Radeon Pro 580X (27"imac) benchmarks

[patearrings]

Hi guys,

Does anyone have any benchmarks of this card yet please using UNIGINE HEAVEN on ULTRA settings?

I see tons of reviews and benchmarks for the Radeon pro vega 48 but the other card in the top spec 27" imac is hardly mentioned anywhere.

Anyone that has one that would be willing to run the benchmark on the above settings, i would be eternally grateful as this is the card i will most likely order.

Many thanks!

 

[Kaelis]

Hi guys,

Does anyone have any benchmarks of this card yet please using UNIGINE HEAVEN on ULTRA settings?

I see tons of reviews and benchmarks for the Radeon pro vega 48 but the other card in the top spec 27" imac is hardly mentioned anywhere.

Anyone that has one that would be willing to run the benchmark on the above settings, i would be eternally grateful as this is the card i will most likely order.

Many thanks!

I don’t have Unigine, but I tried some games with my new iMac with i5 9600K/RX580X.

Total War Warhammer 1 in 1440p High settings ran at 48FPS on Bootcamp.

 

[patearrings]

I don’t have

I don’t have Unigine, but I tried some games with my new iMac with i5 9600K/RX580X.

Total War Warhammer 1 in 1440p High settings ran at 48FPS on Bootcamp.

Im not familiar with that game, were you impressed with the performance?

 

[Abbas]

Ran the test using the "Extreme" preset which made it tun at 1600x900 with 8xAA and the result was 44.4FPS or 1117. Screenshot attached.

 

[Bohemien]

I see tons of reviews and benchmarks for the Radeon pro vega 48 but the other card in the top spec 27" imac is hardly mentioned anywhere.

Could you pls post the values for the i9/Vega (or other) iMacs for comparison? Thx!

 

[patearrings]

Ran the test using the "Extreme" preset which made it tun at 1600x900 with 8xAA and the result was 44.4FPS or 1117. Screenshot attached.

Wow! Thank you so much Abbas, this is EXACTLY what i was after, thank you ever so much for taking the time and trouble to do that, i really appreciate it!

 

[macduke]

Could you pls post the values for the i9/Vega (or other) iMacs for comparison? Thx!

I'm at work now so don't have access to mine but I took this screenshot from a YouTube video. I don't think I've ran Unigine yet on mine. Actually seems like a substantial improvement. Maybe I'm not disappointed with my Vega 48 after all, lol. This was in bootcamp. His score in macOS was 1713. The Vega 56 was 1831 on macOS.

 

[Bohemien]

Thanks for posting this, really a considerable performance gain - however, this benchmark is from a i9/Vega combo, the one above (posted by Abbas) is for the i5/580X.

Sorry for asking, I don't know this benchmark: is is it GPU-only or does the CPU affect the values?

 

[Abbas]

That test's run on Windows instead of Mac (as the screenshot states.) Wonder how much of that score is due to Windows and better drivers for gaming. Would be great if someone could post Mac numbers for a true comparison.

I'm at work now so don't have access to mine but I took this screenshot from a YouTube video. I don't think I've ran Unigine yet on mine. Actually seems like a substantial improvement. Maybe I'm not disappointed with my Vega 48 after all, lol. This was in bootcamp. His score in macOS was 1713. The Vega 56 was 1831 on macOS.

View attachment 830341

 

[Kaelis]

Im not familiar with that game, were you impressed with the performance?

That's a pretty neat result, it's better than my previous dedicated PC gaming desktop with i5 2400/R9 290 MSI OC/8GB RAM/ 512 SSD.

I had some time to try other games since, here are the results :

- Dark Souls 3 : maxed out settings, 45-60FPS 1440P, Bootcamp
- Dead Or Alive 6 : maxed out settings, 50-60FPS 1440P, 25-30FPS 4K, Bootcamp
- Total War Warhammer 2 : High/Ultra settings, 30-50FPS (depending on map, units etc.) 1440P, Bootcamp
- League of Legends : maxed out settings 60FPS 5K, rock-solid, MacOS.

Fan noise level isn't a problem at all unless you play with no sound. No slow-down of performance even after 4 hours on Total War Warhammer 2.

 

[macduke]

That test's run on Windows instead of Mac (as the screenshot states.) Wonder how much of that score is due to Windows and better drivers for gaming. Would be great if someone could post Mac numbers for a true comparison.

I literally did in my post you quoted. I said it was 1713. I didn't post a screenshot because he just had a bunch of graphs up for that number and didn't have the fps or anything else listed but he showed that screenshot for Windows for comparison right after that. Windows makes a difference but it's not THAT big of a difference, lol.

 

[kaintxu]

Quick questions, is Extreme better more demanding than Ultra for Uninge Heave benchmark? I have some i5 580X results for Ultra, so want to compare

 

[kaintxu]

Hi guys,

Have a look at this video.

Could someone with the 580Xrun any of this games (even if only fortnite which is free to play) with same game config and tell us results?

Also do you guys think results would be the same FPS with an i5 instead of an i9?

Cheers

 

[Kaelis]

I ran Unigine Heaven 4 in Ultra preset and 1440p resolution. Ran it on Bootcamp, don’t know if it matters.
i5 9600K / RX580X / 8GB / 512GB SSD

https://imgur.com/gallery/ThlhWwl

50.1 FPS average
1261 Score

 

[Bohemien]

Does anybody have a benchmark for the i9/580X combo under OS X, "Extreme" preset? Thx!

 

[kaintxu]

I ran Unigine Heaven 4 in Ultra preset and 1440p resolution. Ran it on Bootcamp, don’t know if it matters.
i5 9600K / RX580X / 8GB / 512GB SSD

https://imgur.com/gallery/ThlhWwl

50.1 FPS average
1261 Score

What is more demanding, Ultra or extre? Everyone is running the unigine Heaven on diferent settings which is no good for comparisons.

It's really appreciated your help. Could you run it on extreme just so that we can see numbers to compare

 

[Kaelis]

What is more demanding, Ultra or extre? Everyone is running the unigine Heaven on diferent settings which is no good for comparisons.

It's really appreciated your help. Could you run it on extreme just so that we can see numbers to compare

Extreme is more demanding, it adds tesselation, 8X AA etc.

Anyway I ran the benchmark in Extreme preset as asked.

Here is the screenshot : https://imgur.com/gallery/PogxGl0

Result is 60FPS and score is 1512. So FPS wise, the i9/Vega 48 combo is 25% faster than the i5 9600K/RX580X one.

Side note : the result is better than my previous Ultra benchmark because this time I left everything at default preset values. On my Ultra bench I changed the resolution to 1440p. The default for Extreme is 900p resolution so obviously FPS is better even with the higher graphics settings.

 

[Bohemien]

Result is 60FPS and score is 1512. So FPS wise, the i9/Vega 48 combo is 25% faster than the i5 9600K/RX580X one.

Thanks for posting this! I'd have expected more - @macduke reports a score of 1713 above and I think I've seen a score of approx. 1700 in one of the videos, too. Of course there's always variation, and one never knows whether the same settings were used.

 

[cynics]

That OpenGL and Direct3D variable is a big deal. Make sure when comparing to other benchmarks you are comparing OpenGL.

Even then that is only for a rough comparison. Many MacOS games are using Metal which is an option I don't believe Heaven has an option for.

 

[kaintxu]

What is more demanding, Ultra or extre? Everyone is running the unigine Heaven on diferent settings which is no good for comparisons.

Extreme is more demanding, it adds tesselation, 8X AA etc.

Anyway I ran the benchmark in Extreme preset as asked.

Here is the screenshot : https://imgur.com/gallery/PogxGl0

Result is 60FPS and score is 1512. So FPS wise, the i9/Vega 48 combo is 25% faster than the i5 9600K/RX580X one.

Side note : the result is better than my previous Ultra benchmark because this time I left everything at default preset values. On my Ultra bench I changed the resolution to 1440p. The default for Extreme is 900p resolution so obviously FPS is better even with the higher graphics settings.

That for this extremely helpful, though it seem extremely high for the 580X. Could you run same settings but for OpenGL rather than Direct X as it is what we have.

I will make a compilation of all these results to help people

 

[Kaelis]

That for this extremely helpful, though it seem extremely high for the 580X. Could you run same settings but for OpenGL rather than Direct X as it is what we have.

I will make a compilation of all these results to help people

Good catch about the OpenGL, Direct3D difference.

I’ll try the Open GL when I get home in two hours. My +25% result for i9/Vega vs i5/RX580X still stands since the screenshot higher in the thread shows Direct3D11.

I’ll also re-run the benchmark in Direct3D11 to eliminate any random error creeping during the benchmark.

And nice idea about making a compilation in a new thread, it will make the numbers easier to find for the ones hesitating between the different configs !

 

[bondavi]

I read here that the 580X was a rebranded 580.

Specs on AMD page show the same thing: https://www.amd.com/en/RX-series#paragraph-341606

Would be good to hear about the actual differences, benchmark and hardware-wise, overclock (to 100Mhz, as some have posted)...

 

[cynics]

I read here that the 580X was a rebranded 580.

Specs on AMD page show the same thing: https://www.amd.com/en/RX-series#paragraph-341606

Would be good to hear about the actual differences, overclock (to 100Mhz, as some have posted)...

The frequency of the GPU (base and boost) and the memory is reduced to use less power. Base and boost clock speed is ~ 150mhz less and memory is ~300mhz less with the Pro 580X vs RX 580 respectively.

This results in slower memory bandwidth (217 GB/s vs 256 GB/s) and lower RAW performance (5.5 Tflops vs 617 Tflops).

Polaris 20 (XTX), die size (232mm^2), process (14nm), memory size (8gb), shading units (2304), raster operation units (32), compute units (32), PCIe 3.0 x16, etc etc ALL the same between the Pro 580X and RX 580.

The Radeon Pro being IGP (integrated graphic processor) implemented (no display output ports, power connectors, etc) and TDP dialed back for thermals is why people categorize them as mobile GPU's. But they are both the same, saying the Pro 580X is a desktop GPU is just as accurate as saying the RX 580 is a mobile GPU, it just depends on their implementation.

The iMac uses Polaris 20 XTX and the MacBook Pro uses Polaris 21 XT. The 21 is pretty much the 20 divided in half. Half the memory bandwidth, half the compute, half the TDP, half the PCIe lanes, etc. So if we were looking for a mobile variant of Radeon Pro the Polaris 21 would be it.

BTW, AMD sells one GPU for every 10 products they market. AMD's product line is very convoluted to say the least but with Polaris and Vega the LOWER GPU's name the faster it is this runs at odds with the products name. Ex Radeon Pro Vega 48 uses Vega 10 and its a faster the Radeon Pro Vega 20 which uses Vega 12. And why is it the Radeon Pro Vega and not just higher numbers? The polaris wasn't the Radeon Pro Polaris....ugh

 

[kaintxu]

Good catch about the OpenGL, Direct3D difference.

I’ll try the Open GL when I get home in two hours. My +25% result for i9/Vega vs i5/RX580X still stands since the screenshot higher in the thread shows Direct3D11.

I’ll also re-run the benchmark in Direct3D11 to eliminate any random error creeping during the benchmark.

And nice idea about making a compilation in a new thread, it will make the numbers easier to find for the ones hesitating between the different configs !

It might be bothering you for too much, but could you runthe test both on windows and OS X, if we can get the 4 (2 for direct3D11 and Open GL) to compare on both OS would be great

 

[macduke]

Thanks for posting this! I'd have expected more - @macduke reports a score of 1713 above and I think I've seen a score of approx. 1700 in one of the videos, too. Of course there's always variation, and one never knows whether the same settings were used.

I always benchmark at a baseline of all third party apps terminated. No background Dropbox, MagicPrefs, Paste, Bartender, Droplr, 1Password Mini or other utilities. This is usually what professional testers do to be consistent and establish a baseline for comparison. I also wait at least a day on a new machine so Spotlight can stop indexing, iCloud Photos can sync, Time machine can backup, etc. I also unplug drives and external displays. I benchmark after a reboot and wait about 5-10 minutes for everything to settle down. Might explain why mine is faster, but it’s also not a fake result even though you’d be running these apps while working because it removes variables between user configurations when testing. Many people don’t think about this and could be running tests with all kinds of things going on in the background and who knows, maybe even half the Adobe Creative Cloud running, lol. I also check activity monitor before testing to make sure nothing weird is going on and that the CPU is close to 0%. I’d say my results are more likely to be consistent with professional benchmark results.