Author Topic: Apple's SSD milking machine...  (Read 5575 times)

0 Members and 1 Guest are viewing this topic.

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #25 on: July 01, 2022, 09:44:16 am »
No, it's really not my imagination. That's your assumption and your argument.

I shoot 4k 60fps video from DLSR and iPhone and work with 3-5 streams (mix RAW and ProRes). That will be fine on an M1 MacBook Air. On a MacBook M2, nope. It's not about aggregate bandwidth either - each stream has to seek blocks and if you hit swap when you are working you need overhead for that.

My old MacBook Pro, an Intel one was slightly better than the M2 measured performance and it was hitting an IO wall regularly.

Edit: also surely you see the irony of complaining about people complaining about something by creating a tangent on a thread about it. Shock-ception! I have better things to do so that was my last comment on the matter :-//
« Last Edit: July 01, 2022, 09:49:04 am by bd139 »
 

Online DiTBho

  • Super Contributor
  • ***
  • Posts: 3798
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #26 on: July 01, 2022, 10:01:11 am »
it's the second time I think these topics are getting more frustrating than Apple products, which is depressing.

(so you can guess why I am more enthusiast for an home-made optic-Kvm than for that usual big *hits.)

Anyway, I am going to return my Apple Studio to Amazon! I mean, I need a similar machine, but I like neither its SSD nor the Apple crazy policy.
The opposite of courage is not cowardice, it is conformity. Even a dead fish can go with the flow
 
The following users thanked this post: bd139

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #27 on: July 01, 2022, 10:06:02 am »
Indeed. It's just fake meta-outrage.

I don't really care. It's a toaster to me. Buy it, use it until AppleCare runs out then lose it before I incur any risk. If I don't like it to start with I return it like you're doing.

tco = (initial_cost + apple_care - sale_cost) / months_apple_care

£32 a month for a studio. I'm thinking about it. That's couch change.

Edit: although I might just blow that on a spectrum analyser instead at the moment...
« Last Edit: July 01, 2022, 10:09:05 am by bd139 »
 

Offline SilverSolder

  • Super Contributor
  • ***
  • Posts: 6126
  • Country: 00
Re: Apple's SSD milking machine...
« Reply #28 on: July 01, 2022, 10:14:52 am »

Is 256GB enough for video editing anyway?  -  Wouldn't you want 1TB or better?
 
The following users thanked this post: Someone

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #29 on: July 01, 2022, 10:20:46 am »
It's fine for most edits I do which are usually < 10 minutes total. Once I'm happy I render it out and nuke the source media as I don't need it. I archive stuff on a 1TB external disk which is rendered and push it to Amazon S3 for backup.

I've got a 512Gb SSD in my 14" MBP and that's sitting at 351Gb free. That contains everything I've ever done as well as the OS and apps.

Only people editing long ass YT crap and films need 1TB or more (or those with a serious data hoarding problem)
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 4510
  • Country: au
    • send complaints here
Re: Apple's SSD milking machine...
« Reply #30 on: July 01, 2022, 10:23:55 am »
No, it's really not my imagination. That's your assumption and your argument.

I shoot 4k 60fps video from DLSR and iPhone and work with 3-5 streams (mix RAW and ProRes). That will be fine on an M1 MacBook Air. On a MacBook M2, nope. It's not about aggregate bandwidth either - each stream has to seek blocks and if you hit swap when you are working you need overhead for that.
So you'd have some benchmarks to back that up? (against a computer you don't own and don't want to buy)  :-DD

Below is one of the "slow" M2 256GB benchmarks on throughput (with results for equivalent prores and compressed raw data rates) boo hoo hoo, only 400 frames per second of 4k prores. Without going to the obvious solution of proxies you'd need to have more than 5 simultaneous 60fps streams to start worrying, which would....  be less than 4 minutes of storage if it filled the entire drive. Its not realistic, real people would just drop to proxies as you're not editing anything of such value (Hollywood movie etc) in that limited space that you need exact pixel playback in realtime.

You picked the applications, and they aren't significantly impacted. Compiling/software is often IO request rate limited more than throughput, I've done those benchmarks on projects when selecting workstations. Throughput above 1000MB/s is limiting almost nothing.
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 4510
  • Country: au
    • send complaints here
Re: Apple's SSD milking machine...
« Reply #31 on: July 01, 2022, 10:29:52 am »
Is 256GB enough for video editing anyway?  -  Wouldn't you want 1TB or better?
Or more likely, some external drives on the the thunderbolt port(s). Still, the point is pretty simple that high throughput on a small sized drive is useless beyond some contrived examples. Reading through the entire data of the drive in under 3 minutes is not a real world use case.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #32 on: July 01, 2022, 10:33:55 am »
Meh your funeral. Measurable difference here outside synthetic benchmarks which you tell me isn't possible and that I'm stupid. All your "evidence" is sourced from a screenshot and some outrage.

And if you read the thread further back I did want to buy an iMac shaped version of it for my daughter.

Yeah great argument. I'm done. Not getting into another one of these fucking threads again. Life's too short.
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 4510
  • Country: au
    • send complaints here
Re: Apple's SSD milking machine...
« Reply #33 on: July 01, 2022, 10:45:54 am »
Meh your funeral. Measurable difference here outside synthetic benchmarks which you tell me isn't possible and that I'm stupid. All your "evidence" is sourced from a screenshot and some outrage.
Measurable difference of what between M1 256GB and M2 256GB units? You have some measure of this you won't share? The example you did share of 3 different computers doing some task has very little to do with their SSD throughput when the all the other variables are a) not stated and b) likey all over the place, and c) the task isn't stressing/challenging throughput.

If there were real world differences, why aren't people posting examples of it? You say you know the truth but can't substantiate it, given the easily found evidence to the contrary your argument is looking full of holes.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #34 on: July 01, 2022, 10:49:04 am »
Code: [Select]
bd139 ~$ unsubscribe-topic
 
The following users thanked this post: JPortici

Online mariush

  • Super Contributor
  • ***
  • Posts: 4983
  • Country: ro
  • .
Re: Apple's SSD milking machine...
« Reply #35 on: July 08, 2022, 06:10:02 pm »
The endurance of that 256 GB of flash memory is probably around 100-150 TB

I'd be a bit worried about editing with such low amount, due to writes.  For a 10 minute video you could have 30-50 GB of content, and you'd render another 5-10 GB ... so 10-20 youtube videos could potentially burn through 1% of your SSDs endurance, and then you'd have to pay for new memory chips.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #36 on: July 08, 2022, 06:37:42 pm »
No it’s not. It’s 1Pb+ these days.

I’ve written 900TB on consumer SSDs.
 

Online Monkeh

  • Super Contributor
  • ***
  • Posts: 7990
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #37 on: July 08, 2022, 07:01:12 pm »
100-150TB is probably reasonable for the very low end of the market - which is a valid section of the market. It's just not on the same page of the atlas as Apple.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #38 on: July 08, 2022, 07:04:10 pm »
I dunno. We burned up ass end crucial and sandisk ones to see what would happen and it was very uneventful  :-//

Best one was the hitachi enterprise PCIe 6.4TB one. Those do 30PB a go in database servers. But not cheap.
 

Online wraper

  • Supporter
  • ****
  • Posts: 16801
  • Country: lv
Re: Apple's SSD milking machine...
« Reply #39 on: July 08, 2022, 07:06:20 pm »
But indeed it was a stupid move from Apple, sent all reviewers the fast 1TB version so everyone is enthousiastic and then crippling the 256GB version by not installing two 128GB flash chips but a single 256GB flash chip, reducing access speeds with a factor of almost 2.
Actually for SSD, it's industry normal that drives of the same model line have worse performance for smaller size. Same controller, same NAND chips but lower count. And thus situation may easily be that same size of newer generation SSD have worse performance than previous since it uses newer generation of NAND of larger size but less channels are used for the same size models.
To PKTKS
Why do you compare M2 prices with different SSD size? AFAIK performance is worse only for M2 with 256GB SSD when compared to M1 with the same SSD size.
 
The following users thanked this post: bd139

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #40 on: July 08, 2022, 07:13:42 pm »
On the M2 my objection is by the time you buy the 512Gb one it’s £1549. You can get a 14” MacBook Pro with more cores, more SoC bandwidth, more ports, twice as much ram, 4.5x the read/write performance and an XDR screen for less than £200 more on Amazon  :-//
 

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 10576
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #41 on: July 08, 2022, 07:15:54 pm »
Code: [Select]
bd139 ~$ unsubscribe-topic

So, you have got a nice warm redhead waiting!  :)

Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #42 on: July 08, 2022, 07:17:57 pm »
Fortunately no  :-DD

I’m only replying to this as I’m being internally murdered by an Asda burrito.
 

Online Monkeh

  • Super Contributor
  • ***
  • Posts: 7990
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #43 on: July 08, 2022, 07:25:23 pm »
I dunno. We burned up ass end crucial and sandisk ones to see what would happen and it was very uneventful  :-//

Best one was the hitachi enterprise PCIe 6.4TB one. Those do 30PB a go in database servers. But not cheap.

Try some WD Greens for some proper ass-end. About as bad as you can get commercially unless you're PKTKS and buy Aliexpress specials. And yet, they still are fine for basic tasks to replace spin-stabilised data storage..
 
The following users thanked this post: wraper, bd139

Online wraper

  • Supporter
  • ****
  • Posts: 16801
  • Country: lv
Re: Apple's SSD milking machine...
« Reply #44 on: July 08, 2022, 07:25:57 pm »
After a bit of thought, the reason they used a newer generation of NAND instead 2x of older one may be not money skimping at all. M2 will be produced for longer and as NAND manufacturers like to phase out production of their older NAND and mostly produce latest generation, they may need to move to newer NAND in future production of M2. And then it would be a real scandal if they at some point nerfed performance of 256GB M2. EDIT: also clinging to old NAND during chip shortage is not the best idea FWIF.
« Last Edit: July 08, 2022, 07:30:03 pm by wraper »
 
The following users thanked this post: bd139

Online mariush

  • Super Contributor
  • ***
  • Posts: 4983
  • Country: ro
  • .
Re: Apple's SSD milking machine...
« Reply #45 on: July 08, 2022, 07:45:27 pm »

The amount of writes you can do on a SSD varies depending on type ... SLC, MLC, TLC , QLC from best to worse.
Very few SSDs are still MLC, majority of drives these days are either TLC or QLC

Examples of TLC drive

WD SN570 (I think it's 112 layer TLC memory chips) : https://cdn.cnetcontent.com/syndication/feeds/wd/inline-content/C2/2D253893B5A85CE8F7AAFA3C5D3B0CF3E5D96C5B_PRODUCTBRIEFWDBLUESN570NVMESSD_source.PDF

Note it's 150 TBW for the 250 GB model,  300 TBW for the 500 GB model and 600 TB for the 1 TB model.

Same ratings for the  WD SN550, here's datasheet (which uses 96 layer TLC memory if my memory is correct) :  https://documents.westerndigital.com/content/dam/doc-library/en_us/assets/public/western-digital/product/internal-drives/wd-blue-nvme-ssd/data-sheet-wd-blue-sn550-nvme-ssd-idk.pdf

The 112 layer memory is technically a bit worse, but they stretch the amount of writes by using bigger amount of write cache and a newer better controller - they switch some amount of TLC memory to pseudo-SLC mode.

Samsung 980 uses the latest TLC Samsung makes, also has 160/300/600 TB rating : https://semiconductor.samsung.com/resources/data-sheet/Samsung_NVMe_SSD_980_Data_Sheet_Rev.1.1.pdf

Samsung 980 has one of the biggest write caches, if there's enough free space it has up to around 60 GB of pseudo-SLC write cache (for the 1 TB model) ...

QLC memory is much worse .. ex Kingston NV1 drives are rated 240 TB for the 1 TB model, Samsung QVO drives are rated at around 360 TB for the same 1 TB model.

Sure, yes, you can actually write more than these estimated amounts, but some blocks of memory will wear out and the SSD will degrade.


And as for the M1 apple machines... thread reminded me of this, not sure if it was fixed or not : https://twitter.com/marcan42/status/1494213855387734019?lang=en





Quoting from the thread

Hector Martin
@marcan42
Well, this is unfortunate. It turns out Apple's custom NVMe drives are amazingly fast - if you don't care about data integrity.

If you do, they drop down to HDD performance. Thread.
9:35 AM · Feb 17, 2022·Twitter Web App

For a while, we've noticed that random write performance with fsync (data integrity) on Asahi Linux (and also on Linux on T2 Macs) was terrible. As in 46 IOPS terrible. That's slower than many modern HDDs.

We thought we were missing something, since this didn't happen on macOS.

As it turns out, macOS cheats. On Linux, fsync() will both flush writes to the drive, and ask it to flush its write cache to stable storage.

But on macOS, fsync() only flushes writes to the drive. Instead, they provide an F_FULLSYNC operation to do what fsync() does on Linux.

So effectively macOS cheats on benchmarks; fio on macOS does not give numbers comparable to Linux, and databases and other applications requiring data integrity on macOS need to special case it and use F_FULLSYNC.

How bad is it if you use F_FULLSYNC? It's bad.

Single threaded, simple Python file rewrite test:

Macbook Air M1 (macOS):
- flushing: 46 IOPS
- not: 40000 IOPS

x86 iMac + WD SN550 1TB NVMe (Linux):
- flushing: 2000 IOPS
- not: 20000 IOPS

x86 laptop + Samsung SSD 860 EVO 500GB SATA:
- flushing: 143 IOPS
- not: 5000 IOPS

So, effectively, Apple's drive is faster than all the others without cache flushes, but it is more than 3 times slower than a lowly SATA SSD at flushing its cache. Even if all you wrote is a couple of sectors. You pay a huge flush penalty if you do *any* writes.

Here, "flushing" on macOS means F_FULLSYNC and "not" means fsync(); on Linux both are fsync(), but "not flushing" is measured by telling Linux that the drive write cache is write-through (which stops it from issuing cache flushes).

Note that the numbers are filesystem-dependent (and encryption makes things more complicated); e.g. the SATA SSD numbers double on VFAT vs. my root filesystem (ext4 on LVM on dm-crypt), but the pattern is clear.

macOS doesn't even seem to try to proactively issue syncs; you can write a file on macOS, fsync() it, wait 5 seconds, issue a hard reboot (e.g. via USB-PD command), and the data is gone. That's pretty bad.

Of course, in normal usage, this is basically never an issue on laptops; given the right software hooks, they should never run out of power before the OS has a chance to issue a disk flush command. But it certainly is for desktops. And it's a bit fragile re: panics and such.

Unfortunately, this manifests itself as quite visible issues on Linux. For example, apt-get on Asahi Linux is noticeably slow. Making fsync() not really flush on macOS is not fair; lots of portable software is written to assume fsync() means your data is safe.

Our current thinking is we're going to add a knob to the NVMe driver to defer flush requests up to a maximum time of e.g. 1 second. That would ensure that a hard shutdown never loses you more than 1 second of data, which is better than what macOS can claim right now.

Alas, that's still not quite safe. Not flushing means we cannot guarantee ordering of writes, which means you could end up with actual data corruption in e.g. a database, not just data loss. There's no good way around this other than doing full flushes.

So the unfortunate conclusion is that if you're e.g. running a transactional database on Apple hardware, and you need to be able to survive a hard poweroff without data corruption, you're never going to get more than ~46 TPS.

Unless Apple improves their ANS firmware to fix this.

And for what it's worth, I inadvertently triggered a data consistency issue in macOS while testing this. Before running any tests I had GarageBand open. I closed it without saving the open project. After the first hard reboot later, it tried to reopen it and threw up an error.

So I guess the unsaved project file got (partially?) deleted, but not the state that tells it to reopen the currently open file on startup.

Data consistency matters.
 
The following users thanked this post: SilverSolder

Online Monkeh

  • Super Contributor
  • ***
  • Posts: 7990
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #46 on: July 08, 2022, 07:52:00 pm »
Note it's 150 TBW for the 250 GB model,  300 TBW for the 500 GB model and 600 TB for the 1 TB model.

Bearing in mind these are numbers for warranty purposes, which makes them highly conservative.
 

Offline bd139

  • Super Contributor
  • ***
  • Posts: 23018
  • Country: gb
Re: Apple's SSD milking machine...
« Reply #47 on: July 08, 2022, 07:53:43 pm »
Oh not the flush thing again.

1. Consistency is guaranteed.
2. Currency of data is not.
3. Computer has battery.
4. It’s a client computer not a server.

It’s perfect trade off for workstation and desktop use.  :-//
 

Offline Bassman59

  • Super Contributor
  • ***
  • Posts: 2501
  • Country: us
  • Yes, I do this for a living
Re: Apple's SSD milking machine...
« Reply #48 on: July 18, 2022, 06:29:00 pm »

Is 256GB enough for video editing anyway?  -  Wouldn't you want 1TB or better?

Everyone I know who does video and audio edition keeps the assets on external drives.
 
The following users thanked this post: bd139

Offline SilverSolder

  • Super Contributor
  • ***
  • Posts: 6126
  • Country: 00
Re: Apple's SSD milking machine...
« Reply #49 on: July 18, 2022, 07:19:39 pm »

Is 256GB enough for video editing anyway?  -  Wouldn't you want 1TB or better?

Everyone I know who does video and audio edition keeps the assets on external drives.

Not while editing, surely? - you'd want max disk access speed for that?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf