Topaz Video 1.1.x - user benchmarking results

Users can run benchmarking (Process > Benchmark, or, Ctrl/Cmd + B) to compare results across different machines.

No real changes for Nvidia with this update.

```
Topaz Video  v1.1.0
System Information
OS: Windows v11.25
CPU: AMD Ryzen 7 9800X3D 8-Core Processor             63.556 GB
GPU: NVIDIA GeForce RTX 5070 Ti  15.298 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	30.29 fps 	2X: 	14.51 fps 	4X: 	04.29 fps 	
Iris		1X: 	33.97 fps 	2X: 	16.90 fps 	4X: 	04.63 fps 	
Proteus		1X: 	31.95 fps 	2X: 	16.85 fps 	4X: 	04.97 fps 	
Gaia		1X: 	09.45 fps 	2X: 	06.66 fps 	4X: 	04.74 fps 	
Nyx		1X: 	13.68 fps 	2X: 	11.82 fps 	
Nyx Fast		1X: 	33.75 fps 	
Nyx XL		1X: 	30.90 fps 	
Rhea		4X: 	04.43 fps 	
RXL		4X: 	04.31 fps 	
Hyperion HDR		1X: 	32.89 fps 	
4X Slowmo		Apollo: 	38.06 fps 	APFast: 	87.15 fps 	Chronos: 	21.74 fps 	CHFast: 	32.22 fps 	
16X Slowmo		Aion: 	ERR fps 	

```
1 Like

Topaz Video  v1.1.0
System Information
OS: Windows v11.25
CPU: AMD Ryzen 9 5950X 16-Core Processor              63.922 GB
GPU: AMD Radeon RX 7900 XTX  23.984 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	27.69 fps 	2X: 	11.14 fps 	4X: 	02.99 fps 	
Iris		1X: 	36.26 fps 	2X: 	13.36 fps 	4X: 	03.60 fps 	
Proteus		1X: 	39.98 fps 	2X: 	14.77 fps 	4X: 	03.76 fps 	
Gaia		1X: 	17.07 fps 	2X: 	11.34 fps 	4X: 	03.36 fps 	
Nyx		1X: 	13.16 fps 	2X: 	10.23 fps 	
Nyx Fast		1X: 	28.47 fps 	
Nyx XL		1X: 	24.12 fps 	
Rhea		4X: 	03.09 fps 	
RXL		4X: 	03.09 fps 	
Hyperion HDR		1X: 	31.66 fps 	
4X Slowmo		Apollo: 	32.09 fps 	APFast: 	57.35 fps 	Chronos: 	16.99 fps 	CHFast: 	34.57 fps 	
16X Slowmo		Aion: 	33.67 fps 	

Topaz Video  v1.1.0
System Information
OS: Windows v11.24
CPU: AMD Ryzen 9 9900X 12-Core Processor              31.602 GB
GPU: AMD Radeon RX 9070 XT  15.922 GB
Processing Settings
device: 0 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	26.75 fps 	2X: 	13.38 fps 	4X: 	04.00 fps 	
Iris		1X: 	33.53 fps 	2X: 	19.89 fps 	4X: 	04.88 fps 	
Proteus		1X: 	33.24 fps 	2X: 	19.32 fps 	4X: 	05.32 fps 	
Gaia		1X: 	17.57 fps 	2X: 	11.14 fps 	4X: 	05.14 fps 	
Nyx		1X: 	13.00 fps 	2X: 	10.97 fps 	
Nyx Fast		1X: 	27.23 fps 	
Nyx XL		1X: 	40.03 fps 	
Rhea		4X: 	04.17 fps 	
RXL		4X: 	03.95 fps 	
Hyperion HDR		1X: 	29.28 fps 	
4X Slowmo		Apollo: 	39.69 fps 	APFast: 	92.56 fps 	Chronos: 	21.06 fps 	CHFast: 	26.48 fps 	
16X Slowmo		Aion: 	15.4 fps 	

Ignore the Aion number. It only reports an error in the benchmark, but when I ran it at 2X slowmo on a FHD source, it eventually stayed around this number.

1 Like
```
Topaz Video  v1.1.0
System Information
OS: Windows v11.25
CPU: AMD Ryzen 9 5950X 16-Core Processor              63.909 GB
GPU: NVIDIA GeForce RTX 5070 Ti  15.62 GB
Processing Settings
device: -2 vram: 1 instances: 1
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	27.74 fps 	2X: 	12.07 fps 	4X: 	03.04 fps 	
Iris		1X: 	31.66 fps 	2X: 	13.23 fps 	4X: 	03.77 fps 	
Proteus		1X: 	30.07 fps 	2X: 	14.90 fps 	4X: 	03.69 fps 	
Gaia		1X: 	09.12 fps 	2X: 	06.45 fps 	4X: 	03.31 fps 	
Nyx		1X: 	13.14 fps 	2X: 	11.03 fps 	
Nyx Fast		1X: 	32.34 fps 	
Nyx XL		1X: 	22.91 fps 	
Rhea		4X: 	03.31 fps 	
RXL		4X: 	03.18 fps 	
Hyperion HDR		1X: 	12.55 fps 	
4X Slowmo		Apollo: 	24.99 fps 	APFast: 	54.97 fps 	Chronos: 	20.52 fps 	CHFast: 	26.00 fps 	
16X Slowmo		Aion: 	38.93 fps 	

```

Interesting performance differences from my system with the same GPU.

I asked Claude AI to give explanations. Here its answer:

“The processing device setting in Topaz Video determines which hardware component handles the AI processing workload. Let me explain the difference between device “0” and device “-2”:

Device Settings Explained

Device 0 (andymagee-52287’s setting):

  • This typically refers to the primary GPU (Graphics Processing Unit)

  • Device 0 is usually the first discrete graphics card in the system

  • In this case, it’s the RTX 5070 Ti doing all the AI processing

  • GPU processing is generally much faster for AI/ML workloads like video enhancement

Device -2 (macoco’s setting):

  • Negative device numbers in many applications indicate CPU processing or a hybrid mode

  • Device -2 specifically often means “CPU + GPU hybrid” or “automatic device selection”

  • This could involve the CPU (AMD Ryzen 9 5950X) handling part of the workload alongside or instead of the GPU

Why This Matters for Performance

The device setting difference could partially explain the performance gap:

  1. Pure GPU (device 0) is typically faster - AI models like those in Topaz Video are optimized for parallel processing on GPUs with thousands of cores

  2. CPU/Hybrid mode (device -2) may bottleneck - Even though macoco has a 16-core CPU, CPUs have far fewer cores than GPUs and aren’t as efficient for these specific AI tasks

  3. The Hyperion HDR difference is striking - andymagee got 32.89 fps vs macoco’s 12.55 fps (over 2.5x faster). This massive gap strongly suggests that macoco’s system might be falling back to CPU processing for certain tasks, which would be dramatically slower

Possible Reasons for macoco’s -2 Setting

  • Driver or compatibility issue preventing full GPU utilization

  • VRAM management - Perhaps the setting is trying to preserve GPU memory

  • User preference or accidental configuration

  • Topaz Video’s automatic device selection choosing a suboptimal path

The Bottom Line

While andymagee’s newer 9800X3D CPU is superior to macoco’s older 5950X, the device setting difference (0 vs -2) is likely a significant contributor to the performance gap. If macoco switched to device 0 (pure GPU processing), their results would likely improve substantially, though probably still not match andymagee’s due to the CPU architecture differences affecting frame scheduling and data preparation.”

2 Likes

```

Topaz Video v1.1.0

System Information

OS: Mac v26.03

CPU: Apple M2 Ultra 64 GB

GPU: Apple M2 Ultra 51.84 GB

Processing Settings

device: 0 vram: 1 instances: 1

Input Resolution: 1920x1080

Benchmark Results

Artemis 1X: 19.67 fps 2X: 11.08 fps 4X: 03.52 fps

Iris 1X: 08.69 fps 2X: 06.00 fps 4X: 02.56 fps

Proteus 1X: 16.32 fps 2X: 11.63 fps 4X: 03.46 fps

Gaia 1X: 06.56 fps 2X: 04.50 fps 4X: 03.23 fps

Nyx 1X: 08.48 fps 2X: 06.81 fps

Nyx Fast 1X: 21.22 fps

Nyx XL 1X: 43.60 fps

Rhea 4X: 01.76 fps

RXL 4X: 01.90 fps

Hyperion HDR 1X: 62.87 fps

4X Slowmo Apollo: 24.12 fps APFast: 72.08 fps Chronos: 06.16 fps CHFast: 08.29 fps

16X Slowmo Aion: 22.60 fps

```

Awesome! Also I feel that the Mac’s are now working much smoother than before. Today I did benchmarks on both of my Macs. Older Mac Studio with M1 Ultra and Newer Mac Studio with M4 Max. Here are the results

1 Like

1 Like

```

Topaz Video v1.1.0

System Information

OS: Mac v26.02

CPU: Apple M2 Max 32 GB

GPU: Apple M2 Max 24.96 GB

Processing Settings

device: -2 vram: 1 instances: 1

Input Resolution: 1920x1080

Benchmark Results

Artemis 1X: 10.64 fps 2X: 07.07 fps 4X: 01.75 fps

Iris 1X: 11.93 fps 2X: 04.37 fps 4X: 00.95 fps

Proteus 1X: 12.06 fps 2X: 08.03 fps 4X: 02.37 fps

Gaia 1X: 03.39 fps 2X: 02.36 fps 4X: 01.77 fps

Nyx 1X: 03.76 fps 2X: 03.43 fps

Nyx Fast 1X: 09.67 fps

Nyx XL 1X: 46.04 fps

Rhea 4X: 00.94 fps

RXL 4X: 00.97 fps

Hyperion HDR 1X: 47.93 fps

4X Slowmo Apollo: 13.52 fps APFast: 44.91 fps Chronos: 03.79 fps CHFast: 06.36 fps

16X Slowmo Aion: 14.03 fps

```

Currently in the process of installing Topaz Video on my new RTX 5090/9950X3D Windows 11 24H2 build. Will post those results as soon as I have them available. But in the meantime, here are my MacBook Pro results.

2 Likes

```

Topaz Video v1.1.0

System Information

OS: Windows v11.24

CPU: AMD Ryzen 9 9950X3D 16-Core Processor 93.641 GB

GPU: NVIDIA GeForce RTX 5090 31.433 GB

Processing Settings

device: 0 vram: 1 instances: 1

Input Resolution: 1920x1080

Benchmark Results

Artemis 1X: 39.82 fps 2X: 16.75 fps 4X: 04.42 fps

Iris 1X: 52.92 fps 2X: 19.63 fps 4X: 05.14 fps

Proteus 1X: 52.56 fps 2X: 19.23 fps 4X: 05.32 fps

Gaia 1X: 18.78 fps 2X: 12.52 fps 4X: 04.84 fps

Nyx 1X: 26.52 fps 2X: 21.48 fps

Nyx Fast 1X: 48.87 fps

Nyx XL 1X: 31.88 fps

Rhea 4X: 04.42 fps

RXL 4X: 04.72 fps

Hyperion HDR 1X: 31.38 fps

4X Slowmo Apollo: 43.10 fps APFast: 80.68 fps Chronos: 37.64 fps CHFast: 39.57 fps

16X Slowmo Aion: 37.74 fps

```

New Windows PC built. Specs are 9950X3D + ASUS ROG Astral 5090 OC + 96GB DDR5 6000MT RAM.

Also - for those that are curious, Starlight Mini runs at 1.6-1.7 FPS on this setup (that’s running at 1280x720 resolution). At 1920x1080 resolution I’m getting between 0.9-1FPS.

Since I have founder status, I get access to Starlight Mini locally (non-pro license).

This is of my Mac Studio M4 Max - 40 Core GPU with 2Tb ssd and 128Gb Ram, I am going to post soon the results of my Mac Studio M1 Ultra - 64 Core GPU with 8Tb ssd and 128 Ram in a few minutes

```

Topaz Video v1.1.0

System Information

OS: Mac v26.01

CPU: Apple M4 Max 64 GB

GPU: Apple M4 Max 51.84 GB

Processing Settings

device: 0 vram: 1 instances: 1

Input Resolution: 1920x1080

Benchmark Results

Artemis 1X: 13.16 fps 2X: 08.74 fps 4X: 03.09 fps

Iris 1X: 12.21 fps 2X: 05.27 fps 4X: 01.88 fps

Proteus 1X: 13.85 fps 2X: 09.54 fps 4X: 02.77 fps

Gaia 1X: 04.10 fps 2X: 02.98 fps 4X: 02.33 fps

Nyx 1X: 05.26 fps 2X: 05.31 fps

Nyx Fast 1X: 14.64 fps

Nyx XL 1X: 56.76 fps

Rhea 4X: 01.17 fps

RXL 4X: 01.30 fps

Hyperion HDR 1X: 66.68 fps

4X Slowmo Apollo: 16.04 fps APFast: 37.41 fps Chronos: 05.65 fps CHFast: 09.23 fps

16X Slowmo Aion: 17.65 fps

```

I don’t understand what the benchmark values for NYX XL 1920x1080 1x mean.
They are ALWAYS higher than the usual 1x NYX.
It makes no sense!

My Mac Studio with M1 Ultra chip with 64 GPU cores and 128Gb RAM Memory almost beats my new Mac Studio M4 Max with 40 GPU cores and 128Gb Ram Memory in Benchmark Tests, I tried the test again just to confirm and sililar results. Guess what I am gonna keep my old Mac Studio until Apple launches the next gen of Mac Studio with M5 Max or M4 Ultras!

1 Like

```

Topaz Video v1.1.0

System Information

OS: Windows v11.24

CPU: AMD Ryzen 9 9950X3D 16-Core Processor 95.641 GB

GPU: NVIDIA GeForce RTX 5090 31.433 GB

Processing Settings

device: 0 vram: 1 instances: 1

Input Resolution: 1920x1080

Benchmark Results

Artemis 1X: 45.68 fps 2X: 16.70 fps 4X: 04.95 fps

Iris 1X: 55.30 fps 2X: 20.03 fps 4X: 05.19 fps

Proteus 1X: 52.27 fps 2X: 20.39 fps 4X: 05.25 fps

Gaia 1X: 18.88 fps 2X: 12.80 fps 4X: 04.88 fps

Nyx 1X: 26.84 fps 2X: 21.44 fps

Nyx Fast 1X: 46.33 fps

Nyx XL 1X: 30.39 fps

Rhea 4X: 04.85 fps

RXL 4X: 04.72 fps

Hyperion HDR 1X: 29.87 fps

4X Slowmo Apollo: 39.52 fps APFast: 80.07 fps Chronos: 36.34 fps CHFast: 37.21 fps

16X Slowmo Aion: 38.82 fps

```

Decided to test once more with my new build (9950X3D + RTX 5090) after making some BIOS/OS tweaks. Was able to improve my results on the same version.

Starlight Mini Results = 720p @ 1.6-1.8FPS, 1080p @ 0.6-0.8FPS (this isn’t from a benchmark, just my real world results).