Hey everyone,
I’m curious if anyone here is running Topaz Photo AI (or even Gigapixel) on one of these newer AI PCs for big workloads. I shoot a lot of high‑megapixel RAW files, and my current machine chokes whenever I try to batch process hundreds of images.
A few questions for those who’ve made the switch:
- Do you notice a significant speed boost compared to a regular desktop or gaming PC?
- Does the AI hardware actually help with quality or just processing speed?
- Any stability issues when running large batches overnight?
I’m trying to figure out if it’s worth building/buying an AI PC now or just waiting for the next generation of GPUs. Would love to hear your experiences, especially if you’re doing professional photo work.
Thanks in advance — benchmarks, anecdotes, and setup tips welcome!
I’m not sure exactly what constitutes an AI pc, but I have a Ryzen AI 9 HX 370 and the NPU (50 TOPS) just sits there and lets the Nvidia GPU (baby 4060 in my case) do all the work. At the moment the Nvidia 5090 (575 watts, $2k USD) is the best, but they have their issues (right, @Fotomaker?)
As things currently stand you’re better off just getting a good modern pc, and not worrying about the AI marketing rubbish. If your current machine is “choking”, it isn’t because you don’t have a NPU.
CPU, RAM and GPU are still the main things to consider for performance.
Do you mean PC with a AMD Strix Halo (Ryzen AI MAX+ 395 for example)?
PS: It’s not about having a NPU. It’s a system with unified memory between system and graphics so you can run large models needing up to 96GB RAM. It is supposed to be faster than a mobile RTX 4070 GPU.
The example PCs the OP was looking at are in the link in their post.
Ah, NPU. Well, NPU are mostly for some real time AI tasks with low power consumption. Even if the software was designed to use TPU instead of GPU, they are not made for processing large set of data. GPU, like NVIDIA RTX cards, contains tensor core which are made for AI tasks processing large set of data.
Not to say that one can’t be as powerful as the other. But integrated NPU in those PC are like integrated GPU compared to having a dedicated GPU.
Anyway, batch processing individual tasks sequentially doesn’t need more resources than processing one task. If it crash in the middle, it’s because there is a bug in the software (memory leak for example), because the system is overheating or because there is some other kind of instability (For example, the damaged intel 13/14th gen cpu. It may, or not, crash on any given task. So it may happen if you run a single task. But if you run many tasks, it’s more likely that it will happen at one point).
