@kyle.topazlabs Any idea when you guys will be supporting the NPU offloading/assist for the Intel Core Ultra processors? This should provide a big boost. I know you guys are working with them on this.
Interesting AIDA values. Throughput looks great but latency is almost twice as much as it could be it seems (if the values are correct because they are really high). Halving the latency could get u a +10% in the 4X results, i think.
Topaz Video AI v6.0.0
System Information
OS: Mac v15.02
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: 07.78 fps 2X: 04.60 fps 4X: 01.38 fps
Iris 1X: 10.57 fps 2X: 03.01 fps 4X: 00.95 fps
Proteus 1X: 08.12 fps 2X: 05.19 fps 4X: 01.43 fps
Gaia 1X: 02.51 fps 2X: 01.73 fps 4X: 01.31 fps
Nyx 1X: 02.18 fps 2X: 01.96 fps
Nyx Fast 1X: 05.12 fps
Rhea 4X: 00.37 fps
RXL 4X: 00.40 fps
Hyperion HDR 1X: 28.51 fps
4X Slowmo Apollo: 08.65 fps APFast: 33.15 fps Chronos: 02.42 fps CHFast: 04.08 fps
16X Slowmo Aion: 06.10 fps
Yep, a fact of life with that separated memory controller in the package it seems.
Ok, high latency seems to be a disadvantage of CUDIMMs. I am just surprised that it seems to lower image scaling performance in TVAI despite the high read and write performance. High latency usually affects performance negatively when u have a lot of random memory access operations (e.g. gaming). But image scaling should be mostly sequential access which should benefit from high read/write. Maybe there is something to optimize for TVAI.
Not going to bother installing TVAI 6 on my main machine until they fix a few things. For now, here’s from what everyone says is the last stable version:
Topaz Video AI v5.3.6
System Information
OS: Windows v11.24
CPU: AMD Ryzen 9 7900X 12-Core Processor 31.688 GB
GPU: NVIDIA GeForce RTX 4070 Ti 11.715 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 24.64 fps 2X: 15.21 fps 4X: 04.17 fps
Iris 1X: 25.80 fps 2X: 15.66 fps 4X: 04.02 fps
Proteus 1X: 23.83 fps 2X: 16.07 fps 4X: 05.46 fps
Gaia 1X: 08.32 fps 2X: 05.76 fps 4X: 03.81 fps
Nyx 1X: 09.52 fps 2X: 08.42 fps
Nyx Fast 1X: 16.93 fps
Rhea 4X: 03.22 fps
4X Slowmo
Apollo: 37.63 fps
APFast: 76.31 fps
Chronos: 18.80 fps
CHFast: 29.41 fps
16X Slowmo
Aion: 32.52 fps
Thanks for this. Really impressive. I thought the loss of Hyperthreading with the Core Ultra series would be fatal in Topaz Video AI, but looks otherwise!
Actually, TVAI is faster using only 8 out of 16 cores of my AMD 7950. Partially because in that case it only uses a single 8 core CCD and there is no CCD to CCD communication penalty but TVAI can’t really make good use of 16 cores.
Topaz Video AI Beta v6.0.0.1.b
System Information
OS: Windows v11.24
CPU: 12th Gen Intel(R) Core(TM) i7-12700K 31.686 GB
GPU: NVIDIA GeForce RTX 3060 Ti 7.8359 GB
GPU: Intel(R) UHD Graphics 770 0.125 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 11.85 fps 2X: 08.79 fps 4X: 02.97 fps
Iris 1X: 11.62 fps 2X: 07.22 fps 4X: 02.22 fps
Proteus 1X: 11.45 fps 2X: 08.55 fps 4X: 03.05 fps
Gaia 1X: 03.89 fps 2X: 02.67 fps 4X: 01.82 fps
Nyx 1X: 04.70 fps 2X: 04.05 fps
Nyx Fast 1X: 09.26 fps
Rhea 4X: 01.56 fps
RXL 4X: 01.49 fps
Hyperion HDR 1X: 26.50 fps
4X Slowmo Apollo: 17.58 fps APFast: 47.24 fps Chronos: 09.09 fps CHFast: 14.90 fps
16X Slowmo Aion: 20.77 fps
Have you tried using the model where benchmark stops on an actual video?
I have extremely fast.
Topaz Video AI v6.0.0
System Information
OS: Windows v11.24
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: 18.48 fps 2X: 12.31 fps 4X: 04.07 fps
Iris 1X: 18.52 fps 2X: 11.21 fps 4X: 03.18 fps
Proteus 1X: 17.44 fps 2X: 12.25 fps 4X: 03.77 fps
Gaia 1X: 06.10 fps 2X: 04.19 fps 4X: 02.85 fps
Nyx 1X: 07.17 fps 2X: 06.10 fps
Nyx Fast 1X: 13.35 fps
Rhea 4X: 02.35 fps
RXL 4X: 02.10 fps
Hyperion HDR 1X: 18.05 fps
4X Slowmo Apollo: 24.59 fps APFast: 55.01 fps Chronos: 13.03 fps CHFast: 20.62 fps
16X Slowmo Aion: 22.31 fps
Topaz Video AI v6.0.0
System Information
OS: Windows v11.24
CPU: Intel(R) Core(TM) Ultra 9 285K 31.371 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
GPU: Intel(R) Graphics 0.125 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 40.85 fps 2X: 14.69 fps 4X: 03.78 fps
Iris 1X: 37.87 fps 2X: 21.12 fps 4X: 04.59 fps
Proteus 1X: 40.36 fps 2X: 17.51 fps 4X: 04.57 fps
Gaia 1X: 15.36 fps 2X: 10.54 fps 4X: 04.35 fps
Nyx 1X: 16.91 fps 2X: 14.54 fps
Nyx Fast 1X: 30.45 fps
Rhea 4X: 04.35 fps
RXL 4X: 04.31 fps
Hyperion HDR 1X: 21.45 fps
4X Slowmo Apollo: 34.81 fps APFast: 79.71 fps Chronos: 30.96 fps CHFast: 28.85 fps
16X Slowmo Aion: 27.83 fps
Topaz Video AI v6.0.1
System Information
OS: Windows v11.24
CPU: AMD Ryzen 7 5800X3D 8-Core Processor 63.928 GB
GPU: NVIDIA GeForce RTX 4070 Ti SUPER 15.687 GB
Processing Settings
device: -2 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 20.52 fps 2X: 09.07 fps 4X: 01.95 fps
Iris 1X: 20.35 fps 2X: 09.49 fps 4X: 02.55 fps
Proteus 1X: 22.07 fps 2X: 09.61 fps 4X: 02.62 fps
Gaia 1X: 08.29 fps 2X: 05.67 fps 4X: 02.28 fps
Nyx 1X: 09.70 fps 2X: 07.49 fps
Nyx Fast 1X: 17.34 fps
Rhea 4X: 02.24 fps
RXL 4X: 02.08 fps
Hyperion HDR 1X: 16.46 fps
4X Slowmo Apollo: 23.70 fps APFast: 38.83 fps Chronos: 17.20 fps CHFast: 19.51 fps
16X Slowmo Aion: 25.86 fps
Running on a couple year old Asus ROG Zephyrus G14 laptop (plugged in of course). The Radeon was not as bad as I expected, quite workable. But memory keeps making HDR (Hyperion) have issues at surprising times.
Topaz Video AI v6.0.1
System Information
OS: Windows v11.23
CPU: AMD Ryzen 9 6900HS with Radeon Graphics 15.231 GB
GPU: AMD Radeon RX 6700S 7.9558 GB
GPU: AMD Radeon(TM) Graphics 0.47439 GB
Processing Settings
device: 0 vram: 0.95 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 05.50 fps 2X: 03.68 fps 4X: 01.27 fps
Iris 1X: 06.01 fps 2X: 03.38 fps 4X: 01.06 fps
Proteus 1X: 05.43 fps 2X: 03.89 fps 4X: 01.69 fps
Gaia 1X: 02.24 fps 2X: 01.54 fps 4X: 01.12 fps
Nyx 1X: 02.01 fps 2X: 01.71 fps
Nyx Fast 1X: 04.14 fps
Rhea 4X: 00.62 fps
RXL 4X: 00.57 fps
Hyperion HDR 1X: 01.93 fps
4X Slowmo Apollo: 06.16 fps APFast: 22.55 fps Chronos: 03.66 fps CHFast: 06.04 fps
16X Slowmo Aion: ERR fps
Upgraded my sim rig from a 5800X3D to a 9800X3D:
Topaz Video AI v6.0.0
System Information
OS: Windows v10.22
CPU: AMD Ryzen 7 9800X3D 8-Core Processor 61.655 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
GPU: AMD Radeon(TM) Graphics 1.9744 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 45.04 fps 2X: 19.35 fps 4X: 04.66 fps
Iris 1X: 41.98 fps 2X: 21.90 fps 4X: 05.27 fps
Proteus 1X: 44.22 fps 2X: 22.84 fps 4X: 05.68 fps
Gaia 1X: 15.77 fps 2X: 10.90 fps 4X: 05.10 fps
Nyx 1X: 18.82 fps 2X: 15.65 fps
Nyx Fast 1X: 36.88 fps
Rhea 4X: 04.81 fps
RXL 4X: 04.91 fps
Hyperion HDR 1X: 31.71 fps
4X Slowmo Apollo: 46.29 fps APFast: 90.73 fps Chronos: 34.43 fps CHFast: 39.33 fps
16X Slowmo Aion: 32.90 fps
Topaz Video AI v6.0.1
System Information
OS: Windows v11.23
CPU: Intel(R) Core(TM) i7-14700K 31.781 GB
GPU: NVIDIA GeForce RTX 4090 23.576 GB
GPU: Intel(R) UHD Graphics 770 0.125 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 39.98 fps 2X: 19.54 fps 4X: 05.44 fps
Iris 1X: 37.17 fps 2X: 21.78 fps 4X: 06.02 fps
Proteus 1X: 40.55 fps 2X: 21.71 fps 4X: 05.99 fps
Gaia 1X: 15.34 fps 2X: 10.66 fps 4X: 05.51 fps
Nyx 1X: 17.02 fps 2X: 14.07 fps
Nyx Fast 1X: 29.92 fps
Rhea 4X: 05.49 fps
RXL 4X: 04.79 fps
Hyperion HDR 1X: 27.38 fps
4X Slowmo Apollo: 41.68 fps APFast: 79.23 fps Chronos: 32.39 fps CHFast: 36.55 fps
16X Slowmo Aion: 34.99 fps
Topaz Video AI v6.0.1
System Information
OS: Mac v15.02
CPU: Apple M4 Max 64 GB
GPU: Apple M4 Max 48 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 3840x2160
Benchmark Results
Artemis 1X: 03.04 fps 2X: 01.70 fps 4X: 00.58 fps
Iris 1X: 02.59 fps 2X: 01.19 fps 4X: 00.37 fps
Proteus 1X: 02.76 fps 2X: 01.73 fps 4X: 00.58 fps
Gaia 1X: 00.86 fps 2X: 00.62 fps 4X: 00.45 fps
Nyx 1X: 00.93 fps 2X: 00.81 fps
Nyx Fast 1X: 02.11 fps
Rhea 4X: 00.21 fps
RXL 4X: 00.23 fps
Hyperion HDR 1X: 23.15 fps
4X Slowmo Apollo: 02.93 fps APFast: 08.45 fps Chronos: 01.10 fps CHFast: 01.71 fps
16X Slowmo Aion: 04.65 fps
Topaz Video AI v6.0.1
System Information
OS: Windows v10.22
CPU: AMD Ryzen Threadripper 3970X 32-Core Processor 127.87 GB
GPU: NVIDIA GeForce RTX 3090 23.756 GB
Processing Settings
device: -2 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 24.40 fps 2X: 12.39 fps 4X: 03.66 fps
Iris 1X: 21.99 fps 2X: 14.39 fps 4X: 03.65 fps
Proteus 1X: 22.17 fps 2X: 13.81 fps 4X: 03.99 fps
Gaia 1X: 07.99 fps 2X: 05.73 fps 4X: 03.23 fps
Nyx 1X: 10.20 fps 2X: 08.36 fps
Nyx Fast 1X: 20.81 fps
Rhea 4X: 02.96 fps
RXL 4X: 02.86 fps
Hyperion HDR 1X: 20.91 fps
4X Slowmo Apollo: 31.94 fps APFast: 58.65 fps Chronos: 18.71 fps CHFast: 26.59 fps
16X Slowmo Aion: 24.55 fps
Topaz Video AI v6.0.1
System Information
OS: Windows v11.24
CPU: AMD Ryzen 9 9900X 12-Core Processor 31.113 GB
GPU: AMD Radeon RX 7800 XT 15.774 GB
GPU: AMD Radeon(TM) Graphics 0.44899 GB
Processing Settings
device: -2 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis 1X: 13.72 fps 2X: 09.44 fps 4X: 03.70 fps
Iris 1X: 16.01 fps 2X: 09.65 fps 4X: 03.32 fps
Proteus 1X: 15.54 fps 2X: 11.03 fps 4X: 04.40 fps
Gaia 1X: 06.06 fps 2X: 04.16 fps 4X: 02.18 fps
Nyx 1X: 06.36 fps 2X: 05.45 fps
Nyx Fast 1X: 11.99 fps
Rhea 4X: 00.78 fps
RXL 4X: 00.91 fps
Hyperion HDR 1X: 23.74 fps
4X Slowmo Apollo: 21.55 fps APFast: 58.27 fps Chronos: 10.15 fps CHFast: 16.47 fps
16X Slowmo Aion: 39.40 fps