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

Apple’s M1 Positioning Mocks the Entire x86 Business Model

extremetech.com

Post reply on HN
Print view View original

871–880 of 942 posts

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#871

Post by to11mtm »

dmos62 wrote: post #275
XorNot wrote: post #266

Earlier quoted context omitted.

Keep in mind this is all fallout from the 90s/really 00's where "clockspeed was king" which AMD worked hard to disabuse the market of. We've never really recovered from that.

Can you elaborate? I do remember clockspeed being the main thing you looked at and then being surprised that changed, but didn't really have any insight.

As others mentioned, Pipelining resulted in scenarios where Other manufacturer's CPUs had a lower clock rate but 'comparable' performance.

There's two main Eras of this;

During the P5/P6 Days, AMD, Cyrix, and NexGen made CPUs with 'PR' ratings, based on what they felt their CPU's compared to.

Ironically, this first era is probably why folks got so soured on 'PR ratings'; As far as the AMD K5 and Cyrix 6x86 went, These numbers were based more on integer performance, additionally Intel's P5 had a very novel (at the time) pipeline that some Game developers were optimizing for (Quake comes to mind here.) For NexGen the situation was even worse, in that some of their models completely lacked an FPU.

All those factors together made consumers a bit more wary of PR ratings for quite a long time.

Thankfully, AMD Bought out NexGen, took their arch and made it into the K6, which was very competitive with the P5 Clock for clock, and PR ratings went away.

They came back in the days of the Palomino K7s, but what a lot of people might not remember is that the Athlon XP's PR ratings were technically supposed to relate to a Thunderbird core. IOW, an Athlon XP 1800 was supposed to be 'equivalent' to an TBird Athlon running at 1800Mhz.

But, of course, PR ratings drifted again, as they tended to... and now we are in model number hell.

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#872

Post by dilippkumar »

slver wrote: post #713
maerF0x0 wrote: post #645

Earlier quoted context omitted.

https://scicomp.stackexchange.com/a/1395 Explains it better than I could in a HN reply. Assuming our desired work to be done continues to grow, faster core speed* will eventually matter. * technically, net instructions per second after prediction/compute ahead etc

Honestly that's a bit too abstract to make sense of. As a programmer to another, I'd rather ask... what's one example of a problem we have today that needs faster linear performance than our best chips (not in a nice-to-have way but in a must-have way). I'd rule out all casual computing like home PCs, smartphones, and so on, because honestly we've been there for years already. Also due to decades of bias we have seri…

> I'd rule out all casual computing like home PCs, smartphones, and so on, because honestly we've been there for years already.

Casual computing can definitely be a lot better than where we are today[0][1].

The software business has moved to a place where it’s not really practical to program bare metal silicon in assembly to get screaming fast performance. We write software on several layers of abstraction, each of which consumes a ton of compute capacity.

We have resigned to live with 100ms latencies in our daily computing. This is us giving up on the idea of fast computers. It should not be confused with actually having a computer where all interactions are sub 10ms (less than 1 frame refresh period at 90fps).

[0]. https://danluu.com/input-lag/

[1]. https://www.extremetech.com/computing/261148-modern-computer...

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#873

Post by Frondo »

spurgu wrote: post #826
spurgu wrote: post #816

Earlier quoted context omitted.

Sure, I've never used Safari enough myself to have an opinion (since it's never been an option for me due to various things), it might be equal to Chrome or even better (I like Safari's clean UI). I'm mostly comparing it to Firefox which is quite clunky in my opinion (I've given it many chances). And I'm actually typing this on Safari now and it looks like the issue has been fixed - Bitwarden now works in private mod…

Whoa. Maybe an hour into testing Safari and I look at Activity Monitor and see that https://calendar.google.com is consuming 1 GB of RAM and https://docs.google.com 865 MB. :o Don't know if the numbers are comparable but in Chrome's Task Manager the numbers are 224 MB and 121 MB respectively. Edit: This might be an old (unfixed) issue https://discussions.apple.com/thread/6640430 ... This sucks because I have a specif…

I recently got an M1 mac myself and tried Safari out again -- I seldom used it on my 2016 Pro (preferred Firefox) but it's supposed to have stellar battery life, so why not give it a go?

Nice that they've got built-in tracking protection now, I guess, but I left Twitter running in a tab overnight and when I woke up Safari was reporting over a gb of memory use for that tab alone. Something in how they cache or handle JS for some of these long-running services, or maybe having to do with service workers ?, Safari just seems to consume a lot of memory. It's not a huge issue I guess, with the way the M1 never seems to have any memory pressure issues, but it's part of why I remember switching off Safari in the first place.

If anyone knows why Safari seems to consume more memory with these long-running processes, I'd be real curious to know why...

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#874

Post by hajile »

whizzter wrote: post #98
morpheuskafka wrote: post #64

Earlier quoted context omitted.

That struck me as an extremely odd metric to differentiate products by, given its low relevance to non-technical users who don't know what a GPU core even is (except gamers, but they are not buying iMacs). Additionally, most people are going to think adding one more core is hardly worth the price upgrade, and its quite strange to see an odd number of cores.

Depends, remember that the 5nm processors overall are fairly new so we have no idea of yields. But if 1/5th of the processors are flawless, 1/5th have one faulty GPU and the rest have more errors (in either CPU or two or more GPU failures) then having twice the amount of CPU's available to sell (at a slightly lower pricepoint) might make perfect sense.

TSMC claimed last year that 5nm yields are better than 7nm.

https://www.anandtech.com/show/16028/better-yield-on-5nm-tha...

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#875

Post by hajile »

londons_explore wrote: post #265
kingosticks wrote: post #141

Earlier quoted context omitted.

> I don't think they have the fab capacity to be able to do that Do you have a source for there being any fab capacity shortage for Apple (or anyone, for that matter) at 5nm? Older nodes are a different story because that's still where most of TSMC's customers are doing their high-volume production.

Plenty of people would like to die-shrink their 7nm designs down to 5nm if only capacity existed.

SRAM scaling from 7nm to 5nm is pitiful. While the CPU transistors see 50-70% increases, the SRAM is only shrinking 20-30%.

For chips with massive cache, that isn’t super cost effective (I suspect as a cost saving measure that we’ll see L2/L3 cache moving to a separate chip on a larger process while the rest shrinks down).

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#876

Post by spurgu »

spurgu wrote: post #816
mvanbaak wrote: post #814

Earlier quoted context omitted.

That’s a personal preference thing. Personally I think chrome UI/UX is horrible and slow. Ut I’m a safari user so again: personal preference.

Sure, I've never used Safari enough myself to have an opinion (since it's never been an option for me due to various things), it might be equal to Chrome or even better (I like Safari's clean UI). I'm mostly comparing it to Firefox which is quite clunky in my opinion (I've given it many chances). And I'm actually typing this on Safari now and it looks like the issue has been fixed - Bitwarden now works in private mod…

Meh.

New tabs are opening in weird places, really faaaar to the right skipping between 6-7 (unrelated) tabs. Really cumbersome to navigate.

Horrible selection of extensions still. Really missing searching for any tab with Cmd+Shift+K (I often have ~100 tabs open across multiple windows) that I get with the Tab Switcher extension.

There's a Noscript equivalent extension but it costs $3...

And I can't initiate a search with "url bar -> you[tab][enter]" (for a Youtube search).

Meh. Safari just doesn't work for me. Maybe if I sat down for a week or two and wrote a bunch of extensions myself but I don't see the value in that.

On the plus side Safari does seem snappy but then again I only have ~30 tabs open (across 6 windows) currently. So I might give it a win on speed but it definitely loses in overall usability.

Will try again in a year or so.

Edit: OH SHIT this is a definite dealbraker - apparently I can't select multiple tabs by Shift-clicking to pull them out into a new window. That completely kills my workflow of separating topics into different windows, and/or closing tabs in bulk.

Edit2: Wtf, how didn't I notice this earlier - I cannot see the URL when hovering over a link. How does anyone consider this within the realm of good UX?

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#877

Post by spurgu »

Frondo wrote: post #873
spurgu wrote: post #826

Earlier quoted context omitted.

Whoa. Maybe an hour into testing Safari and I look at Activity Monitor and see that https://calendar.google.com is consuming 1 GB of RAM and https://docs.google.com 865 MB. :o Don't know if the numbers are comparable but in Chrome's Task Manager the numbers are 224 MB and 121 MB respectively. Edit: This might be an old (unfixed) issue https://discussions.apple.com/thread/6640430 ... This sucks because I have a specif…

I recently got an M1 mac myself and tried Safari out again -- I seldom used it on my 2016 Pro (preferred Firefox) but it's supposed to have stellar battery life, so why not give it a go? Nice that they've got built-in tracking protection now, I guess, but I left Twitter running in a tab overnight and when I woke up Safari was reporting over a gb of memory use for that tab alone. Something in how they cache or handle…

Battery life has been the reason for me trying it in the past as well.

And yeah I mean even though Activity Monitor is reporting high RAM usage I don't seem to have any memory pressure issues so it might not be that big of a deal (MBP 2015 w/ 8 GB RAM).

I'm actually considering getting an M1 myself next week. Have been waiting for the initial kinks to get sorted out and most apps ported to it and I think it's time now. :) SSD wear seems to have been resolved with the latest Big Sur versions and fewer Rosetta apps and now even ST4 beta has an M1 build.

My justifications are that I'll get an excuse to upgrade to 16 GB RAM and I'll get touch ID so I can have a more secure password and not have to type it all the time. And with the stellar performance and battery life of the M1 I can just keep using Chrome (Brave). :P

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#878

Post by smt88 »

sillysaurusx wrote: post #652
bayindirh wrote: post #622

Earlier quoted context omitted.

I use an M1 powered 16GB MacBook Air and I've never seen any freezing, even under heavy load. I don't use Google Chrome though. Also, system is writing to the SSD with the same rate of my Linux machines: 4TB/year.

Heavy load in my case is a few dozen terminal tabs, six of which are running TPU training jobs, 40 chrome tabs, a few of which are playing twitch streams and YouTube videos, a half dozen instances of pycharm, one webstorm, one clion, around 35 macvim windows, and a few dozen PDFs open in Preview. I’m not exactly representative, but I think the stddev of usage patterns is high across developers. The unfortunate part i…

Apple doesn't and shouldn't care about optimizing for your use-case. At some point, the user should be responsible for stopping processes when they're not using them anymore.

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#879

Post by skohan »

foobiekr wrote: post #829
skohan wrote: post #821

Earlier quoted context omitted.

The worst case performance is very bad though. I have done some playing around with pushing Swift in terms of performance, and it's very hard because of ARC.

Code in the fastpath must not generate substantial garbage whether using ARC or a gc. That's just a coding error.

It's not about garbage creation, is about reference cost. With ARC, you pay a penalty every time you pass a reference type, not only for memory churn. You can avoid this using value types, but this often results in a lot of unnecessary copying.

It's not unavoidable, but Swift doesn't give good tools for avoiding ARC-based performance cliffs, so it takes a lot of profiling and effort compared to other languages.

Re: Apple’s M1 Positioning Mocks the Entire x86 Business Model

#880

Post by totalZero »

stevenwliao wrote: post #868
totalZero wrote: post #861

Earlier quoted context omitted.

I can't think of a technological product where that mentality applies. Can you? In fact, it seems that the opposite is true; as a product gets better, people care more about the specs. Whether you're buying a Wusthof knife or a luxury car, you want to know what makes the product good enough to justify its price and position in the market.

It's like how no one cares about which chip is in their smart TV, regardless of price range. They just want to watch Netflix without UI bugs.

TVs are sold on refresh rate, color accuracy, contrast, backlight localization, and panel resolution. This stuff is written on the box and promoted in marketing material. They are unavoidably differentiated on technical specs, even in the eyes of a layperson. Some of those specs depend on the panel (itself a semiconductor product) and others depend on processors/ICs.

I'm not suggesting that no parts can ever be a commodity (like capacitors in a laptop), but as I spend more, I increasingly tend to look for a technical advantage that justifies the marginal price.

I'll give you an example. If you want to buy a docking station for an Apple TB3 enabled MacBook, you have a couple of controller options at the higher end: Alpine Ridge and Titan Ridge. Better chips exist but they haven't found their way into truly well engineered consumer docks. I have a multi monitor setup with one superultrawide screen that can do 120Hz, so I opted for the Titan Ridge dock. It was buggy, so I ended up returning it and buying an Alpine Ridge dock that lacks the ability to push my big monitor at its best-looking resolution. And that's for a sub-$300 peripheral.

Like most of us here on HN, I'm one of the "few" that GP mentioned. But these are consumer products and they are mass marketed based on their technical specs.

Apple itself markets technical innovations in its 6k IPS monitors. The back is designed to dissipate heat, the monitor works with TB3 (ie, an Intel chip)...they even market the glass treatment in a deeply specific way.

Post reply on HN