Do you know what differences in render time you’d be looking at between a w6800x duo and w5700x X2? And i can’t fathom what increases you’d get between the 6800x duo and a 6900x would be?
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 7,1 W6800x Duo benchmarks vs Vega ii duo, 2x w5700x, 2x 6900xt
- Thread starter rondocap
- Start date
- Sort by reaction score
There are tons of benchmark values for different things contained in this thread, but the below link might help.
Pro W6800 versus RX 6000 series in Mac Pro 2019
real world speed test results for performance minded Macintosh users
barefeats.com
I would look at all the benchmark articles at Barefeats, the link above is for the retail W6800; so for GPU accelerated apps that can utilize multiple GPUs then at worst you can probably multiply whatever value is for the W6800 * 1.5 and that should be a rough ballpark for the W6800X Duo.
I think most have concluded that a W6800X Duo will be faster than a W6900X for any apps that can utilize multiple GPUs.
But also watch all the YouTube videos about the cards; there might also be better info specific for whatever your app in question is on the application's forum page.
nice, I provided the quad w6800x results for this too. Love this site!
Pro W6800X Quadruple MPXs versus other GPUs in Mac Pro 2019
real world speed test results for performance minded Macintosh users
barefeats.com
Interesting times in the world of Apple MPX GPUs, with some calling the w6800x duo “crap” and others calling the duo x 2 “the ultimate MPX GPU setup”, all with seemingly good evidence for their point of view.
Looking forward to seeing how this evolves over time.
Looking forward to seeing how this evolves over time.
Be interesting to see a post comparing the new cards to the dual W5700X!
Hopefully I can help with the w5700x vs W6800x test, waiting for the W6800x to ship still
In my opinion the W6800x Duo is a good GPU for the price. First, I do think TG Pro has some issue with the temperature sensor and it is not reporting it correctly, so I will not confirm this as a real issue until we have some resolution.
Second, as someone who has also tested the Vega ii Duo a good amount - the Duo GPUs will thermal throttle at a lower threshold than the single GPUs. The Vega ii Duo acts in a similar way to the W6800x Duo when really pushed thermally.
For example, two single Vega ii will maintain a cooler temperature just because they're two separate GPUs and get access to more fans on the Mac Pro.
My two solo Vega II run cooler and a bit quieter on average than both Vega II Duo or W6800x Duo GPUs.
You can't compare the Duo GPUs to the single versions, even if doubled up, as "Apples to Apples" due to their different design.
In my previous video I compared W6800X DUO to 2x Vega II, I don't have any W5700X.
My Mac Pro fans are controlled by system itself, not by TG Pro. Apple's system itself recognized overheating card and because of that ramped up fans almost to max! So any update to any TG Pro can't fix it when TG Pro has nothing to do with it. And no, I'm not idiot to compare DUO card to single one and expect same thermals as single card. Obviously single card will have better thermals. But I can compare it in terms of performance and that's what I did in my reviews.
I just traded out my GPU's from a 580X and two Radeon VII's to a 580x and a Radeon Pro Vega II Duo. I also bought a 6800Duo but I plan to return it because I think the Vega II Duo is a better value for me. My only real interest is in how it affects FCPX performance. FCPX appears to be designed to rely on render files for playback performance. It doesn't matter whether I have my monitors plugged into a the 580X or the Radeon, playback performance is the same. My computer has a problem with pulling down 60p longGOP into a 24p timeline and it will skip frames at the same rate no matter what card the monitoring is running from. However, once the timeline is rendered or an optimized media is created, I have no issue with playback or scrubbing.
Therefore, my primary concern is how to setup the system for background rendering. I have noticed that FCPX appears to use both cards in the Vega II Duo for background rendering whereas I only remember it using one of the Radeon VII's. I would guess that it is programmed to use the Infinity fabric and recognize the two cards as a single processor. It may be that two of the 6800Duos connected would use all four cards for background rendering and speed up the process considerably. General background rendering, however, appears to be more reliant on the CPU than on the GPU. At least that is what I am seeing in my performance monitoring if I make color correction or stabilization changes. What I've discovered is that if I use Command Post to force the computer to keep background rendering running continuously, the Vega II Duo is able to keep the timeline rendered almost as fast as I can play it back. Even if I had slightly faster graphics cards, it isn't going to make a huge difference for editing since I am only relying on them for rendering rather than for realtime playback. If I was doing more exporting or graphics work, faster cards might improve my workflow. As of right now, though, it looks like a 580X to run both of my monitors, a Vega II Duo to run continuous background rendering, and Command Post to keep background rendering from pausing during playback is good enough.
Like I said, I have a 6800 Duo on standby, but I don't think it is even worth taking it out of its packaging for testing since I don't think the fractional improvement render times would be worth the $2000 more that it would cost me as a replacement for my Vega II Duo. This is especially true since Morgonaut has shown that it has poorer heat transfer. Even under full load on the Heaven test, the fans never ramped up more than halfway with the Vega II Duo.
ProRes encoding is CPU based and not GPU based. BTW you should ditch FCPX and use Resolve where you can easily assign which GPU to drive displays and on what GPUs to do compute tasks.
if you want to speed up ProRes decoding then yes, but don't forget Apple Afterburner accelerates ProRes decoding, not encoding, that's again CPU based
I really wish other softwares would get a clue and do this. *cough* Adobe!
Once life slows down again I will be taking a serious look at Resolve.
Unfortunately I am too invested in it right now. And I actually like After Effects, probably the only app I actually "really" like.Adobe is dyingStay away from it
I do not however, like FCP X; in fact I hate it; so Premiere made the most sense at the time the switch was made.
Resolve is gaining my interest though.
Ummm, no. I am not sure what *poor* performance you are talking about? For some things it stomps, other things it is about the same. Some things are a tad slower.
Premiere vs. Resolve vs. FCP X running on the same hardware, any performance discrepancies are mostly going to be software coding/optimization issues, not hardware.
I think if we keep it in perspective, the performance is good overall, across the entire scope of the GPU's abilities. Here are a few points:
1. The W6800x Duo is cheaper than the Vega ii Duo was
2. For 3D work/games, the newer GPU is faster for sure
3. For video work, the difference is minor, but there are also numerous codecs and workflows, each with their own differences, so it is difficult to quantify.
4. For some things, the memory bandwidth does make a difference - for example, if they were used for crypto mining, the Vega II is still faster - 58 mh/s vs 79 mh/s on the Vega ii.
Will software optimizations improve performance? It is very possible, I mean FCP didn't even utilize multiple GPUs until fairly recently - so anything can happen.
One thing is for sure though, even with 4 W6800x, the performance gain vs 2 Vega II is there in video work, but not massive at all for the price increase. They're really meant for 3D when using 4.
Now, I do think 2 GPUs is the sweet spot for video imo. 2 W5700x, W6800x Duo, Vega ii Duo, etc. Much better than a single GPU, that's where you'll see the most gains in FCP and Resolve
@Morgonaut said the 6800 duo wasn't worth it compared to the Vega 2 duo. I was just trying to understand why it is underperforming in comparison.
I recall someone from either this forum or from the video editing forum mentioned previously that they had to do the re-do the thermal paste and re-seat the thermal pads inside the MPX units to get better thermal and temperature. Has anyone tried that on the 6800x duo. I also recall someone posting benchmark of 2 afterburners and 1 Solo Vega being very fast for FCP. I don’t have a Mac Pro 2019 yet but it would be good if someone could re-do the thermal paste work and try to measure the temp again.
Like others have mentioned: I'd suggest to wait for Monterey GM & then some driver update action. Clearly, Apple already have bad form here. Whichever, the price is ridiculous and as always, prioritises spin & shareholders over customers. Greedy. I'm using a pair of Vega IIs, does the job well enough for Resolve Studio, but apart from the MP7.1, I do not use any other Apple products and stay well out of that 'walled garden' wherever possible.