https://www.cnn.com/2021/07/22/tech/ftc-right-to-repair/index.html

On 2021-07-22 10:39, RichA wrote:
> https://www.cnn.com/2021/07/22/tech/ftc-right-to-repair/index.html
>

Good.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

On Thursday, 22 July 2021 at 17:54:08 UTC+1, Alan Browne wrote:
> On 2021-07-22 10:39, RichA wrote:
> > https://www.cnn.com/2021/07/22/tech/ftc-right-to-repair/index.html
> >
>
>
> Good.

be interesting to see how they go about it. Sounds good in theory.

>
>
>
> --
> "...there are many humorous things in this world; among them the white
> man's notion that he is less savage than the other savages."
> -Samuel Clemens

On Thursday, 22 July 2021 at 21:12:17 UTC-4, Whisky-dave wrote:
> On Thursday, 22 July 2021 at 17:54:08 UTC+1, Alan Browne wrote:
> > On 2021-07-22 10:39, RichA wrote:
> > > https://www.cnn.com/2021/07/22/tech/ftc-right-to-repair/index.html
> > >
> >
> >
> > Good.
> be interesting to see how they go about it. Sounds good in theory.
I figure if Apple can't use adhesives, it may make some of their products thicker (screws).

In article <cbcd201d-eba6-486b-bd78-55921e71a68bn@googlegroups.com>,
RichA <rander3128@gmail.com> wrote:

>
> I figure if Apple can't use adhesives, it may make some of their products thicker (screws).

don't tell microsoft, who has the prestige of being the first to get a
*zero* for repairability on a scale of 1 to 10.

<https://www.extremetech.com/computing/251046-ifixit-labels-surface-lapt
op-glue-filled-monstrosity>
According to iFixit, the Surface Laptop isn¹t repairable at all. In
fact, it got a 0 out of 10 for repairability and was labeled a
³glue-filled monstrosity.² Ouch. That¹s never happened before.
The lowers scores previously were a 1 out of 10 for all previous
iterations of the Surface Pro and the 2013 HTC One Android
smartphone.

It¹s clear that Microsoft never intended for the Surface Laptop to
be repaired because it¹s a completely sealed device. There aren¹t
even any screws to take out, so iFixit had to slice the fabric cover
open to peel it away from the metal chassis. That¹s never going back
together. The inner metal shield is also devoid of screws, relying
instead upon spot welds and glue. Again, this is probably not going
to be reassembled.

<https://www.ifixit.com/Teardown/Microsoft+Surface+Laptop+Teardown/92915
Verdict: The Surface Laptop is not a laptop. It¹s a glue-filled
monstrosity. There is nothing about it that is upgradable or
long-lasting, and it literally can¹t be opened without destroying
it. (Show us the procedure, Microsoft, we¹d love to be wrong.)

On 2021-07-22 21:12, Whisky-dave wrote:
> On Thursday, 22 July 2021 at 17:54:08 UTC+1, Alan Browne wrote:
>> On 2021-07-22 10:39, RichA wrote:
>>> https://www.cnn.com/2021/07/22/tech/ftc-right-to-repair/index.html
>>>
>>
>>
>> Good.
>
> be interesting to see how they go about it. Sounds good in theory.

Would be best if Apple, themselves, would sell parts directly to
consumers as well as design products to be more easily repairable or
upgradeable.

I recently updated the RAM in a Mac Mini. This could have been:

- remove bottom plate
- remove ram cover/retention clip
- add / change RAM
- replace bottom plate.

Instead had to disconnect many connectors, two being difficult to
re-connect, remove the mother board, various parts to even get close to
the RAM. Re-assembly was not easy and I damaged the power-on light
connector a little (more cosmetic than anything...). Took nearly half
an hour for what takes about 2 minutes on a 2012 iMac...

Of course if Apple charged competitive prices for RAM in the first place
I would have ordered it in 32 GB instead of 8.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

Am 23.07.2021 um 16:58 schrieb Alan Browne:
> Of course if Apple charged competitive prices for RAM in the first place
> I would have ordered it in 32 GB instead of 8.

Why would they since there are people willing to pay their prices...
--
Alfred Molon

Olympus 4/3 and micro 4/3 cameras forum at
https://groups.io/g/myolympus
https://myolympus.org/ photo sharing site

In article <dZBKI.201924$ea3.34058@fx10.ams1>, Alfred Molon
<alfred_molon@yahoo.com> wrote:

> > Of course if Apple charged competitive prices for RAM in the first place
> > I would have ordered it in 32 GB instead of 8.
>
> Why would they since there are people willing to pay their prices...

they're competitive with other manufacturers. for example, lenovo's
prices for memory are similar, if not more expensive. i haven't checked
in several years, but last time i did it was *more* than apple for the
same capacity.

buying from the manufacturer also includes a full warranty, which 3rd
party memory does not. it's quite common for warranty service to remove
3rd party memory, and if the problem goes away, it's considered to be
fixed, which means you either need to buy new memory or somehow
convince them it's not a memory issue, assuming it isn't. either way,
it's a significant hassle.

also, most people aren't interested in opening up their computer to
save a few dollars.

On 2021-07-23 12:54, nospam wrote:
> also, most people aren't interested in opening up their computer to
> save a few dollars.

In the case of Apple v. 3rd party memory, it is not a "few" dollars.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

In article <bRCKI.37397$5Y6.37010@fx10.iad>, Alan Browne
<bitbucket@blackhole.com> wrote:

> > also, most people aren't interested in opening up their computer to
> > save a few dollars.
>
> In the case of Apple v. 3rd party memory, it is not a "few" dollars.

you're missing the point, that people want to take the computer out of
the box, plug it in and start using it. they do not want the hassle of
finding the correct memory, ordering it separately, installing it and
hoping everything works properly, and if it doesn't, trying to figure
out the problem (memory or computer) before taking it in for service.

another advantage is that with the memory on the soc, as it is with the
new m1 macs, it's *much* faster than anything that can be added later,
especially when *very* few people upgrade after purchase. meanwhile,
they benefit from a significantly faster computer.

On Jul 23, 2021, nospam wrote
(in article<230720211341138387%nospam@nospam.invalid>):

> In article<bRCKI.37397$5Y6.37010@fx10.iad>, Alan Browne
> <bitbucket@blackhole.com> wrote:
>
> > > also, most people aren't interested in opening up their computer to
> > > save a few dollars.
> >
> > In the case of Apple v. 3rd party memory, it is not a "few" dollars.
>
> you're missing the point, that people want to take the computer out of
> the box, plug it in and start using it. they do not want the hassle of
> finding the correct memory, ordering it separately, installing it and
> hoping everything works properly, and if it doesn't, trying to figure
> out the problem (memory or computer) before taking it in for service.
>
> another advantage is that with the memory on the soc, as it is with the
> new m1 macs, it's *much* faster than anything that can be added later,
> especially when *very* few people upgrade after purchase. meanwhile,
> they benefit from a significantly faster computer.

While researching my M1, I saw reviews where the base M1 (8GB/256GB) was
out-benchmarking the $35,000 Mac Pros in some video related testing. You can
get that for $650 or even less now.

In article <0001HW.26AB3E37004D2ADD30DF4338F@news-us.newsgroup.ninja>,
Bill W <nothing@nowhere.com> wrote:

> > another advantage is that with the memory on the soc, as it is with the
> > new m1 macs, it's *much* faster than anything that can be added later,
> > especially when *very* few people upgrade after purchase. meanwhile,
> > they benefit from a significantly faster computer.
>
> While researching my M1, I saw reviews where the base M1 (8GB/256GB) was
> out-benchmarking the $35,000 Mac Pros in some video related testing. You can
> get that for $650 or even less now.

yea, it's kinda crazy, and that's with an *entry* *level* computer and
first generation chips.

the next generation processors (in a couple of months) are going to be
even faster, probably by quite a bit.

nevertheless, even with an entry level m1 mac, unzipping files is twice
as fast as a mac pro, which shows just how fast its unified memory
architecture actually is. for the record, xcode is somewhere around a
12 gig zip file, if i recall.
<https://techcrunch.com/wp-content/uploads/2020/11/xcode-unzip-time.jpg>

compiling webkit (the rendering engine of safari and other browsers) is
about as fast as on a mac pro (3.3gz xeon):
<https://techcrunch.com/wp-content/uploads/2020/11/webkit-compile-time.j
pg>

and compiling webkit doesn't use much battery:
<https://techcrunch.com/wp-content/uploads/2020/11/webkit-compile-batter
y1.jpg>

On 2021-07-23 13:41, nospam wrote:
> In article <bRCKI.37397$5Y6.37010@fx10.iad>, Alan Browne
> <bitbucket@blackhole.com> wrote:
>
>>> also, most people aren't interested in opening up their computer to
>>> save a few dollars.
>>
>> In the case of Apple v. 3rd party memory, it is not a "few" dollars.
>
> you're missing the point, that people want to take the computer out of
> the box, plug it in and start using it. they do not want the hassle of
> finding the correct memory, ordering it separately, installing it and
> hoping everything works properly, and if it doesn't, trying to figure
> out the problem (memory or computer) before taking it in for service.
>
> another advantage is that with the memory on the soc, as it is with the
> new m1 macs, it's *much* faster than anything that can be added later,

On the M1 Macs the memory is not on the SOC - and the memory is
configured no differently (clocking, etc.) than a conventional
CPU:memory setup.

The M1 SOC and the memory chip carriers are mounted to a common pcb as
an integrated module.

Indeed some Chinese modders upgraded a Mac Mini to 16 from 8 GB by
desoldering the memory and replacing it...

> especially when *very* few people upgrade after purchase. meanwhile,
> they benefit from a significantly faster computer.

It would be no slower if the memory were in its own chip carrier. Since
.... that's exactly how an M1 is set up...

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

In article <w9EKI.63021$dp5.42272@fx48.iad>, Alan Browne
<bitbucket@blackhole.com> wrote:

> >
> > another advantage is that with the memory on the soc, as it is with the
> > new m1 macs, it's *much* faster than anything that can be added later,
>
> On the M1 Macs the memory is not on the SOC - and the memory is
> configured no differently (clocking, etc.) than a conventional
> CPU:memory setup.

it is on the soc:

<https://en.wikipedia.org/wiki/Apple_M1#Other_features>
The M1 uses 4266 MT/s LPDDR4X SDRAM in a unified memory
configuration shared by all the components of the processor. The
SoC and RAM chips are mounted together in a system-in-a-package
design. 8 GB and 16 GB configurations are available.

the major benefit is that it's a unified memory architecture.

> The M1 SOC and the memory chip carriers are mounted to a common pcb as
> an integrated module.

yep, which means no dimm slots.

> Indeed some Chinese modders upgraded a Mac Mini to 16 from 8 GB by
> desoldering the memory and replacing it...

that is *not* an easy task.

> > especially when *very* few people upgrade after purchase. meanwhile,
> > they benefit from a significantly faster computer.
>
> It would be no slower if the memory were in its own chip carrier. Since
> ... that's exactly how an M1 is set up...

yes it very definitely would, and no that's not how an m1 is set up.
you're ignoring the unified memory architecture.

just look at the various benchmarks, where 8 gig m1 macs outperform
16-32 gig intel macs, including a mac pro in some tasks.

On 2021-07-23 15:21, nospam wrote:
> In article <w9EKI.63021$dp5.42272@fx48.iad>, Alan Browne
> <bitbucket@blackhole.com> wrote:
>
>>>
>>> another advantage is that with the memory on the soc, as it is with the
>>> new m1 macs, it's *much* faster than anything that can be added later,
>>
>> On the M1 Macs the memory is not on the SOC - and the memory is
>> configured no differently (clocking, etc.) than a conventional
>> CPU:memory setup.
>
> it is on the soc:
=== No. It most definitely is not.

> <https://en.wikipedia.org/wiki/Apple_M1#Other_features>
> The M1 uses 4266 MT/s LPDDR4X SDRAM in a unified memory
> configuration shared by all the components of the processor. The
> SoC and RAM chips are mounted together in a system-in-a-package
=== === ======= ===================

> design. 8 GB and 16 GB configurations are available.
>
> the major benefit is that it's a unified memory architecture.

Marketing speak for "we soldered the memory carriers to the same little
board as the CPU chip carrier." 0 performance difference.

https://valkyrie.cdn.ifixit.com/media/2021/01/28102657/m1_chip-900x900.jpeg

>> The M1 SOC and the memory chip carriers are mounted to a common pcb as
>> an integrated module.
>
> yep, which means no dimm slots.

Doesn't make any difference - they are separate chip carriers with
signals running through the common PCB.

>
>> Indeed some Chinese modders upgraded a Mac Mini to 16 from 8 GB by
>> desoldering the memory and replacing it...
>
> that is *not* an easy task.

Agree'd. Though I know many technicians who are very able to do it -
given the right equipment. I also know a repair shop who are capable of
it - though they don't guarantee it for more than 60 days - too hard to
do reliably in all cases.

>>> especially when *very* few people upgrade after purchase. meanwhile,
>>> they benefit from a significantly faster computer.
>>
>> It would be no slower if the memory were in its own chip carrier. Since
>> ... that's exactly how an M1 is set up...
>
> yes it very definitely would, and no that's not how an m1 is set up.
> you're ignoring the unified memory architecture.

It's no different than if it were in a slot.

>
> just look at the various benchmarks, where 8 gig m1 macs outperform
> 16-32 gig intel macs, including a mac pro in some tasks.

Irrelevant. Would do just as well in slots.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

On 2021-07-23 15:21, nospam wrote:

> yes it very definitely would, and no that's not how an m1 is set up.
> you're ignoring the unified memory architecture.

Forgot to mention: the "unified" advantage comes from the CPU integrated
I/O not needing separate memory as it would have in a common intel
setup. As such there is much less movement of memory from OS/user
allocated memory to I/O allocated memory.

This is where much of the speed improvement comes from. Not the
physical arrangement of the memory to the CPU.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

In article <AOEKI.42175$UR4.34883@fx37.iad>, Alan Browne
<bitbucket@blackhole.com> wrote:

> >>> another advantage is that with the memory on the soc, as it is with the
> >>> new m1 macs, it's *much* faster than anything that can be added later,
> >>
> >> On the M1 Macs the memory is not on the SOC - and the memory is
> >> configured no differently (clocking, etc.) than a conventional
> >> CPU:memory setup.
> >
> > it is on the soc:
> === No. It most definitely is not.
>
> > <https://en.wikipedia.org/wiki/Apple_M1#Other_features>
> > The M1 uses 4266 MT/s LPDDR4X SDRAM in a unified memory
> > configuration shared by all the components of the processor. The
> > SoC and RAM chips are mounted together in a system-in-a-package
> === === ======= ===================

versus separate dimm slots elsewhere on the logic board. i called it an
soc, wikipedia clarifies it as siap. that doesn't actually change much.

as i said, the key difference is the unified memory architecture.

> > design. 8 GB and 16 GB configurations are available.
> >
> > the major benefit is that it's a unified memory architecture.
>
> Marketing speak for "we soldered the memory carriers to the same little
> board as the CPU chip carrier." 0 performance difference.

numerous benchmarks show that to be totally false.

On 2021-07-23 15:41, nospam wrote:
> In article <AOEKI.42175$UR4.34883@fx37.iad>, Alan Browne
> <bitbucket@blackhole.com> wrote:
>
>>>>> another advantage is that with the memory on the soc, as it is with the
>>>>> new m1 macs, it's *much* faster than anything that can be added later,
>>>>
>>>> On the M1 Macs the memory is not on the SOC - and the memory is
>>>> configured no differently (clocking, etc.) than a conventional
>>>> CPU:memory setup.
>>>
>>> it is on the soc:
>> === No. It most definitely is not.
>>
>>> <https://en.wikipedia.org/wiki/Apple_M1#Other_features>
>>> The M1 uses 4266 MT/s LPDDR4X SDRAM in a unified memory
>>> configuration shared by all the components of the processor. The
>>> SoC and RAM chips are mounted together in a system-in-a-package
>> === === ======= ===================
>
> versus separate dimm slots elsewhere on the logic board. i called it an
> soc, wikipedia clarifies it as siap. that doesn't actually change much.
>
> as i said, the key difference is the unified memory architecture.
>
>>> design. 8 GB and 16 GB configurations are available.
>>>
>>> the major benefit is that it's a unified memory architecture.
>>
>> Marketing speak for "we soldered the memory carriers to the same little
>> board as the CPU chip carrier." 0 performance difference.
>
> numerous benchmarks show that to be totally false.

Doesn't change the physical memory speed. See my other comment - the
unified memory is about the avoidance of moving data around because the
i/o portions of the soc don't need things moved as they would be in a
conventional set up.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

In article <s6FKI.36654$Yv3.19040@fx41.iad>, Alan Browne
<bitbucket@blackhole.com> wrote:

> >> Marketing speak for "we soldered the memory carriers to the same little
> >> board as the CPU chip carrier." 0 performance difference.
> >
> > numerous benchmarks show that to be totally false.
>
> Doesn't change the physical memory speed. See my other comment - the
> unified memory is about the avoidance of moving data around because the
> i/o portions of the soc don't need things moved as they would be in a
> conventional set up.

that's one reason why it's faster.

again, benchmarks show a dramatic difference using less memory. your
claim of 0 performance difference is false.

you're also going to have to correct the tech journalists who call it
soc.

<https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro-is-powerfu
l-but-its-the-battery-life-that-will-blow-you-away/>
Much ado has been made of Apple including only 16GB of memory
on these first M1 machines. The fact of it, however, is that I have
been unable to push them hard enough yet to feel any effect of this
due to Apple¹s move to unified memory architecture. Moving RAM to
the SoC means no upgradeability ‹ you¹re stuck on 16GB forever. But
it also means massively faster access to that memory by chips on the
system that need it most.
....
While running many applications simultaneously, the M1 performed
extremely well. Because this new architecture is so tightly arranged,
with memory being a short hop away next door rather than out over
a PCIe bus, swapping between applications was zero trouble. Even
while tasks were run in the background ‹ beefy, data heavy tasks ‹
the rest of the system stayed flowing.

On 2021-07-23 16:17, nospam wrote:
> In article <s6FKI.36654$Yv3.19040@fx41.iad>, Alan Browne
> <bitbucket@blackhole.com> wrote:
>
>>>> Marketing speak for "we soldered the memory carriers to the same little
>>>> board as the CPU chip carrier." 0 performance difference.
>>>
>>> numerous benchmarks show that to be totally false.
>>
>> Doesn't change the physical memory speed. See my other comment - the
>> unified memory is about the avoidance of moving data around because the
>> i/o portions of the soc don't need things moved as they would be in a
>> conventional set up.
>
> that's one reason why it's faster.

It's the only reason (aside from it also being a faster memory
technology to begin with.

>
> again, benchmarks show a dramatic difference using less memory. your
> claim of 0 performance difference is false.

You'd have to benchmark the memory i/o, not application performance.
Application performance is simply the unified architecture for the
reasons above.

>
> you're also going to have to correct the tech journalists who call it
> soc.

I can't help journalists who don't know the difference between a SOC and
memory soldered to a PCB close to the CPU/SOC. Indeed this is the sort
of thing _you_ normally quibble over....

>
> <https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro-is-powerfu
> l-but-its-the-battery-life-that-will-blow-you-away/>
> Much ado has been made of Apple including only 16GB of memory
> on these first M1 machines. The fact of it, however, is that I have
> been unable to push them hard enough yet to feel any effect of this
> due to Apple¹s move to unified memory architecture. Moving RAM to
> the SoC means no upgradeability ‹ you¹re stuck on 16GB forever. But
> it also means massively faster access to that memory by chips on the
> system that need it most.
> ...
> While running many applications simultaneously, the M1 performed
> extremely well. Because this new architecture is so tightly arranged,
> with memory being a short hop away next door rather than out over
> a PCIe bus, swapping between applications was zero trouble. Even
> while tasks were run in the background ‹ beefy, data heavy tasks ‹
> the rest of the system stayed flowing.

Clearly they don't understand how memory works. Whether soldered to the
PCB or in a slot would make 0 difference in memory to CPU bandwidth.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

In article <KXVKI.24290$Nq7.14064@fx33.iad>, Alan Browne
<bitbucket@blackhole.com> wrote:

> > again, benchmarks show a dramatic difference using less memory. your
> > claim of 0 performance difference is false.
>
> You'd have to benchmark the memory i/o, not application performance.
> Application performance is simply the unified architecture for the
> reasons above.

unless you do nothing but run benchmarks all day, memory benchmarks are
not particularly interesting.

the *only* metrics that matter is real world performance for the apps
someone actually uses, which will vary depending on what that someone
is doing.

some of the more demanding tasks, such as rendering video and compiling
apps, are comparable in speed and in some cases faster than even a mac
pro, and that's with a 'paltry' 16 gig (or 8 gig in some cases).

your claim of '0 performance difference' is not supported by people who
actually own and use apple silicon macs.

> > you're also going to have to correct the tech journalists who call it
> > soc.
>
> I can't help journalists who don't know the difference between a SOC and
> memory soldered to a PCB close to the CPU/SOC. Indeed this is the sort
> of thing _you_ normally quibble over....

the difference is irrelevant and i'm not the one quibbling over it.

the fact is that the memory and processor is one unit and part of why
m1 is much faster than previous designs.

> > <https://techcrunch.com/2020/11/17/yeah-apples-m1-macbook-pro-is-powerfu
> > l-but-its-the-battery-life-that-will-blow-you-away/>
> > Much ado has been made of Apple including only 16GB of memory
> > on these first M1 machines. The fact of it, however, is that I have
> > been unable to push them hard enough yet to feel any effect of this
> > due to Apple1s move to unified memory architecture. Moving RAM to
> > the SoC means no upgradeability Ð you1re stuck on 16GB forever. But
> > it also means massively faster access to that memory by chips on the
> > system that need it most.
> > ...
> > While running many applications simultaneously, the M1 performed
> > extremely well. Because this new architecture is so tightly arranged,
> > with memory being a short hop away next door rather than out over
> > a PCIe bus, swapping between applications was zero trouble. Even
> > while tasks were run in the background Ð beefy, data heavy tasks Ð
> > the rest of the system stayed flowing.
>
> Clearly they don't understand how memory works. Whether soldered to the
> PCB or in a slot would make 0 difference in memory to CPU bandwidth.

clearly *you* don't understand how memory works, since what you are
claiming is simply false.

tl;dr 16 gig on m1 is different than 16 gig on intel.

from the link above, an m1 mac mini with 16 gig was slightly faster in
compiling than the mac pro with 48 gig of memory and a 12 core xeon
processor:
<https://techcrunch.com/wp-content/uploads/2020/11/webkit-compile-time.j
pg>

here's what apple has to say (the link now redirects, but it's still in
the wayback machine):
<https://web.archive.org/web/20201110184847/https://www.apple.com/mac/m1
/>
M1 also features our unified memory architecture, or UMA. M1
unifies its high-bandwidth, low-latency memory into a single pool
within a custom package. As a result, all of the technologies in the
SoC can access the same data without copying it between multiple
pools of memory. This dramatically improves performance and power
efficiency. Video apps are snappier. Games are richer and more
detailed. Image processing is lightning fast. And your entire system
is more responsive.

<https://www.theregister.com/2020/11/19/apple_m1_high_bandwidth_memory_p
erformance/>
Let's take a closer look at how Apple uses high-bandwidth memory
in the M1 system-on-chip (SoC) to deliver this rocket boost.

High-bandwidth memory (HBM) avoids the traditional CPU socket-memory
channel design by pooling memory connected to a processor via an
interposer layer. HBM combines memory chips and gives them closer
and faster access to the CPU as the distance to the processor is only
a few micrometer units. This on its own speeds data transfers.
....
SDRAM operations are synchronised to the SoC processing clock
speed. Apple describes the SDRAM as a single pool of high-bandwidth,
low-latency memory, allowing apps to share data between the CPU, GPU,
and Neural Engine efficiently.

but what really matters is *real* *world* *performance*.

users who actually use the m1 macs and are not blindly reading numbers
off a spec sheet realize that 16 gig on an m1 is not anywhere near as
limiting as 16 gig on intel.

there are numerous reports of people opening many hundreds of browser
tabs or launching hundreds of apps all at the same time and the m1 macs
don't even flinch, while the intel macs keel over in pain.

as noted above, an m1 mac with 16 gig of memory is comparable to a mac
pro with 48 gig of memory and a xeon processor when compiling apps,
which is cpu and memory intensive.

here's more:
<https://www.techrepublic.com/article/the-2020-macbook-pro-with-m1-is-in-
a-word-astonishing/>
Upon getting the M1 MacBook Pro set up, one of the first things I
did was open Safari and start hitting the internet. My immediate
impression? I have never seen websites load this quickly. 

Ever. Period.

On paper, the specs on my desktop blow the MacBook Pro away, and
yet, it cannot hold a candle to how blazingly fast Safari renders
websites--that's across the board. Every site I go to appears almost
instantly. We're talking blink of an eye fast. Once a site is cached,
it makes me laugh how quickly they load.

I'm not exaggerating--no matter what site I visit, if I blink I'll
miss the loading.

Next up was email. Open Apple Mail and the missives are dumped
into the inbox with incredible speed. Comparing that to my previous
MacBook and the difference is eye-opening.

That's a lot of hyperbole. It's also truth.

<https://www.inputmag.com/reviews/macbook-air-m1-review-windows-laptops-
are-so-screwed>
I¹ve used plenty of MacBooks and Windows laptops with 8GB of RAM
and they just barely cut it the moment you install Chrome or run
Adobe¹s Creative Cloud apps like Photoshop or Lightroom. So I was
cautiously skeptical how well the M1 MacBook Air with 8GB would
perform. Turns out, its 8GB of ³unified memory² is not equal to the
same amount of RAM in an Intel-powered MacBook. It¹s better.

In almost every use case, the M1 MacBook Air with 8GB of RAM smoked
my 2019 13-inch MacBook Air with 16GB of RAM. I¹m not going to get
into the technical details, but essentially macOS Big Sur and apps
optimized for the M1 chip are able to ramp up and down performance
while using less RAM. This is just like how iPhones and iPads work.

On paper, iOS devices have less RAM than Android devices so it seems
like they're worse specs-wise, but as Apple has proven over and over,
iOS does more with less. Apps built for Intel (x86) chips come with
all kinds of baggage that require more RAM to run smoothly. By
integrating the memory on the M1 silicon, Big Sur and optimized apps
don¹t need as much RAM. This all but puts the kibosh on the whole
³There¹s no 32GB RAM option for the M1 Macs.²

On my MacBook Pro with 16GB of RAM, I frequently run out of RAM
with Chrome open with two dozen tabs and Slack, Notes, Tweetdeck,
Photoshop, Lightroom, Final Cut Pro, and Handbrake open. If the M1
MacBook Air could talk, it would have chuckled at me when I threw the
same workload at it and said ³Is that all you got?² It took almost 50
Chrome tabs and a few more open apps like Spotify and VLC Player on
top of that workload to finally bring the 8GB of RAM in the M1
MacBook Air to its knees. And even then, it was because of Lightroom.
(Because of course.) The photo editing app isn¹t optimized for the M1
so it¹s running ³translated² via emulation with Rosetta 2. I can¹t
even be mad that an Adobe app froze and crashed my MacBook Air
because they¹re notorious for performing poorly on even Intel Macs.

That's not to say Adobe apps don't run at all. With fewer apps and
processes open, Photoshop and Lightroom actually run faster than on
my Intel MacBook Pro. It boggled my mind to see Lightroom instantly
import 80 RAW photos and not so much as show a loading bar. That's
just nuts! I'm so conditioned to stretching my legs out and grabbing
tea while waiting for RAW photos to import that I thought something
was wrong. Nope, the M1's just super fast.

On 2021-07-24 14:34, nospam wrote:
> In article <KXVKI.24290$Nq7.14064@fx33.iad>, Alan Browne
> <bitbucket@blackhole.com> wrote:
>
>>> again, benchmarks show a dramatic difference using less memory. your
>>> claim of 0 performance difference is false.
>>
>> You'd have to benchmark the memory i/o, not application performance.
>> Application performance is simply the unified architecture for the
>> reasons above.
>
> unless you do nothing but run benchmarks all day, memory benchmarks are
> not particularly interesting.
>
> the *only* metrics that matter is real world performance for the apps
> someone actually uses, which will vary depending on what that someone
> is doing.
>
> some of the more demanding tasks, such as rendering video and compiling
> apps, are comparable in speed and in some cases faster than even a mac
> pro, and that's with a 'paltry' 16 gig (or 8 gig in some cases).
>
> your claim of '0 performance difference' is not supported by people who
> actually own and use apple silicon macs.

You're going out of your way to distort things.

Unified memory's benefits do not come from the physical location of the
memory. They come from not needing to move data because it is well
integrated by the CPU to be where it is needed for both processing and
i/o. Impossible to do with, eg, intel.

>
>>> you're also going to have to correct the tech journalists who call it
>>> soc.
>>
>> I can't help journalists who don't know the difference between a SOC and
>> memory soldered to a PCB close to the CPU/SOC. Indeed this is the sort
>> of thing _you_ normally quibble over....
>
> the difference is irrelevant and i'm not the one quibbling over it.

Exactly: it is irrelevant to the speed improvement: that is all the
unified memory architecure. Not how the RAM is implemented.

>
> the fact is that the memory and processor is one unit and part of why
> m1 is much faster than previous designs.

As stated (by Apple) the unified design means data does not have to be
moved from processor memory to i/o memory or v-v. It's already in
place. Just address it there. (For all i/o that is directly controlled
on the M1. Which is a lot.

>> Clearly they don't understand how memory works. Whether soldered to the
>> PCB or in a slot would make 0 difference in memory to CPU bandwidth.
>
> clearly *you* don't understand how memory works, since what you are
> claiming is simply false.

No it isn't. The memory b/w would be the same because all of the timing
signals demanded by the chip are the same. This is off-the-shelf memory.

>
> tl;dr 16 gig on m1 is different than 16 gig on intel.
>
> from the link above, an m1 mac mini with 16 gig was slightly faster in
> compiling than the mac pro with 48 gig of memory and a 12 core xeon
> processor:
> <https://techcrunch.com/wp-content/uploads/2020/11/webkit-compile-time.j
> pg>

Which has nothing to do with the memory config. It has everything to do
with unified memory. (Which is not the memory config).

>
> here's what apple has to say (the link now redirects, but it's still in
> the wayback machine):
> <https://web.archive.org/web/20201110184847/https://www.apple.com/mac/m1
> />
> M1 also features our unified memory architecture, or UMA. M1
> unifies its high-bandwidth, low-latency memory into a single pool
> within a custom package. As a result, all of the technologies in the

package is not SOC.

> SoC can access the same data without copying it between multiple

See that "copying" thing. That is the unified memory. Not how it is
soldered to a PCB that is beside the CPU.

> pools of memory. This dramatically improves performance and power

Yes: exactly: the pools of memory meaning data does not have to be moved
as often, saving bags of time.

> efficiency. Video apps are snappier. Games are richer and more
> detailed. Image processing is lightning fast. And your entire system
> is more responsive.
>
> <https://www.theregister.com/2020/11/19/apple_m1_high_bandwidth_memory_p
> erformance/>
> Let's take a closer look at how Apple uses high-bandwidth memory
> in the M1 system-on-chip (SoC) to deliver this rocket boost.
>
> High-bandwidth memory (HBM) avoids the traditional CPU socket-memory
> channel design by pooling memory connected to a processor via an
> interposer layer. HBM combines memory chips and gives them closer
> and faster access to the CPU as the distance to the processor is only
> a few micrometer units. This on its own speeds data transfers.

Even if there is an advantage there - it is NOTHING compared to the
unified memory above. And actually - the timings are no different.

> ...
> SDRAM operations are synchronised to the SoC processing clock
> speed. Apple describes the SDRAM as a single pool of high-bandwidth,
> low-latency memory, allowing apps to share data between the CPU, GPU,
> and Neural Engine efficiently.

Indeed this is the unification part that reduced (eliminates) data
transfers between functional parts. When you don't have to move data -
that really improves speed.

>
> but what really matters is *real* *world* *performance*.

It is what is. Because: unified. Not location .

tl-dr snipped away.

The speed improvement is not because of the memory location - but
because data is not moved around as much. Big improvement. Even if the
physical memory were faster - it wouldn't matter because so much data
"moving" is no longer happening at all. That is the essence of the
unified memory. Not the packaging.

--
"...there are many humorous things in this world; among them the white
man's notion that he is less savage than the other savages."
-Samuel Clemens

In article <xl_KI.48842$Nz.21751@fx22.iad>, Alan Browne
<bitbucket@blackhole.com> wrote:

> >>> again, benchmarks show a dramatic difference using less memory. your
> >>> claim of 0 performance difference is false.
> >>
> >> You'd have to benchmark the memory i/o, not application performance.
> >> Application performance is simply the unified architecture for the
> >> reasons above.
> >
> > unless you do nothing but run benchmarks all day, memory benchmarks are
> > not particularly interesting.
> >
> > the *only* metrics that matter is real world performance for the apps
> > someone actually uses, which will vary depending on what that someone
> > is doing.
> >
> > some of the more demanding tasks, such as rendering video and compiling
> > apps, are comparable in speed and in some cases faster than even a mac
> > pro, and that's with a 'paltry' 16 gig (or 8 gig in some cases).
> >
> > your claim of '0 performance difference' is not supported by people who
> > actually own and use apple silicon macs.
>
> You're going out of your way to distort things.

nope. i'm going out of my way to clarify things, which is apparently a
waste of time since you refuse to learn anything.

> Unified memory's benefits do not come from the physical location of the
> memory. They come from not needing to move data because it is well
> integrated by the CPU to be where it is needed for both processing and
> i/o. Impossible to do with, eg, intel.

the fact is that m1 is faster using less memory, due to unified and
high bandwidth (due to shorter distance).

your claim of '0 performance difference' is false.

stop reading spec sheets and try actually using one.

On Friday, 23 July 2021 at 10:59:01 UTC-4, Alan Browne wrote:
> On 2021-07-22 21:12, Whisky-dave wrote:
> > On Thursday, 22 July 2021 at 17:54:08 UTC+1, Alan Browne wrote:
> >> On 2021-07-22 10:39, RichA wrote:
> >>> https://www.cnn.com/2021/07/22/tech/ftc-right-to-repair/index.html
> >>>
> >>
> >>
> >> Good.
> >
> > be interesting to see how they go about it. Sounds good in theory.
> Would be best if Apple, themselves, would sell parts directly to
> consumers as well as design products to be more easily repairable or
> upgradeable.
>
> I recently updated the RAM in a Mac Mini. This could have been:
>
> - remove bottom plate
> - remove ram cover/retention clip
> - add / change RAM
> - replace bottom plate.
>
> Instead had to disconnect many connectors, two being difficult to
> re-connect, remove the mother board, various parts to even get close to
> the RAM. Re-assembly was not easy and I damaged the power-on light
> connector a little (more cosmetic than anything...). Took nearly half
> an hour for what takes about 2 minutes on a 2012 iMac...
>
> Of course if Apple charged competitive prices for RAM in the first place
> I would have ordered it in 32 GB instead of 8.
> --
> "...there are many humorous things in this world; among them the white
> man's notion that he is less savage than the other savages."
> -Samuel Clemens

Apple: 43% profit margins on sales. Most other companies? Half that.

In article <fd069bca-9c38-4bc8-b482-9934ed6ee531n@googlegroups.com>,
RichA <rander3128@gmail.com> wrote:

> Apple: 43% profit margins on sales. Most other companies? Half that.

no