Announcement

Collapse
No announcement yet.

Apple M1 Ultra With 20 CPU Cores, 64 Core GPU, 32 Core Neural Engine, Up To 128GB Memory

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    Originally posted by ezst036 View Post

    Are M1's CPU internals really the challenge though with widely known ARM specs? Isn't the GPU portion of the M1 a bigger challenge?

    Without specifications, can anybody make M1 GPU/Vulkan/OpenGL/etc drivers that rise any higher than Nouveau considering the same difficulties? For example, can we do re-clocking on the M1 GPU? We can't do that with Nouveau on newer GPUs to my knowledge.

    I'm not being critical, I honestly don't know so I'm curious about it. Perhaps the M1 GPU is worse in one respect though. Nvidia at least publishes a binary Linux driver to achieve full performance. Does Apple publish a binary Linux GPU driver?
    NVidia doesn't offer the firmware to the opensource community. Not everybody wants a binary driver. On the other hand, Apple loads the firmware themselves before the OS of your choice starts.

    Comment


    • #42
      Originally posted by piotrj3 View Post
      What about very poor ecosystem for GPU programming with very small number of libs (comparing to let's say CUDA)?
      Apple gives people the best hardware and it's on them to use it (or not). CUDA is not everything. Developers actually often use libraries running above CUDA, not CUDA itself. These libraries are being extended now with a codepath for M1. Also, the same argument you can use against AMD and Intel - Apple is not worse than others.

      Comment


      • #43
        Originally posted by piotrj3 View Post
        to make good overclockable chip you need relativly high wattage leakage.
        I think you've got it backwards. Intel bins its chips so that the highest-spec SKUs have the lowest leakage of the dies they make. So, if you took a lower-spec model, the corresponding perf/W graph would probably look worse over its entire range.

        Originally posted by piotrj3 View Post
        if you compare more mild designs of Intel like 12400 (non-k) vs AMD 5600, Intel wins in efficiency easly.
        Are you saying Intel would win vs. the M1 Ultra, or you're just saying that Intel CPU beats the AMD one?

        Anyway, the main thing you seem to be missing is how their graph shows significantly better efficiency, even at the top of their performance envelope, which also happens to be far above the top of Intel's performance envelope. Granted, it's comparing a 20-core/20-thread CPU to a 16-core/24-thread one.

        Originally posted by piotrj3 View Post
        It is very funny to read this graph when you know Intel recently posted similar type of graph for Laptop CPUs (which are not designed to be overclocked) and compared it succesfully to Apple's laptop M1s.
        It all depends on what benchmarks you use, and whether the M1 is running them in Rosetta. Got a link to these benchmarks?

        Originally posted by piotrj3 View Post
        3rd. What about raytracing apple?
        Good question. However, Apple has an IP license for Imagination Tech's IP. Imagination demonstrated HW-accelerated raytracing several years before Nvidia launched RTX. Granted, I'm sure they were much less capable, but they were also running in mobile power envelopes.

        Originally posted by piotrj3 View Post
        What about AI denoising like Optix or Intel's solution.
        That 32-core Neural unit is probably good for something.

        Originally posted by piotrj3 View Post
        What about very poor ecosystem for GPU programming with very small number of libs (comparing to let's say CUDA)?
        Well, there's Metal and MoltenVK (i.e. Vulkan on top of Metal), for graphics.

        Good question about GPU-compute, though. I do wonder how many do GPU-compute on macs, regardless.

        Originally posted by piotrj3 View Post
        4th. Current GPU designs are fairly long on market (more then 2 years in fact). Comparing them to next Nvidia 5nm might make Apple look like a fool very very fast.
        Yeah, but Apple is still yet to launch their new Mac Pro. So, we don't yet have their ultimate solution.

        Anyway, I think Apple isn't really competing head-to-head with Intel, AMD, or Nvidia. Most people buying Macs wouldn't even consider anything else, as long as Apple had a product that could meet their needs. And most people not buying Macs wouldn't buy one because it's too expensive for what you get. So, if they don't need a Mac, they're not going to buy one.

        Basically, Apple just needs to keep people in its garden ...and keep making them feel special, for being there.

        Comment


        • #44
          Originally posted by Lbibass View Post
          I'm very excited to see this machine in testers hands. The maxed out ram / cpu / gpu model starts out at 200 dollars less than the base mac pro. This is insanely affordably and incredibly powerful. I like it a lot. So long as you can take advantage of the fixed-function hardware, it's a no-brainer.
          assuming you never need to add memory, or change processor, change GPU, .......

          Comment


          • #45
            Originally posted by nranger View Post

            My bet would be Apple bets on more accelerators on chip in future Apple Silicon generations rather than massively expand cpu and gpu core counts. More, bigger, fancier media engines, AI processing, matrix math blobs, etc. would do more to accelerate the specific work loads Apple wants customers to focus on, and do it on specific apps like Final Cut Pro, Logic Pro, Photoshop, etc. They can get more performance, with less power, while also locking customers into the Mac ecosystem, while also avoiding direct comparisons with PC software.
            I think this is a key element.
            I don't think they have much advantages (besides maybe half or a process node) if compared piece by piece to PC HW/SW.
            However, they get massive system gains by avoiding traditional interfaces/bottlenecks for ram, gpu, ...

            Comment


            • #46
              I would really like to have a system with one of the high end M1 chips, but the prices are... well, I'd rather buy a new Quadro for work, let's put it that way.

              Originally posted by coder View Post
              Good question. However, Apple has an IP license for Imagination Tech's IP. Imagination demonstrated HW-accelerated raytracing several years before Nvidia launched RTX. Granted, I'm sure they were much less capable, but they were also running in mobile power envelopes.
              Imagination is one of the companies where I ask myself, "If 3dfx hadn't dominated the early 3D accelerator market with Glide, just how good would PowerVR be now?"


              Comment


              • #47
                Originally posted by mppix View Post
                I don't think they have much advantages (besides maybe half or a process node) if compared piece by piece to PC HW/SW.
                However, they get massive system gains by avoiding traditional interfaces/bottlenecks for ram, gpu, ...
                Okay, let's look at Apple's Firestorm core (A14, M1) that launched more than a year before Intel's Golden Cove (Alder Lake):

                Parameter Apple Firestorm Intel Golden Cove
                Decoder Ports* 8 6
                Reorder Buffer 630 512
                Max uOps Dispatched/cycle 17? 12

                * x86 decoders tend to be restricted. For instance, Sunny Cove had 1 complex + 4 simple decoders. GC's decode block certainly has similar restrictions.

                Sources:
                Now, according to Anandtech, Alder Lake has higher single- & multi- thread performance than the M1 Max (although SPECfp2017 MT is basically equal!). However, it does so at considerably higher power consumption. So, Apple's wider & more-sophisticated microarchitecture manages to deliver much better perf/W, and by a margin you can't reasonably attribute to the delta between TSMC 5nm vs. Intel 7.
                Last edited by coder; 09 March 2022, 02:36 AM.

                Comment


                • #48
                  Originally posted by Ladis View Post
                  NVidia doesn't offer the firmware to the opensource community. Not everybody wants a binary driver. On the other hand, Apple loads the firmware themselves before the OS of your choice starts.
                  That's a nice bit of corporate-speak, but unless Apple is releasing GPU specifications the only driver equivalent you're going to get is something akin to Nouveau.

                  Comment


                  • #49
                    Originally posted by ezst036 View Post

                    That's a nice bit of corporate-speak, but unless Apple is releasing GPU specifications the only driver equivalent you're going to get is something akin to Nouveau.
                    The opensource defelopers are catching the commands the driver in macOS calls to the GPU. It's easy since they run the original macOS in a virtual machine running inside the bare metal Linux. The problem of Nouveau is the missing relocking because of the missing firmware. That also demotivates more developers to work on the opensource driver.

                    Comment


                    • #50
                      Originally posted by piotrj3 View Post
                      I think Apple went very high on marketing here. RTX3090 has 940GB/s memory bandwidth only for itself (RAM is seperate thing for GPU) and without NUMA like topology between few chips and RTX3090 is actually handicapped often by memory speed even in games (and productivity way more).
                      Win some. Lose some.
                      Yes the GPU gets (very slightly) less bandwidth. But it has access to 128GB, rather than the max of 48GB on a Quattro.
                      Professionals tell me they care about that vastly more than about a slightly lower bandwidth…

                      And, how often does this need to be stated, this is still not end of the line! This is Apple operating at the iMac Pro level (the machine it replaces), not at the Mac Pro level. That will come with CXL and a whole new vision of how to modularize and design a professional machine.

                      Comment

                      Working...
                      X