Video AI v3.3.X - User Benchmarking Results

[Alex.Schneider]

Topaz Video AI v3.3.0
System Information
OS: Windows v11.22
CPU: AMD Ryzen 9 7950X 16-Core Processor 63.729 GB
GPU: NVIDIA GeForce RTX 4080 15.688 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 3840x2160
Benchmark Results
Artemis		1X: 	06.32 fps 	2X: 	02.94 fps 	4X: 	00.80 fps 	
Proteus		1X: 	05.68 fps 	2X: 	02.82 fps 	4X: 	00.81 fps 	
Gaia		1X: 	02.31 fps 	2X: 	01.57 fps 	4X: 	00.84 fps 	
4X Slowmo		Apollo: 	12.15 fps 	APFast: 	16.79 fps 	Chronos: 	05.14 fps 	CHFast: 	09.94 fps 	

[07broom.fibulae (Andy)]

Thanks for including results for an SD input resolution. Could I ask a favour, please - any chance of uploading the results of the 720x480 again with memory reduced to the minimum 10% to see the increase in speed.

Many thanks.

Andy

[Alex.Schneider]

Topaz Video AI  v3.3.0
System Information
OS: Windows v11.22
CPU: AMD Ryzen 9 7950X 16-Core Processor              63.729 GB
GPU: NVIDIA GeForce RTX 4080  15.688 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	30.74 fps 	2X: 	12.22 fps 	4X: 	03.25 fps 	
Proteus		1X: 	26.53 fps 	2X: 	11.95 fps 	4X: 	03.18 fps 	
Gaia		1X: 	10.87 fps 	2X: 	07.39 fps 	4X: 	03.38 fps 	
4X Slowmo		Apollo: 	38.52 fps 	APFast: 	55.73 fps 	Chronos: 	23.96 fps 	CHFast: 	32.14 fps 	

[Alex.Schneider]

MAX. MEMORY USAGE (%): 100

Topaz Video AI  v3.3.0
System Information
OS: Windows v11.22
CPU: AMD Ryzen 9 7950X 16-Core Processor              63.729 GB
GPU: NVIDIA GeForce RTX 4080  15.688 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis		1X: 	147.54 fps 	2X: 	68.45 fps 	4X: 	17.24 fps 	
Proteus		1X: 	124.14 fps 	2X: 	60.90 fps 	4X: 	17.36 fps 	
Gaia		1X: 	60.58 fps 	2X: 	42.68 fps 	4X: 	23.04 fps 	
4X Slowmo		Apollo: 	170.20 fps 	APFast: 	325.95 fps 	Chronos: 	131.59 fps 	CHFast: 	162.67 fps 	

MAX. MEMORY USAGE (%): 10

Topaz Video AI  v3.3.0
System Information
OS: Windows v11.22
CPU: AMD Ryzen 9 7950X 16-Core Processor              63.729 GB
GPU: NVIDIA GeForce RTX 4080  15.688 GB
Processing Settings
device: 0 vram: 0.1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis		1X: 	99.34 fps 	2X: 	62.36 fps 	4X: 	19.44 fps 	
Proteus		1X: 	89.50 fps 	2X: 	59.29 fps 	4X: 	16.76 fps 	
Gaia		1X: 	38.82 fps 	2X: 	28.90 fps 	4X: 	18.79 fps 	
4X Slowmo		Apollo: 	98.84 fps 	APFast: 	164.27 fps 	Chronos: 	71.75 fps 	CHFast: 	85.17 fps 	

[adrian.wyard]

Mac Studio M2 Max, benchmarks upthread were with Max memory set to 100%. Now set to 10% memory setting:

Why would speed increase for SD in this case?!?

FHD

Topaz Video AI  v3.3.0
System Information
OS: Mac v13.04
CPU: Apple M2 Max  64 GB
GPU: Apple M2 Max  48 GB
Processing Settings
device: 0 vram: 0.1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	08.07 fps 	2X: 	05.63 fps 	4X: 	02.38 fps 	
Proteus		1X: 	07.99 fps 	2X: 	05.59 fps 	4X: 	02.23 fps 	
Gaia		1X: 	02.47 fps 	2X: 	01.82 fps 	4X: 	01.43 fps 	
4X Slowmo		Apollo: 	07.15 fps 	APFast: 	30.80 fps 	Chronos: 	02.85 fps 	CHFast: 	04.23 fps 	

720x480

Topaz Video AI  v3.3.0
System Information
OS: Mac v13.04
CPU: Apple M2 Max  64 GB
GPU: Apple M2 Max  48 GB
Processing Settings
device: 0 vram: 0.1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis		1X: 	52.56 fps 	2X: 	34.65 fps 	4X: 	11.74 fps 	
Proteus		1X: 	42.30 fps 	2X: 	33.49 fps 	4X: 	11.09 fps 	
Gaia		1X: 	15.60 fps 	2X: 	11.50 fps 	4X: 	09.43 fps 	
4X Slowmo		Apollo: 	41.59 fps 	APFast: 	154.06 fps 	Chronos: 	18.86 fps 	CHFast: 	26.78 fps 	

[07broom.fibulae (Andy)]

Thanks very much, Adrian. The speed increase for SD at 10% memory is because it forces the GPU cores to be used for processing rather than only the CPU cores. There is a response from Topaz which gives some explanation:

Processing preferences produce preposterous performance

Processing on M1/M2 Macs is always Neural Engine(NE) first, in almost all cases the NE has similar performance to the GPU at a fraction of power, so it is favored. On the M1/M2 Macs since everything is on a single chip, selecting CPU or M1/M2 will have the same internal selections. This seems like a bug. It would be great if you can provide more information about this problem. Can you send me logs with memory set to 100% and whatever lower number that is giving you better performance. Memory …

Thanks again.

Andy

[michaelandrews (michael.A)]

Topaz Video AI  v3.3.1
System Information
OS: Windows v11.22
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: 	193.50 fps 	2X: 	74.84 fps 	4X: 	21.38 fps 	
Proteus		1X: 	140.10 fps 	2X: 	67.18 fps 	4X: 	20.82 fps 	
Gaia		1X: 	83.83 fps 	2X: 	61.09 fps 	4X: 	27.53 fps 	
4X Slowmo		Apollo: 	216.39 fps 	APFast: 	429.27 fps 	Chronos: 	157.62 fps 	CHFast: 	192.57 fps 	

Topaz Video AI  v3.3.1
System Information
OS: Windows v11.22
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: 	38.91 fps 	2X: 	15.68 fps 	4X: 	04.33 fps 	
Proteus		1X: 	30.04 fps 	2X: 	15.98 fps 	4X: 	04.08 fps 	
Gaia		1X: 	15.94 fps 	2X: 	11.03 fps 	4X: 	04.42 fps 	
4X Slowmo		Apollo: 	39.86 fps 	APFast: 	68.99 fps 	Chronos: 	33.31 fps 	CHFast: 	32.92 fps 	

[david.123 (New Profile Picture!)]

Topaz Video AI  v3.3.1
System Information
OS: Windows v11.22
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: 	41.19 fps 	2X: 	14.48 fps 	4X: 	03.81 fps 	
Proteus		1X: 	31.61 fps 	2X: 	14.09 fps 	4X: 	03.75 fps 	
Gaia		1X: 	15.72 fps 	2X: 	10.63 fps 	4X: 	03.82 fps 	
4X Slowmo		Apollo: 	42.90 fps 	APFast: 	65.31 fps 	Chronos: 	32.27 fps 	CHFast: 	38.47 fps 	

[sergiohzph (Sergio)]

Awesome performance as always…

Topaz Video AI  v3.3.1
System Information
OS: Windows v11.22
CPU: 13th Gen Intel(R) Core(TM) i9-13900K  31.685 GB
GPU: NVIDIA GeForce RTX 4090  22.096 GB
GPU: Intel(R) UHD Graphics 770  0.125 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 3840x2160
Benchmark Results
Artemis		1X: 	08.42 fps 	2X: 	03.56 fps 	4X: 	00.58 fps 	
Proteus		1X: 	06.43 fps 	2X: 	03.30 fps 	4X: 	00.62 fps 	
Gaia		1X: 	03.21 fps 	2X: 	02.23 fps 	4X: 	00.91 fps 	
4X Slowmo		Apollo: 	14.42 fps 	APFast: 	20.47 fps 	Chronos: 	07.07 fps 	CHFast: 	12.07 fps 	

[XR.FAN]

Topaz Video AI  v3.3.1
System Information
OS: Windows v10.2
CPU: AMD Ryzen 9 3900X 12-Core Processor              127.91 GB
GPU: Radeon RX Vega  7.9299 GB
GPU: Radeon RX Vega  7.9299 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	06.43 fps 	2X: 	04.00 fps 	4X: 	01.59 fps 	
Proteus		1X: 	06.19 fps 	2X: 	03.88 fps 	4X: 	01.57 fps 	
Gaia		1X: 	03.05 fps 	2X: 	01.97 fps 	4X: 	01.35 fps 	
4X Slowmo		Apollo: 	06.73 fps 	APFast: 	21.54 fps 	Chronos: 	04.51 fps 	CHFast: 	07.12 fps 	

[Westyjester]

Topaz Video AI  v3.3.1
System Information
OS: Windows v11.22
CPU: 13th Gen Intel(R) Core(TM) i9-13900KF  63.845 GB
GPU: NVIDIA GeForce RTX 4090  23.59 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	37.99 fps 	2X: 	14.64 fps 	4X: 	03.35 fps 	
Proteus		1X: 	31.88 fps 	2X: 	13.30 fps 	4X: 	03.55 fps 	
Gaia		1X: 	15.06 fps 	2X: 	10.00 fps 	4X: 	03.86 fps 	
4X Slowmo		Apollo: 	34.92 fps 	APFast: 	62.31 fps 	Chronos: 	31.21 fps 	CHFast: 	28.48 fps 	

[jo.vo]

Interesting finding
See below for benchmarks on the Mac Studio M2 ultra.

I do have the impression that Video AI isn’t fully optimized for those chips yet and sometimes shows quite strange/erratic behaviour.

With the current 3.3.1 there is a bug when upscaling a video and having RAM set to 100% that it’ll take forever before the actual upscaling process starts as before it’ll not only fill the complete RAM (all 64GB!) but after that also fill the swap space leading to this unnecessary delay and also then slowing things down because of the unnecessary use of swap…

[jo.vo]

Mac Studio 2023 with M2 ultra (24c CPU / 60c GPU):

Upscaling DVD with RAM at 100%:

Topaz Video AI  v3.3.1
System Information
OS: Mac v13.04
CPU: Apple M2 Ultra  64 GB
GPU: Apple M2 Ultra  48 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis		1X: 	44.21 fps 	2X: 	25.56 fps 	4X: 	09.66 fps 	
Proteus		1X: 	41.97 fps 	2X: 	22.98 fps 	4X: 	06.17 fps 	
Gaia		1X: 	23.84 fps 	2X: 	15.78 fps 	4X: 	12.48 fps 	
4X Slowmo		Apollo: 	56.44 fps 	APFast: 	181.62 fps 	Chronos: 	12.14 fps 	CHFast: 	16.45 fps

And then with RAM set to 10%:

Topaz Video AI  v3.3.1
System Information
OS: Mac v13.04
CPU: Apple M2 Ultra  64 GB
GPU: Apple M2 Ultra  48 GB
Processing Settings
device: 0 vram: 0.1 instances: 1
Input Resolution: 720x480
Benchmark Results
Artemis		1X: 	60.40 fps 	2X: 	43.74 fps 	4X: 	15.16 fps 	
Proteus		1X: 	63.79 fps 	2X: 	36.45 fps 	4X: 	12.88 fps 	
Gaia		1X: 	23.02 fps 	2X: 	17.31 fps 	4X: 	12.77 fps 	
4X Slowmo		Apollo: 	52.54 fps 	APFast: 	227.39 fps 	Chronos: 	20.42 fps 	CHFast: 	30.01 fps

So that’s quite a BIG difference.

[jo.vo]

With Full HD sources the picture changes (in the benchmark):

100% RAM:
Topaz Video AI v3.3.1
System Information
OS: Mac v13.04
CPU: Apple M2 Ultra 64 GB
GPU: Apple M2 Ultra 48 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 15.27 fps 2X: 08.90 fps 4X: 02.94 fps
Proteus 1X: 15.19 fps 2X: 07.91 fps 4X: 02.50 fps
Gaia 1X: 04.78 fps 2X: 03.38 fps 4X: 02.65 fps
4X Slowmo Apollo: 10.52 fps APFast: 33.48 fps Chronos: 05.25 fps CHFast: 07.20 fps

10% RAM:
Topaz Video AI v3.3.1
System Information
OS: Mac v13.04
CPU: Apple M2 Ultra 64 GB
GPU: Apple M2 Ultra 48 GB
Processing Settings
device: 0 vram: 0.1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 10.06 fps 2X: 07.11 fps 4X: 02.71 fps
Proteus 1X: 09.90 fps 2X: 07.09 fps 4X: 02.33 fps
Gaia 1X: 04.03 fps 2X: 02.95 fps 4X: 01.91 fps
4X Slowmo Apollo: 08.72 fps APFast: 36.18 fps Chronos: 03.97 fps CHFast: 05.67 fps

[07broom.fibulae (Andy)]

Thanks very much for that most enlightening comparison. I think Apple Silicon users upscaling SD are being let down by Topaz - they are not even taking the easiest step - to inform users on maximising performance. Even worse - their online user guide still states, without exception, that increasing memory increases performance. That is demonstratively incorrect (and the opposite) when upscaling SD on Apple Silicon. I’m going to raise that in Bugs and Issues.

Thanks.

Andy

[jsjjsj]

Topaz Video AI  v3.3.1
System Information
OS: Mac v13.0401
CPU: Apple M2 Ultra  128 GB
GPU: Apple M2 Ultra  96 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	17.09 fps 	2X: 	09.58 fps 	4X: 	03.10 fps 	
Proteus		1X: 	16.22 fps 	2X: 	08.19 fps 	4X: 	02.71 fps 	
Gaia		1X: 	06.32 fps 	2X: 	04.32 fps 	4X: 	03.22 fps 	
4X Slowmo		Apollo: 	11.61 fps 	APFast: 	40.13 fps 	Chronos: 	06.02 fps 	CHFast: 	07.47 fps 	

[XR.FAN]

Topaz Video AI  v3.3.0
System Information
OS: Windows v10.2
CPU: AMD Ryzen 9 3900X 12-Core Processor              127.91 GB
GPU: NVIDIA GeForce RTX 4080  15.688 GB
GPU: NVIDIA GeForce GTX 1080 Ti  10.859 GB
GPU: NVIDIA GeForce RTX 4080  15.688 GB
Processing Settings
device: 3 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	20.32 fps 	2X: 	07.61 fps 	4X: 	01.97 fps 	
Proteus		1X: 	19.35 fps 	2X: 	06.91 fps 	4X: 	01.85 fps 	
Gaia		1X: 	12.32 fps 	2X: 	05.74 fps 	4X: 	00.79 fps 	
4X Slowmo		Apollo: 	10.83 fps 	APFast: 	32.66 fps 	Chronos: 	13.37 fps 	CHFast: 	10.07 fps 	

[XR.FAN]

Topaz Video AI  v3.3.0
System Information
OS: Windows v10.2
CPU: AMD Ryzen 9 3900X 12-Core Processor              127.91 GB
GPU: NVIDIA GeForce RTX 4080  15.688 GB
GPU: NVIDIA GeForce RTX 4080  15.688 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	23.00 fps 	2X: 	08.04 fps 	4X: 	02.16 fps 	
Proteus		1X: 	15.26 fps 	2X: 	07.52 fps 	4X: 	02.03 fps 	
Gaia		1X: 	10.46 fps 	2X: 	07.16 fps 	4X: 	02.02 fps 	
4X Slowmo		Apollo: 	27.20 fps 	APFast: 	41.34 fps 	Chronos: 	22.38 fps 	CHFast: 	21.80 fps 	

[XR.FAN]

Topaz Video AI  v3.3.0
System Information
OS: Windows v10.2
CPU: AMD Ryzen 9 3900X 12-Core Processor              127.91 GB
GPU: NVIDIA GeForce GTX 1080 Ti  10.859 GB
GPU: NVIDIA GeForce GTX 1080 Ti  10.859 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	04.10 fps 	2X: 	02.72 fps 	4X: 	01.07 fps 	
Proteus		1X: 	03.86 fps 	2X: 	02.60 fps 	4X: 	01.04 fps 	
Gaia		1X: 	01.50 fps 	2X: 	01.00 fps 	4X: 	ERR fps 	
4X Slowmo		Apollo: 	04.88 fps 	APFast: 	19.43 fps 	Chronos: 	02.43 fps 	CHFast: 	03.81 fps 	

[elfamosisimojon]

is this the 60 core or 76 core version?