Topaz Video AI Linux Beta v5.0.3.0.b

Well … the problem is :

  • Topaz decided to go single GPU for all : the worst idea ever.
  • Hunting "pro "user with a Windows (that is known to badly handle more than 16 threads) and Tensor focus is also figuring a lack of developpers. and strategy : No repsectable “pro” use Windows bug nest. They are either MAC or Linux seeing UNIX. As owner of Threadripper you even can’t imagin my frustation seeing my kickass tuff sleeping for tasks
  • ALL CPU are far more morefull under Linux and this scale up to 100% with AMD.
  • ALSO Linuw can’t handle very easily 24 GPU while problems starts at 4 with Windows.

→ We definitely can’t talk about “pro” or any other wordsl ike like with such stuff.
This app. is for geek short use, and not real deep work. Either it will take years due to the limitations they force us, either it crashes or bug. EOS

I’ve mostly been running Artemis on Linux lately for noise cleanup. I get around 23 fps for 4:3 HD and 19 fps for 16:9 HD. I’m using an NVIDIA RTX 4070 card. So the performance is OK.

I only use Linux as well, and so without Linux support, I will simply stop using TopazLabs. I really can’t be bothered to maintain two PCs. That’s too much space, effort and money wasted.

Considering that TVAI runs fairly well on Linux, from command line at least, it is possible to run it on cloud/HPC hardware. I think that would be a decent market in itself, and boost their efforts here, but it seems not.

1 Like

Linux support was the second most voted for feature in the ideas section with 65 votes. Considering only a tiny percentage of users know about this forum, let alone register for it, and can actually navigate to the ideas section and actually place a vote, I’d say that’s a pretty strong demand.

They really should step up with the Linux version, who knows how many lost sales it’s costing them?

@tony.topazlabs

When you know how far Linux leave Windows behind with such program, if I was Topaz I would have jumped on it.

This program needs multiple CPU cores, high mem bandwidth, many GPU, everything that Linux supports greatly where Windows is a gaming OS.

2 Likes

@tony.topazlabs : it’s been quite a while now. Is there an update on the way? It would be helpful to have a closer alignment with the regular releases; the Linux build is several major releases behind it seems.

1 Like

I would like to think a Linux version is built automatically alongside the other versions, in a build pipeline. Obviously I don’t know anything of their internal build process, but typically its all automated. Maybe being low priority, the manual testing of Linux versions just isn’t done on a frequent cadence, and therefore they can’t confidently release?

Anyway, I too would like to see more recent versions. I recently decided to switch over to atomic Fedora from Windows. I’m waiting a bit longer before renewing my Video Enhance subscription.

Also a flatpak instead of or alongside the deb would be helpful.

Hi @gregory.maddra, @tony.topazlabs, @David.123

I get the feeling Topaz team are not developing Linux any more, but just in case we are looking for updates or advice if Linux is dead. The lack of communication and updates for months really gives the impression that Linux isn’t a priority for them. Other places to look, I suppose. It might be time for us to explore alternatives that cater more to the Linux community.

Based on the forum discussion, it seems like Topaz has not been very supportive or collaborative with Linux beta testers recently:

  • The last Linux beta release was version 5.0.3.1.b on May 1, 2023, with no new releases since.
  • Multiple users expressed frustration over the lack of updates and communication:
    • On July 4, a user asked for an update on the next beta.
    • On July 25, another user asked about an updated Linux build.
    • On July 28, someone directly asked Gregory Maddra for release timing.
    • On August 5, a user detailed several bugs encountered.
    • On August 12, disappointment was expressed over the three months of silence.

Testers are now questioning whether it’s worth continuing to use the Linux version given the lack of support. There have been calls for Topaz to bring the Linux version in line with Windows and Mac, with concurrent releases, but communication from Gregory Maddra has been absent since late June.

Comments should probably move back to Facebook or the like, where there might be broader discussion or insights. Development seems slow and more about GUI improvements for Crappentosh and Adobe users. Personally, I use DaVinci on Linux, and it’s been solid.

5.0.3.1.b was released April 30th of this year.

Still a long time ago now though…

1 Like

Thanks David, I had a search for most senior and you had the Leader badge ( which I now know is a Topaz Owner, not employee of Topaz) . But just in case any names of Topaz team that would help here. Maybe you know if Gregory on leave or unwell?

I came to the same conclusion. I installed Windows 10, now use Davinci Resolve and Topaz Video AI and blender on Windows instead of linux. The performance of the system as an editing station is inferior to linux on Resolve and blender, but Topaz performs a bit faster – especially any AI model that has been optimized for TensorFlow.

It is a shame, but there is no sense in swimming against the current. I still use linux for my personal computing, but have to dual boot for my video editing workstation now. Sad.

1 Like

Topaz Linux Support & Product Direction Concerns

Hi Topaz Team and Fellow Forum Members,

It seems pretty clear that Topaz Linux support is all but dead – no updates, no maintainers, and minimal engagement from the team. I’m left wondering why we still have the Linux forum or beta when it feels like it’s been abandoned, especially as more focus seems to have shifted towards the Pro version of TVAI. Now I lost my dream of multigpu unless I fork out. Is this where the resources are being redirected? I am glad the TVAI Pro users ares finding the faults with multigpu now, and not me, I stop testing and asking back in version 4. There was hope in version 3.

Meanwhile, open-source development on platforms like GitHub, with tools like ComfyUI, is moving fast, especially with AI-driven projects. The flexibility to run on AMD, NVIDIA, or even CPUs for both images and videos, with full customisation and upscale options, fills a significant gap that Topaz once catered to. While I still admire Topaz’s achievements in terms of making these tools accessible to the wider market, I can’t help but feel that focus has shifted too heavily towards new GUIs and marketing, rather than keeping up with real innovation.

I also find the Apple-centric focus an interesting one, especially considering the hardware limitations of the platform. I’ve been frustrated with Pro version features like multi-GPU support, which has never worked as advertised in the last couple of years.

While cloud-based apps aren’t really my thing, I have to admit that fibre optic internet has changed the game where I live. However, it’s clear that Linux, Windows, and cloud are still the true workhorses of creativity, and it feels like Topaz has left a gap here. I’m enjoying my time exploring non-Topaz products and hope to see Topaz catch up.

Looking forward to hearing others’ thoughts.

Best,
Billy

1 Like

Don’t worry, my multiGPU Pro version does work very badly with latest releases …
Staying in 5.0.4 was max I can expect. All new models Rhea, RXL and Hyperion, turn CUDA completely nut.

  • GPU badly selected
  • Memory allocation impossible to understand
  • GPU thread are random and chaotic
  • Leads to almost instant failure if multiple files

All of this with random “Error”, “AI error” or “Resolution error”

TVAI is a single high end CPU, single high end GPU app., that’s a lol folks
Add a “Windows” layer on, and you add crashes :slight_smile:

I have been using the software since back in version 2.X.X and it’s been a great experience with Topaz. Until there complete cluster on Linux builds. Providing broken, unusable builds, and having ghosted us for months afterward.

I completely purged Windows when Linux betas were provided. And I no longer have the willpower to re-learn how to make windows usable, or deal with setting up QEMU for a windows VM. This is even more disappointing after the Ryzen 7000 launch making the tech news for the ridiculous performance discrepancies windows has, getting even worse with the newest versions of Win11.

I’d think this would have developers re-consider the issues Windows and [a.] have, but apparently they’re happy with the limitations they have. I will be looking for a new solution, there’s an open source software I’ve been meaning to look into that supports Linux. Other software may work better with crossover as well.

a. I don’t recall what Apple calls there OS now, OSX? iOS? High something? Whatever.

2 Likes

I actually switched to Linux from Windows in part because long running upscale tasks would be completely ruined by Windows Update deciding that it really had to, absolutely must, right then and there, without let or hindrance and disregarding all CPU and GPU activity…reboot itself 3 days in to a multi-day task.
After the 3rd time this happened, I threw my hands up and asked what use it was putting up with that garbage, and spent a while exploring Linux options. That Topaz had a Linux version was really quite helpful in making that jump and I’d really hoped they’d keep it up. It’s a crying shame they seem to have silently decided to abandon the effort, and I’m coming up to renewal and will simply let it drop, it seems. It’s not as though it works under WINE (except on CPU which is so slow as to be worthless).

3 Likes

I still use linux for my daily computing, for all the reasons I chose it in the first place.

Yes WINE is great for CPU only tasks, not useful for Topaz.

I am not a gamer, but Steam has successfully shoehorned several popular GPU intensive games to work with a modified WINE (called Proton). Each game requires its own adjustments and hacks to get it working well in Steam’s modified WINE (Proton.) Getting this to work for TopazVideoAI with CUDA and Tensor GPU instructions would be challenging. Way more challenging than the alternatives.

If you are super motivated, and you have BOTH an integrated GPU on your motherboard and a discrete GPU you can use QEMU to install a windows virtual machine inside of linux that will use the integrated GPU for the VM and then it can pass-through your NVIDIA GPU (or whatever it is) to the Windows VM. Then you can run Windows Topaz in that virtual machine with good speeds – close to a bare metal installation.

However, it is a really huge pain, and for me I just found it to be less of a waste of my time to just install windows on a separate SSD and dual boot.

Part of becoming middle-aged, I guess. Now i just grumble under my breath and do the most pragmatic thing to use the tool I need to use to get a job done.

1 Like

Any plans for a RHEL release? We are running Rocky Linux 8 over here.

2 Likes

I pinged support recently and they indicated that work continues on Linux, but only as resource permits (which seems to be the key issue). They acknowledge the release lag and suggested waiting for now until visible activity picks up, before spending any more money.

So that’s really it - no real commitment. Situation acknowledged. Carry on with your day.

3 Likes

1). Is api a possible solution?
2). What’s the best practice

I don’t know if TVAI’s translation layer can run normally on LINUX, but I have tried using ESRGAN’s real-life and anime models to enhance anime and real-life videos on the UBANTU server.

1 Like