Skip to main content
Participant
July 17, 2026
Question

Cannot decide which GPU to take 5060 ti 16gb or 5070, the price diffrence is not that much about 50-60 eur, and I would take 5070, but what's holding me up is 16gb.

  • July 17, 2026
  • 1 reply
  • 60 views

What I do in LIghtroom: I have small photo catalogues, usually I work with clients via dng files, so my catalogues are very small. My main use is masks, a lot of masks when I edit, also denoise. Do I really need that full support of preview generation or better take faster 5070?
I have 265K + 32GB RAM

    1 reply

    D Fosse
    Community Expert
    Community Expert
    July 17, 2026

    A few functions in Lightroom Classic requires 16 GB VRAM, like “use GPU for preview generation”. ACR/LrC masking is generally very memory-intensive, so that’s probably the overriding concern.

     

    I don’t do very much masking in Lightroom, I prefer the added control - and, admittedly, extra time - of manual masking in Photoshop, but that said, I’m very happy with the RTX 5060 Ti 16 GB. It performs splendidly and everything is snappy and responsive. Denoise is very fast, a 60 MP file is done in about 20 seconds, which is considerably faster than the 3060 I had before (not to mention older GTX models which would take several minutes).

    Participant
    July 17, 2026

    as far as I understood this stuff, masking uses NVIDIA’s tensor cores? I might be wrong here, but if so - then 5070 should perform faster in this situations. Although - I do zoom in and out a lot during edits, maybe here 16gbs + preview generation helps. I wish Adobe was more clear on this. 

    PS: as memory crisis goes on, I really think a lot of users just dont have 16gbs and pushing for this is ludicrous in my opinion, maybe optimize your app better?! Just ranting….

    D Fosse
    Community Expert
    Community Expert
    July 17, 2026

    I should think the 5060 has plenty of tensor cores. But judging by what I read here, masking eats memory like crazy, which quickly becomes a problem for those with integrated GPUs that use system memory.

     

    As for the memory shortage, don't worry, the AI bubble is not sustainable and will inevitably pop pretty soon, and then you'll have it thrown at you 🙂