MP 7,1 Need your input! What benchmarks and tests would you like to see on Mac Pro with 6900XT?

[rondocap]

Would you mind telling me where you bought them? And how loud is it when idle vs full usage? The near dead silence of the W5700X is one of my favorite things about it? And lastly, would there be a noticeable advantage to using both my current W5700X installed alongside the 6900XT?

Thank you for your time!

I got one from amd, one from eBay, right now it’s tough to get them at msrp.

it’s louder than the w5700x at full load, but it’s similar to the Vega ii which ramps up the fans more.
I wouldn’t mix the two cards for video editing. But for octane x or 3D it can work better together

 

[demitri]

Thank you! My apologies, but I have two more follow up questions... When it's idle is the 6900XT as silent as the w5700X, and since I do video editing AND 3D, it would be best to use both cards? There wouldn't be a performance hit for video editing?

 

[rondocap]

Thank you! My apologies, but I have two more follow up questions... When it's idle is the 6900XT as silent as the w5700X, and since I do video editing AND 3D, it would be best to use both cards? There wouldn't be a performance hit for video editing?

Yes when Idle it is silent.

I just rendered 12K braw and Red 8K with 2x 6900xt, and it wasn't too loud, but you can hear them a bit. I had W5700x in there previously too, big performance improvement over them for sure.

Which video editing software do you use? and which 3d?

Final cut and resolve work better with 2 of the same cards, but resolve can use 1 w5700x and 1 6900xt somewhat - sometimes you don't get the best results though. So 2 of the same is much better. It can be a performance hit depending on the codec with 2 different cards, and sometimes it won't matter.

 

[rondocap]

Posted in the other thread, but here for reference:


i ve done extensive benchmarks on basically all of the possible GPU combinations in pairs. Heaviest stuff I could throw at them, 12K Braw to 4444 XQ, 8K R3d Raw, etc.

1. 6900 XT *2 by the fastest in pretty much everything
2. Vega ii Solo *2 Neck and neck with the duo, but slower than 6900 XT
3. Vega ii Duo Same as the duo, could not find a bottleneck with the pcie bandwidth in the real world
4. W5700x *2 - almost as good as the duo Vegas in most cases

There isn't a huge video rendering difference between all of those, usually seconds or a minute or two for larger files - but can be more in some cases.

If you order the list in terms of price to performance, either the 6900xt or W5700x will be 1 and 2, with the Vegas not a great price to performance anymore.

If you find a 6900xt at $999 msrp, that's awesome - but most likely real world pricing is $1400-1500 right now, still a good value for the performance you get.

Also, running 6900 XT is pretty quiet - in fact near a Duo. Fans do spin up with heavy tests, the W5700x is the most quiet due to lower power.

Otherwise, 6900xt works almost flawless in Mac Os, aside from not having extra thunderbolt 3 ports. USB C cable to LG 5k Display. Sometimes waking up from sleep can hang, so I unplug and plug, but overall fine.


I do like the vega ii duo still though, just because it is such an awesome piece of hardware - but price to performance loses to the new 6900xt.

The new MPX W6800 should be awesome with 32GB of vram

 

[demitri]

Yes when Idle it is silent.

I just rendered 12K braw and Red 8K with 2x 6900xt, and it wasn't too loud, but you can hear them a bit. I had W5700x in there previously too, big performance improvement over them for sure.

Which video editing software do you use? and which 3d?

Final cut and resolve work better with 2 of the same cards, but resolve can use 1 w5700x and 1 6900xt somewhat - sometimes you don't get the best results though. So 2 of the same is much better. It can be a performance hit depending on the codec with 2 different cards, and sometimes it won't matter.

That's GREAT to hear about it being silent when idle. When of my favorite reasons to work on a Mac is how silent it is when not doing anything taxing.

I use Final Cut Pro for all my editing, and Cinema 4D for 3D, which as you may know has a CPU-based render engine, but I'm starting to use Redshift more, which is GPU-based. My main job, which I use 90% of the time is After Effects, and a lot of the effects I use rely on GPUs for speed - in that instance, for After Effects, 16GB of VRAM is already overkill.

The majority of my Final Cut footage is ProRes and or h264, so I'm not sure if that would mean a performance hit with the mixed cards. I can't justify the cost right now of two 6900XT, but I can definitely justify one. It sounds like maybe it would be best for my setup to just use the one 6900XT, and possibly get a second one sometime in the future?

 

[OkiRun]

Posted in the other thread, but here for reference:


i ve done extensive benchmarks on basically all of the possible GPU combinations in pairs. Heaviest stuff I could throw at them, 12K Braw to 4444 XQ, 8K R3d Raw, etc.

1. 6900 XT *2 by the fastest in pretty much everything
2. Vega ii Solo *2 Neck and neck with the duo, but slower than 6900 XT
3. Vega ii Duo Same as the duo, could not find a bottleneck with the pcie bandwidth in the real world
4. W5700x *2 - almost as good as the duo Vegas in most cases

There isn't a huge video rendering difference between all of those, usually seconds or a minute or two for larger files - but can be more in some cases.

If you order the list in terms of price to performance, either the 6900xt or W5700x will be 1 and 2, with the Vegas not a great price to performance anymore.

If you find a 6900xt at $999 msrp, that's awesome - but most likely real world pricing is $1400-1500 right now, still a good value for the performance you get.

Also, running 6900 XT is pretty quiet - in fact near a Duo. Fans do spin up with heavy tests, the W5700x is the most quiet due to lower power.

Otherwise, 6900xt works almost flawless in Mac Os, aside from not having extra thunderbolt 3 ports. USB C cable to LG 5k Display. Sometimes waking up from sleep can hang, so I unplug and plug, but overall fine.


I do like the vega ii duo still though, just because it is such an awesome piece of hardware - but price to performance loses to the new 6900xt.

The new MPX W6800 should be awesome with 32GB of vram

Will there be an infinity link duo option.

 

[Korican100]

Otherwise, 6900xt works almost flawless in Mac Os, aside from not having extra thunderbolt 3 ports. USB C cable to LG 5k Display. Sometimes waking up from sleep can hang, so I unplug and plug, but overall fine.

i have two monitors and have this issue sometimes. but instead of unplug and plug, press cmd+ f1 to mirror, and then press it again to unmirror. That usually turns it back on for me. But not sure if it works with a single monitor

 

[Tala]

Hi there
Maybe I can get some help here,

Im using the 7.1 Mac Pro with the 580X 8GB and an Rx Vega 64.
I want to upgrade to the 6900 XT, but not sure it will be the right thing for me.

I work with two screens, LG 5k 27" and an Apple Thunderbolt Display, my main work is on After Effects and a little bit of Cinema4D.
When working on after effects it feels that most of the load goes to the 580x and not to the Vega64.

So I was wondering, maybe it would be better for me to upgrade to the w5700X which Both screens will be connected too?

Cinema4D (Octane) can utilize any GPU to the max with no problem, so my main concern is After Effects.

It feels stupid to buy the w5700X when the 6900 XT is so superior...

Id rather buy one new Graphics card,
What should I do?

Any help will be much appreciated
Thanks

 

[Tala]

That's GREAT to hear about it being silent when idle. When of my favorite reasons to work on a Mac is how silent it is when not doing anything taxing.

I use Final Cut Pro for all my editing, and Cinema 4D for 3D, which as you may know has a CPU-based render engine, but I'm starting to use Redshift more, which is GPU-based. My main job, which I use 90% of the time is After Effects, and a lot of the effects I use rely on GPUs for speed - in that instance, for After Effects, 16GB of VRAM is already overkill.

The majority of my Final Cut footage is ProRes and or h264, so I'm not sure if that would mean a performance hit with the mixed cards. I can't justify the cost right now of two 6900XT, but I can definitely justify one. It sounds like maybe it would be best for my setup to just use the one 6900XT, and possibly get a second one sometime in the future?

Hi Demitri,
I saw you were using After Effects on the Mac Pro, so will be nice to get your point of view...


Im using the 7.1 Mac Pro with the 580X 8GB and an Rx Vega 64.
I want to upgrade to the 6900 XT, but not sure it will be the right thing for me.

I work with two screens, LG 5k 27" and an Apple Thunderbolt Display, my main work is on After Effects and a little bit of Cinema4D.
When working on after effects it feels that most of the load goes to the 580x and not to the Vega64.

So I was wondering, maybe it would be better for me to upgrade to the w5700X which Both screens will be connected too?

Cinema4D (Octane) can utilize any GPU to the max with no problem, so my main concern is After Effects.

It feels stupid to buy the w5700X when the 6900 XT is so superior...

Id rather buy one new Graphics card,
What should I do?

Any help will be much appreciated
Thanks

 

[rondocap]

Hi Demitri,
I saw you were using After Effects on the Mac Pro, so will be nice to get your point of view...


Im using the 7.1 Mac Pro with the 580X 8GB and an Rx Vega 64.
I want to upgrade to the 6900 XT, but not sure it will be the right thing for me.

I work with two screens, LG 5k 27" and an Apple Thunderbolt Display, my main work is on After Effects and a little bit of Cinema4D.
When working on after effects it feels that most of the load goes to the 580x and not to the Vega64.

So I was wondering, maybe it would be better for me to upgrade to the w5700X which Both screens will be connected too?

Cinema4D (Octane) can utilize any GPU to the max with no problem, so my main concern is After Effects.

It feels stupid to buy the w5700X when the 6900 XT is so superior...

Id rather buy one new Graphics card,
What should I do?

Any help will be much appreciated
Thanks

I would do either 2 5700 XT, or one 6900 XT. Mac Os doesn't seem to like different multi GPUs unless it is like Octane X, but for other video stuff, keeping it homogenous is better.

W5700x is great, especially two of them for the price imo.

 

[Tala]

I would do either 2 5700 XT, or one 6900 XT. Mac Os doesn't seem to like different multi GPUs unless it is like Octane X, but for other video stuff, keeping it homogenous is better.

W5700x is great, especially two of them for the price imo.

Thanks for the reply rondocap!

Did you mean a setup of two Apple MPX w5700X or Two regular Radeon RX 5700 XT?

If I get one 6900 XT, Id have to also use the 580X card, which I have, to be able to connect the screens to the Mac Pro's thunderbolt connections.
(the 580X doesn't have thunderbolt connections, only two Hdmi ports)

Do you have an idea how well this setup could work?


Im leaning on getting the 6900 XT because I think it would be relevant for longer than the 5700 XT and will keep its resale value better...

 

[diamond.g]

Thanks for the reply rondocap!

Did you mean a setup of two Apple MPX w5700X or Two regular Radeon RX 5700 XT?

If I get one 6900 XT, Id have to also use the 580X card, which I have, to be able to connect the screens to the Mac Pro's thunderbolt connections.
(the 580X doesn't have thunderbolt connections, only two Hdmi ports)

Do you have an idea how well this setup could work?


Im leaning on getting the 6900 XT because I think it would be relevant for longer than the 5700 XT and will keep its resale value better...

Eeeehhhhh if the RDNA3 rumors are true our 6900xt's are going to lose quite a bit of value next year.

 

[rondocap]

Thanks for the reply rondocap!

Did you mean a setup of two Apple MPX w5700X or Two regular Radeon RX 5700 XT?

If I get one 6900 XT, Id have to also use the 580X card, which I have, to be able to connect the screens to the Mac Pro's thunderbolt connections.
(the 580X doesn't have thunderbolt connections, only two Hdmi ports)

Do you have an idea how well this setup could work?


Im leaning on getting the 6900 XT because I think it would be relevant for longer than the 5700 XT and will keep its resale value better...

At this point two W5700x makes sense if you plan to keep long term, as the used 5700 XT is still pretty high in price right now on the used market. You do get more TB ports and better cooling with the MPX modules too

 

[jerrygladh]

Thx for your efforts and tests rondocap!
Sold my 2 Radeon VII for more then I paid for them.
Bought and installed 2 brand new 6900xt AMD reference cards from eBay att a very good price.
Runs like a charm

 

[jasonmvp]

I'm sincerely curious about the performance difference of RDNA2's hardware encoder vs the older encoders on the Vega II. For instance, the best the Vega II can hope for in HEVC 4K/60 is real time. In other words, a 10 minute video will take about 10 minutes to encode into HEVC. I'm pretty sure the h.264 encoder performance is better, but I'm not positive. And by "better" I mean: how long does it take to encode?

Any way you can get a 10 minute chunk of 4K/60 footage and export it out into HEVC? Purely hardware encoding? Ideally the footage is a low-stress sort of format so you're not working the GPU even harder. The time it takes to do the output is what I'm after.

h.264 hardware 4k/60
h.265 (HEVC) hardware 4k/60

The single vs dual card doesn't matter as much with respect to this specific test because only one encoder will be used at a time by Resolve. So a single 6900XT vs the single Vega II.

If you have the time and footage, I'd appreciate that.

 

[rondocap]

I'm sincerely curious about the performance difference of RDNA2's hardware encoder vs the older encoders on the Vega II. For instance, the best the Vega II can hope for in HEVC 4K/60 is real time. In other words, a 10 minute video will take about 10 minutes to encode into HEVC. I'm pretty sure the h.264 encoder performance is better, but I'm not positive. And by "better" I mean: how long does it take to encode?

Any way you can get a 10 minute chunk of 4K/60 footage and export it out into HEVC? Purely hardware encoding? Ideally the footage is a low-stress sort of format so you're not working the GPU even harder. The time it takes to do the output is what I'm after.

h.264 hardware 4k/60
h.265 (HEVC) hardware 4k/60

The single vs dual card doesn't matter as much with respect to this specific test because only one encoder will be used at a time by Resolve. So a single 6900XT vs the single Vega II.

If you have the time and footage, I'd appreciate that.

Sure, I can do that. I have the w6800x duo landing tomorrow too most likely, so I can also test those.

what footage do you prefer for me to use? I can use anything from the Sony a7siii, canon r5, Panasonic s1h, etc

 

[jasonmvp]

Sure, I can do that. I have the w6800x duo landing tomorrow too most likely, so I can also test those.

what footage do you prefer for me to use? I can use anything from the Sony a7siii, canon r5, Panasonic s1h, etc

It doesn't matter the source footage. Ideally something low-stress. The goal here is to exercise the encoder, less so the decoding of the source. The kind of footage that takes up a mountain of disk space because it's not compressed to hell and gone. That sort of footage.

And the W6800X Duo is precisely the card I'd be considering, so if you have the time, yeah: test that sucker.

Thanks!

 

[rondocap]

It doesn't matter the source footage. Ideally something low-stress. The goal here is to exercise the encoder, less so the decoding of the source. The kind of footage that takes up a mountain of disk space because it's not compressed to hell and gone. That sort of footage.

And the W6800X Duo is precisely the card I'd be considering, so if you have the time, yeah: test that sucker.

Thanks!

Ok I’ll do that. I’ll throw some r3d raw and export it to h264 and h265, maybe braw or canon raw too. That should do the trick

 

[rondocap]

It doesn't matter the source footage. Ideally something low-stress. The goal here is to exercise the encoder, less so the decoding of the source. The kind of footage that takes up a mountain of disk space because it's not compressed to hell and gone. That sort of footage.

And the W6800X Duo is precisely the card I'd be considering, so if you have the time, yeah: test that sucker.

Thanks!

Could you tell me a specific setting to test in resolve?
I tried to run a baseline test on the Vega ii with red 4k60p, but a 10 minute clip exports in around 5 minutes, h264 and h265

if I turn off hardware acceleration in resolve, the time goes a lot longer than 10 minutes

6k res was around 7:15 for both h264 and h265, but with hardware acceleration

 

[jasonmvp]

Could you tell me a specific setting to test in resolve?
I tried to run a baseline test on the Vega ii with red 4k60p, but a 10 minute clip exports in around 5 minutes, h264 and h265

Excellent. Perhaps I was incorrect regarding the hardware acceleration on the Vega II. I thought it was only capable of, at most: 4K/60 real time. These are what I use:

 

[rondocap]

Excellent. Perhaps I was incorrect regarding the hardware acceleration on the Vega II. I thought it was only capable of, at most: 4K/60 real time. These are what I use:

View attachment 1816333

Ok great I'll run some numbers today and tomorrow - what camera do you typically use and codecs? I can see if I can match it or have the most similar

 

[jasonmvp]

Ok great I'll run some numbers today and tomorrow - what camera do you typically use and codecs? I can see if I can match it or have the most similar

My footage is all h.265 4K/60 game footage, so it leans heavily into the hardware decoder as well. What I'm uncertain of is: if you're beating the snot out of the decoder, does that hurt the encoder's performance? That's why I was curious as to what you're seeing with a low-stress input CODEC.

 

[csxlab]

I would like your input, I am testing a Mac Pro with a few new GPU configurations, and I would like for you to tell me what you would like me to test.

I’ve already done a lot of testing in final cut pro, as well as da Vinci resolve, but I can do more specific ones and whatever resolution or codecs that you would like to see as well. Raw, different cameras, etc.

I would also love suggestions for other non-video editing type of benchmarks I can do, stuff like 3-D rendering, machine learning, or anything else that GPUs can be used for.

Let me know specific requests and I would love to do it!

Final goal is to put together all of the benchmarks in one place, that way people can have an idea of performance as the years go on since I think a lot of people will start to get into this Mac Pro as it gets cheaper.

Specs:
28 core Xeon w3275m
96gb ram
Afterburner

Available gpus to compare:

2x w5700x mpx
2x 6900 XT
1 Vega ii mpx
(580x mpx too if anyone wants numbers)

Would Love to see a Redshift Benchmark
I am thinking geting a macpro for it

 

[rondocap]

My footage is all h.265 4K/60 game footage, so it leans heavily into the hardware decoder as well. What I'm uncertain of is: if you're beating the snot out of the decoder, does that hurt the encoder's performance? That's why I was curious as to what you're seeing with a low-stress input CODEC.

Red 6k to H264 in Resolve, 10 min clip (H265 is similar, slightly faster)

Vega II 07:15
W6800x Duo: 06:34

 

[jasonmvp]

Red 6k to H264 in Resolve, 10 min clip (H265 is similar, slightly faster)

Vega II 07:15
W6800x Duo: 06:34

Thanks for that; that's awesome. 45 second improvement is nothing to sneeze at. The hard part is knowing how much of that was due specifically to the encoder, vs the rendering brute of the newer card.