I had a feeling in early November when Apple showed charts of Topaz comparison, maybe only M4 Ultra will have significant increase.
Topaz Video AI v5.5.0
System Information
OS: Windows v11.23
CPU: AMD Ryzen 9 5950X 16-Core Processor 127.92 GB
GPU: NVIDIA GeForce RTX 3080 9.8174 GB
Processing Settings
device: -2 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 20.19 fps 2X: 09.85 fps 4X: 02.96 fps
Iris 1X: 19.62 fps 2X: 12.02 fps 4X: 03.61 fps
Proteus 1X: 19.70 fps 2X: 12.93 fps 4X: 03.76 fps
Gaia 1X: 06.78 fps 2X: 04.74 fps 4X: 03.14 fps
Nyx 1X: 08.56 fps 2X: 06.93 fps
Nyx Fast 1X: 16.60 fps
Rhea 4X: 02.72 fps
4X Slowmo Apollo: 24.56 fps APFast: 53.89 fps Chronos: 15.07 fps CHFast: 25.17 fps
16X Slowmo Aion: 34.15 fps
Topaz Video AI v5.5.0
System Information
OS: Mac v15.0101
CPU: Apple M2 Max 32 GB
GPU: Apple M2 Max 21.333 GB
Processing Settings
device: -2 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 11.13 fps 2X: 06.71 fps 4X: 01.68 fps
Iris 1X: 12.78 fps 2X: 04.09 fps 4X: 01.12 fps
Proteus 1X: 11.69 fps 2X: 07.57 fps 4X: 01.66 fps
Gaia 1X: 03.40 fps 2X: 02.34 fps 4X: 01.77 fps
Nyx 1X: 03.06 fps 2X: 03.32 fps
Nyx Fast 1X: 09.14 fps
Rhea 4X: 00.88 fps
4X Slowmo Apollo: 13.02 fps APFast: 45.41 fps Chronos: 03.64 fps CHFast: 06.13 fps
16X Slowmo Aion: 12.84 fps
MacBook Pro M2 Max Results ^
Very strangeā¦honestly not much faster than my M2 Max MacBook Proā¦I expected better results out of an M4 Max.
Waiting patiently for the first Intel Core Ultra 9 285k + CUDIMM DDR5-8200+ with nVidia RTX 4090 results. 
1 Like
Seeing for the 1st time, in Appleās history since Jobās return, the MacBook M4 seriesā prices has been slashed by hundreds pretty much tells the TRUE story of that lines performance & sales.
There have been complaints that 5.5 is slower than previous versions. I checked 5.3 benchmarks and they are faster for the M4 Max. You might want to try 5.4 since I think 5.5 big feature is cloud rendering.
Topaz Video AI v5.5.0
System Information
OS: Windows v11.23
CPU: Intel(R) Xeon(R) w5-2465X 127.25 GB
GPU: NVIDIA GeForce RTX 4070 11.73 GB
GPU: NVIDIA RTX A4000 15.79 GB
Processing Settings
device: 0 vram: 0.9 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 18.14 fps 2X: 12.09 fps 4X: 03.52 fps
Iris 1X: 18.48 fps 2X: 11.38 fps 4X: 03.25 fps
Proteus 1X: 17.80 fps 2X: 12.17 fps 4X: 04.07 fps
Gaia 1X: 06.16 fps 2X: 04.24 fps 4X: 02.88 fps
Nyx 1X: 07.22 fps 2X: 06.12 fps
Nyx Fast 1X: 13.46 fps
Rhea 4X: 02.36 fps
4X Slowmo Apollo: 22.68 fps APFast: 55.57 fps Chronos: 13.07 fps CHFast: 20.43 fps
16X Slowmo Aion: ERR fps
Yep, slower on PC and Mac for me.
Topaz Video AI v5.5.0
System Information
OS: Mac v14.0701
CPU: Intel(R) Xeon(R) W-3245 CPU @ 3.20GHz 128 GB
GPU: AMD Radeon RX 6900 XT 15.984 GB
GPU: AMD Radeon Pro W5700X 15.984 GB
GPU: AMD Radeon Pro W5700X 15.984 GB
Processing Settings
device: 3 vram: 1 instances: 1
Input Resolution: 768x576
Benchmark Results
Artemis 1X: 09.93 fps 2X: 08.28 fps 4X: 02.61 fps
Iris 1X: 16.22 fps 2X: 10.22 fps 4X: 02.13 fps
Proteus 1X: 15.91 fps 2X: 11.78 fps 4X: 02.71 fps
Gaia 1X: 05.31 fps 2X: 02.55 fps 4X: 02.44 fps
Nyx 1X: 06.81 fps 2X: 05.58 fps
Nyx Fast 1X: 11.67 fps
Rhea 4X: 03.11 fps
4X Slowmo Apollo: 10.96 fps APFast: 35.59 fps Chronos: 02.59 fps CHFast: 04.27 fps
16X Slowmo Aion: ERR fps
Topaz Video AI v5.5.0
System Information
OS: Mac v14.0701
CPU: Intel(R) Xeon(R) W-3245 CPU @ 3.20GHz 128 GB
GPU: AMD Radeon RX 6900 XT 15.984 GB
GPU: AMD Radeon Pro W5700X 15.984 GB
GPU: AMD Radeon Pro W5700X 15.984 GB
Processing Settings
device: 3 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 05.87 fps 2X: 03.10 fps 4X: 01.05 fps
Iris 1X: 05.42 fps 2X: 03.48 fps 4X: 00.99 fps
Proteus 1X: 06.27 fps 2X: 04.70 fps 4X: 01.23 fps
Gaia 1X: 02.28 fps 2X: 01.71 fps 4X: 01.21 fps
Nyx 1X: 03.48 fps 2X: 03.25 fps
Nyx Fast 1X: 07.62 fps
Rhea 4X: 01.28 fps
4X Slowmo Apollo: 07.40 fps APFast: 19.81 fps Chronos: 01.11 fps CHFast: 02.32 fps
16X Slowmo Aion: ERR fps
Topaz Video AI v5.5.0
System Information
OS: Windows v10.21
CPU: AMD EPYC 7R13 Processor 63.134 GB
GPU: NVIDIA L40S 44.124 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 24.39 fps 2X: 09.32 fps 4X: 02.80 fps
Iris 1X: 20.00 fps 2X: 13.29 fps 4X: 03.00 fps
Proteus 1X: 32.44 fps 2X: 12.54 fps 4X: 03.34 fps
Gaia 1X: 10.11 fps 2X: 07.87 fps 4X: 02.95 fps
Nyx 1X: 12.75 fps 2X: 09.51 fps
Nyx Fast 1X: 21.99 fps
Rhea 4X: 02.80 fps
4X Slowmo Apollo: 29.52 fps APFast: 49.25 fps Chronos: 21.96 fps CHFast: 25.66 fps
16X Slowmo Aion: 24.97 fps
```cloud gpu use . l40s .
1 Like
My 4/5 year old PC
Topaz Video AI v5.5.0
System Information
OS: Windows v10.22
CPU: AMD Ryzen 9 3900X 12-Core Processor 95.95 GB
GPU: NVIDIA GeForce RTX 2060 SUPER 7.8184 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 09.37 fps 2X: 06.36 fps 4X: 01.84 fps
Iris 1X: 09.69 fps 2X: 05.53 fps 4X: 01.77 fps
Proteus 1X: 09.41 fps 2X: 06.93 fps 4X: 02.22 fps
Gaia 1X: 02.90 fps 2X: 02.02 fps 4X: 01.38 fps
Nyx 1X: 03.73 fps 2X: 03.27 fps
Nyx Fast 1X: 07.86 fps
Rhea 4X: 01.18 fps
4X Slowmo Apollo: 04.27 fps APFast: 40.36 fps Chronos: 07.19 fps CHFast: 11.50 fps
16X Slowmo Aion: 13.23 fps
Nice benchmark. Happy to see something like this land in here.
This surprised me but seems to check out. Doing RHEA 480p - 1080P was 1.5-1.8fps for a single process on my M2Pro Mini.
If I run 4 concurrent, itās 0.7 per (2.8fps total). I would be very curious how the ULTRA chips scale in the regard. I remember reading somewhere that Apples AI implementation is slow and wide vs narrow and fast on NVIDIA.
If anyone has an M1/M2 Ultra and can compare and post single vs 2,3 and 4 process would be very interested to see the results, since ULTRA chips look disappointing on single benchmarks.
1 Like
Iām also waiting for this!
Hi, with 720x480 source using Rhea 4x, I can get a combined 5.8 fps on my M4 Pro Mac Mini (24 GB, 16 Core GPU). The āsecretā combination to achieve this:
Run 1 process in low power mode and another in normal mode. The M4 Pro Neural Engine really comes into its own. In low power mode, which prioritises the Neural Engine with minimal GPU - is ~ 30% faster than normal mode - maximum GPU with āhelpā from the Neural Engine. The low power mode process is running at 3.3 fps and the full power mode (using 100% GPU) is only at 2.5 - 2.6 fps.
I havenāt yet checked if I can get any further increase by changing the max memory setttings. In this example of 2 processes, it was set to 33% for both.
Thanks.
Andy
How do you specify one in low power mode and not the others?
Wow LOW POWER mode is legit??? Iām getting a combined 5fps on 2 process 480p RHEA nowā¦
You can change the processor settings (low power mode, and/or max memory) after starting a process. Process(es) already running generally wonāt be affected by the new settings.
However, there is an exception if you have the pause feature enabled: A resume after pause will use any changed processor settings, rather than the settings the process started with.
Regards
Andy
1 Like
