Topaz Video AI v4.1.0
System Information
OS: Windows v11.23
CPU: AMD Ryzen 9 5900X 12-Core Processor 31.922 GB
GPU: NVIDIA GeForce RTX 3080 Ti 11.816 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 22.55 fps 2X: 12.99 fps 4X: 03.14 fps
Iris 1X: 21.53 fps 2X: 13.57 fps 4X: 03.98 fps
Proteus 1X: 22.25 fps 2X: 14.88 fps 4X: 04.21 fps
Gaia 1X: 08.01 fps 2X: 05.53 fps 4X: 03.53 fps
Nyx 1X: 09.77 fps 2X: 08.03 fps
Nyx Fast 1X: 18.89 fps
4X Slowmo Apollo: 33.30 fps APFast: 57.56 fps Chronos: 18.12 fps CHFast: 28.19 fps
Topaz Video AI v4.1.0
System Information
OS: Windows v11.23
CPU: Intel(R) Xeon(R) w5-2465X 127.25 GB
GPU: NVIDIA GeForce RTX 4070 11.744 GB
GPU: NVIDIA RTX A4000 15.804 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 18.43 fps 2X: 12.94 fps 4X: 03.97 fps
Iris 1X: 19.15 fps 2X: 11.78 fps 4X: 03.34 fps
Proteus 1X: 18.13 fps 2X: 12.64 fps 4X: 04.51 fps
Gaia 1X: 06.40 fps 2X: 04.44 fps 4X: 03.00 fps
Nyx 1X: 07.51 fps 2X: 06.28 fps
Nyx Fast 1X: 13.78 fps
4X Slowmo Apollo: 27.37 fps APFast: 55.02 fps Chronos: 13.55 fps CHFast: 21.01 fps
RTX4070 75%TDP, update Nyx fast score
Topaz Video AI v4.1.0
System Information
OS: Windows v10.22
CPU: AMD Ryzen 9 7950X 16-Core Processor 31.118 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
GPU: AMD Radeon(TM) Graphics 0.47445 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 43.21 fps 2X: 18.74 fps 4X: 05.29 fps
Iris 1X: 43.96 fps 2X: 21.90 fps 4X: 06.11 fps
Proteus 1X: 42.66 fps 2X: 22.00 fps 4X: 06.05 fps
Gaia 1X: 15.69 fps 2X: 10.90 fps 4X: 05.13 fps
Nyx 1X: 17.56 fps 2X: 16.28 fps
Nyx Fast 1X: 31.50 fps
4X Slowmo Apollo: 51.55 fps APFast: 111.72 fps Chronos: 34.26 fps CHFast: 41.58 fps
DDR5 6200@CL30 optimized RAM which speeds up 2X and 4X processing compared to unoptimized RAM
You sir have a dream machineā¦sadly I just built my system back in April of 2021 so I have to try and squeeze a few more years of life out of it before building another (at least I promised my wife to try and get at least 4-5 years out of each build 
)
My 1X results are very similar to the other 7950X + 4090 system here. Not sure why your 1X results are lower because it should depend primarily on the GPU (my GPU was running @ 2800Mhz with voltage lock).
When comparing the 2X and 4X results, the 6200 RAM seems to make the difference compared to the other 7950X system. Intels memory subsystem is a little more efficient and has more headroom compared to AMD, but that requires some fast RAM and tuning work.
So then tell her you DID try hard - but it just didnāt work out 
1 Like
Topaz Video AI v4.1.0
System Information
OS: Mac v14.0201
CPU: Apple M3 Max 128 GB
GPU: Apple M3 Max 96 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 12.59 fps 2X: 07.59 fps 4X: 02.77 fps
Iris 1X: 08.63 fps 2X: 04.86 fps 4X: 01.84 fps
Proteus 1X: 12.33 fps 2X: 08.19 fps 4X: 02.56 fps
Gaia 1X: 03.50 fps 2X: 02.28 fps 4X: 01.59 fps
Nyx 1X: 05.10 fps 2X: 01.90 fps
Nyx Fast 1X: 12.81 fps
4X Slowmo Apollo: 12.62 fps APFast: 46.48 fps Chronos: 04.10 fps CHFast: 06.69 fps
Topaz Video AI v4.1.0
System Information
OS: Mac v13.0603
CPU: Intel(R) Core(TM) i5-7600K CPU @ 3.80GHz 32 GB
GPU: AMD Radeon RX 480 8 GB
GPU: AMD Radeon Pro 580 8 GB
Processing Settings
device: 2 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 02.97 fps 2X: 01.77 fps 4X: 00.79 fps
Iris 1X: 02.04 fps 2X: 01.15 fps 4X: 00.41 fps
Proteus 1X: 02.88 fps 2X: 01.63 fps 4X: 00.74 fps
Gaia 1X: 01.09 fps 2X: 00.77 fps 4X: 00.58 fps
Nyx 1X: 01.51 fps 2X: 01.58 fps
Nyx Fast 1X: 03.33 fps
4X Slowmo Apollo: 02.61 fps APFast: 09.31 fps Chronos: 00.44 fps CHFast: 01.13 fps
Yes indeed. The benchmark utility needs reworking too, as in some cases the āsystemā files used for upscale benchmarking are clearly too short in length to produce realistic results. Comparable between different systems maybe (or maybe not), but producing fps results far higher than real world processing. This can be misleading to potential new purchasers currently running the free trial.
Perhaps even more important, the relative fps between the different models is often misleading e.g. models with the fastest TVAI benchmarks can be the slowest by a long way, in real world production. I for one have to do my own time consuming testing, it would be nice to be able to use the TVAI benchmarks even if they as a result took longer to run.
Topaz Video AI v4.1.0
System Information
OS: Mac v14.0201
CPU: Apple M1 Max 32 GB
GPU: Apple M1 Max 21.333 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 09.38 fps 2X: 05.55 fps 4X: 01.98 fps
Iris 1X: 06.63 fps 2X: 03.24 fps 4X: 01.33 fps
Proteus 1X: 11.77 fps 2X: 06.54 fps 4X: 01.73 fps
Gaia 1X: 02.69 fps 2X: 01.96 fps 4X: 01.43 fps
Nyx 1X: 03.62 fps 2X: 03.18 fps
Nyx Fast 1X: 10.05 fps
4X Slowmo Apollo: 09.50 fps APFast: 33.22 fps Chronos: 03.08 fps CHFast: 05.26 fps
Base M3, but the 10 GPU Core version. Nice to see Iris x2 being more inline with expectations than it was before.
Topaz Video AI v4.1.0
System Information
OS: Mac v14.0201
CPU: Apple M3 24 GB
GPU: Apple M3 16 GB
Processing Settings
device: 0 vram: 0.74 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 05.97 fps 2X: 03.98 fps 4X: 01.51 fps
Iris 1X: 04.32 fps 2X: 02.44 fps 4X: 00.67 fps
Proteus 1X: 05.03 fps 2X: 03.05 fps 4X: 01.12 fps
Gaia 1X: 01.85 fps 2X: 01.33 fps 4X: 00.98 fps
Nyx 1X: 01.76 fps 2X: 01.19 fps
Nyx Fast 1X: 03.76 fps
4X Slowmo Apollo: 07.09 fps APFast: 28.35 fps Chronos: 02.25 fps CHFast: 03.26 fps
And the same with 100% memory usage
Topaz Video AI v4.1.0
System Information
OS: Mac v14.0201
CPU: Apple M3 24 GB
GPU: Apple M3 16 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 06.52 fps 2X: 04.20 fps 4X: 01.55 fps
Iris 1X: 04.36 fps 2X: 02.43 fps 4X: 00.80 fps
Proteus 1X: 05.62 fps 2X: 03.29 fps 4X: 01.27 fps
Gaia 1X: 02.01 fps 2X: 01.40 fps 4X: 01.08 fps
Nyx 1X: 01.86 fps 2X: 01.21 fps
Nyx Fast 1X: 03.85 fps
4X Slowmo Apollo: 07.57 fps APFast: 26.50 fps Chronos: 02.26 fps CHFast: 03.44 fps
Topaz Video AI v4.1.0
System Information
OS: Windows v10.22
CPU: Intel(R) Xeon(R) CPU X5690 @ 3.47GHz 23.99 GB
GPU: Radeon RX 580 Series 7.9795 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 02.50 fps 2X: 01.57 fps 4X: 00.56 fps
Iris 1X: 02.55 fps 2X: 01.43 fps 4X: 00.44 fps
Proteus 1X: 02.48 fps 2X: 01.61 fps 4X: 00.57 fps
Gaia 1X: 01.03 fps 2X: 00.50 fps 4X: ERR fps
Nyx 1X: 00.58 fps 2X: 00.77 fps
Nyx Fast 1X: 02.03 fps
4X Slowmo Apollo: 02.81 fps APFast: 08.19 fps Chronos: 01.33 fps CHFast: 02.21 fps
This is my cMac Pro running windows. Am I limted mainly by my CPU, memory or GPU? Asked another way, would I get a significant boost in speed by just upgrading the videocard in this classic mac pro or should I give up and build a new computer?
rfrankway, have you tested the M2 Max at various memory levels? Iāve read a few postings that say a higher memory setting does not necessarily get you faster results.
Thatās only true for SD content; there having RAM set to 10 can he considerably faster.
But for FHD benchmarks the RAM at high values / 100 does give the highest results.
Topaz Video AI v4.1.0
System Information
OS: Mac v14.03
CPU: Apple M2 Ultra 64 GB
GPU: Apple M2 Ultra 48 GB
Processing Settings
device: 0 vram: 0.85 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 16.71 fps 2X: 08.39 fps 4X: 02.73 fps
Iris 1X: 09.34 fps 2X: 04.79 fps 4X: 02.09 fps
Proteus 1X: 13.70 fps 2X: 10.02 fps 4X: 03.18 fps
Gaia 1X: 04.80 fps 2X: 03.25 fps 4X: 02.68 fps
Nyx 1X: 05.22 fps 2X: 04.96 fps
Nyx Fast 1X: 14.53 fps
4X Slowmo Apollo: 16.05 fps APFast: 55.97 fps Chronos: 05.22 fps CHFast: 06.94 fps
While the cheaper AppleSilicon variants mostly perform as they should, the Ultra variants donāt really seem to be fully used - especially when comparing the Max vs. the Ultra chips where the Max are nearly as fast as the Ultra (though having only half the CPU/GPU power).
I guess TVAI is not optimized to take advantage of the extra cores. I hope Topaz Labs could optimize TVAI
Wow!!!
The Apple M3 Max has the same benchmark as the Apple M2 Max:
I was actually just about to get a MaxBook Pro with Apple M3 Max, but if it canāt anymore at allā¦.
The bad thing about apple is the memory, while performance itsnt a problem, until you move to 4K, memory is and apple want a lot of money for memory you cant upgrate for yourself.
And because of this i didnt did get an apple device, i would want to bc of the mobility, but i would need to pay 2K for a very limited machine (4K for Optimum).
1 Like
Iād say au contrare: memory is a big point in favour of Apple Silicon.
Having a massive 48GB VRAM would be unaffordable with any Nvidia or AND Gfx card.
You really could use large models directly on the GPU without ever having to swap.
But TVAI does not really take advantage of this as it only uses half of the available RAM at max here.
And @mcfly777 the differences between M1,M2and M3 arenāt really that big so itās not surprising that thereās little difference between a M2Max vs. a M3Max.
