Apple ARM is superior?

[Bombastinator]

13 minutes ago, Sauron said:

Only if that example doesn't actually support your point like in this case.

Excuse me? Did you even bother to open the link?

It happens when you make a statement that is just wrong and then start arguing definitions because you can't bring yourself to admit it or at least explain what point you were trying to make with that incorrect statement.

Yes. If a Dell laptop is a "PC" then so is the Mac Mini. A PC is still a PC regardless of what operating system it runs because the term just refers to a computer that is practical for personal use. Just because in one ad campaign from Apple they used "PC" just to refer to Windows computers doesn't mean that's what the term means now. A PC is not any less of a PC just because it's running macOS or Linux. Windows didn't even exist when the IBM PC came out and yet you claim that is your point of reference so clearly you didn't accept that definition either...

 

I could have just ignored it you had used it just to refer to Windows computers because it's a relatively common colloquial use of the word, but your sentence wouldn't have made any sense in that context.

Still waiting for you to explain how it's different. You literally took a common use word out of a trademark and claimed it specifically referred to that trademark.

Yay.  Cut quotes.  People made me stop using those.  Complained that they created posts that were too long.  I don’t recall who did the complaining.

re: accuracy of examples:  false.

 

re: dictionary definition: I did.  You used it though like you were attempting to define PC.  Which it didn’t.   A weird discontinuity.  I start to wonder if there is a difference of understanding here.

 

re: the accusation about intent regarding I’m not even sure what.

My statement iirc is that the term “PC” was used to refer to either IBM PCs or IBM compatibles as differentiates from personal computers that did not run microsoft windows.  Are You are saying this is not true? Of are you saying this was not the thrust of the point?

[Bombastinator]

12 minutes ago, NunoLava1998 said:

This seems to be M1-specific from what I've seen though.. Intel Macs (even with SSDs) swap a lot, but they're usually much more conservative doing it than M1 Macs.

Also, the SSD (and pretty much everything else) is soldered onto the board, so if the SSD goes out, your system becomes e-waste. Can't even worry about buying apple-branded SSDs when you can't upgrade them in the first place

Ouch.  So it’s even worse that that.  Soldered on SSDs in general are bad ideas period imho. SSDs wear out and can do it extremely quickly.  If this goes as you describe someone or other in the next few months is going to run through their SSD which will then require a motherboard replacement and things will get epic. 

[NunoLava1998]

2 minutes ago, Bombastinator said:

Ouch.  So it’s even worse that that.  Soldered on SSDs in general are bad ideas period imho. SSDs wear out and can do it extremely quickly.  If this goes as you describe someone or other in the next few months is going to run through their SSD which will then require a motherboard replacement and things will get epic. 

Well, it'll likely take years if the SMART data on endurance that it gives out is true (about 2 PBW for 250GB). But still, not good design

[Bombastinator]

2 minutes ago, NunoLava1998 said:

Well, it'll likely take years if the SMART data on endurance that it gives out is true (about 2 PBW for 250GB). But still, not good design

Then a few years instead of a few months.  complaints will still likely be epic though.   A bomb on a longer timer but no less explosive.

[Roswell]

5 hours ago, Bombastinator said:

I’ve seen this referred to as “abuse of swap” but I don’t understand how this could make something fast.  Swap has a rep for being incredibly slow.   Did it do it by forcing the other app to use swap thus making itself look fast I. Comparison?

kernel_task can be anything. It can even be writes to memory or simply drive indexing. The activity monitor really needs a revamp in macOS to more accurately report information.

[igormp]

2 minutes ago, Vitamanic said:

kernel_task can be anything. It can even be writes to memory or simply drive indexing. The activity monitor really needs a revamp in macOS to more accurately report information.

It seems that the culprit is the swapping. If you want to read a bit more about some findings, have fun:

 

[Roswell]

13 minutes ago, igormp said:

It seems that the culprit is the swapping. If you want to read a bit more about some findings, have fun:

 

I imagine cold booting the machine (and establishing the swap file) 256 times in 2 months is the overwhelming majority of their 600GB of writes. There's always context to be had.

 

They have 191 hours of total usage which comes out to them cold booting the laptop about every 40 minutes. Why?!

[Zodiark1593]

5 hours ago, Bombastinator said:

Then a few years instead of a few months.  complaints will still likely be epic though.   A bomb on a longer timer but no less explosive.

Tbh, I was hypothesizing that Apple was using pretty aggressive swapping with their iPhones as well, hence the implementation of custom PCIe storage controllers in a phone. Even without swapping though, the fast storage reduces the need to keep apps in memory. 

 

From what I recall, some SSDs use a SLC cache to speed up performance before committing writes to slower TLC/SLC storage. Do these SSD drives lack an SLC cache, and if so, could Apple specify Swap to go to only the SLC portion?

[Bombastinator]

18 minutes ago, Vitamanic said:

I imagine cold booting the machine (and establishing the swap file) 256 times in 2 months is the overwhelming majority of their 600GB of writes. There's always context to be had.

 

They have 191 hours of total usage which comes out to them cold booting the laptop about every 40 minutes. Why?!

Doesn’t matter.  People will do that.

[Bombastinator]

I also remember something about Apple talking about adding another “L” layer of ram.  If this ultra heavy swap stuff is being run to a RAM disk it solves the wear issue. 

[Roswell]

13 minutes ago, Bombastinator said:

Doesn’t matter.  People will do that.

Yeah people will do weird stuff. That said, cold booting a machine 256 times in 60 days is going to cause significant writes regardless of the platform.

[Arika]

6 hours ago, Bombastinator said:

The question is whether “PC” means personal computer or a specific kind of personal computer.  

And the answer is absolute semantics. If the entire argument hinges on the fact of whether or not the name Personal Computer or PC is an eponym or not, then the argument itself has no real weight to it.

 

it's like if someone scraped their knee and they ask for a band-aid and your response is:

"well, do you mean the Band-Aid® brand adhesive medical strips, or are you using the name "band-aid" in the generic way to say adhesive medical strips?"

[Bombastinator]

29 minutes ago, Vitamanic said:

Yeah people will do weird stuff. That said, cold booting a machine 256 times in 60 days is going to cause significant writes regardless of the platform.

True.  Probably trying to do some sort of stress test.  60days or 60 weeks though soldered SSDs are still a problem.  People pay more for macs because they last longer.  If you halve the life of a Mac you halve it’s value.  I remember some idiot at Apple trying to figure out how to make their machines last a shorter amount of time because they think they will sell more machines that way.  They won’t.  If macs don’t last any longer than other machines but cost more why buy one?  It’s idiocy.  The high resale value of macs is based on their long useful life.  iPhone is creaming android devices.  Why?  They’re cheaper.  They cost more but they last longer. 

[saltycaramel]

Apple bring on a 32GB + 4TB M1X Mac so I don’t have to worry about SSD-writes-gate!

 

4 weeks left before the best all-in-one desktop of all times, I hope they bring the resolution to 6K to match the XDR display but I’m afraid it will still be 5K to leave 6K as an high end exclusive. 

 

Now, if it’ll be 5K 120Hz, I wouldn’t complain..

[Spindel]

5 hours ago, igormp said:

It seems that the culprit is the swapping. If you want to read a bit more about some findings, have fun:

 

I hope this guy doesn't use a web browser, wait until he realizes how much writes to disk the cache causes, or any program with a data base, or anything that writes to disk. 

 

Seriously this is a nothing burger. 

 

With the advent of SSDs swap is actually something useful, not something to dread like it was/is with HDDs. 

 

Oh and for the record my old iMac from 2013 had a 256 GB SSD, I retired it in november 2020 when I got my M1 Mini and during those 7 years nothing negative has happened to the SSD and it has 1000s of TB of writes. 

[leadeater]

37 minutes ago, saltycaramel said:

4 weeks left before the best all-in-one desktop of all times, I hope they bring the resolution to 6K to match the XDR display but I’m afraid it will still be 5K to leave 6K as an high end exclusive. 

Well that is every iMac basically ever, all other all-in-ones suck so hard, omg. Even the HP Elite 8000 series business ones from a while ago were just horrid compared to the tower and SFF models. External power bricks can go to hell for those.

[leadeater]

4 minutes ago, Spindel said:

Oh and for the record my old iMac from 2013 had a 256 GB SSD, I retired it in november 2020 when I got my M1 Mini and during those 7 years nothing negative has happened to the SSD and it has 1000s of TB of writes. 

Are you sure it's actually that high, my 256GB 840 Pro only has 40TB writes on it, it is however just an OS disk and I use other storage for majority of things but general downloads etc go on it still though. 1PB written on a 256GB SSD is actually a lot, that's 1.56 DWPD which is significantly higher than many server SSDs of that vintage and even today for Read Intensive and Mixed Use (though many are this and higher today rating wise).

 

I suspect you were more using hyperbole, like I totally agree with what you're saying, just multiple PB on a 256GB SSD of 7 years ago is actually extreme.

[Spindel]

6 minutes ago, leadeater said:

Are you sure it's actually that high, my 256GB 840 Pro only has 40TB writes on it, it is however just an OS disk and I use other storage for majority of things but general downloads etc go on it still though. 1PB written on a 256GB SSD is actually a lot, that's 1.56 DWPD which is significantly higher than many server SSDs of that vintage and even today for Read Intensive and Mixed Use (though many are this and higher today rating wise).

 

I suspect you were more using hyperbole, like I totally agree with what you're saying, just multiple PB on a 256GB SSD of 7 years ago is actually extreme.

A bit of a hyperbole true

 

But it has a lot of writes, I can not go into details in this forum, but for a while it was also used for heavy downloading. This means writing a lot of compressed files, and then rewriting them while uncompressing them.

Ingesting photos from my cameras. Autosaves for all kinds of things I work on. Browser cache, software updates etc. It all adds up. 

[igormp]

10 hours ago, Vitamanic said:

I imagine cold booting the machine (and establishing the swap file) 256 times in 2 months is the overwhelming majority of their 600GB of writes. There's always context to be had.

Take a look at the replies to the thread, many people have gone over 5TB usage in less than 2 months.

Also, it's not due to the "establishment of the swap file", since at boot it usually starts at 1gb and increases dynamically as needed.

Quote

They have 191 hours of total usage which comes out to them cold booting the laptop about every 40 minutes. Why?!

He is a dev working on asahi linux, a linux port to the M1, so that machine is meant to be just a test bed for his patches.

 

9 hours ago, Vitamanic said:

Yeah people will do weird stuff. That said, cold booting a machine 256 times in 60 days is going to cause significant writes regardless of the platform.

That has nothing to do with the amount of boots. Here's another proper example:

 

For comparison, here's the 2 month old NVMes in my desktop, where I heavily use Docker, an Android VM, and do tons of data processing:

 

5 hours ago, Spindel said:

I hope this guy doesn't use a web browser, wait until he realizes how much writes to disk the cache causes, or any program with a data base, or anything that writes to disk. 

 

Seriously this is a nothing burger. 

 

With the advent of SSDs swap is actually something useful, not something to dread like it was/is with HDDs. 

 

Oh and for the record my old iMac from 2013 had a 256 GB SSD, I retired it in november 2020 when I got my M1 Mini and during those 7 years nothing negative has happened to the SSD and it has 1000s of TB of writes. 

Have a look at the above 

[Spindel]

4 minutes ago, igormp said:

Take a look at the replies to the thread, many people have gone over 5TB usage in less than 2 months.

Also, it's not due to the "establishment of the swap file", since at boot it usually starts at 1gb and increases dynamically as needed.

He is a dev working on asahi linux, a linux port to the M1, so that machine is meant to be just a test bed for his patches.

 

That has nothing to do with the amount of boots. Here's another proper example:

 

For comparison, here's the 2 month old NVMes in my desktop, where I heavily use Docker, an Android VM, and do tons of data processing:

 

Have a look at the above 

And here is my 3 months old M1 Mac Mini for comparison

 

EDIT:// But as I said in the news thread about this, I don't trust these reporting tools for the M1 macs. First thing that raises suspicion is the power on time reported. I might also add that I have a 16 GB M1 Mini and I almost never see it using swap so it's not the swapping that is driving my disk usage (which in it self for my case is not something I'm worried about).

[Ashley MLP Fangirl]

On 2/15/2021 at 2:08 AM, svmlegacy said:

The M1 is merely comparable to the x86 competition. 

but it's lightyears ahead of competing ARM designs... 

[igormp]

13 minutes ago, Spindel said:

And here is my 3 months old M1 Mac Mini for comparison

Almost 15x less hours than my disks, and almost the same amount of writes, welp.

 

14 minutes ago, Spindel said:

But as I said in the news thread about this, I don't trust these reporting tools for the M1 macs.

That's information from the drive's own controller, so either it's lying or broken somehow. 

 

14 minutes ago, Spindel said:

First thing that raises suspicion is the power on time reported.

Yeah, yours is truly weird. Could you doublecheck with smartctl?

 

15 minutes ago, Spindel said:

I might also add that I have a 16 GB M1 Mini and I almost never see it using swap so it's not the swapping that is driving my disk usage (which in it self for my case is not something I'm worried about).

Sadly you can't control the swappiness parameter on MacOS, so the system chooses when to page stuff in and out of memory. 

 

Anyway, here's another extreme report, where the drive itself reports that it has gone through 10% of its lifetime, meaning that it supposedly should die in another 2~3 years:

 

[Spindel]

9 minutes ago, igormp said:

Almost 15x less hours than my disks, and almost the same amount of writes, welp.

 

That's information from the drive's own controller, so either it's lying or broken somehow. 

 

Yeah, yours is truly weird. Could you doublecheck with smartctl?

 

Sadly you can't control the swappiness parameter on MacOS, so the system chooses when to page stuff in and out of memory. 

 

Anyway, here's another extreme report, where the drive itself reports that it has gone through 10% of its lifetime, meaning that it supposedly should die in another 2~3 years:

 

Can't do it right now. 

 

But apparently @just_dave got the same results with smartmontools as DriveDx (the one I used). 

 

I'm still suspicious about these tools, they report stuff like power on time that in my case is obviously wrong, also my reported temperatures is wrong, impossible that the temps have been fixed at 28 degrees . 

 

But go read this thread where we discuss this thing

 

[igormp]

12 minutes ago, Spindel said:

Can't do it right now. 

 

But apparently @just_dave got the same results with smartmontools as DriveDx (the one I used). 

 

I'm still suspicious about these tools, they report stuff like power on time that in my case is obviously wrong. 

 

But go read this thread where we discuss this thing

 

Didn't know there was a thread for that, good to know. 

I'll follow the discussions there, thanks!

[steelo]

On 2/14/2021 at 8:10 PM, emosun said:

i'd say this bit is debatable. Regardless of method , if the systems aren't much faster in similar tasks i wouldn't consider them that far ahead.

Speed relative to power consumption is another factor , but in the desktop industry people dont care much for efficiency over speed.

I guess another issue would be what third party software would be supported? Would you run mobile apps on it? I have doubts software developers are going to invest in rewriting code meant to work on x86 CPU's unless they know it will take off.