melgross
About
- Username
- melgross
- Joined
- Visits
- 127
- Last Active
- Roles
- member
- Points
- 10,981
- Badges
- 2
- Posts
- 33,729
Reactions
-
Apple silicon Macs to support Thunderbolt despite shift to ARM
See, this is interesting. Apple is saying that they developed it together. But shortly after the technology became out, Intel said that it wasn’t true. They said that Apple came to them with the idea of a fast port, but that Intel did all the work, and that Apple had nothing to do with the development. So this statement is interesting.rcfa said:People forget that Apple and Intel developed TB TOGETHER. It’s not like a PROTOCOL is depending on a specific CPU 🤦🏻♂️
additionally, as far as TB 4 is concerned, VT-D is the reason I was concerned. While it’s true that TB is part of PCIE, VT-D is a technology inside Intel’s’ x86 chipsets. My concern was how Apple would implement an x86 technology. I guess we’ll find out.
info from Intel on VT-D:
https://software.intel.com/content/www/us/en/develop/articles/intel-virtualization-technology-for-directed-io-vt-d-enhancing-intel-platforms-for-efficient-virtualization-of-io-devices.html
-
Display expert again claims 'iPhone 12 Pro' will lack 120Hz ProMotion
The truth is that Apple’s devices already have smooth motion without 120Hz displays. Read, over the years, of Apple’s “buttery smooth” scrolling, for example, compared to the jerky jerky scrolling of Android. Just take your iPhone out and try it. I don’t see the big deal here. It’s a matter of the power of the SoC, and the fact that Apple’s work runs, as the expression goes, “close to the metal”, rather than the slow Java-like implementation of Android, which adds extra software layers. Additionally, they gave these horrible “skins” that slow things down further. Samsung has been pointed out as being particularly egregious in this, with their skin being pretty slow.
i’m really not sure whether we actually need 120Hz, or even 90Hz, particularly as it badly affects battery life. I don’t see what Apple can do that would be so special as to negate that problem. Other than to have an even bigger battery.
-
Why the Mac's migration to Apple Silicon is bigger than ARM
Of course there’s no evidence. I don’t know what people are thinking. Either they didn’t watch the keynote, or just don’t listen carefully. Apple clearly said, as I keep telling people, that they created a chip line SPECIFICALLY for the Mac. That should tell them something. And Craig said recently that the A12z will never be in a shipping product (other than the developer machines). And it’s two years old.joeljrichards said:
Sure. Fujistu and Ampere do it with custom ARM cores. The reference ARM N1 core looks like a contender. The (current) fastest computer on the planet runs on Fujistu's ARM-based chips.darthw said:Will it be possible, eventually, for Apple to make faster SoCs than the fastest most powerful intel Xenon chips?
My understanding is that Apple has the best single threaded performance of ANY version of the ARMv8 core—and until this week it was a mobile-only processor! Even if they only clock it moderately higher, the tech to scale it up to 40+ cores is already accessible to ARM licensees. Imagine 60-80 Lightening cores running on a single SoC. How many Xenon chips would be competitive with that?
I hear a lot of moaning about "Apple can't compete in the high end" if they transition. There's just not evidence for that. I'm less sure how they'll do it (chiplets, license HPC IP, etc.) but they certainly CAN.
-
Apple Silicon Macs are needed for consumers and pro users alike
We’re talking about 8 years.SpamSandwich said:
I think even in the state of California a new car or product must have replacement parts for repairs available for at least 3 years after purchase.melgross said:
No manufacturer cares about an owner of such an old machine. It’s time to think about buying a new one. If you can’t afford one, buy a newer model that’s used or refurbished.pslice said:Is Apple going to offer an AppleCare extension to bridge Intel users to the Apple Silicon? My AppleCare expired on 6/15. I had hoped Apple would announce new iMacs. Right now I feel naked without AppleCare coverage.
-
Apple Silicon Mac mini dev kit looks like a desktop iPad Pro
I certainly hope that on a high performance Mac, the gpu will NOT share RAM with the rest of the SoC. Shared RAM is always going to be slower than graphics RAM. It’s not just a faster version of regular RAM. It’s different.rob53 said:After watching the WWDC video "Explore the New System Architecture of Apple Silicon Macs" I'm beginning to understand just how powerful the A-series SOCs really are and how the will change the way Macs work in the future. The graphic showing multiple components in an Intel-based Mac while everything is on a SOC (not sure about all RAM) demonstrates how the AS SOC can run faster and more efficiently than the Intel configuration. Graphic is over-simplistic but AS SOC uses the same memory for both the CPU and GPU. No more having to worry about specifying a GPU with more memory since its able to use as much shared-memory as it needs. The video didn't mention PCIe memory in regards to the AS drawing, it uses whatever internal memory bus the SOC has. I don't know how the SOC will connect to external devices like SSDs, (probably?) RAM, and all I/Os but the more information that comes out about the AS Mac the more it looks like it's going to be much more powerful than Intel Macs.
As for discrete GPUs with their huge heatsinks, I wonder how many GPUs Apple will be able to add to their SOC and whether they will create specialized GPU-only "SOCs" that work with the main CPU SOC over some kind of ultra fast bus. When I look at a traditional slot-mounted GPU, I see a reasonable size GPU chip with a lot of other electronic components all hidden by a heatsink the size of a Mac mini (just joking but not by much). I'm sure the GPU manufacturer is overclocking the GPU, creating enough heat to heat a small house and definitely your office. Will Apple be able to create a GPU with the same number of cores as something like the high end Nvidia TITAN RTX with 4608:288:96:576:72 (72) (6) cores (That's Main Shader Processors : Texture Mapping Units : Render Output Units : Tensor Cores (or FP16 Cores in GeForce 16 series) : Ray-tracing Cores (Streaming Multiprocessors) (Graphics Processing Clusters)) without needing a refrigerator to cool it?
