They’ve already beaten Intel, they’re doing 16 cores to beat themselves and provide a more significant delta front the current 8 core machines.
Single threaded. Not multiple threaded.
They’ve already beaten Intel, they’re doing 16 cores to beat themselves and provide a more significant delta front the current 8 core machines.
But why wouldn’t Apple do it. They’re not in race with intel. Why benchmark what they do with cores against what the fastest intel chips will do? If they can make a machine twice as fast as the new entry level machines customers will pay for it.Single core performance is also higher, the current M1 is already faster than the i9 in the MBP16 both in single & multi threaded performance, and this is with the m1 having just 4 high performance cores, the other 4 being low powered. I dont think they would be doing 16-core to beat Intel, as they can already pull way ahead with 6 or 8 high performance cores today.
No. He’s right. Apple has already seeded macOS 11.1, a minor update to 11.0. Apple appears to be breaking away from the 10.X.X convention and starting down the macOS 11, 12, 13... naming scheme. Makes sense, though I am disappointed. This will put them in more consistency along their other xOS releases naming schemes.
Doesn’t matter. Apple’s goal isn’t “beating Intel”, they’re already beaten as indicated by the single threaded score. The goal now is to beat their current processors.Single threaded. Not multiple threaded.
Um, yes they are I guess. Look at Intel, you could say the same about their engineers. Also, Apple will never outsource components to others, they’ll keep the competitive advantage and dominate all profitable parts of the market. M1 GPU is no joke, watch some reviews if you like, you need to get the $800 gpu upgrade on the 16” Mbp just to beat the MBA in gpu performance. It’s all about the performance per watt, the rest is scaling it up. I confidently predict Apple will have the fastest gaming laptop (under 5lbs) next year.Not at all likely. Do you really think Apple engineers are that much smarter than AMD or NVidia GPU designers? If that were true, Apple would begin designing graphics cards that could also be used in the other 90% of the PC market held by Windows machines.
I think they will go with another naming convention than M because the SOC will be very different to accommodate the Pro series with added expandability.This is a meaningless statement. If Apple calls it an M, it’s an M. When you refer to an “M series chip” do you mean ”a chip named M?” or “the same chip that already exists” or “a chip based on the same SoC blocks?”
if it’s the last option, that’s surely wrong. The pros will definitely get a chip with the same core designs, but a lot more of them (Possibly also clocked faster).
You just add a small intermediate step involving eBay.Add to this the upgrade-ability. Most gamers on desktop solutions upgrade CPUs (when the Mobo doesn't change sockets) and discrete graphics cards every year or two to maintain top-end gaming performance, and that's something you can't do with a SoC.
They can just build a chip for example with 17 cores and 34 GPU cores and sell it as 16+32 cores to get a very high yield.Bigger chips and more cores will obviously consume more power and produce more heat... but Apple has a massive advantage here of requiring less than half power compared to their competitors. An 8-core Apple GPU draws 10 watts at it's peak. An 128-core Apple GPU will therefore draw around 150-180 watts. That's 16384 shading units!
Apple's senior chip designers have left the company precisely since Apple didn't intent to enter the server market.
They could probably fit 32 core GPU, a 12+4 core CPU and a large shared cache on a 300mm2 die, which is not too large, compared to the current GPUs.
A 128-core Apple GPU would be faster than anything currently on the market.
What is your basis for this statement? Apple should be able to make relatively large chips while still being commercially viable (even if the yield rate is low, it is still going to be cheaper than buying them from Intel+AMD). And there is always the chiplet route if they feel that a single die is not large enough.
That is an entirely different design than M1. Basically you throw away all the out-of-order processing, branch prediction etc. and relY in hyper threading to fill the execution units with work. You don’t get good single thread performance. You get mediocre performance per thread, but that mediocre performance is multiplied by 224 making it an excellent server chip, capable of serving 224 users simultaneously. Completely useless for a MacBook.Will Apple be able to implement hyperthreading on these cores? I seem to remember a report on a 56-core ARM server processor that had 4x hyperthreading. I can’t find it now, and all ARM server chips seem to be 1-core / 1-thread designs.
You misunderstood something. M1 with 4 cores is right between 6 and 8 core Intel processors. 8 core M1 would be close to 14 core Intel, it runs rings around 8 Intel cores. 16 core M1 is very close to the current top end MacPro with 28 cores. 32 core M1 beats anything Intel has.Sounds great, but am I the only one here who thinks that needing that many cores to surpass current performance is a bit worrying for the future? The current Macbook Pro 16" has eight cores, right? They're doing a 16-core to beat Intel? I suppose it doesn't matter how you get that higher performance as long as you get there, but I don't know, this doesn't seem like a scalable architecture.
You misunderstood something. M1 with 4 cores is right between 6 and 8 core Intel processors. 8 core M1 would be close to 14 core Intel, it runs rings around 8 Intel cores. 16 core M1 is very close to the current top end MacPro with 28 cores. 32 core M1 beats anything Intel has.
And this is news . . . how, exactly?
Apple is working on a series of new custom Apple silicon processors to power upgraded versions of the MacBook Pro, new iMacs, and a new Mac Pro for introduction as early as next year, according to a new report by Bloomberg.
Apple is said to be working on several successors to the M1 custom chip, its first Mac main processor that debuted in November in a new Mac mini, MacBook Air, and 13-inch MacBook Pro.
I am curious what tools you use about which you can say that with such confidence? I have found many tools that do not have macOS versions that do have iOS/iPadOS version (meaning they now run on Apple Silicon macOS systems), and many more that have moved to the web/Electron. If Apple has machines that are seriously competitive with the highest end systems, I would not be surprised to see other software ported either native to macOS or to Windows on ARM as a VM on macOS.Pros need the tools to be available though, in my case a lot of the tools that are use are limited to Windows and will never be ported to Mac - of any flavour. therefore I fully rely on being able to run Win10 in a VM and with the M1 chip that is currently not possible. hence I cannot stay with Apple - which I prefer for the rest of my workflow.
Well as long as they ship with Catalina, and a 34" nano screen I'm interestedThe story is obvious and unnecessary, but what an awesome breath of fresh air for the Mac. Exciting time for Apple hardware.
Some people won’t bother to read the rest of the thread, let alone the article they’re allegedly reacting to. But their reading comprehension allows them to analyze the headline and marvel at their own cleverness.Yeah I really don't get what's so obvious about this article. First of all, while it's likely, it's not a given that Apple will simply scale up its SoC to more cores, especially 128! It's also not at all obvious that Apple will use their own GPUs at the high end, that will depend on if their GPU architecture can scale that far without a big enough logarithmic falloff before they need to redesign it again. This provides some info there, too.
Given that this article has multiple mentions of core counts, with interesting info about both CPUs and GPUs, there's plenty to learn if you read carefully all the way through.
The GPU is among the most interesting to me since IMO it represents the biggest challenge:
One of the best well-balanced, high-performance laptops you can buy right now was released a few months ago and features an Nvidia 2080 Super Max-Q with a Geekbench graphics benchmark about 4x that of the 8-core M1. Assuming the GPU scales linearly, without any complications, Apple will need at least 32 cores, and the GPU will have a TDP about half the 2080smq at the same performance. By the time they release that, AMD will have a mobile variant of new GPUs with big improvements, along with Nvidia's 30-series mobile (I'm expecting only a moderate jump).
Assuming the M1-based architecture, that will necessitate even more cores to stay caught up, say 40 or maybe 48 to pull ahead. That might put Apple at a 70W TDP, which is arguably too high for the MacBook Pro. And again, this is assuming their GPU scales linearly to multiple times that of the M1 iGPU, but also its "only" an M1X instead of an M2 or whatever.
See how all this is not obvious?