593
this post was submitted on 18 Apr 2024
593 points (95.7% liked)
Technology
59378 readers
2612 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Your 64 gigs of ram probably uses 10x the power and takes up significantly more space than the single memory chip that's on the M1-M3s die. And yet it still has less bandwidth than the M1, and on top of that the M1 utilizes it more efficiently than a "normal" desktop or laptop can since there's one pool of memory for RAM RAM and VRAM.
https://en.wikipedia.org/wiki/Apple_M1#:~:text=While%20the%20M1%20SoC%20has%2066.67GB/s%20memory%20bandwidth
Chat GPT guestimates 57GB/s for dual channel DDR4 at 3600mhz
$1000 for 8 gigs of RAM in the Air is whatever. $1200 for 8 gigs of ram in the Pro was not great. But 1600 for 8 gigs of ram in the new M3 MBP is really awful.
Memory bandwidth is useless if you run out of memory and need to swap.
GPU not having it's own pool of memory is really going to help to.
Pigs fly in apple land.
That's not how it works, unfortunately.
A UMA (unified memory architecture) enables zero-copy texture uploads and frame buffer access, but that's not likely to constitute notable memory savings outside games or GPU-accelerated photo editing. Most of the memory is going to be consumed by applications running on the CPU anyway, and that's not something that can be improved by sharing memory between the CPU and GPU.
It's by necessity that the M1 has higher memory bandwidth. UMA comes with the drawback of the GPU and CPU having to share that memory, and there's only so much bandwidth to go around. GPU cores are bandwidth hungry, which is mitigated by either using a pile of L2 cache or by giving the system better memory bandwidth.