Read more.Single and multi-core GB5 tests outpace the 16-inch MacBook Pro with Intel Core-i9 CPU.
Read more.Single and multi-core GB5 tests outpace the 16-inch MacBook Pro with Intel Core-i9 CPU.
Meh,more useless benchmark suites. Apple also is know for profiling certain applications to tune performance in,so it won't be repesentative of all software IMHO. Also the fact Apple is basically thowing transistors at their SOCs. How much of the improvement is because Apple is on 5NM and Intel is still stuck on 14NM??
Meh, all that tells me is that the a14x has more cores than the a14, I'd say 10 versus the 6 in the a14.
I've said it before geekbench is only really good for comparing like for like but I'm sure Apple will wheel out some slides with huge gains in carefully selected situations.
While I'm sure the transition will be smoother on os-x than windows (os architectural differences and 'complete control over hardware' help) I still wouldn't count on arm being faster with 'professional' programs.
Or maybe they swapped out the 4 energy efficient ".little" cores for 4 more big ones. A sort of "big.big" variant
Edit: I've thought Apple CPUs were overly wide issue for just mobile use for years, I presumed leading up to this moment. They probably had a bunch of knobs they could tune for this die lined up, so it could actually be quite quick. It won't get me buying into the Apple ecosystem, but I have to be impressed with their CPU design skills so far.
ofc the benchmarks are probably going to be "at 10W power level" or similar, a game Intel have been known to play when it was at their advantage.
Macs are great computers for doing very little. As soon as you start pushing, they show they were not made to work hard. Changing the silicon wont make any difference to most users, as current Mac owners do little more than browse and read emails.
Anyone that still does more than text on a Mac really should reconsider their life choices.
I can't wait to see comments elsewhere from people saying that the Apple chips are faster than i9s... Of course the numbers are still interesting to see.
Ah yes, a geekbench benchmark versus a thermally throttling i9, truly indicative of how far they've come!
Great... an ecosystem locked device is somewhat better than a gimped, overheating, under performing chip in a synthetic benchmark. I'm going to go back to windows and AMD, now.
Geekbench really is just nonsense across platforms, isn't it?
Just makes you wonder how efficient these huge ARM cores will be. Intel struggled with atom to get a low powered competitive SOC going for the handheld market. I can imagine Apple will have the same issue upscaling their design to fit the need for high end. Might be fine for a reg MacBook.
Excellent performance, but in a super specific test. My concern as a Software Dev, is its compatibility with various compilers, development environments, tools, etc... Something tells me I will have to stick with the Intel variant of Macbook Pro.
TLDR: x86 is a mess that doesn't scale down; ARM is comparatively clean and has no theoretical barrier to scaling up.
To be fair the later Atom cores are a fair bit better (as were the equivalent AMD Cat cores), the ones after Intel gave up on phones and tablets and generally labelled "Celeron". But that is partly because they aren't so scaled down as the early Atom chips. AMD64 is a really bad instruction set, which also drags the even worse 386 and the truly dire 286 modes behind it.
The dreadful variable length CISC instruction set and ancient baggage is what stops x86 from scaling down. Even if you can manage with the energy requirement to decode those instructions, it takes more silicon and the low end market is *really* price sensitive and won't tolerate an extra square mm.
ARM V8 is OTOH a more recent instruction set. You can't run all code from the original ARM chips, they don't even try. Throwing out the old baggage allows a cleaner approach aided by the fact that the starting point of ARM V1 is a heck of a lot better than starting from the 8086.
We haven't seen a *really* fast ARM implementation yet, but it is very possible and seems to be now very much on the cards. ARM have just been talking about the A78C where they take the usual 4+4 big.LITTLE setup and make it 8 core big.big with maxed out L3 cache. Amazon have their 64 core Graviton server chips, and as that chip is now a year old I presume they will have a Graviton 3 with a more recent N2 core. For desktop/laptop style use we now have the V1 cores.
https://www.anandtech.com/show/16073...neoverse-v1-n2
I look forward to my ARM V1 based Raspberry Pi
I've not seen core estimates broken out, but you have to remember the A14 chips have a really big tensor co processor and pretty big GPU as well as all the camera DSPs etc in their transistor count that the likes of Zen 3 just doesn't have making comparison hard.
If you want to do some back of the envelope really rough calculations, there are some numbers here you can multiply out: https://www.tomshardware.com/uk/news...ionic-revealed
I would myself, but it's the weekend and I'm eager to finish Crysis 3. I've been playing through my backlog of never played Steam games bought on sale, and when I've finished Crysis it might be FO4 next. You wouldn't want to keep me from that
I have wondered before if it is possible to produce a pseudo x86 based cpu with some of the legacy instructions removed and still have a largely functional PC.
If Windows and office work then that is 95% off people covered. I wonder what software out there uses instructions that are considered obsolete.
Have Intel or AMD worked on something like that in the past?
I've heard unsubstantiated rumours, but nothing I believe and not for a long time (like a 386 that couldn't do 8086 mode).
The thing is, it is the variable instruction length and available addressing modes that hurt you. The very CISC essence of x86/AMD64, not just the baggage.
Any attempts to run old Windows software on modern hardware always fails for me, so I don't think it would be any practical loss, just not a real win either.
VIA did do a RISC mode in their CPUs, some 7 byte fixed length thing that gave better access to the underlying uOPS for testing purposes.
There are currently 1 users browsing this thread. (0 members and 1 guests)