10 ms·

I don’t think this is going to happen, but: It would be extremely funny if Apple strapped a heavy copper heatsink and a turbo fan to an M1 and just overclocked

• bserge 5y ago
• gjsman-1000 5y ago
  Apple (if they were willing to expose a more wild side) should make an "M1 Pro Max Unlocked" board (or something like that) and just sell it as a bare, microATX board. It could cost exactly the same as the Mac mini for all I care, but if PC builders could use it in crazy cases and with liquid (or even liquid nitrogen) cooling, that would be so cool.
  • vbezhenar 5y ago
    Just rip the board from Mac Mini?
  • barkingcat 5y ago
    I bet Woz would totally do it, and he has the clout to do it too - as a retro-mod calling back to the Apple I or II
  • devin 5y ago
    This will never happen.
    • gjsman-1000 5y ago
      Nope, it won't, Apple will never be willing to show a more wild side. I'm just dreaming.
    • qbasic_forever 5y ago
      It would not surprise me to see some future Mac pro use the M1 in a desktop form with exotic high wattage but near silent cooling (liquid, air or otherwise). These chips are fantastic for video and media work and really need to get put into more workstation-like form factors.
    • imwillofficial 5y ago
      I remember hell freezing over with iTunes for Windows.
      • copperx 5y ago
        iTunes for Windows was not unimaginable. Now, iMessage for Android ...
  • asdff 5y ago
    Would love a liquid cooled mac.
    • ipython 5y ago
      They have already done it, almost twenty years ago now: https://www.computerworld.com/article/2565453/new-power-mac-uses-liquid-cooling-on-high-end-model.amp.html https://www.computerworld.com/article/2565453/new-power-mac-...
• LeoPanthera 5y ago
  > I don’t think this is going to happen, but: It would be extremely funny if Apple strapped a heavy copper heatsink and a turbo fan to an M1 and just overclocked the ever-lasting crap out of it. I fully expect this is what the new Mac Pro will look like, probably with multiple M1s.
  • kmeisthax 5y ago
    M1 Max has a cross-die interconnect they photoshopped out of the marketing die shots, so I fully expect Apple to ship dual-Max configs on the Pro. (Also, RIP one of the most tinkerer-friendly Mac configs in a very long time.)
    • marcan_42 5y ago
      The internal design is 2-die max, so that'll be a somewhat puny Pro by Mac Pro standards. Maybe a Mac Mini Pro? I expect them to come up with something much crazier with a newer design, perhaps in the M2 generation. Doesn't mean much, but the IRQ controller architecture and driver in these machines claim to scale to 8 dies, even though the existing implementation in the M1 Max is sized for 2. That's no guarantee they're looking at a product like that, but they might be.
• monocasa 5y ago
  marcan seems to have a pretty good handle on the clock tree. There's probably enough breadcrumbs in asahi to void some warranties on a Mac mini, go nuts with cooling, and see how high it can go. The pmic might not be configurable though.
  • marcan_42 5y ago
    We actually know very little about the clock tree; we know where the clock mux registers are but not how they connect to each other, and that's mostly about peripherals, not the CPUs. Also, the PLL registers are locked out and read only. The CPU DVFS stuff is all done automatically (you just select a performance level in one register), and I wouldn't be surprised if the underlying tables and configs for that are also locked post boot. The PMIC is accessible to some extent, but runs its own firmware anyway which might lock things out, and I don't think CLVRs (final voltage regulator stage, the one you care about) are directly accessible (on the Pro/Max) and that should all be controlled by the SMC or hardware logic, which again has a good chance of being locked off. Basically, all of this is configured at boot and much of it locked off. Apple have done a good job making it difficult to nuke these things from the OS level, for better or for worse :). I once overwrote all the GPIO registers and didn't even break anything (though the Type C controllers did crash, but a hard shutdown and boot fixed it - this was on an M1 Pro MacBook). It's in one of my streams somewhere...
    • monocasa 5y ago
      Oh, word. Thanks for the correction. I guess I extrapolated too much from the combo of you knowing enough of the clock inputs to the peripherals to make them work and a statement of yours about finding a pll bypass signal. I want to say your main point your statements on the pll bypass was that the cores must be mainly static logic because they ran stable at whatever rate the crystal is, but I'm half remembering and could conflating that with something else.
      • marcan_42 5y ago
        Ah, the PLL bypass was just p-state 0, but that was still using the high level interface. macOS never uses that mode, but it's there.
• wmf 5y ago
  Sadly, I don't think the M1 family will clock any higher than it's already going. The rumored M1 Max Duo and M1 Max Quadro will have more cores but they'll still trail Intel in single-thread performance (although not enough for most people to notice).
  • smoldesu 5y ago
    Precisely. ARM has traditionally had issues scaling it's clock speed even when given ample cooling and power, history suggests that the only way they'll significantly improve performance is with new fab technology, engineering a better design, or expanding horizontally (eg. adding more cores).
    • FateOfNations 5y ago
      My understanding is that Apple isn't using ARM IP cores (i.e. ARM Cortex) but just implement the ARM instruction set architecture using their own proprietary CPU core design. Edit: Since the A6-era for CPU cores and and A10/11 for the GPU cores.
    • gjsman-1000 5y ago
      Apple has an architectural license, which is why they are so insistent on calling them "Apple Silicon" in branding instead of ARM. The only thing the Apple M1 has in common with ARM is the instruction set. The actual core design? Completely Apple. No Cortex X1s or A73s in there. Apple does use the stock ARM M3 in a few places IIRC but that's a very-low-power ~100mhz part for managing some internals. Also, as for who can push the limits of ARM, Apple appears to know ARM better than ARM knows themselves. Both based on Apple's higher-performing core designs than ARM stock, and also because Apple cofounded ARM with VLSI and Acorn, and an Apple VP was the first CEO. Edit: On that note, it's kind of ironic how every Android phone has the DNA of a company Apple cofounded and initially led.
      • gjsman-1000 5y ago
        Also, for every Android commentator talking about ARM, consider that Apple even named the company "Advanced RISC Machines," because they opposed having the name of a former competitor in there (because it was originally going to be "Acorn RISC Machine"). Apple also provided almost all the funding and initial leadership for the venture, while Acorn supplied most of the engineers and VLSI the silicon tooling.
• chias 5y ago
  Overclocking isn't just limited by heat dissipation, and dealing with the additional heat generation is one of the easier issues to overcome. Much more limiting is the fact that the electronic components in the hardware (everything from RAM to the more "integrated" components on the PCBs) have certain amounts of time which they need to take in order to tie voltages high or low, to either read or output a bit. This amount of time is typically very small as compared to the amount of time that a voltage stays tied high or low, and so isn't a problem, but while you can shorten the amount of time that a voltage needs to stay high or low when overclocking, generally the transition time isn't as negotiable. Increasing clock speeds to the point where these transition times dominate means you're going to get garbage reads and garbage writes out of a component. This is part of why if you're overclocking your computer, you end up experiencing errors or unexpected crashes even when temperature doesn't go too high. Past reasonable tolerances, you can't just cool it more and expect it to work.
  • nielsbot 5y ago
    cryogenic cooling tho?
    • chias 5y ago
      Wouldn't solve the problem. Basically you need the integrated components to have shorter transition times. These are an inherent property of the component: speeding up the transition time basically means "replace this component with a different component that has a shorter transition time"
      • dragontamer 5y ago
        Overvolting the component and sending more current to any transistor tends to speed it up, though at the cost of severely more watts used. Overclocking is often a combination of overvolting (leading to more amps), allowing you to increase clock speeds.
        • bserge 5y ago
• KennyBlanken 5y ago
  > in single-threaded performance, it's really no faster than a dippy little Macbook Air. Maybe in terms of raw FLOPS or something, but more importantly it has enormous levels of memory and storage bandwidth, which is what's actually important for most user workloads these days. That's not even accounting for the performance of the GPU cores which can be used for a ton of multimedia workloads.
  • smoldesu 5y ago
    > more importantly it has enormous levels of memory and storage bandwidth, which is what's actually important for most user workloads these days. I've heard this tossed around a lot, but is there any evidence that it's actually true? I struggle to imagine a regular workload that benefits from 32gbps of bandwidth, much less a scenario where the average layman could saturate a 200gbps or 400gbps bus.
    • sroussey 5y ago
      Working with media files requires lots of bandwidth. No so much for productivity.
    • inkyoto 5y ago
      > I struggle to imagine a regular workload that benefits from 32gbps of bandwidth, much less a scenario where the average layman could saturate a 200gbps or 400gbps bus. Since this is HN, many of us like… compiling, you know, stuff. It is a regular workload for many of us on here. The main beneficiary of the M1 Max wide memory architecture appears to be the Haskell GHC compiler, which is very, very voracious when it comes to the memory consumption and the data bus width. Regular Haskell builds shift the data around at 35-40 Gbps on average, with the Haskell optimiser hovering around 50 Gpbs and easily peaking out at 65 Gbps on medium size projects. Since cabal (the Haskell build toolchain) for some reason can't fully utilise the «-j10» parameter, roughly a 1/2 of CPU cores are not utilised. Which means we can kick off another large Haskell build in parallel without affecting memory transfer speeds of the first build. Rust release builds are a runner-up with 30-35 Gbps transfer speeds on average. Static type analysis and type inference in Haskell and Rust is hard, is complex and requires a lot of fast memory and benefits vastly from large CPU caches. I have not attemped a Rust LTO build yet, but I would expect the LTO builds to like the wide memory bus, like, A LOT. In general, a LTO build (e.g. large C++ builds) is heavily memory bound and shifts data blocks, small and large, around extensively. But an application is not very useful with a dataset, so many of us then like training our data models. Data model training can easily saturate 200 Gbps per a M1 Max CPU cluster. Then the application needs to load the data in and usually do something with it. For example, issue complex queries to a graph database. Large CPU caches coupled with a wide data bus is a winner here. All of the above are examples of real, practical and regular workloads.
• marcan_42 5y ago
  They won't overclock it to any major extent; that's not their game and it's not designed to scale to insanity. But word is the cooling system in the new Mac Mini is pretty nice. At the very least I'd expect it to ship with an M1 Max with both neural engines active (the laptops have one disabled for some reason) and wouldn't put it past them throwing in the rumored (and all but confirmed to exist based on what I know about the architecture) M1 Max Duo.
• inter_netuser 5y ago
  copper body macbook pro?
  • bacchusracine 5y ago
    >copper body macbook pro? Apple Marketing: This year Apple is going green in an all new way!
• lostlogin 5y ago
  > I don’t think this is going to happen, but: It would be extremely funny if Apple strapped a heavy copper heatsink and a turbo fan to an M1 and just overclocked the ever-lasting crap out of it. If this was released and allowed for relatively straightforward overclocking, user replaceable RAM and a full size GPU of one’s choosing, what could they charge? The M1 Mad Macs edition would be loved by all 9 of us.
• runeks 5y ago
  > The crazy thing about using a Macbook Pro (with an M1 Pro or M1 Max CPU) is that, in single-threaded performance, it's really no faster than a dippy little Macbook Air. That “dippy little M1 MacBook Air” is damn fast. Even for single core tasks. I’ve owned one for a year now, and recently got the new M1 Pro with eight of the same cores, and single thread performance is not low by any sensible measure.
• ribit 5y ago
  It is likely that 3.2ghz is close to practical ceiling for these CPU cores for some technical reason. It's not about thermal headroom, but the very ability of the circuitry to run at faster speeds. It has been speculated that one way how Apple achieves their impressive perf/watt is to use logic gate arrangements that can only run at low frequency but bring massive wins in power consumption.