Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple M1 CPU & GPU speed is very disappointing
- Thread starter Appletoni
- Start date
- Sort by reaction score
- Status
Apple was an investor, but when Steve Jobs came back he essentially shut us down. Apple would have been our main customer, but he also stopped the macos licensing to clone makers, who had placed lots of orders.
Our Austin office became EVSX, which then became Intrinsity, which then was bought by Apple.
When Exponential went under, I stuck around for a month, then went to Sun for a few months before ending up at AMD for a very long time.
Same sentiment from myself. In a previous post I said "suspending @cmaier from MacRumors is like banning the Pope from the Vatican". The MR Pope has returned, so I'll be happy to continue to sit under the learning tree, once again.
I would note that @cmaier wasn't wrong about WWDC and new laptops, because he always said that the sources that he "may or may not have" inside Apple's semiconductor design team aren't involved in manufacturing or marketing. So, they completed the work on their side of things, passed it along, and its up to Apple's other moving parts to decide when to release products that include those designs.
Well I’ll accept that I was wrong. I guess the supply chain issues may have caused a change of plans, or maybe it was never the plan. Who knows. I’ll say the design of these new MBPs has been ready for awhile now, so whatever the issue is is above the pay grade of anyone who pushes transistors around for a living.
(p.s. - apparently don’t call something a ”screed.”)
Well, the M1 is a 5-10W chip. So it actually over performs compared to the U series that has at least 30% more power room. This said, it will depend on the specific system it is used in: iPad, MBA, MBP will have different TDP configurations.
*in multicore, with half the threads. iIf I measure it against the integrated gpu or in cinebench single core will that shut you up?Apple silicon needs to be on one node smaller to barely keep up with AMD 15W U series.
AMD U Series operate at higher power limit for the first couple of minutes. You are not comparing a 15W 4800U to a 15W M1. You are comparing a 30-35W 4800U to a 15W M1. Come on, man, I’ve explain it to you like five times already.
Image taken from https://www.hardwaretimes.com/intel-tiger-lake-u-cpus-have-a-pl2-value-max-boost-power-of-64w/
M1 powered MacBook Pro, 256GB/16GB: $1,699
HP EliteBook 845 with 5859U, 256GB/16GB: $2,118
Again, you post a lot of primafacia technical details that, like the proverbial Emporers new clothes, are uttetly empty and pointless.
Its no secret that you dislike Mac. However these posts do nothing to prove any of your points.
AMD U Series operate at higher power limit for the first couple of minutes. You are not comparing a 15W 4800U to a 15W M1. You are comparing a 30-35W 4800U to a 15W M1. Come on, man, I’ve explain it to you like five times already.
Image taken from https://www.hardwaretimes.com/intel-tiger-lake-u-cpus-have-a-pl2-value-max-boost-power-of-64w/
Everyone has explained it to him about 5 times. His problem isn’t that he doesn’t know. He knows.
That's a popular ~$140 CPU option right now. Do you happen to know your CPU package power during benchmark? You can get it with HWiNFO64 sensors.
LOL. Ok, that’s just false. Most task run-times are determined by single core, not multi-core, performance.
And the good news is that for the rare workload where multi core matters, Apple can come out with new chips with more cores. I assure you that adding more cores to improve multi-core performance is oodles easier than improving each core to improve single core performance.
Wow, while using only 105W itself in CB R20, having 4 more threads and way more clock speed , very impressive... ?
Messungen: Leistungsaufnahme und Temperatur - Seite 10 - Hardwareluxx
Die wohl beste Rakete: Der Intel Core i5-11400F im Test.
Even if You use ONLY safari webrowser all cores are active ! Look at CPU history. I cant find any my apps use only one core cpu.
I have many apps running on my mac right now which activity monitor says are running on a single thread.
And you are confusing how multiprocessing works. A multithreaded process may run on multiple cores, if multiple cores are available. But it can also run on a single core. Or on a subset of available cores, depending on things like heat, the amount of available processing cycles, etc. Even if all your apps were running on all your cores (show us a screen shot, by the way - my safari currently has 6 threads, which is fewer than the number of cores), that doesn’t mean the time it takes to complete any process has anything to do with multi core performance (which, you’ll note, is what I said). Things run in parallel, but that doesn’t mean your work gets done any faster. Once again: the time it takes to run tasks, for almost ALL tasks, is a function of single-core, not multi-core, performance. Very few tasks are highly parallelizable.
I have to admit part of me (the petty part) is annoyed that Apple didn’t release the new MacBook Pros just to shut this particular line of argumentation down (it still slower than a threadripper in multi core!
) so we can go back to what really matters: chess benchmarks 
Let's get real there...
Special compiles of Stockfish, cFish, etc. are already optimizing compiles for using M1:s NEON, etc. But they are still less than half the speed of similar priced, similar-sized, computers. The fact of the matter is that M1 isn't that fast as the usual influencer-types (fanbois) make it out to be. A similar priced modern CPU from AMD runs circles around it. For certain use-cases, it may be "ok" for its "watt" but let's keep it real.. the CPU is faster on Apple PowerPoint.-presentations than it is in real-life performance. It's more or less a glorified std. ARM big-little phone CPU with a focus mainly on the low-power slow "little"-cores and relies heavily on optimized code to even be comparable to intel/AMD these days. The CPU is overrated and underperforming.. (not only for chess). Just compare it with amazing new stuff like the AMD 5700G and realize that anyone looking for "real" performance of CPU+GPU for the dollar should look elsewhere than the fruity un-open company these days.
Apple M1 chip is much better at everything but very slow in chess · Issue #3529 · official-stockfish/Stockfish
What are the reasons? How many things in detail cause this slowdown? -M1 chip -macOS Big Sur -Stockfish engine (source code) Any ideas how to fix the one or another problem? Would it be possible to...github.com
No AVX2
No AVX512 (VNNI)
No Hyperthreading
…
…
…
How to compensate the losses?
Hyperthreading runs 2 schedules on one cpu to keep it busy. For a lot of compute bound parallel workloads, you shouldn't see a big difference as one thread may keep a given core sufficiently busy.
AVX512 has a lot of implementation bugs. It's a nice interface, but it's under-utilized as a result of this. It tends to push downclocking to a degree, and unaligned memory access becomes expensive compared to AVX2, as any misalignment splits cache line boundaries.
On ARM, you use Neon rather than AVX or AVX2. They have differences in handling of predication and one is wider than the other, but the purpose is the same. AVX2 provides fewer register names, but it works well for cases where the compiler can fold single use loads into an op.
Stockfish may not be fully tuned for some of the differences, but this doesn't make it a bad architecture.
I would also point out, auto-vectorization on Clang has come a long way, so not all of this may need to be explicitly rewritten in coming years. We're probably within five years of my never bothering with intrinsics again, based on the current state of research. Most of the active work has a about a 5 year gap to implementation in a stable branch, and stuff coming in 2016 and later is pretty great in that regard.
Well, I certainly hope that a desktop Rocket Lake chip would outperform Apple‘s M1. Otherwise it would be really embarrassing for Intel.
- Status