this post was submitted on 27 Mar 2026
116 points (99.2% liked)
Apple
20371 readers
78 users here now
Welcome
to the largest Apple community on Lemmy. This is the place where we talk about everything Apple, from iOS to the exciting upcoming Apple Vision Pro. Feel free to join the discussion!
Rules:
- No NSFW Content
- No Hate Speech or Personal Attacks
- No Ads / Spamming
Self promotion is only allowed in the pinned monthly thread
Communities of Interest:
Apple Hardware
Apple TV
Apple Watch
iPad
iPhone
Mac
Vintage Apple
Apple Software
iOS
iPadOS
macOS
tvOS
watchOS
Shortcuts
Xcode
Community banner courtesy of u/Antsomnia.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
User replaceable components come at a performance costs. Everything in tech is a trade-off.
I've been working on Apple Silicon until about a year ago. When I switched to jobs, I switched to Windows operated machines. But I kept doing the same thing. The windows machine I use now has "basically the same spec" (same amount of RAM, same price range etc.) than the laptop before.
I am going nuts. I completely forgot the waiting and the "lemme just freeze for some time" behaviour. The difference between unified memory and standard bus is not only detectable in benchmarks.
Yes, I was pissed when Apple Silicon came out and they dropped repairability. I'm starting to think it might have been the right thing to do. But I haven't been using windows for a decade. Maybe it's just that fucking operating system. My company refuses to let me install Linux on that crappy laptop though.
A lot of that may just be the bloated shambling corpse that is windows.
Even my main PC immediately felt more responsive when I shifted to Linux a few months ago.
Unified memory is awesome. It lets you do things like handing over tasks to the GPU with zero overhead.
That means you can choose whether to do something on CPU or GPU based on which is more suitable for the task. With a traditional discrete GPU with it’s own VRAM you sometimes do things on the CPU even though the GPU would be better suited as the overhead of copying data to and from VRAM would negate any performance benefits of letting the GPU do it.
I feel like discrete GPUs could be so much faster at these kind of cross compute tasks through ReBAR/SAM if not for all the proprietary crap and half baked drivers. Unified is a view into that perfect world of what could be if we stopped with the monopolistic BS.
ReBAR may make things bit more efficient, but you still need to move data back and forth to VRAM over the slow PCIe bus.
This is not a huge problem for games, as it’s mostly sending data to VRAM that stays there for a long time, but it is a major bottleneck for many GPGPU tasks.
Since even Nvidia doesn’t seem to care about games anymore (and consoles have always used unified memory) I expect the discrete GPU to go the way of the dodo.
It's just the operating system windows is bloated beyond belief. Most apps themselves are too. Especially if they are windows only
It's just the operating system windows is bloated beyond belief. Most apps themselves are too. Especially if they are windows only
Computers have been good enough for years though, we just keep writing worse code. There is no good reason for a laptop with 8gb ram running Windows to need six seconds to rename a file, and that's basically normal performance... especially if you have to have OneDrive on for work.
It’s not a direct correlation, though. It’s not always the case.