Video AI v3.2.X - User Benchmarking Results

TPX · June 1, 2023, 4:22pm

A new AMD Driver with optimisations for DirectML is out, targeted at 79XX GPUs.

Could someone test?

david.123 · June 2, 2023, 8:22am

Topaz Video AI  v3.2.8
System Information
OS: Windows v11.2009
CPU: AMD Ryzen 9 7950X 16-Core Processor              63.14 GB
GPU: NVIDIA GeForce RTX 4090  23.59 GB
GPU: AMD Radeon(TM) Graphics  0.47446 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	39.84 fps 	2X: 	14.30 fps 	4X: 	03.80 fps 	
Proteus		1X: 	36.23 fps 	2X: 	12.97 fps 	4X: 	03.65 fps 	
Gaia		1X: 	15.72 fps 	2X: 	10.79 fps 	4X: 	03.84 fps 	
4X Slowmo		Apollo: 	42.04 fps 	APFast: 	73.26 fps 	Chronos: 	32.10 fps 	CHFast: 	38.51 fps

TPX · June 2, 2023, 8:33am

I am testing a new approach with TVAI and with my W6800 it seems to be a waste of time and power.

michaelandrews · June 2, 2023, 12:09pm

Topaz Video AI  v3.2.8
System Information
OS: Windows v11.2009
CPU: 13th Gen Intel(R) Core(TM) i9-13900K  127.75 GB
GPU: NVIDIA GeForce RTX 4090  23.59 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis		1X: 	200.06 fps 	2X: 	76.84 fps 	4X: 	22.93 fps 	
Proteus		1X: 	146.49 fps 	2X: 	69.54 fps 	4X: 	22.19 fps 	
Gaia		1X: 	82.66 fps 	2X: 	58.58 fps 	4X: 	29.45 fps 	
4X Slowmo		Apollo: 	227.22 fps 	APFast: 	434.52 fps 	Chronos: 	170.21 fps 	CHFast: 	205.18 fps 	

Looks like new version is slower

michaelandrews · June 2, 2023, 12:10pm

Just got my mac mini last night. Looks how bad it is.

Topaz Video AI  v3.2.8
System Information
OS: Mac v13.04
CPU: Apple M2 Pro  16 GB
GPU: Apple M2 Pro  10.667 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis		1X: 	33.10 fps 	2X: 	20.45 fps 	4X: 	05.72 fps 	
Proteus		1X: 	31.81 fps 	2X: 	19.27 fps 	4X: 	06.29 fps 	
Gaia		1X: 	10.62 fps 	2X: 	08.08 fps 	4X: 	06.45 fps 	
4X Slowmo		Apollo: 	31.84 fps 	APFast: 	112.23 fps 	Chronos: 	10.75 fps 	CHFast: 	14.66 fps

michaelandrews · June 2, 2023, 12:23pm

Topaz Video AI  v3.2.8
System Information
OS: Windows v11.2009
CPU: 13th Gen Intel(R) Core(TM) i9-13900K  127.75 GB
GPU: NVIDIA GeForce RTX 4090  23.59 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	39.74 fps 	2X: 	18.56 fps 	4X: 	04.57 fps 	
Proteus		1X: 	35.09 fps 	2X: 	15.01 fps 	4X: 	04.43 fps 	
Gaia		1X: 	15.93 fps 	2X: 	10.94 fps 	4X: 	04.77 fps 	
4X Slowmo		Apollo: 	43.06 fps 	APFast: 	78.21 fps 	Chronos: 	33.49 fps 	CHFast: 	35.15 fps

michaelandrews · June 2, 2023, 12:24pm

Topaz Video AI  v3.2.8
System Information
OS: Mac v13.04
CPU: Apple M2 Pro  16 GB
GPU: Apple M2 Pro  10.667 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	07.44 fps 	2X: 	05.09 fps 	4X: 	01.75 fps 	
Proteus		1X: 	07.59 fps 	2X: 	05.19 fps 	4X: 	01.63 fps 	
Gaia		1X: 	02.49 fps 	2X: 	01.74 fps 	4X: 	01.39 fps 	
4X Slowmo		Apollo: 	05.48 fps 	APFast: 	20.78 fps 	Chronos: 	02.45 fps 	CHFast: 	04.05 fps

DaveL · June 2, 2023, 2:30pm

I’ve found that too, on my 3060 ti setup - but only slightly. Of more concern is that for at least the last two releases (I didn’t check before that), my real time results on actual video longer than 10 seconds, are 10% to 20% slower than the benchmarks at 2x and 4x on Artemis High. Proteus is much slower than that - 40% slower than benchmark. They match the benchmarks for a few seconds but rapidly stabilize at the lower values, maybe there’s a bit of a CPU issue as it’s at 80% and the GPU 80% on Art 2x, CPU 75% and GPU 50% on Prot 2x, all 1080p. But anyway, as some 480p and 1920p benchmarks have been posted recently, here are mine - albeit higher than the real world on Art and Prot.

Topaz Video AI v3.2.8
System Information
OS: Windows v10.2009
CPU: AMD Ryzen 7 5800X 8-Core Processor 31.906 GB
GPU: NVIDIA GeForce RTX 3060 Ti 7.8496 GB
Processing Settings
device: -2 vram: 0.99 instances: 0
Input Resolution: 720x480
Benchmark Results
Artemis 1X: 75.32 fps 2X: 44.08 fps 4X: 11.80 fps
Proteus 1X: 70.47 fps 2X: 39.32 fps 4X: 12.09 fps
Gaia 1X: 24.18 fps 2X: 17.40 fps 4X: 12.32 fps
4X Slowmo Apollo: 61.84 fps APFast: 150.94 fps Chronos: 51.38 fps CHFast: 70.55 fps
…
…
…
Topaz Video AI v3.2.8
System Information
OS: Windows v10.2009
CPU: AMD Ryzen 7 5800X 8-Core Processor 31.906 GB
GPU: NVIDIA GeForce RTX 3060 Ti 7.8496 GB
Processing Settings
device: -2 vram: 0.99 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 12.32 fps 2X: 08.18 fps 4X: 02.25 fps
Proteus 1X: 12.88 fps 2X: 08.39 fps 4X: 02.20 fps
Gaia 1X: 04.18 fps 2X: 02.88 fps 4X: 01.99 fps
4X Slowmo Apollo: 17.65 fps APFast: 39.88 fps Chronos: 09.40 fps CHFast: 15.95 fps

Also, comparing recent benchmarks, it looks like for upscaling on Art/Prot, I can only expect a speed uplift of less than 2x if I upgrade to a an 4090 (the one result a fraction higher is a generally more powerful system than mine). Slomo benchmarks produce a much greater improvement 3060 ti > 4090 and would justify such an upgrade - but definitely not the poorer upscaling results.

Seems to me that Topaz may have a LOT of speed optimizing to do on Artemis and Proteus for the RTX 4000 series and perhaps elsewhere too (3000s). For now, any upgrade I do will be for Stable Diffusion performance improvements, as I can’t justify £/$ 1500+ for TVAI for a 2x or less speed uplift.

ForSerious · June 2, 2023, 4:53pm

Topaz Video AI  v3.2.8
System Information
OS: Windows v11.2009
CPU: AMD Ryzen 9 5900X 12-Core Processor              31.929 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: 	22.18 fps 	2X: 	09.67 fps 	4X: 	02.63 fps 	
Proteus		1X: 	15.53 fps 	2X: 	09.33 fps 	4X: 	02.53 fps 	
Gaia		1X: 	08.04 fps 	2X: 	05.55 fps 	4X: 	02.68 fps 	
4X Slowmo		Apollo: 	28.95 fps 	APFast: 	42.90 fps 	Chronos: 	17.72 fps 	CHFast: 	24.77 fps

They should have made a new benchmarks topic on the version that changed how the benchmarks were running. Anyway, I haven’t seen any real world decrease in speeds… but I also never upscale to more than FHD.

memo90061 · June 5, 2023, 4:16am

I also bought a Mac Mini. It’s slightly faster than the 2k base model Mac Studio.

Topaz Video AI  v3.2.8
System Information
OS: Mac v13.04
CPU: Apple M2 Pro  16 GB
GPU: Apple M2 Pro  10.667 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	07.65 fps 	2X: 	05.17 fps 	4X: 	01.21 fps 	
Proteus		1X: 	07.79 fps 	2X: 	05.11 fps 	4X: 	01.21 fps 	
Gaia		1X: 	02.49 fps 	2X: 	01.82 fps 	4X: 	01.39 fps 	
4X Slowmo		Apollo: 	05.46 fps 	APFast: 	20.66 fps 	Chronos: 	02.43 fps 	CHFast: 	03.95 fps

vamyu0202 · June 5, 2023, 8:59am

Topaz Video AI  v3.2.8
System Information
OS: Windows v10.2009
CPU: AMD Ryzen 9 5900X 12-Core Processor              31.923 GB
GPU: Intel(R) Arc(TM) A750 Graphics  7.9063 GB
Processing Settings
device: 0 vram: 0.9 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	07.47 fps 	2X: 	03.31 fps 	4X: 	00.96 fps 	
Proteus		1X: 	04.95 fps 	2X: 	02.58 fps 	4X: 	00.80 fps 	
Gaia		1X: 	04.30 fps 	2X: 	02.87 fps 	4X: 	01.75 fps 	
4X Slowmo		Apollo: 	06.79 fps 	APFast: 	18.68 fps 	Chronos: 	04.96 fps 	CHFast: 	07.98 fps

michaelandrews · June 7, 2023, 1:37am

535.98 Nvidia Studio Driver
W11 debloated with latest updates
Not a gamer so only adobe suite and topaz software
Asus z790 MB
1300 watt PS
4 (2tb gen 4 m2 drives)
128 Gig DDR 5
1 boot, 1 cache, 1 media, 1 mixdown

Topaz Video AI v3.2.9
System Information
OS: Windows v11.2009
CPU: 13th Gen Intel(R) Core™ i9-13900K 127.75 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis 1X: 167.42 fps 2X: 77.56 fps 4X: 21.24 fps
Proteus 1X: 151.50 fps 2X: 67.86 fps 4X: 20.60 fps
Gaia 1X: 87.20 fps 2X: 60.32 fps 4X: 27.37 fps
4X Slowmo Apollo: 214.15 fps APFast: 402.76 fps Chronos: 166.41 fps CHFast: 204.59 fps

Topaz Video AI v3.2.9
System Information
OS: Windows v11.2009
CPU: 13th Gen Intel(R) Core™ i9-13900K 127.75 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 38.23 fps 2X: 15.67 fps 4X: 04.76 fps
Proteus 1X: 31.88 fps 2X: 15.32 fps 4X: 04.08 fps
Gaia 1X: 15.85 fps 2X: 10.24 fps 4X: 04.12 fps
4X Slowmo Apollo: 40.01 fps APFast: 71.19 fps Chronos: 33.09 fps CHFast: 33.80 fps

Topaz Video AI v3.2.8
System Information
OS: Windows v11.2009
CPU: 13th Gen Intel(R) Core™ i9-13900K 127.75 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis 1X: 200.06 fps 2X: 76.84 fps 4X: 22.93 fps
Proteus 1X: 146.49 fps 2X: 69.54 fps 4X: 22.19 fps
Gaia 1X: 82.66 fps 2X: 58.58 fps 4X: 29.45 fps
4X Slowmo Apollo: 227.22 fps APFast: 434.52 fps Chronos: 170.21 fps CHFast: 205.18 fps

Topaz Video AI v3.2.8
System Information
OS: Windows v11.2009
CPU: 13th Gen Intel(R) Core™ i9-13900K 127.75 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 39.74 fps 2X: 18.56 fps 4X: 04.57 fps
Proteus 1X: 35.09 fps 2X: 15.01 fps 4X: 04.43 fps
Gaia 1X: 15.93 fps 2X: 10.94 fps 4X: 04.77 fps
4X Slowmo Apollo: 43.06 fps APFast: 78.21 fps Chronos: 33.49 fps CHFast: 35.15 fps

david.123 · June 7, 2023, 5:58am

I agree, I think the 4000 series has more performance to unlock. It may be up to Nvidia to release driver optimizations as well.

miyconst · June 8, 2023, 8:38am

Topaz Video AI  v3.2.9

System Information
OS: Windows v10.2009
CPU: Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz  15.91 GB
GPU: Intel(R) HD Graphics 4600  0.10986 GB
GPU: NVIDIA Tesla P4  7.8701 GB

Processing Settings
device: 1 vram: 1 instances: 1

Input Resolution: 480x360
Benchmark Results
Artemis		1X: 	13.97 fps 	2X: 	09.11 fps 	4X: 	03.43 fps 	
Proteus		1X: 	13.77 fps 	2X: 	08.60 fps 	4X: 	03.33 fps 	
Gaia		1X: 	05.32 fps 	2X: 	03.41 fps 	4X: 	02.52 fps 	
4X Slowmo		Apollo: 	14.63 fps 	APFast: 	56.79 fps 	Chronos: 	09.52 fps 	CHFast: 	13.90 fps

Topaz Video AI  v3.2.9

System Information
OS: Windows v10.2009
CPU: Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz  15.91 GB
GPU: Intel(R) HD Graphics 4600  0.10986 GB
GPU: NVIDIA Tesla P4  7.8701 GB

Processing Settings
device: 1 vram: 1 instances: 1

Input Resolution: 640x480
Benchmark Results
Artemis		1X: 	08.98 fps 	2X: 	06.44 fps 	4X: 	02.43 fps 	
Proteus		1X: 	08.78 fps 	2X: 	06.27 fps 	4X: 	02.38 fps 	
Gaia		1X: 	02.86 fps 	2X: 	01.91 fps 	4X: 	01.48 fps 	
4X Slowmo		Apollo: 	08.54 fps 	APFast: 	39.35 fps 	Chronos: 	05.71 fps 	CHFast: 	08.87 fps

lptoronto · June 10, 2023, 12:45am

Those are insane results. I will be getting the base model m2 ultra so will see what results I receive on tuesday on the benchmarks. I bet it won’t be closer to what you receive. @topaz257

JesusSheep · June 10, 2023, 4:43am

Topaz Video AI Beta  v3.3.0.0.b
System Information
OS: Mac v13.04
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: 	02.87 fps 	2X: 	01.69 fps 	4X: 	00.70 fps 	
Proteus		1X: 	02.83 fps 	2X: 	01.71 fps 	4X: 	00.69 fps 	
Gaia		1X: 	01.07 fps 	2X: 	00.71 fps 	4X: 	00.55 fps 	
4X Slowmo		Apollo: 	02.13 fps 	APFast: 	07.71 fps 	Chronos: 	00.42 fps 	CHFast: 	01.06 fps

TPX · June 10, 2023, 9:31am

The P4 has no Tensor cores, the T4 and L4. would be interesting since these are inferrence GPUs.

L4 half numbers are the real numbers (without Sparsity).

FP32	30.3 teraFLOPs
TF32 Tensor Core	120 teraFLOPS*
FP16 Tensor Core	242 teraFLOPS*
BFLOAT16 Tensor Core	242 teraFLOPS*
FP8 Tensor Core	485 teraFLOPs*
INT8 Tensor Core	485 TOPs*
GPU memory	24GB
GPU memory bandwidth	300GB/s
NVENC	NVDEC
Max thermal design power (TDP)	72W
Form factor	1-slot low-profile, PCIe
Interconnect	PCIe Gen4 x16 64GB/s

y-komura · June 11, 2023, 1:20pm

Topaz Video AI  v3.2.9
System Information
OS: Windows v10.2009
CPU: AMD Ryzen Threadripper 3970X 32-Core Processor   127.87 GB
GPU: NVIDIA GeForce RTX 3090  23.77 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	24.20 fps 	2X: 	09.68 fps 	4X: 	02.71 fps 	
Proteus		1X: 	23.08 fps 	2X: 	09.56 fps 	4X: 	02.69 fps 	
Gaia		1X: 	08.50 fps 	2X: 	05.69 fps 	4X: 	02.68 fps 	
4X Slowmo		Apollo: 	29.46 fps 	APFast: 	50.22 fps 	Chronos: 	18.69 fps 	CHFast: 	25.03 fps

GTOMAN · June 13, 2023, 10:43am

Hi~! I share my benchmark results.

This is my PC information.

CPU: AMD Ryzen 7 3700X 8-Core Processor
GPU: AMD Radeon RX 5700 XT , 8GB
MAINBOARD : GIGABYTE AORUS ELITE X570
MEMORY : DDR4 , 64GB

And I recently replaced VGA from RX 5700 XT to RTX4090.

2023.05.28
Topaz Video AI v3.2.7
System Information
OS: Windows v10.2009
CPU: AMD Ryzen 7 3700X 8-Core Processor 63.952 GB
GPU: AMD Radeon RX 5700 XT 7.9605 GB

MAINBOARD : GIGABYTE AORUS ELITE X570
MEMORY : DDR4 , 64GB

Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 06.04 fps 2X: 03.87 fps 4X: 01.38 fps
Proteus 1X: 05.56 fps 2X: 03.56 fps 4X: 01.35 fps
Gaia 1X: 02.84 fps 2X: 01.88 fps 4X: 01.25 fps
4X Slowmo Apollo: 07.08 fps APFast: 24.58 fps Chronos: 04.09 fps CHFast: 06.70 fps

2023.05.31
Topaz Video AI v3.2.8
System Information
OS: Windows v10.2009
CPU: AMD Ryzen 7 3700X 8-Core Processor 63.952 GB
GPU: NVIDIA GeForce RTX 4090 23.59 GB

MAINBOARD : GIGABYTE AORUS ELITE X570
MEMORY : DDR4 , 64GB

Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 24.79 fps 2X: 08.01 fps 4X: 02.08 fps
Proteus 1X: 19.37 fps 2X: 07.38 fps 4X: 01.94 fps
Gaia 1X: 15.52 fps 2X: 07.68 fps 4X: 02.14 fps
4X Slowmo Apollo: 27.19 fps APFast: 37.32 fps Chronos: 24.75 fps CHFast: 22.44 fps

2023.06.07
Topaz Video AI  v3.2.9
System Information
OS: Windows v10.2009
CPU: AMD Ryzen 7 3700X 8-Core Processor               63.952 GB
GPU: NVIDIA GeForce RTX 4090  23.59 GB

MAINBOARD : GIGABYTE AORUS ELITE X570
MEMORY : DDR4 , 64GB

Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	25.25 fps 	2X: 	08.20 fps 	4X: 	02.07 fps 	
Proteus		1X: 	15.06 fps 	2X: 	07.41 fps 	4X: 	01.95 fps 	
Gaia		1X: 	15.47 fps 	2X: 	08.10 fps 	4X: 	02.13 fps 	
4X Slowmo		Apollo: 	27.42 fps 	APFast: 	32.58 fps 	Chronos: 	23.90 fps 	CHFast: 	22.85 fps

I was wondering how much performance 2X would improve if I upgraded my CPU to AMD 5950X. Please let me know if anyone knows.

ForSerious · June 13, 2023, 4:31pm

It’s hard to say exactly since people have only posted results from 4090s with 7000 series CPUs.
What I have noticed, is all the benchmark results posted with AMD 3000 series CPUs, are significantly slower than other CPUs.
My guess, would be your 8s would turn into 12s, but that’s just a guess. I could be very wrong.