Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

twitter.com

Post reply on HN
Print view View original

411–420 of 529 posts

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#411

Post by unsigner »

vgchh wrote: post #88
cercatrova wrote: post #26

Is anyone else thinking, what the f*ck? Are we in a new era of computing? It certainly feels that way when looking at these desktop class ARM chips, where performance doubled every year or so, just like back in the 80s and 90s.

This is the Apple equivalent of a Tesla showing up and out competing much more expensive cars on the performance. It feels like a huge disruption and probably Apple’s chance to gain significant PC market share.

The difference being that Apple can actually manufacture the numbers needed to make a difference.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#412

Post by jonfw »

egao1980 wrote: post #180
jonfw wrote: post #115

Earlier quoted context omitted.

Given the build quality, screen, performance, and battery life- Apple isn't really premium pricing anymore. The Dell XPS 13 starts at 999, the Lenovo X1 carbon starts at 949- at equivalent pricing- and the new Air outperforms both of those.

HP x360 13 with Ryzen - 799 and you have a 2-in-1 that you can use at work and play recent games at home. Install Docker and run Linux if you like that option.

That's a 1080p display and only $200 difference... still very comparable value propositions IMO.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#413

Post by Symmetry »

gumby wrote: post #184
whalesalad wrote: post #48

I love everything about the move to Apple Silicon with the exception of the decision to put memory on-die or in-package (not sure how it is configured). They call it 'Unified Memory'. It makes a lot of sense but I don't know if they are going to be able to pack enough memory in there. A lot of folks are fixated on CPU performance lately (which is rad) but I think that there is a tendency to ignore memory. I have 32gb…

> memory on-die or in-package (not sure how it is configured). They call it 'Unified Memory' It’s in the package; RAM on the die is called “cache”. “Unified memory” has nothing to do with packaging. It’s the default for how computer memory has worked since, well, the 1950s: all the parts talk to the same pool of memory (and you can DMA data for any device). That’s why the term of art for, say, GPUs having their own m…

In a NUMA system all the memory is in the same address space even if it's faster for a core to talk to some places than other. Traditionally GPUs work in a completely different address space and doesn't use virtual memory. Yes you can DMA to it but if you DMA a pointer it will break on the other end.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#415

Post by soganess »

naikrovek wrote: post #407
soganess wrote: post #406

This was pretty sparse on details. Can anyone verify they were compiling for the same target? (IE at least one of them was cross-compiling).

Yeah. Aren't Rust compiles IO heavy? Like really IO heavy?

That is a pretty good point. IOPS may have a hand in this as well. That said the benchmarks on the SSD seem similar.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#416

Post by Symmetry »

wtallis wrote: post #373
heisenbergs wrote: post #141

Earlier quoted context omitted.

According to anandtech the memory throughput is off the charts at 68.25GB/s [1]. That's twice as fast as high speed ddr4 memory (DDR4-4000 at 32GB/s). In other words: they totally trounced and took it to the next level with regards to memory, because they can. If anything, memory control is their biggest advantage. Scaling the amount of ram won't be an issue. Increasing the bandwidth perhaps, but it'll still be way q…

It's twice the speed of one module of high-speed DDR4 memory. Mainstream consumer PC platforms all support dual-channel memory. Dual-channel DDR4-4266 would provide the same theoretical bandwidth as the M1's 128-bit wide collection of LPDDR4X-4266. Intel's LPDDR support has been lagging far behind what mobile SoCs support (largely because of Intel's 10nm troubles), but their recently-launched Tiger Lake mobile proces…

Just to be pedantic DDR4-4266 is non-standard and so won't be found in any mainstream OEM's laptops. LPDDR4X-4266, soldered to the board instead of socketed and with a lower voltage, is indeed an official thing though.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#417

Post by jjtheblunt »

whalesalad wrote: post #48

I love everything about the move to Apple Silicon with the exception of the decision to put memory on-die or in-package (not sure how it is configured). They call it 'Unified Memory'. It makes a lot of sense but I don't know if they are going to be able to pack enough memory in there. A lot of folks are fixated on CPU performance lately (which is rad) but I think that there is a tendency to ignore memory. I have 32gb…

Maybe ARM binaries are smaller than x86_64?

ARM has some sophistication in its ISA, THUMB mode instructions for example, that might be in play here.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#418

Post by qayxc »

gigel82 wrote: post #107
tomduncalf wrote: post #7

Earlier quoted context omitted.

TechCrunch give figures for compiling WebKit on the 13" Pro in their review: https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro... - the gains are not as drastic, but it is still faster than the 16" and as fast as a Mac Pro, while using hardly any battery. Very impressive stuff. I only got a new 16" last year but a lot of my time is spent compiling, I might find it hard to resist upgrading next yera when th…

This feels fishy somehow; we need a more thorough in-depth look at exactly what is being compiled in both instances (I've seen projects that build half the object files on different architectures). It's likely not a 1:1 comparison on actual amount of code being compiled and linked. It will be awesome if accurate, but I think we need more data.

The results check out - code compilation is very sensitive to cache and the M1 has about 4 times more L1 cache per core than Intel CPUs.

The i7 9750H has 32kb data- and 32kb instruction cache plus 256kb L2 cache per core.

The M1 has 192kb instruction cache and 64kb data cache per core as well as 12MB L2 cache (shared) for the high performance cores.

This alone is enough to explain significantly better code compilation performance.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#419

Post by tootie »

rsynnott wrote: post #175
tootie wrote: post #18

Nearly universal praise for the M1 versus it's predecessor but I don't see how this really changes Apple's position the way he implies. Their machines are still far more expensive than Windows and can't use high-powered GPUs like AMD/Nvidia. The people who choose PC, still have the same incentives to do so. For most users, the faster performance will probably not be very noticeable. The reduced power consumption is t…

> Their machines are still far more expensive than Windows Which $1000 Windows laptop has this performance? > and can't use high-powered GPUs like AMD/Nvidia. Not relevant to all markets (particularly ultra bookish laptops, which never have discrete GPUs, but more broadly many people just don't need one). I don't think it changes anything much immediately, but it's extremely bad news for Intel in the long run.

The difference is that you can buy something specced to your needs. Most users and even most developers won't be bottlenecked by even an i5. Getting a quad-core Windows/Intel laptop with a 1080 screen, 16GB and upgrade slots for RAM or disk, plus all the output ports for an external monitor is easily under $1000 for even a name brand. My $800 Acer does fine and has an onboard ethernet port because my most resource-intensive application that I use daily is Zoom. And it has an entry-level Nvidia card.

Re: 16-inch MBP 2x slower than M1 MacBook Air in a real-world Rust compile

#420

Post by bgorman »

the8472 wrote: post #369
JeremyNT wrote: post #269

Earlier quoted context omitted.

I can't help but chuckle whenever I read these comments on HN about how doomed Intel is. Intel just does this . Every now and then, they get so far ahead that the rest of the market just totally disintegrates, which allows them to screw around and juice up their margins while failing to actually innovate. Their brand is so strong it takes years for it to erode, even when they do suck, and when they have actual compet…

14nm was delayed, 10nm was massively delayed, 7nm is delayed again. That's a fairly long run of underperformance. Sure, they sit on enough money that they can recover if they do it right. But there currently aren't any signals that they're getting things right.

Lots of executivees have been fired and Intel has a new CEO. That is certainly a sign that things will... be different.
Post reply on HN