Skip to content
 
If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(Ars Technica)   Intel: check out our new mobile processors' compute advantage over AMD's when running on batteries. Reality: advantage only for the first 10 or so seconds, then the Ryzen's clock throttles up, delivering superior computing while using less power   (arstechnica.com) divider line
    More: Fail  
•       •       •

417 clicks; posted to STEM » on 24 Nov 2020 at 7:50 AM (7 weeks ago)   |   Favorite    |   share:  Share on Twitter share via Email Share on Facebook



20 Comments     (+0 »)
 
View Voting Results: Smartest and Funniest
 
2020-11-24 6:22:08 AM  
AMD's approach makes sense.  I wonder how long before Intel has issues with CPUs dying due to heat.
 
2020-11-24 8:02:40 AM  

Merltech: AMD's approach makes sense.  I wonder how long before Intel has issues with CPUs dying due to heat.


Probably never. The bigger issue is thermal throttling, so the performance drops when the temps get too high.
 
2020-11-24 8:16:27 AM  
Somebody's getting desperate.  When does Intel start suing people?
 
2020-11-24 8:52:58 AM  
This is unsurprising after the presentation that they revealed this group of processors and mentioned AMD more than their own product:

Fark user imageView Full Size


For comparison, Nvidia didn't mention the competition at all in their reveal presentation, instead keeping it focused on their own products and showing the improvements over last gen.
 
2020-11-24 9:48:36 AM  
From some of the comments, it sounds like while this is an AMD standard reference design decision, manufacturers have the option of tweaking the delay? Maybe there could be a BIOS option or utility to adjust things if the user wanted a faster ramp-up. I mean, for the most part, it would only be noticeable in computer start-up (apparently, the Tiger Lake boot-up is insanely quick) and perhaps the initial loading of things like games (but not the actual play), but of course there are always people who will complain because their computer "feels slow" when it loads something in .8 seconds versus .3 of their friend's. (For the standard bad "car analogy", this is complaining that your car can only do 0-60 in 4.9 seconds, while your buddy's car does it in 4.2, even though their car uses far more gas and doesn't handle nearly as well at highway speeds or on twisty roads.)
 
2020-11-24 11:01:05 AM  
Has anybody mentioned that the time lag before clocking up only applies when you are "on battery"?

howtogeek.comView Full Size
 
2020-11-24 11:19:51 AM  

BullBearMS: Has anybody mentioned that the time lag before clocking up only applies when you are "on battery"?

[howtogeek.com image 650x381]


I am surprised you have not mention the Apple M1 smokes them in all tests.
 
2020-11-24 12:17:08 PM  

theflatline: BullBearMS: Has anybody mentioned that the time lag before clocking up only applies when you are "on battery"?

[howtogeek.com image 650x381]

I am surprised you have not mention the Apple M1 smokes them in all tests.


Does it? The Apple M1 hits well above its weight as a entry level chip by crushing all other entry and most midlevel chips, but my understanding was the top of the line chips from AMD have a small advantage over the M1. Or was that top of the line AMD chip the desktop version?
 
2020-11-24 12:27:20 PM  

Bob The Nob: Somebody's getting desperate.  When does Intel start suing people?


They probably won't bother. They were in a far worse situation than this when AMD released the A64 architecture, absolutely crushing Intel chips at the time. The Intel chips, which were often jokingly referred to as space heaters, couldn't compete except in very obscure tests that wouldn't matter to 99.999% of end users. Fast forward a couple years and the only thing that saved Intel from more embarrassment was the laptop chip team came up with a design (Core) that could be scaled to a desktop with excellent performance.

Nice to see the AMD team get a good chip out there to keep competition alive.

Merltech: AMD's approach makes sense.  I wonder how long before Intel has issues with CPUs dying due to heat.


Intel has rarely, if ever, had an issue with chips frying due to heat at standard clock timing. Even waayyy back you could run an Intel Pentium D (aka the space heater) with a seized fan and it would start throwing errors and shutdown when it passed the "redline" for heat. Older Pentium (socket) chips were able to just use a (stock) heatsink to run for hours in a 70'F room without crashing. I watched people try to kill them by removing the heatsink and they'd still run for 30m without seizing up or burning out. Just let it cool off, and run it for another 30 minutes. You had to worry more about PCB delamination due to heat than burning the chip out. On the flip side, AMD has made chips that did burn themselves out (in less than 30 minutes) if the fans seized, much less being run without a heatsink. The most egregious offender here for AMD was the K6-2 chips.
 
2020-11-24 1:04:23 PM  

Heliodorus: theflatline: BullBearMS: Has anybody mentioned that the time lag before clocking up only applies when you are "on battery"?

[howtogeek.com image 650x381]

I am surprised you have not mention the Apple M1 smokes them in all tests.

Does it? The Apple M1 hits well above its weight as a entry level chip by crushing all other entry and most midlevel chips, but my understanding was the top of the line chips from AMD have a small advantage over the M1. Or was that top of the line AMD chip the desktop version?


Per core?  It depends on the test.  Intel gets spanked and Zen 3 and the M1 go back and forth on which can run faster.

Multicore?  You can buy way more Zen 3 cores in a single socket.

However, the M1 is Apple's version of the i3.  It's for entry level products only.If you compare an M1 to Intel and AMD's entry level ultrabook chips, you're still talking about a dual core i3 and a quad core Zen 2 chip.  At that point it's a bit of a spanking.
 
2020-11-24 1:06:50 PM  

Heliodorus: Or was that top of the line AMD chip the desktop version?


It was a desktop version. I was astonished when the benchmarks came out- I really wasn't expecting the M1 to kick that much ass, and I'm an Apple user. The only thing that could reliably beat it was an AMD Ryzen- the desktop version. And it took way the hell more power to even match the M1 since those were desktop chips, not mobile. The only reason I haven't put in an order for an M1 MacBook Pro is because I'd rather have one with more ports- they've only released the lower end models. Makes one wonder what the hell is going to happen when the fast ones drop.

The CPU wars have returned. With a farking vengeance. It's going to be interesting.
 
2020-11-24 2:06:50 PM  
Intel: hope you're a fan of fan noises
 
2020-11-24 4:54:16 PM  

Heliodorus: The Apple M1 hits well above its weight as a entry level chip by crushing all other entry and most midlevel chips, but my understanding was the top of the line chips from AMD have a small advantage over the M1. Or was that top of the line AMD chip the desktop version?


Not even close to the top of the line AMD desktop CPUs.

It does (M1) very well with single core scores, besting Ryzen mobile processors.
It does not match up well in multicore benchmarks.
 
2020-11-24 5:59:40 PM  

Abe Vigoda's Ghost: Heliodorus: The Apple M1 hits well above its weight as a entry level chip by crushing all other entry and most midlevel chips, but my understanding was the top of the line chips from AMD have a small advantage over the M1. Or was that top of the line AMD chip the desktop version?

Not even close to the top of the line AMD desktop CPUs.

It does (M1) very well with single core scores, besting Ryzen mobile processors.
It does not match up well in multicore benchmarks.


Against Zen 3 cores, M1 always wins in floating point math and sometimes wins on integer math.

However, Ryzen mobile processors are still on Zen 2 cores, and the M1 always beats Zen 2 cores.

In SPEC 2006 M1 wins on integer and floating point.

images.anandtech.comView Full Size


In SPEC 2017 Zen 3 wins on integer, but not floating point.

images.anandtech.comView Full Size


Anandtech also has scores for the individual integer and floating point workloads that make up the two versions of SPEC.
 
2020-11-24 7:04:08 PM  

BullBearMS: Abe Vigoda's Ghost: Heliodorus: The Apple M1 hits well above its weight as a entry level chip by crushing all other entry and most midlevel chips, but my understanding was the top of the line chips from AMD have a small advantage over the M1. Or was that top of the line AMD chip the desktop version?

Not even close to the top of the line AMD desktop CPUs.

It does (M1) very well with single core scores, besting Ryzen mobile processors.
It does not match up well in multicore benchmarks.

Against Zen 3 cores, M1 always wins in floating point math and sometimes wins on integer math.

However, Ryzen mobile processors are still on Zen 2 cores, and the M1 always beats Zen 2 cores.

In SPEC 2006 M1 wins on integer and floating point.

[images.anandtech.com image 650x750]

In SPEC 2017 Zen 3 wins on integer, but not floating point.

[images.anandtech.com image 650x600]

Anandtech also has scores for the individual integer and floating point workloads that make up the two versions of SPEC.

.


The only people I have seen use SPEC benchmarking is Apple.


The M1 is a great chip, and I expect that future generations of Apple silicone are going to be extremely powerful, based on what the M1 has shown. But it's not comparable to desktop CPUs right now.
 
2020-11-24 7:30:53 PM  

Abe Vigoda's Ghost: The only people I have seen use SPEC benchmarking is Apple.


SPEC has literally been the industry standard benchmark for comparing completely different processor architectures for the last several decades.

The only people who complain about it are people who don't like who is getting the best scores.

The M1 is not only comparable to desktop scores, it's beating them while using a fraction of the power.

Hell, Apple's current gen iPhone chip is beating everything that Intel makes on a per core basis while only burning 5 watts.

images.anandtech.comView Full Size


In the overall SPEC2006 chart, the A14 is performing absolutely fantastic, taking the lead in absolute performance only falling short of AMD's recent Ryzen 5000 series.

The fact that Apple is able to achieve this in a total device power consumption of 5W including the SoC, DRAM, and regulators, versus +21W (1185G7) and 49W (5950X) package power figures, without DRAM or regulation, is absolutely mind-blowing.
 
2020-11-25 2:19:54 PM  
I'm just gonna take a moment to mention the 800-lb gorilla nobody's talking about.
encrypted-tbn0.gstatic.comView Full Size

These guys own the IP that goes into just about every non-x86 CPU on the market, including the M1. There's no stopping a competitor (Quaalcom, Samsung, Nvidia, etc) from making an equivalent. Also, Nvidia is in the process of buying the company. Interesting times lie ahead.
 
2020-11-25 4:13:44 PM  

pheelix: I'm just gonna take a moment to mention the 800-lb gorilla nobody's talking about.
[encrypted-tbn0.gstatic.com image 288x175]
These guys own the IP that goes into just about every non-x86 CPU on the market, including the M1. There's no stopping a competitor (Quaalcom, Samsung, Nvidia, etc) from making an equivalent. Also, Nvidia is in the process of buying the company. Interesting times lie ahead.


Apple does license ARM tech, but the technology on chip is all Apple. It will be interesting to see where they go with it. Because of the ARM architecture, I'm guessing they will use dual processors for the Pro side of their computers.
And no, the 13" MacPro is not really a 'pro' machine.

And I'm posting this on a 16" MacBook Pro, so I'm not an Apple hater.
I just like looking at the tech in a realistic manner.
 
2020-11-25 6:28:33 PM  

Abe Vigoda's Ghost: pheelix: I'm just gonna take a moment to mention the 800-lb gorilla nobody's talking about.
[encrypted-tbn0.gstatic.com image 288x175]
These guys own the IP that goes into just about every non-x86 CPU on the market, including the M1. There's no stopping a competitor (Quaalcom, Samsung, Nvidia, etc) from making an equivalent. Also, Nvidia is in the process of buying the company. Interesting times lie ahead.

Apple does license ARM tech, but the technology on chip is all Apple. It will be interesting to see where they go with it. Because of the ARM architecture, I'm guessing they will use dual processors for the Pro side of their computers.
And no, the 13" MacPro is not really a 'pro' machine.

And I'm posting this on a 16" MacBook Pro, so I'm not an Apple hater.
I just like looking at the tech in a realistic manner.


You're contradicting yourself. Yes, Apple designed the M1, but the tech on chip can't be all Apple because ARM owns the IP for the cpu cores. Apple most likely pays someone else for the USB controller, the cellular modem, the wifi, and a whole bunch of other stuff too. That's how chips are made these days.
 
2020-11-25 7:35:15 PM  

pheelix: Abe Vigoda's Ghost: pheelix: I'm just gonna take a moment to mention the 800-lb gorilla nobody's talking about.
[encrypted-tbn0.gstatic.com image 288x175]
These guys own the IP that goes into just about every non-x86 CPU on the market, including the M1. There's no stopping a competitor (Quaalcom, Samsung, Nvidia, etc) from making an equivalent. Also, Nvidia is in the process of buying the company. Interesting times lie ahead.

Apple does license ARM tech, but the technology on chip is all Apple. It will be interesting to see where they go with it. Because of the ARM architecture, I'm guessing they will use dual processors for the Pro side of their computers.
And no, the 13" MacPro is not really a 'pro' machine.

And I'm posting this on a 16" MacBook Pro, so I'm not an Apple hater.
I just like looking at the tech in a realistic manner.

You're contradicting yourself. Yes, Apple designed the M1, but the tech on chip can't be all Apple because ARM owns the IP for the cpu cores. Apple most likely pays someone else for the USB controller, the cellular modem, the wifi, and a whole bunch of other stuff too. That's how chips are made these days.


ARM owns the instruction set and Apple licensed that.

The cores in Apple's chips are 100% Apple's own design and do not work like other people's ARM cores (including tho ones designed by ARM) except for the part where they can execute the same code.

Just like Intel and AMD can execute the same x86 code, but their cores are not the same.

Anandtech published a laundry list of ways in which Apple's ARM implementation is doing things that AMD and Intel's cores can't match up to, much less the cores ARM has licensed so far.
 
Displayed 20 of 20 comments

View Voting Results: Smartest and Funniest

This thread is closed to new comments.

Continue Farking





On Twitter



  1. Links are submitted by members of the Fark community.

  2. When community members submit a link, they also write a custom headline for the story.

  3. Other Farkers comment on the links. This is the number of comments. Click here to read them.

  4. Click here to submit a link.