Topaz Video 1.0.4 does not use my RTX 5070 TI GPU

Topaz Video Bugs and Issues
1

Hi, how are you? I wanted to ask for your help, please.
I’m using Topaz Video 1.0.4. I had a GTX 1660 Ti and it worked fine. I upgraded to an RTX 5070 Ti and it stopped working. The program recognizes it, no problem with that, but it doesn’t use it. I did a clean driver installation (I tried several, the four latest ones), I did a clean reinstall of Topaz Video, and there’s no way to get it to use the GPU. The CPU usage goes above 80% and the GPU usage is at 3%. With the 1660, I didn’t have to do any configuration; I just installed Topaz Video and that was it. Do I need to do any extra configuration with the 5070? By the way, if I don’t select any model and do an export, then the GPU usage goes to 70% and the CPU drops to 66%.

Thank you so much for your help!

(Sorry for my bad English, I’m using a translator)

2

I would first check that the GPU is selected in the preferences (don’t leave it on “Auto”).

Next, if you’re currently using Task Manager to monitor the GPU, swap to a third party utility like GPU-Z.

1 Like
3

Thank you so much for your interest in my case!
I tried all the options, just in case (auto, single, and multi).
With GPU-Z, using Proteus, I get an average usage of 12%. Using Iris, I get an average GPU usage of 12% and CPU usage of 88%. Using Starlight, I get approximately 100%, but for a 720x480 MPEG video, 33 minutes long, it says it takes more than a day (which doesn’t seem reasonable to me).
I’m seeing strange behavior, much worse than what I had with a GTX 1660. Thanks for your help!

4

Can you pull the logs from the app to send to the support team for review along with the system profile report? help@topazlabs.com

How To Obtain Your System Profile (Windows)

To gather logs, please select Help > Logging > Get Logs for Support and attach the zip file to your reply.

1 Like
5

If you can get a 33-minute video processed by Starlight done that quickly, you should be jumping with joy! :hushed_face:

3 Likes
6

Hi! I use a translator for both reading and writing, so I’m not sure if I understood correctly or if I explained myself well. Using Starlight, it says the processing takes more than a day; the system is going crazy, and I don’t know how many more hours it will take, in addition to the day. Thanks!

7

Hi, thank you so much for your interest in my case!
I’ve been running some new tests. I used the Iris model, without scaling, and it doesn’t matter whether I use CPU or GPU; the number of frames (approximately 19) it processes is the same. If I scale it by 2x, it runs at 10.5fps on CPU and 9.9fps on GPU.
Data below:Hi, thank you so much for your interest in my case!
I’ve been running some new tests. I used the Iris model, without scaling, and it doesn’t matter whether I use CPU or GPU; the number of frames (approximately 19) it processes is the same. If I scale it by 2x, it runs at 10.5fps on CPU and 9.9fps on GPU.
Data below:

System Information:

  Time of this report: 12/10/2025, 13:14:28
         Machine name: JMO
           Machine Id: {035F9E7B-8B34-4081-A1AE-C5BD38F90488}
     Operating System: Windows 10 Pro 64-bit (10.0, Build 19045) (19041.vb_release.191206-1406)
             Language: Spanish (Regional Setting: Spanish)
  System Manufacturer: Gigabyte Technology Co., Ltd.
         System Model: Z87X-UD3H
                 BIOS: BIOS Date: 01/17/14 10:55:00 Ver: 04.06.05 (type: BIOS)
            Processor: Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz (8 CPUs), ~3.5GHz
               Memory: 32768MB RAM
  Available OS Memory: 32718MB RAM
            Page File: 5656MB used, 68020MB available
          Windows Dir: C:\Windows
      DirectX Version: DirectX 12
  DX Setup Parameters: Not found
     User DPI Setting: 96 DPI (100 percent)
   System DPI Setting: 96 DPI (100 percent)
      DWM DPI Scaling: Disabled
             Miracast: Available, no HDCP

Microsoft Graphics Hybrid: Not Supported
DirectX Database Version: 1.0.8
DxDiag Version: 10.00.19041.5794 64bit Unicode

LOG Topaz Video:

2025-12-10 12:54:06 11948 INFO: Windows DisplayVersion: 22H2
2025-12-10 12:54:06 11948 INFO: WINDOWS VERSION IS 10.22
2025-12-10 12:54:06 11948 INFO: VNNI: 0 AVX: 1 CPUName: Intel(R) Core™ i7-4770K CPU @ 3.50GHz Thread Count: 8
2025-12-10 12:54:06 11948 INFO: MEM TOTAL: 31.9503 AVAIL: 28.6305
2025-12-10 12:54:06 11948 INFO: Machine Id: 834d2068-9afc-42c8-af29-7e6798c50ee3
2025-12-10 12:54:06 11948 INFO: Available GPU LIST
2025-12-10 12:54:07 11948 INFO: 0 Memory Info 80714137600015874392064
2025-12-10 12:54:07 11948 INFO: 0 Device Id: 0 Name: NVIDIA GeForce RTX 5070 Ti VRAM: 15.5342 GB DataType: 2 Serial: 000000000000a401 Visible: 1 Index: 0
2025-12-10 12:54:07 11948 INFO: 1 Memory Info 83081615360016347887616
2025-12-10 12:54:07 11948 INFO: 1 Device Id: 1 Name: Microsoft Basic Render Driver VRAM: 0 GB DataType: 2 Serial: 000000000000cba0 Visible: 0 Index: 0
2025-12-10 12:54:07 11948 INFO: === System Information ===
2025-12-10 12:54:07 11948 INFO: OS Windows Version 10.22
2025-12-10 12:54:07 11948 INFO: CPU Intel(R) Core™ i7-4770K CPU @ 3.50GHz Threads 8 AVX 1 AVX2 1 VNNI 0
2025-12-10 12:54:07 11948 INFO: Is Apple Processor 0
2025-12-10 12:54:07 11948 INFO: RAM 31.9503 GB Total / 3.31985 Free/Used
2025-12-10 12:54:07 11948 INFO: Machine ID: 834d2068-9afc-42c8-af29-7e6798c50ee3
2025-12-10 12:54:07 11948 INFO: Device count: 1
2025-12-10 12:54:07 11948 INFO: - Index 0 Name NVIDIA GeForce RTX 5070 Ti Cores 0
2025-12-10 12:54:07 11948 INFO: VRAM 15.5342 GB Total / 0 GB Used DataType 2
2025-12-10 12:54:07 11948 INFO: Serial 000000000000a401 ComputeLevel: 1200 CUDA ID: 0 Visible: 1 Legacy: 0
2025-12-10 12:54:07 11948 INFO: ==========================2025-12-10 12-54-07.303 Thread: 11948 Warning qrc:/models/TOutputModel.qml:62: TypeError: Cannot call method ‘map’ of undefined

2025-12-10 12:54:52 11612 WARNING: ModelManager: Ignoring model: C:\ProgramData\Topaz Labs LLC\Topaz Video\models\thm-1.json

2025-12-10 12:54:52 11612 WARNING: Checking file “C:\ProgramData\Topaz Labs LLC\Topaz Video\models\video-encoders.json” failed with error Unknown exception

2025-12-10 12:54:52 11612 INFO: Loaded models info map

2025-12-10 12:54:52 11612 INFO: Model found in list iris-3

2025-12-10 12:54:52 11612 INFO: Windows DisplayVersion: 22H2

2025-12-10 12:54:52 11612 INFO: WINDOWS VERSION IS 10.22

2025-12-10 12:54:52 11612 INFO: VNNI: 0 AVX: 1 CPUName: Intel(R) Core™ i7-4770K CPU @ 3.50GHz Thread Count: 8

2025-12-10 12:54:52 11612 INFO: MEM TOTAL: 31.9503 AVAIL: 27.6751

2025-12-10 12:54:52 11612 INFO: Machine Id: 834d2068-9afc-42c8-af29-7e6798c50ee3

2025-12-10 12:54:52 11612 INFO: Available GPU LIST

2025-12-10 12:54:53 11612 INFO: 0 Memory Info 80714137600015874392064

2025-12-10 12:54:53 11612 INFO: 0 Device Id: 0 Name: NVIDIA GeForce RTX 5070 Ti VRAM: 15.5342 GB DataType: 2 Serial: 000000000000a401 Visible: 1 Index: 0

2025-12-10 12:54:53 11612 INFO: 1 Memory Info 83081615360016347887616

2025-12-10 12:54:53 11612 INFO: 1 Device Id: 1 Name: Microsoft Basic Render Driver VRAM: 0 GB DataType: 2 Serial: 000000000000cba0 Visible: 0 Index: 0

2025-12-10 12:54:53 11612 INFO: === System Information ===

2025-12-10 12:54:53 11612 INFO: OS Windows Version 10.22

2025-12-10 12:54:53 11612 INFO: CPU Intel(R) Core™ i7-4770K CPU @ 3.50GHz Threads 8 AVX 1 AVX2 1 VNNI 0

2025-12-10 12:54:53 11612 INFO: Is Apple Processor 0

2025-12-10 12:54:53 11612 INFO: RAM 31.9503 GB Total / 4.27523 Free/Used

2025-12-10 12:54:53 11612 INFO: Machine ID: 834d2068-9afc-42c8-af29-7e6798c50ee3

2025-12-10 12:54:53 11612 INFO: Device count: 1

2025-12-10 12:54:53 11612 INFO: - Index 0 Name NVIDIA GeForce RTX 5070 Ti Cores 0

2025-12-10 12:54:53 11612 INFO: VRAM 15.5342 GB Total / 0 GB Used DataType 2

2025-12-10 12:54:53 11612 INFO: Serial 000000000000a401 ComputeLevel: 1200 CUDA ID: 0 Visible: 1 Legacy: 0

2025-12-10 12:54:58 11612 INFO: TOTAL INSTANCES: 1

2025-12-10 12:54:58 11612 INFO: Init backend runner 1
2025-12-10 12-54-58.604 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:58 11612 INFO: Preflight frame index 2 4
2025-12-10 12-54-58.608 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:58 11612 INFO: OVM INITING
2025-12-10 12-54-58.614 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:58 11612 INFO: OV device selection: CPU
2025-12-10 12-54-58.652 Thread: 11948 Info FF Process Output: 4 Intel(R) Core™ i7-4770K CPU @ 3.50GHz

2025-12-10 12:54:58 11612 INFO: OpenVino device string is CPU index -1
2025-12-10 12-54-58.835 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:58 11612 INFO: - Loaded network successfully for deviceid CPU index -1

2025-12-10 12:54:58 11612 INFO: ExeNetwork optimal requests: 1

2025-12-10 12:54:58 11612 INFO: OVM FINISHED INITING
2025-12-10 12-54-58.864 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:58 11612 INFO: Preflight frame index 3 4
2025-12-10 12-54-58.892 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:58 11612 INFO: Preflight frame index 2 4
2025-12-10 12-54-58.918 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:58 11612 INFO: Preflight frame index 1 4
2025-12-10 12-54-59.042 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:59 11612 INFO: —StandardTilingStrategy::StandardTilingStrategy W: 384 H: 480 C: 2 R: 1 X: 324 Y: 0

2025-12-10 12:54:59 11612 INFO: —StandardTilingStrategy::StandardTilingStrategy W: 384 H: 480 C: 2 R: 1 X: 324 Y: 0
2025-12-10 12-54-59.248 Thread: 11948 Info FF Process Output: 4

2025-12-10 12:54:59 10500 INFO: License heartbeat successful

2025-12-10 12:54:59 10500 CRITICAL: Heartbeat failed frame= 1 fps= 0

2025-12-10 12-54-59.327 Thread: 11948 Info FF Process Output: 4 frame= 5 fps= 50.5261

Video Info:
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : Main@L3
Format settings : CABAC / 4 Ref Frames
Format settings, CABAC : Yes
Format settings, Reference : 4 frames
Codec ID : V_MPEG4/ISO/AVC
Duration : 25 min 14 s
Bit rate : 1 200 kb/s
Width : 708 pixels
Height : 480 pixels
Display aspect ratio : 4:3
Frame rate mode : Constant
Frame rate : 23.976 (24000/1001) FPS
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 0.147
Stream size : 217 MiB (77%)
Writing library : x264 core 120
Encoding settings : cabac=1 / ref=1 / deblock=1:0:0 / analyse=0x1:0x111 / me=hex / subme=2 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=0 / me_range=16 / chroma_me=1 / trellis=0 / 8x8dct=0 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=0 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=3 / b_pyramid=2 / b_adapt=1 / b_bias=0 / direct=1 / weightb=1 / open_gop=0 / weightp=1 / keyint=240 / keyint_min=24 / scenecut=40 / intra_refresh=0 / rc_lookahead=10 / rc=2pass / mbtree=1 / bitrate=1200 / ratetol=1.0 / qcomp=0.60 / qpmin=3 / qpmax=69 / qpstep=4 / cplxblur=20.0 / qblur=0.5 / ip_ratio=1.40 / aq=1:1.00
Default : Yes
Forced : No
Color range : Limited
Color primaries : BT.601 NTSC
Transfer characteristics : BT.709
Matrix coefficients : BT.601

8

The translation seems to have worked well - I understood what was written!

Starlight is a very slow model, even on higher-end GPUs than the 5070Ti. Depending upon the resolution of your source video and the upscale factor you choose, it can take much longer than the projected time that you’ve cited.

For example, I have seen a 5-minute video take nearly 3 full days to complete, doing a 3X upscale of a 720p source.

The estimated time to complete seems to be fairly accurate, as long as you’re not using your system for other purposes while it’s rendering.

What’s not clear is what you mean by “the system is going crazy”. Could it be that you’re actually using the Starlight Sharp variant, rather than Starlight? With Starlight Sharp, the fps display and the estimated time to completion is less stable and changes several times while it’s running.

1 Like
9

Okay, great that the translation works!

I used the Starlight (mini) with a 720x480 video. With so many tests, I can’t remember if I used 2x scaling (most likely) or left it unscaled. I also don’t remember how many hours it displayed, besides the whole day.
Regarding the counter going crazy, I meant that it kept increasing. It kept adding hours; it never finished calculating. I immediately stopped the process, since what I wanted to check wasn’t how long it took, but the GPU usage.

Thank you so much for your interest!

10

You may want to try again with a much shorter test clip, and leave the output resolution at the default minimum. This should give you a better estimate of the time it will take, and it should switch from the estimate to run and remaining time before too long. (It can take a few minutes before it’s obvious that it’s running. Try to wait long enough so that you can feel confident that it’s running properly.) It will eventually settle down and list the ‘Time remaining’ in your display.

You can gain confidence with short, simple tests and work your way up, knowing what to expect.

1 Like
11

What is your NVIDIA driver version?

1 Like
12

Thank you so much for your interest in my case!

I tried the last 4 or 5 drivers (studio).

13

I think the old CPU is the case, but I might be wrong.

1 Like
14

Thank you so much for your interest in my case!

I thought the same thing, but the strange thing is that the GPU doesn’t even turn on its fans; it’s like it’s barely working.

15

Thanks again!
The thing is, I’m not worried about the timing; I understand it can be long. What concerns me is the almost nonexistent GPU involvement in the process.

16

Can you run OCCT or GPU-Z and take a screenshot of a GPU load under TVAI processing?

1 Like
17

Thank you once again for your interest and help!

Here are two screenshots. The first shows the Iris model selected, with 2x scaling. The second is a screenshot of the performance without selecting any model. The video resolution is 640x480, so it’s not a very large file.

sin modelo
sin modelo1920×1080 191 KB

modelo iris escalado x2
modelo iris escalado x21920×1080 252 KB

18

Your card is utilised properly.

Iris is a lightweight model, compared to Starlight, it uses a bit of VRAM.

1 Like
19

Once again, thank you so much for following up on my case!
I’m experiencing the same issue with Proteus. The GPU usage seems very low to me. With a GTX 1660 Ti I had before, the usage was higher. I see other cases that utilize the GPU more effectively. It seems strange to me; perhaps I’m mistaken and it’s just my impression.

20

Ah, that’s what the case: your 1660Ti was so outdated that any modern AI models were using it improperly.

A had a bunch of old Nvidia Quadros (even Kepler and Maxwell) at my job that performed the same way: maximum load, minumum FPS.

All Pascal, Turing cards are not suitable for modern AI software with models trained upon Nvidia A100 GPU or similar Blackwell GPUs.

So, having Blackwell GPU you’ve unlocked true potential of Topaz, congratulations!

Now you have to upgrade your CPU for it to feed data-hungry 5070Ti.

For example, my work setup us Core i7-10700, my home PC is Ryzen 7 5700x. Both CPUs are bit outdated but fine.

1 Like