These videos are so very helpful, especially the discussions that draw from your work. Thanks -- and looking forward to more in the coming year. Best wishes for 2025.
The m3’s graphics card is integrated into the cpu. It’s not an actual gpu. Only an integrated gpu which is not comparable to an actual gpu. Amd is a very close competitor with NVIDIA. I think that amd is going to push past NVIDIA soon, because NVIDIA is adopting the intel model of just shoving as much power as you can through it. That being said NVIDIA have a new line of gpus rumored for 2026 so maybe they’ll shock us.
Did you enable CUDA support in PyTorch because that NVIDIA run looks slow. My old cheap refurbed NVIDIA Quadro running Topaz just beat a brand new MacBook Pro by minutes upscaling the same images because they support CUDA fine tuning.
The gpu on the Mac is fully integrated with the cpu. You’re testing an apu’s integrated graphics against a top of the line gpu. Laptops are great at what they do, but they aren’t made for what you’re testing.
But you have said 0 about what type of M3 you have, not even the number of CPU or GPU cores… And yes, you’re comparing a laptop with a desktop, why don’t you use RTX 3070 Mobile which is NVIDIA GPU for laptops?
M3 has weak GPU and bandwidth. It is like mobile version of 3050. You should compare M4 Pro to 3070. Plus, 3070 eats way more energy, and form factor is not the same.
These videos are so very helpful, especially the discussions that draw from your work. Thanks -- and looking forward to more in the coming year. Best wishes for 2025.
That's such a mind blowing difference! 🤯🤯🤯🤯
Nice video. For those Apple Fan bois, just be aware Nvidia GPUs can be clustered very efficiently, remember that.
Well well, my Macbook pro M1 Max gives me 16 seconds while burning a few watts compared to the System76 beast.
nice work bro i am learning nn too and wanted to buy a machine u cleared the doubt which one is faster so thankyou
I seriously appreciate this type of content, really wonder if AMD GPUs can at least touch that level of performance
The m3’s graphics card is integrated into the cpu. It’s not an actual gpu. Only an integrated gpu which is not comparable to an actual gpu. Amd is a very close competitor with NVIDIA. I think that amd is going to push past NVIDIA soon, because NVIDIA is adopting the intel model of just shoving as much power as you can through it. That being said NVIDIA have a new line of gpus rumored for 2026 so maybe they’ll shock us.
that is an insane difference! question though, why didn't you opt for a RTX 40 series GPU...pretty sure that would have been even faster :p
Did you enable CUDA support in PyTorch because that NVIDIA run looks slow. My old cheap refurbed NVIDIA Quadro running Topaz just beat a brand new MacBook Pro by minutes upscaling the same images because they support CUDA fine tuning.
I'd love to see it on a M4 Max
The gpu on the Mac is fully integrated with the cpu. You’re testing an apu’s integrated graphics against a top of the line gpu. Laptops are great at what they do, but they aren’t made for what you’re testing.
But you have said 0 about what type of M3 you have, not even the number of CPU or GPU cores…
And yes, you’re comparing a laptop with a desktop, why don’t you use RTX 3070 Mobile which is NVIDIA GPU for laptops?
Because this isn't an Apple bubble marketing ad, but an eyeopener what the right tool for the job can do.
my gpu rtx 3070 ti laptop got 11.02 sec
M3 has weak GPU and bandwidth. It is like mobile version of 3050. You should compare M4 Pro to 3070. Plus, 3070 eats way more energy, and form factor is not the same.
I think its unfair to compare a pc with a laptop. They really shrunk that GPU to fit
You can use any windows laptop with mobile gpu and it will be a massacre
Yea it’s not even an actual gpu. It’s an Integrated gpu.