M2 MacBook Pro SSD read/write speeds compared to M1

[mattyjoe]

Okay, there are a couple YouTubers who have done BlackMagic Disk Speed Tests on the M2 13inch MacBook Pro, and have found that the speeds are SLOWER than the M1 13inch. I've done some investigating and discovered what I think the difference is.

MaxTech got a write of 1463 MB/s and read of 1446 MB/s, give or take a few MB between each run. His model is a 256GB. He opened it up, and showed that Apple is now using one 256GB NAND chip for it, while the M1 has 2x 128GB NAND chips.

zoneoftech also got similar results on his 256GB model.

Meanwhile, zollotech got very fast speeds on his 512GB M2 model.

So basically, it appears Apple is using a single NAND chip for 256GB models, and likely 2x 256GB chips for 512GB models, which is the reason for the faster speeds with 512GB. It's likely Apple is using 2 NAND chips for every capacity over 256, so as long as you get a model with 512GB or above, you should not see a reduction in speed.

MaxTech, for one, is already using this to say that Apple made the M2 machine worse. For the 256GB model, yes. But someone pointed out to me that 128GB NAND chips are being made much less anyway, and the majority of them are made for phones. It may just be a supply issue that caused Apple to go with single 256GB NANDs.

Most people, I would think, would get 512GB or above anyway.

 

[BeatCrazy]

Most people, I would think, would get 512GB or above anyway.

I think you underestimate. The standard retail configuration for non-Apple resellers is the 256GB. Which means all those stocked/bought at Best Buy, Nebraska Furniture Mart and similar is what the majority of people end up with.

 

[mattyjoe]

Here are links to the YouTube videos, with timestamps.

MaxTech:

zoneoftech:

zollotech:

 

[mattyjoe]

I think you underestimate. The standard retail configuration for non-Apple resellers is the 256GB. Which means all those stocked/bought at Best Buy, Nebraska Furniture Mart and similar is what the majority of people end up with.

You're probably right. So, my hope is prospective buyers see this and if 2000MB/s+ read and write is important to them, they will get a 512GB model or higher.

 

[kappisto]

In principle that's kinda crappy. In practice, the venn diagram between people who will actually notice those disk speeds and people who only need 256GB storage is going to be pretty small.

 

[FSMBP]

Damn - that's quite a huge difference in speed too. Not trying to defend Apple, but I imagine people getting base machines aren't tasking their computers with read/writes; probably like most people I've seen, they're using Chrome/Zoom for everything.

 

[drecc]

Damn - that's quite a huge difference in speed too. Not trying to defend Apple, but I imagine people getting base machines aren't tasking their computers with read/writes; probably like most people I've seen, they're using Chrome/Zoom for everything.

If you're swapping out to RAM, then being able to quickly read a GB or two from RAM as you switch apps would be critical.

 

[drecc]

Does anyone else get significantly different SSD read/write speeds between Blackmagic and AmorphousDiskMark tests?

I set both to 4GB on my 2018 MBP, and the results are as follows:

 

[mattyjoe]

Does anyone else get significantly different SSD read/write speeds between Blackmagic and AmorphousDiskMark tests?

I set both to 4GB on my 2018 MBP, and the results are as follows:

View attachment 2023547

View attachment 2023548

I also have the same discrepancy on my M1 Max 16inch.

 

[drecc]

View attachment 2023553
View attachment 2023554
I also have the same discrepancy on my M1 Max 16inch.

Thanks for checking! This makes me a little optimistic that the new M2 Air will have a better SSD than my 2018 MBP.

 

[Krypton Deer]

my hope is prospective buyers see this and if 2000MB/s+ read and write is important to them, they will get a 512GB model or higher.

the venn diagram between people who will actually notice those disk speeds and people who only need 256GB storage is going to be pretty small.

It's the Pro name that triggers me. This M2 MBP is cheaper than M2 Air at the same config (M2 MBP starts at 10c) and I agree 8/256 MBP users are not likely to task the system enough to notice the difference.

But 8GB ram requires swapping to SSD whenever system is tasked and two-spec BTO (16/512) pushes the price uncomfortably to MBP 14 level. Given the supply constraints and inflation, maybe Apple will further increase M2 MBP 14/16 prices next year sigh.

 

[mattyjoe]

Keeping my eye out for more reviews to see if this is consistent across 256 and 512 GB models.

 

[JPack]

The rule for M1 was to buy 8/512 or 16/256 given the swap characteristics. It's clear that rule is more important now than ever.

 

[AAPLGeek]

Does anyone else get significantly different SSD read/write speeds between Blackmagic and AmorphousDiskMark tests?

I set both to 4GB on my 2018 MBP, and the results are as follows:

View attachment 2023547

View attachment 2023548

Are you using the same test data samples on both these apps? Amorphous shows 4GiBx1 and it should be the same on Blackmagic to get an accurate assessment.

In any case, modern SSD speeds tested with Blackmagic are completely pointless since it only test sequential read and write speeds. Sequential R/W performance is the least important metric when gauging day to day performance of a SSD.

 

[drecc]

Are you using the same test data samples on both these apps? Amorphous shows 4GiBx1 and it should be the same on Blackmagic to get an accurate assessment.

In any case, modern SSD speeds tested with Blackmagic are completely pointless since it only test sequential read and write speeds. Sequential R/W performance is the least important metric when gauging day to day performance of a SSD.

Yes, AmorphousDiskMark only does powers of 2, and Blackmagic does 1..5, so I set them both to 4GB.

> Sequential R/W performance is the least important metric

I'm guessing that when you swap back memory when switching to an app or to tabs that you've not used in a while, those are multi-MB sequential reads. The sequential writes happening in the background probably aren't critical, but I'd guess that the sequential reads are.

 

[drecc]

View attachment 2023553
View attachment 2023554
I also have the same discrepancy on my M1 Max 16inch.

Now this is very strange - I found a blog post here https://blog.greggant.com/posts/2021/03/25/amorphousdiskmark-is-crystaldiskmark-for-mac.html that shows that the RND4KQD64 read performance on an M1 Air is better (1144 MB/s) than on the M1 Max MBP 16 (674MB/s)....

Although the blog post is for 1GB and the benchmark from this thread on the M1 Max is at 4GB, so I wonder if the M1 Air was returning cached data from RAM... If so, then maybe all of these benchmarks are useless.

 

[antiprotest]

I think you underestimate. The standard retail configuration for non-Apple resellers is the 256GB. Which means all those stocked/bought at Best Buy, Nebraska Furniture Mart and similar is what the majority of people end up with.

Correct. Most people buy the cheapest configuration and try to live with it even though it's not practical enough. And then they face all kinds of limitations and keep trying to talk themselves into believing that's it's acceptable, even the smarter choice because they "saved" money and they could use cloud storage etc. It's really Apple's fault but people shouldn't make themselves accept it.

 

[BeatCrazy]

Correct. Most people buy the cheapest configuration and try to live with it even though it's not practical enough. And then they face all kinds of limitations and keep trying to talk themselves into believing that's it's acceptable, even the smarter choice because they "saved" money and they could use cloud storage etc. It's really Apple's fault but people shouldn't make themselves accept it.

For the price/demographic 256GB + optional cloud storage really is enough. It's not like Apple's not letting you buy more storage. Unless I'm buying for myself, I'm very price conscious and trying to get keep the total investment as low as possible.

My wife and kids each get by just fine with 256GB + cloud. Do they really need full-res versions of their 300 cat videos on the device? They're not editing dozens of 500MB photos in Premiere. Nor are they downloading dozens of 2hr movies or albums. All that stuff just streams nowadays.

 

[erikkfi]

Sure there may be reasons for this, but ultimately this is a cut corner on Apple's part.

 

[TheMacDaddy1]

Okay, there are a couple YouTubers who have done BlackMagic Disk Speed Tests on the M2 13inch MacBook Pro, and have found that the speeds are SLOWER than the M1 13inch. I've done some investigating and discovered what I think the difference is.

View attachment 2023492

MaxTech got a write of 1463 MB/s and read of 1446 MB/s, give or take a few MB between each run. His model is a 256GB. He opened it up, and showed that Apple is now using one 256GB NAND chip for it, while the M1 has 2x 128GB NAND chips.

View attachment 2023493

zoneoftech also got similar results on his 256GB model.

View attachment 2023495

Meanwhile, zollotech got very fast speeds on his 512GB M2 model.

View attachment 2023496

So basically, it appears Apple is using a single NAND chip for 256GB models, and likely 2x 256GB chips for 512GB models, which is the reason for the faster speeds with 512GB. It's likely Apple is using 2 NAND chips for every capacity over 256, so as long as you get a model with 512GB or above, you should not see a reduction in speed.

MaxTech, for one, is already using this to say that Apple made the M2 machine worse. For the 256GB model, yes. But someone pointed out to me that 128GB NAND chips are being made much less anyway, and the majority of them are made for phones. It may just be a supply issue that caused Apple to go with single 256GB NANDs.

Most people, I would think, would get 512GB or above anyway.

99.999999999% of Macbook users would never know. Their usage probably never allows the disk to reach maximum speed because the disk use it lite and bursty.

 

[JPack]

Sure there may be reasons for this, but ultimately this is a cut corner on Apple's part.

The only reason is cost. There is no other legitimate reason.

 

[Trungn1]

Does the m1 air have 2 NAND card as well?

 

[yitwail]

The only reason is cost. There is no other legitimate reason.

It's the practical reason for doing this, one less part to stock and install, but it's debatable how legitimate it is to sell a pro machine with half the ssd speed of a base M1 MBA. Here's something else questionable I just noticed. In Apple's M2 MBP page, they claim it's up to 1.4x faster than the M1 model, but if you read the fine print, they compared a 24GB M2 to a 16GB M1. 🤦‍♂️

 

[JPack]

Press Center - NAND Flash Market Oversupplied in 3Q22, Price Quotes to Drop by 0~5%, Says TrendForce | TrendForce - Market research, price trend of DRAM, NAND Flash, LEDs, TFT-LCD and green energy, PV

According to TrendForce research, as output from Kioxia and WDC grows month by month, it has become obvious that production capacity is sufficient to meet increased bit demand. However, post-pandemic demand for consumer electronics such as laptops will no longer lead to flagging orders. Coupled...

www.trendforce.com

 

[ian87w]

Can't wait for Apple to disable new features on future macOS on these Macs because the SSD speed is not up to their "satisfaction."