I mean, in virtually every conceivable metric they run circles around the intel machines. The fact that you are able to find one sole metric (random 4K writes) where they perform marginally better is hardly enough evidence to the contrary to be salient.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Disk speed testing on Apple Silicon: AmorphousDiskMark, Blackmagic, etc. [MERGED]
- Thread starter jdb8167
- Start date
- Sort by reaction score
I mean, in virtually every conceivable metric they run circles around the intel machines. The fact that you are able to find one sole metric (random 4K writes) where they perform marginally better is hardly enough evidence to the contrary to be salient.
Good points. Also the screen is just amazing on the 14" (not just size but quality).
You know that you're in the thread that was specifically made for comparing those other metrics, right? 😉
And in that specific metric, the Intel iMac having 7.4 times the performance of the M2 Pro MBP is hardly a "marginal" difference.
4K reads / writes are at the core of OS/Database operations. It's not just SSD performance, ramdisk performance is greatly lagging in 4k performance, pointing to a larger issue that wasn't corrected with the M2 chip.
Perhaps this is why Apple focuses on pre-rendered media operations with large file transfers?
Apple Silicon uses 16 KiB pages. I would not be surprised if the 4 KiB reads/writes are really 16 KiB behind the scenes, because there is never really a reason to read only 4 KiB.
They are located in my specs below….. if you can’t read them, or have them turned off in your settings, here they are..Don't forget to tell us the device/specs for that test result.
14” MacBook Pro M2 Max 32c cpu 38c gpu 1 tb ssd 👍
Done in hardware and always encrypted. Just depends on whether you add your own password to the mix.
I don’t have the 1TB but here are some 2TB results:
Attachments
Anyone tested AmorphousDiskMark with 16" M2 Pro/Max and 8TB?
Only found this online (in comments): https://www.macrumors.com/2023/01/24/hands-on-m2-pro-mac-mini/
"Mac Mini M2 Pro at 2TB, 4TB, or 8TB SSD is about 6,000 MB/s. (This needs to be confirmed.)"
and this:
"We can also choose a MacBook Pro 14" or 16" M2 Pro or M2 Max model now, with a 512GB SSD drive having maximum 3,000 MB/s read/write Sequential speeds, or upgrade to a 1TB or larger SSD model, with maximum 6,000 MB/s read/write Sequential speeds"
Would love confirmation with a screenshot here.
Hopefully Apple will keep on with this minimum speed for larger disks in the future.
Cheers.
Only found this online (in comments): https://www.macrumors.com/2023/01/24/hands-on-m2-pro-mac-mini/
"Mac Mini M2 Pro at 2TB, 4TB, or 8TB SSD is about 6,000 MB/s. (This needs to be confirmed.)"
and this:
"We can also choose a MacBook Pro 14" or 16" M2 Pro or M2 Max model now, with a 512GB SSD drive having maximum 3,000 MB/s read/write Sequential speeds, or upgrade to a 1TB or larger SSD model, with maximum 6,000 MB/s read/write Sequential speeds"
Would love confirmation with a screenshot here.
Hopefully Apple will keep on with this minimum speed for larger disks in the future.
Cheers.
Thank you, I still can’t find a test on the 1tb! I’m trying to decide what to buy.
This is my work 16” mbp M1 Pro 16gb ram and 512gb base model
Everything besides the 512GB base model is pretty much the same speed. You will generally see slight increases as you get larger and larger capacities, but this is minor. There will be almost as much variation just between different units of the same drive due to manufacturing tolerances etc. Just because someone with a 1TB posts their speeds, it doesn’t mean you will get exactly those speeds if you buy the exact same config. Also, there are a lot of other factors that will affect results, such as how full the drive is, what background processes may or may not be running, etc.
If you get 1TB or higher, you can pretty much count on at least 6000 read/write speeds. I have seen as high as 7100 for reads and 7700 for writes, but there is no guarantee for anything beyond 6000 or so. And just because you get 7700 once on a test doesn’t mean you will constantly be getting that speed either. There‘s much more at play and while these tests results are a useful tool, it seems like people are starting to place an outsized importance on them and are in some cases even drawing reductive conclusions from them.
Yeah that's very awesome speeds to be honest. Feels reassuring since I'm leaning toward the laregest one (8TB). How come there's slight increases as we get larger capacities? What's the technical reason for it? Just curious.
There are numerous reasons why and it's a bit more complicated than this, but to be a bit reductive, two big factors are cache and spare area. We don't know the specifics of how Apple's unique combo of bare flash chips and SoC-based controller are manufactured, but with SSDs in general, what is typical is that they usually have a smaller portion of the drive made up of more expensive, faster performing flash (perhaps SLC or MLC cells) and then a larger portion made up of cheaper, slower performing flash (perhaps TLC or QLC cells). The drive first writes to this faster flash, but for especially large transfers, once it is filled up, the drive then has to fall back to writing to the slower cells and this often causes a big drop-off in performance. Since the amount dedicated to the faster cells is proportional, larger drives will have a larger area of faster flash.
Another reason is spare area. Since drive manufacturers know some flash cells will become unusable due to wear during the lifespan of the drive, they set aside "spare area" of extra flash that is not counted in the total capacity. As cells are worn down due to wear, they are sealed off so to speak and replaced with fresh new cells from this spare area. Larger drive capacities will have more spare area, and therefore will have less slowdown with heavy use over time due to more cells set aside for wear levelling.
Finally, some of the same concepts that apply to HDDs apply here too. Even though SSDs are usually much faster at random reads and writes than HDDs, even SSDs are faster at sequential than random. So as a drive gets more or mostly filled up, (especially for large files) if there are not enough contiguous portions of free space, even though the data could be written sequentially in theory, because there is not a big enough chunk of contiguous space free, it has to be written to different areas of the drive anyway. SSDs will not be affected as badly as HDDs by this, but they will still perform worse than they would if they had tons of free space in this case. Also, because of the way that SSDs write data, doing something like defragmentation is not so simple, and in many cases could actually further decrease the performance of the drive, by increasing wear. Finally, with SSDs, if someone's drive is showing 30% free, but just before that it was 99% full, and they just deleted 29% of the content, then that drive is still going to be almost as badly fragmented as it was when it was 99% full.
I should also note that with modern versions of macOS, it is less apparent unless you are very savvy how much drive space is actually free, as macOS reports used (but purgeable) space as free space. So with APFS file duplication methods, snapshots, caches other background processes going on, often a lot more of the drive is actually filled than what meets the eye.
Thanks for posting both of these - curious to know if you find the M2 Macbook any faster than the M1? Looking at CPU benchmarks it seems to be only 10-20% faster but that seems to be negated by the reduced SSD performance. So I am guessing there is no noticeable difference in everyday use?
I never used them at the same time, so it's hard to tell. I don't believe I will ever do anything that will push the limits of the M2's SSD speed, so it's basically a non issue for me.
I'm actually more worried about the M2 running hotter than the M1, because of the higher clock speed, increased power consumption and smaller heat sink. I'm not sure if that's the case, but I have felt the M2 getting slightly warm on a couple of occations, whereas I never did that with the M1. But I only had the M1 for a few days, so I didn't use it that much.
Make sure you have the latest version. I had an old version and the results were inconsistent with the new version.
NOTE: Please do not repeat the read/write measurements unnecessarily. Repeating read/write measurements may shorten the lifespan of your storage device.
AmorphousDiskMark for macOS - measures storage read/write performance
AmorphousDiskMark measures storage read/write performance in MB/s and IOPS.www.katsurashareware.com