80% or thereabouts.

Overall, not terribly impressive.

Please try the 60fps without DoF blur. Thanks.

 

[dclive]

Please try the 60fps without DoF blur. Thanks.

I did. Those are my results, above.

 

[pastrychef]

My issue with "just" the M1 Max is performance - it's just nothing special anymore. Sure, 3 years ago it would be incredible, but right now, it's about as fast as an i5-12400F ($300 upgrade for most in this thread; keep your existing RAM) + AMD 5700: Cinebench for me is about 11700 range, and for the M1 Max about 12300 range; about the same. If I jumped to the Max, i believe my GPU performance would take a hit in most day to day things, as I'm currently using an AMD 5700 (Metal: around 64,000), and the M1 Max only gives about 66,000 or so; about the same, in Metal. That's not compelling - that's just a sidegrade, plus my performance in Intel apps will be negatively impacted.

The Ultra, OTOH, would be an interesting upgrade - about double the performance of the i5-12400F (Cinebench of around 23,500) with an OpenCL score of 77100, compared to the AMD 5700's 58,700. (I've not seen Metal scores for the Ultra yet.). But at $4k+TT&L.

Alas, after reviewing these scores, maybe it's not as impressive after all. Hackintosh still has legs for day-to-day use, although for specialized ProRes, 8K, etc. work, obviously, the higher M1s walk all over it. I just don't do that.

Here are my Geekbench CPU search results:

The i5-12400F trails the M1 Pro, Max, and Ultra.

The i9-12900K beats the M1s in single core but gets trounced by the M1 Ultra in multi-core.



Here are my Geekbench GPU search results (there were no M1 Ultra results):

RX 5700 XT results:

The RX 5700 Xt beats both M1 Pro and Max. The difference between the RX 5700 XT and M1 Max seems very small.



For me, I think the Mac Studio with M1 Max is a very good value. CPU performance that's competitive with an i9-12900K and GPU performance approximately on par with an RX 5700 XT.

When I was looking for a video card about a week or two ago, I decided on an RX 6600 XT (which has near identical performance with an RX 5700 XT) because I felt it was more than enough for my needs.

If I needed more GPU power, the M1 Ultra is an option. Assuming that GPU scores will be approximately double that of the M1 Max since it has twice the number of GPU cores, it would be pretty competitive against an RX 6800 XT...

RX 6800 XT results:

Other things that a hackintosh will never have are:

• Neural Engine. This is an X-factor that can improve the performance of apps beyond what CPU/GPU specs indicate. (Please see here for an example of what's possible with the Neural Engine.)
• Dedicated ProRes encoder/decoder.
• Longer macOS support. Who knows how much longer Apple will continue to support the X86 platform after the MacPro7,1 has been discontinued?
• Much better efficiency. Remember, the M1 Max can run on a battery. How large of a battery will an i9-12900K w/ RX 5700 XT require?
• Hassle free updates. While hackintoshing has gotten much easier, things can break at anytime when running a Z690 system since no real Mac ever used that platform.
• Ability to run iOS apps.

 

[pastrychef]

I did. Those are my results, above.

Hmm... When I play that video, my CPU usage is around 35%.

I still get dropped frames but it's worlds better than my old Vega 56 which was essentially a slide show.

 

[dclive]

Interesting. I can’t immediately explain, then.