Does the RTX 3090 support machine learning (ML) acceleration?

[memory02]

hello ~

I bought an RTX4090 a few weeks ago and tested it, but there was no ML acceleration in Flame 2023.3 (other accelerations were fast enough)
I have a lot of skin correction work, so I definitely need ML acceleration function
RTX4090 seems to be not supported yet because it is Ada architecture
Does the RTX 3090 (Ampere) support ML acceleration in Flame 2022.3 or higher?

This is a non-English speaking area
There are some clumsy parts because it was written using a translator.

thank you

[PlaceYourBetts]

My go to guy for hardware discussions is always @friendlyape in fact he should have his own patron page where we help fund his continuing study of Linux Flame Hardware.

Flame GPU Benchmarks

Hey team, I ran a basic Flame GPU benchmark on a bunch of cards I’ve got in my office today. Idea was yoinked from randy in the M1 thread! My methodology was incredibly simple as I wanted to 100% use the GPU exclusively. I fed colour noise @ 4K and @ 8K into a motion analysis node, which then fed into a render node (rendering both forward and backward vectors). I timed how long it took to render 200 frames @ 4K and 100 frames @ 8K. The cards in question are: A5000, 3090, 3080 and 3060. These …

I think I remember @MLandon having a 3090 unless I am mistaken

[randy]

While the 30xx and 40xx may work with MOST of Flame, technically they aren’t supported.

Flame Family 2023 release system requirements | Flame Products 2023 |...

I believe @Sinan had a 3090 in his Flame and reported similar issues with Machine Learning tools not working which, isn’t surprising.

[memory02]

Thank you for your comment
When using RTX 4090, ML works in version 2023.3
But it doesn’t seem to have any acceleration.
For example on the Mac 6900XT the ML accelerates very smoothly when you drag the time bar
But under Linux (Rocky 8.5) RTX4090 runs 5x slower than 6900XT (very stuttering)
So I wonder if the RTX 3090 products are also this slow?

[Sinan]

I have an A6000. My gripe was it wasn’t supported before 2023, but now it is. Can’t elaborate on 30XX or 40XX series…

[paul_round]

What about the RTX 6000?

[randy]

This is listed as a supported card in the Flame System Requirements docs.

[MLandon]

Hey there,
I got a great deal on a 3090ti, which is essentially the same card as an A6000 but with less VRAM. I too wasn’t able to use the ML tools until 2023. They seem to work fine now. I’d love to check out the performance of a 4090, but I can’t justify the expense!

[memory02]

hello
RTX 4090 tested by me
Unfortunately, the ML acceleration function is slow. It works normally.
5x slower compared to Hackintosh 6900XT and ML acceleration performance only
I don’t know if this slow is normal because I haven’t compared it to other products.