M3 core counts and performance

[Jimmyjames]

Don‘t come at me with pitchforks if it doesn’t live up to expectations

Too late. Pitchfork purchased…just in case!

 

[theorist9]

Don‘t come at me with pitchforks if it doesn’t live up to expectations

We plan to hold you personally responsible for the hardware. That includes the colors.

 

[rdrr]

We plan to hold you personally responsible for the hardware. That includes the colors.

Especially if that color is 1980 IBM beige.

 

[leman]

I’ll make sure to underfeed my cats so that they are extra hungry and angry then. We’ll be ready for you mob

Some other potential info, likely to be included in M3 series:

1. New branch prediction mechanism aimed at improving energy usage, die area, and suggestive of wider core overall (explanation from Maynard Handley)

2. Error corrected RAM using LPDDR5: https://patentscope.wipo.int/search/en/detail.jsf?docId=US403904970&_cid=P12-LLBQX0-80508-2

If these patents really describe M3 (and the timing suggests they might be), it’s shaping to be an extremely exiting release!

 

[B01L]

Some other potential info, likely to be included in M3 series:

1. New branch prediction mechanism aimed at improving energy usage, die area, and suggestive of wider core overall (explanation from Maynard Handley)

2. Error corrected RAM using LPDDR5: https://patentscope.wipo.int/search/en/detail.jsf?docId=US403904970&_cid=P12-LLBQX0-80508-2

If these patents really describe M3 (and the timing suggests they might be), it’s shaping to be an extremely exiting release!

M3 Ultra with hardware ray-tracing, ECC RAM, & (possibly) double the PCIe bandwidth would be a much more compelling Mac Pro, even more so if Apple manages to produce a M3 Extreme variant...?!?

 

[theorist9]

M3 Ultra with hardware ray-tracing, ECC RAM, & (possibly) double the PCIe bandwidth would be a much more compelling Mac Pro, even more so if Apple manages to produce a M3 Extreme variant...?!?

It's been more-or-less confirmed that there will not be any Extreme variant of the M3 (and likely not through the M5) (see post #88):

Samsung 5k 27" & Dell 6k 32" monitors announced at CES

At that price point, I don't see why anyone would purchase the Samsung over the Apple monitor. Same here. Personally, I haven't found the Studio Display stand limiting. Maybe if I were 6'4" in height. And, I prefer the crispness of a non-matte display (I can somewhat control the lighting in...

techboards.net

 

[KingOfPain]

I’ll make sure to underfeed my cats so that they are extra hungry and angry then. We’ll be ready for you mob

Now that's just cheating! Should give us at least some kind of fighting chance...

 

[dada_dave]

@leman ’s cats when they’re hungry … or anytime?

You’re not you when you’re hungry joke? Or in this case you’re more you when you’re hungry?

 

[Yoused]

Ahh, Tiger, I remember it fondly. It was smooth and clean and still could run Classic. Leopard was sort of better, but only in small ways, and worse in some.

 

[leman]

@leman ’s cats when they’re hungry … or anytime?

View attachment 25371

You’re not you when you’re hungry joke? Or in this case you’re more you when you’re hungry?

Well, right now they are royally pissed since we’ve been in a car for last 15 hours…

 

[throAU]

any news on whether they're going to be including nested virtualisation capabilities in M3?

Not that I'm currently hurt too much by the lack of it, but it would be nice to be able to run VMs inside of VMs which unless I'm mistaken currently isn't possible. Which means you can't run things like QEMU inside of Linux VMs (e.g., GNS3 appliance locally - or to just run network device emulation inside a regular install of Linux in a VM).

Its one of the few things X64 can do right now that my MacBook Pro can't.

 

[dada_dave]

any news on whether they're going to be including nested virtualisation capabilities in M3?

Not that I'm currently hurt too much by the lack of it, but it would be nice to be able to run VMs inside of VMs which unless I'm mistaken currently isn't possible. Which means you can't run things like QEMU inside of Linux VMs (e.g., GNS3 appliance locally - or to just run network device emulation inside a regular install of Linux in a VM).

Its one of the few things X64 can do right now that my MacBook Pro can't.

Apple introduced that in the M2. macOS simply doesn’t support it yet. Asahi is introducing support for it.

Hector Martin (@marcan@treehouse.systems)

Did you know Asahi Linux is introducing support for some brand new Apple Silicon features faster than macOS? * The M1 has a virtual GIC interrupt controller for enhanced virtualization performance. Linux supports it, macOS does not. * The M2 introduced Nested Virtualization support. The patches...

social.treehouse.systems

 

[Jimmyjames]

Apple introduced that in the M2. macOS simply doesn’t support it yet. Asahi is introducing support for it.

Hector Martin (@marcan@treehouse.systems)

Did you know Asahi Linux is introducing support for some brand new Apple Silicon features faster than macOS? * The M1 has a virtual GIC interrupt controller for enhanced virtualization performance. Linux supports it, macOS does not. * The M2 introduced Nested Virtualization support. The patches...

social.treehouse.systems

Interesting.

I am probably in a minority here, but I find Hector’s “macOS is shit, have you considered Linux?” thing incredibly annoying. He also has some very strange views on file systems. Rant over.

 

[casperes1996]

I am probably in a minority here, but I find Hector’s “macOS is shit, have you considered Linux?” thing incredibly annoying. He also has some very strange views on file systems. Rant over.

I've watched som of his live development streams, but don't really come across his writing much. What thoughts on file systems are you referring to?
In his livestreams he generally doesn't give off a "macOS is shit" vibe. He points out interesting or odd design decisions. Often gives ideas for why those odd decisions may have been made. But any system has weird decisions. Linux too. It comes with project age. The Asahi team has done a lot of great stuff for the Apple community and are spreading the word on how good the hardware is.
And getting info about the hardware's capabilities like this is just good for everyone. It informs us what future macOS releases may give us as well.

 

[Jimmyjames]

I've watched som of his live development streams, but don't really come across his writing much. What thoughts on file systems are you referring to?
In his livestreams he generally doesn't give off a "macOS is shit" vibe. He points out interesting or odd design decisions. Often gives ideas for why those odd decisions may have been made. But any system has weird decisions. Linux too. It comes with project age. The Asahi team has done a lot of great stuff for the Apple community and are spreading the word on how good the hardware is.
And getting info about the hardware's capabilities like this is just good for everyone. It informs us what future macOS releases may give us as well.

I have never seen a single positive thing he has said about macOS.

With regard to file systems, I’ll try and find the tweet, but he said that apfs, zfs, btrfs and pretty much all local, modern file systems are terrible. He said he runs some bizarre distributed fs, designed for multiple servers, as his local fs. I don’t pretend to know anything like as much as him concerning hardware, or low level stuff, but I do know modern copy on write file systems pretty well. He views are…interesting. And that’s as polite as I can put it.

Edit: still looking for the tweet. If I recall the discussion started after he berated the apple fs team for some decision about write syncs in the firmware of the ssds. After a huge discussion with many people, it emerged that he misunderstood how much of this stuff works. He never admitted that of course.

Edit2: Sadly all tweets are deleted. Fortunately a quote was captured of the kind of thing he stated. The apfs having widespread corruption being particularly nonsensical.

 

[casperes1996]

I have never seen a single positive thing he has said about macOS.

With regard to file systems, I’ll try and find the tweet, but he said that apfs, zfs, btrfs and pretty much all local, modern file systems are terrible. He said he runs some bizarre distributed fs, designed for multiple servers, as his local fs. I don’t pretend to know anything like as much as him concerning hardware, or low level stuff, but I do know modern copy on write file systems pretty well. He views are…interesting. And that’s as polite as I can put it.

Edit: still looking for the tweet. If I recall the discussion started after he berated the apple fs team for some decision about write syncs in the firmware of the ssds. After a huge discussion with many people, it emerged that he misunderstood how much of this stuff works. He never admitted that of course.

Edit2: Sadly all tweets are deleted. Fortunately a quote was captured of the kind of thing he stated. The apfs having widespread corruption being particularly nonsensical.
View attachment 25380

The bugs he's talking about there were real issues. Apple addressed (at least some of) them. I remember specifically there was something about the usage of diskutil and bless for preparing the Asahi boot partition that needed some weird workaround that got fixed at some point.
Relatively widespread APFS corruptions does come off strange. I guess the word 'relatively' could be pulling a lot of weigh there Though I will believe the statement if what he means is that, when preparing the Asahi partition, bugs in diskutil or similar caused corruptions in APFS that are not the fault of APFS but the fault of direct disk writing software breaking things.

 

[Jimmyjames]

The bugs he's talking about there were real issues. Apple addressed (at least some of) them. I remember specifically there was something about the usage of diskutil and bless for preparing the Asahi boot partition that needed some weird workaround that got fixed at some point.
Relatively widespread APFS corruptions does come off strange. I guess the word 'relatively' could be pulling a lot of weigh there Though I will believe the statement if what he means is that, when preparing the Asahi partition, bugs in diskutil or similar caused corruptions in APFS that are not the fault of APFS but the fault of direct disk writing software breaking things.

I can’t recall exactly because all his tweets are gone, but because file systems are a particular interest, I paid interest. IIRC, he claimed actual widespread corruption in apfs. Also he claimed it in zfs, which is even stranger given its stellar reputation even in the Linux community, and its widespread use of checksums. I’ll just say it: there is no widespread corruption in either apfs or zfs and he damages his credibility by stating such things.

 

[dada_dave]

I have never seen a single positive thing he has said about macOS.

With regard to file systems, I’ll try and find the tweet, but he said that apfs, zfs, btrfs and pretty much all local, modern file systems are terrible. He said he runs some bizarre distributed fs, designed for multiple servers, as his local fs. I don’t pretend to know anything like as much as him concerning hardware, or low level stuff, but I do know modern copy on write file systems pretty well. He views are…interesting. And that’s as polite as I can put it.

Edit: still looking for the tweet. If I recall the discussion started after he berated the apple fs team for some decision about write syncs in the firmware of the ssds. After a huge discussion with many people, it emerged that he misunderstood how much of this stuff works. He never admitted that of course.

Edit2: Sadly all tweets are deleted. Fortunately a quote was captured of the kind of thing he stated. The apfs having widespread corruption being particularly nonsensical.
View attachment 25380

I can’t speak to the file system issue but he has definitely said nice things about macOS. He praises its security model (like seriously he lavishes praise on how well macOS designed its security model, not saying it’s impenetrable or anything but how well they balanced security and flexibility) and kernel decisions in particular. Some drivers/firmware Apple wrote he likes, some he found interesting but flawed, and at least one he downright hated (to be fair seemingly with reason in the case of the display driver). Much of the criticism he levels at macOS he levels at Linux (bugs/missing features) and he does so mostly in the context of responding to people claiming Linux will never run well on Apple Silicon because reasons (mostly that macOS only runs well because it was designed in conjunction with the silicon, the silicon is actually crap, Apple sucks - a narrative he often pushes back against it, etc …) . He obviously is proud when his team does something he feels is better than the “native” version. I think that’s understandable.

I would argue that he criticizes Linux and its ecosystem way more than macOS - probably because he has to deal with its problems far more often. Thinking about it further and definitely the number of sheer broken things he rants about desktop Linux is waaay higher - I repost very little of it here - but many more of his posts are about how standard desktop Linux and its ecosystem is fundamentally broken on most hardware and how resistant the community is to changing.

 

[leman]

I also have so far perceived Hector as a knowledgeable expert who shares deep insights without suffering from particular bias (unlike many others). Of course, one has to keep in mind that he is a security hacker with interest in low-level stuff, so those are the things he particularly focuses at; they might be less relevant or interesting to most people and some of what he says can be perceived as nitpicking for the same reason.

and at least one he downright hated (to be fair seemingly with reason in the case of the display driver).

I think the display controller driver design is very cool, and the GPU driver probably is designed in a way that's not too different. I understand why Asahi team would hate it, as it makes reverse-engineering drivers and communicating with the firmware a total nightmare, but for Apple this sounds like a great development model.

 

[Jimmyjames]

I can’t speak to the file system issue but he has definitely said nice things about macOS. He praises its security model (like seriously he lavishes praise on how well macOS designed its security model, not saying it’s impenetrable or anything but how well they balanced security and flexibility) and kernel decisions in particular. Some drivers/firmware Apple wrote he likes, some he found interesting but flawed, and at least one he downright hated (to be fair seemingly with reason in the case of the display driver). Much of the criticism he levels at macOS he levels at Linux (bugs/missing features) and he does so mostly in the context of responding to people claiming Linux will never run well on Apple Silicon because reasons (mostly that macOS only runs well because it was designed in conjunction with the silicon, the silicon is actually crap, Apple sucks - a narrative he often pushes back against it, etc …) . He obviously is proud when his team does something he feels is better than the “native” version. I think that’s understandable.

I would argue that he criticizes Linux and its ecosystem way more than macOS - probably because he has to deal with its problems far more often. Thinking about it further and definitely the number of sheer broken things he rants about desktop Linux is waaay higher - I repost very little of it here - but many more of his posts are about how standard desktop Linux and its ecosystem is fundamentally broken on most hardware and how resistant the community is to changing.

Oh I accept mine is the minority opinion!