4 ms·

Speed is still relative. Yes the benchmark numbers for the M1 are impressive, but I am curious to see non-benchmark, real-world workloads compared. I'm remind

by mikece 6y ago

Speed is still relative. Yes the benchmark numbers for the M1 are impressive, but I am curious to see non-benchmark, real-world workloads compared. I'm reminded of a YouTube video (link below) in which the comparison was made between an $11,000 Mac Pro and a PC kit that also cost $11,000. In this case the PC was rocking a Ryzen chip with 32 (or was it 64?) cores and smoked the Mac Pro on every benchmark. But when comparing comparable workloads, like doing a final render on a large video project using Final Cut Pro on the Mac and Premiere Pro on the PC, the Mac Pro did narrowly edge the PC. So I would be interested in seeing the M1 MBP do side-by-side, real-life workload tests like final-rendering a large video project, compiling a multi-gigabyte XCode project, applying a complex set of filters to an image, etc. Benchmarking is nice to an extent but I think lacks real-world significance in this context. https://www.youtube.com/watch?v=jzT0-t-7-PA https://www.youtube.com/watch?v=jzT0-t-7-PA

rimliu 6y ago
Well, see this e.g.: https://www.youtube.com/watch?v=HxH3RabNWfE https://www.youtube.com/watch?v=HxH3RabNWfE
- mikece 6y ago
  Awesome -- that's what I'm looking for... but now I wish there was some explanation as to why a $699 machine is doing things that previously you needed a $4000 machine to do. I look forward to learning more!
  - doublepg23 6y ago
    Anandtech had some theories based on the A14s design. https://www.anandtech.com/show/16226/apple-silicon-m1-a14-deep-dive https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de...
OberstKrueger 6y ago
TechCrunch's review includes a compile of WebKit, Final Cut export, as well as some traditional benchmarks. It includes performance and battery impact of the tasks. https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro-is-powerful-but-its-the-battery-life-that-will-blow-you-away/ https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro...
mpweiher 6y ago
WebKit compile / speed / power: https://images.macrumors.com/t/xjpsPEkXCCuNsW90kwPvVl0xfC0=/1600x0/filters:quality(90)/article-new/2020/11/WebKit-Compile-Battery-Remaining.jpg https://images.macrumors.com/t/xjpsPEkXCCuNsW90kwPvVl0xfC0=/... Percent battery remaining after compiling WebKit. 16" MacBookPro 2019: 61%, 13" MacBook Pro: 24%. M1 mini: 91%, M1 Air: 91%. From: https://www.macrumors.com/2020/11/17/apple-silicon-m1-compiles-code-as-fast-as-mac-pro/ https://www.macrumors.com/2020/11/17/apple-silicon-m1-compil...
- philliphaydon 6y ago
  M1 mini doesn’t have a battery. I assume you mean the M1 MacBook?
  - mpweiher 6y ago
    Yep, MBP, sorry.
- MengerSponge 6y ago
  First, that's absolutely bonkers. Second, I think you've got a copy-paste error somewhere... I thought the mini was mains-powered. What does 91% battery mean there?
  - mpweiher 6y ago
    MBP, sorry.
- varispeed 6y ago
  Is battery consumption linear though? For example I can use my phone for half a day and battery will go down to 50%, then the remaining 50% lasts for maybe an hour.
sz4kerto 6y ago
SPEC benchmarks are your friend. > non-benchmark, real-world workloads compared Many (most) benchmarks are designed to reflect real-world workloads.
- markonen 6y ago
  Sure, but not macOS workloads. What most (all?) of these benchmarks fail to capture is the difference made by the specific optimizations for Apple's software stack. Things like the five times faster ARC retain/release cycle or the dedicated objc_msgSend branch predictor are a big deal when your software stack hits these hotspots constantly. The SPEC sample workloads won't.
ricw 6y ago
So you want someone to compare a starting at $699 m1 Mac to a $10k top of the line pro machine? That you are even interested in this shows how great the m1 chip performs, but as a comparison This is pretty useless.
- mikece 6y ago
  No, more like comparing the M1 MacBook Pro to the i7 (or i9?) based MacBook Pro. Comparing the $699 Mac mini to the Mac Pro seems absurd... or at least I thought so until someone else replied with a link to a video where the video-dude is comparing the mini to a $4000 machine: https://youtu.be/HxH3RabNWfE?t=626 https://youtu.be/HxH3RabNWfE?t=626
bstar77 6y ago
I saw that (snazzy guy video?) as well. Thing is, that video was explaining how a highly integrated system can perform better when software optimizes for it. These Macs will only see better performance as more software is optimized over time.

by mikece 6y ago

rimliu 6y ago

Well, see this e.g.: https://www.youtube.com/watch?v=HxH3RabNWfE https://www.youtube.com/watch?v=HxH3RabNWfE

mikece 6y ago
Awesome -- that's what I'm looking for... but now I wish there was some explanation as to why a $699 machine is doing things that previously you needed a $4000 machine to do. I look forward to learning more!
- doublepg23 6y ago
  Anandtech had some theories based on the A14s design. https://www.anandtech.com/show/16226/apple-silicon-m1-a14-deep-dive https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de...

OberstKrueger 6y ago

TechCrunch's review includes a compile of WebKit, Final Cut export, as well as some traditional benchmarks. It includes performance and battery impact of the tasks. https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro-is-powerful-but-its-the-battery-life-that-will-blow-you-away/ https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro...

mpweiher 6y ago

WebKit compile / speed / power: https://images.macrumors.com/t/xjpsPEkXCCuNsW90kwPvVl0xfC0=/1600x0/filters:quality(90)/article-new/2020/11/WebKit-Compile-Battery-Remaining.jpg https://images.macrumors.com/t/xjpsPEkXCCuNsW90kwPvVl0xfC0=/... Percent battery remaining after compiling WebKit. 16" MacBookPro 2019: 61%, 13" MacBook Pro: 24%. M1 mini: 91%, M1 Air: 91%. From: https://www.macrumors.com/2020/11/17/apple-silicon-m1-compiles-code-as-fast-as-mac-pro/ https://www.macrumors.com/2020/11/17/apple-silicon-m1-compil...

philliphaydon 6y ago
M1 mini doesn’t have a battery. I assume you mean the M1 MacBook?
- mpweiher 6y ago
  Yep, MBP, sorry.
MengerSponge 6y ago
First, that's absolutely bonkers. Second, I think you've got a copy-paste error somewhere... I thought the mini was mains-powered. What does 91% battery mean there?
- mpweiher 6y ago
  MBP, sorry.
varispeed 6y ago
Is battery consumption linear though? For example I can use my phone for half a day and battery will go down to 50%, then the remaining 50% lasts for maybe an hour.

sz4kerto 6y ago

SPEC benchmarks are your friend. > non-benchmark, real-world workloads compared Many (most) benchmarks are designed to reflect real-world workloads.

markonen 6y ago
Sure, but not macOS workloads. What most (all?) of these benchmarks fail to capture is the difference made by the specific optimizations for Apple's software stack. Things like the five times faster ARC retain/release cycle or the dedicated objc_msgSend branch predictor are a big deal when your software stack hits these hotspots constantly. The SPEC sample workloads won't.

ricw 6y ago

So you want someone to compare a starting at $699 m1 Mac to a $10k top of the line pro machine? That you are even interested in this shows how great the m1 chip performs, but as a comparison This is pretty useless.

mikece 6y ago
No, more like comparing the M1 MacBook Pro to the i7 (or i9?) based MacBook Pro. Comparing the $699 Mac mini to the Mac Pro seems absurd... or at least I thought so until someone else replied with a link to a video where the video-dude is comparing the mini to a $4000 machine: https://youtu.be/HxH3RabNWfE?t=626 https://youtu.be/HxH3RabNWfE?t=626

bstar77 6y ago

I saw that (snazzy guy video?) as well. Thing is, that video was explaining how a highly integrated system can perform better when software optimizes for it. These Macs will only see better performance as more software is optimized over time.