Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Completely irrelevant... the Air doesn’t have the performance to leverage the speed of the current SSD anyway.
Would be great if people here stop whining and bring a single scenario that they use on their Air in which they would notice the slower SSD. I think even archive extraction is CPU limited on the Air...

It is, but that wasn't my point, fact is Apple uses more expensive MLC chips, I myself wouldn't touch TLC, just my preference.
I wouldn’t generally say TLC or even QLC is bad. In fact, MLC these days lasts longer than 1st gen SLC SSDs did. And a decent TLC SSD will live longer than cheap MLC SSDs. Personally, I would prefer if the device is delivered with a 1024 GB QLC SSD, practically, that’s the same as 512GB MLC. After a while the SSD will age and have trouble operating in QLC. Then it should be possible to select MLC or even SLC Mode in the firmware, which should extend the useful life of the NAND chips well beyond the rest of the laptop at the cost of actual storage dropping to 512 or 256GB respectively. That said... you really need to be a heavy user to exhaust modern SSDs.
 
So you are telling me, based on your anecdote, that Anandtech and several other respectable tech reviewing sites are wrong to use IOPs?

Please enlighten me.

No, they're nto wrong to use IOPS. it's just a stat that represents a perspective ideal capability of the device. But, there are always other factors at play, such as how the OS/Software performs it's IO. Latency, device type. type of load, etc. IOPS is great number, but it can be extremely misleading if applied without understanding the whole picture.

In my anecdote. While the Nutanix cluster could handle millions of IOPS a second. And via many benchmarks I am able to reproduce. They do their disks writes Asynchronously. As it writes to multiple disks over interconnect, rather than completely local. Because of that, software that requires synchronous disk writes (some database engines) receive a quite noticable latency in their disk IO performance. This resulted in the same database, running near the same parity as 10 year old production hardware. Not a wise investment considering the technological upgrades that would have coincided. I managed to build out a better system, for a fraction of the cost. While it didn't have quite as much IOPS (i don't remember the number I got it too, but I may have overprovisioned. the CPU bottlenecks before storage). It now runs the same back end much faster. Overnight processing down from 8 hours to 2, and backups now take 5 minutes instead of 1 hour!


Just pointing out that sometimes, picking the highest number is not always the most beneficial, nor valuable thing to do.
 



The 2019 MacBook Air, refreshed last week, appears to have a slower SSD than the 2018 MacBook Air, according to testing by French site Consomac. Using testing with the Blackmagic Disk Speed benchmarking test, the site found that the read speeds of the new SSD are lower.

A test of the 2019 MacBook Air with 256GB of storage demonstrated write speeds of 1GB/s and read speeds of 1.3GB/s. An equivalent model released in 2018 featured write speeds of 920MB/s and read speeds of 2GB/s. While write speeds are on par with the older machine (and are even slightly better), read speeds have dropped 35 percent.

blackmagicdiskspeedtest-800x500.jpg

Consomac also saw write speeds of 500MB/s in the 128GB 2019 MacBook Air and read speeds of 1.3GB/s, but this is similar to the performance of the 128GB 2018 MacBook Air as that machine also featured large differences between read and write performance. Higher capacity SSDs were not tested, but may display the same slight decline in performance.

The 2019 MacBook Air features an updated True Tone display and a price drop, starting at $1,099 instead of $1,199. Students are able to get the new machine even cheaper, with the MacBook Air now priced at $999 with educational pricing.

It's possible Apple went with slower SSD performance in order to drop the MacBook Air's price to a more affordable level, and it's not a change that most MacBook Air users are likely to notice in day to day usage of the machine, especially those upgrading from a much older model.

Article Link: Apple's 2019 256GB MacBook Air Includes Slower SSD Than 2018 Model
[doublepost=1563229077][/doublepost]Love the updated MacBook Pro! Went with the 256gb storage. The quad core is an amazing upgrade and plus the price drop and education discount with free beats!! ☺️ Sorry realize this is a MacBook Air post but had to share!
 
The read and write speeds of SSD's are overrated. What matters a lot more in real world use is the access times.

Pretty much ANY SSD has access speeds 100 times that of a spinning hard drive, and that's where the real payoff is.

Only if you're copying super large files might you start to see the benefit of crazy fast read/write speeds. And even then, only if you're copying that super large file from and equally fast drive (otherwise the source drive will be the bottleneck).
Well said, though i am about to say the same thing. :D You take the cake.
Just like internet download/upload speed barely matters most of the time, but delay and jitter matters (as well as content access). Instead, people are super hyped about speed and conveniently ignore everything else, just like CNET testing 5G download speed (not even uploading!) and ignore everything else.
Nice to see logic and brains make a comment for once.
And this becomes rarer over time as general public’s brains are losing power to form comprehensive and formal opinions. Sad nature.
One way or the other Apple will screw you. Thanks Tim!
Who the f god knows how many times Apple will screw us behind the scene that we may never know?
Oh, apple stops providing new MacBooks that can be downgraded to High Sierra. Sad.
 
  • Like
Reactions: BigBoy2018
But.... Muh work flow! Im a "pro" who needs everything i have is the absolute best, even when buying the cheapest and slowest machine available from Apple. And then i xomplain it is too expensive, so people dont realize im a spoiled rich dude.

**SARCASM WARNING**
 
  • Like
Reactions: michaelsviews
The logic people are using here is interesting. The point isn't whether or not people NEED the speed or can tell the difference, or blah, blah, blah. The point is, it wouldn't have cost Apple but pennies to use a decent SSD. In case you're not aware, the cost of SSD's has been plummeting lately. So there really isn't a reason to put in a slower SSD.

What's even more embarassing is Apple wants $200 for an extra 128GB of SLOW SSD!!! LOL! Enjoy!!!
 
No, they're nto wrong to use IOPS. it's just a stat that represents a perspective ideal capability of the device. But, there are always other factors at play, such as how the OS/Software performs it's IO. Latency, device type. type of load, etc. IOPS is great number, but it can be extremely misleading if applied without understanding the whole picture.

In my anecdote. While the Nutanix cluster could handle millions of IOPS a second. And via many benchmarks I am able to reproduce. They do their disks writes Asynchronously. As it writes to multiple disks over interconnect, rather than completely local. Because of that, software that requires synchronous disk writes (some database engines) receive a quite noticable latency in their disk IO performance. This resulted in the same database, running near the same parity as 10 year old production hardware. Not a wise investment considering the technological upgrades that would have coincided. I managed to build out a better system, for a fraction of the cost. While it didn't have quite as much IOPS (i don't remember the number I got it too, but I may have overprovisioned. the CPU bottlenecks before storage). It now runs the same back end much faster. Overnight processing down from 8 hours to 2, and backups now take 5 minutes instead of 1 hour!


Just pointing out that sometimes, picking the highest number is not always the most beneficial, nor valuable thing to do.

In this case, an SSD is tested on Windows 10 which is similar in performance in IOPs to OS X when it comes to random 4k reads. For exmaple, Samsungs EVO 860s have IOPs of 90k (same for WD Blue SSDs). I'm betting Apple is using some form of Toshiba SSD which has comparable perfomance due to controller. Oh, and these are laptops for consumers, not databases or servers; using rated IOPs is fine.

It's a limes to lemons comparisson in the end, but it holds true nonetheless.
 
  • Like
Reactions: LordVic
I knew there was a catch! Surely Apple hadn't changed their business model that much with the price drops on MBA/MBP and cutting prices on customizations. Something had to give

I kind of feel people here had it coming. people kept complaining about storage prices, and comparing it to ssd prices online. They usually were comparing to much slower options. So Apple said fine... if u want to compare to inferior parts, we will give em to you, for cheaper price. lol.
 
I kind of feel people here had it coming. people kept complaining about storage prices, and comparing it to ssd prices online. They usually were comparing to much slower options. So Apple said fine... if u want to compare to inferior parts, we will give em to you, for cheaper price. lol.
[doublepost=1563229077][/doublepost]Love the updated MacBook Pro! Went with the 256gb storage. The quad core is an amazing upgrade and plus the price drop and education discount with free beats!! ☺️ Sorry realize this is a MacBook Air post but had to share!
considering the slim price difference... I am sure more will purchase the MBP who were maybe on the fence? ... I know I was....
 
More penny pinching from Timmy? who would have thought! Look at the bright side, this laptop was made for "you".
 
In a previous job I worked in a software development firm. I can safely claim that 90% of the developers might be good at writing code, but almost all of them couldn't tell how a computer actually works, or what each component of a computer is for. as long as the software had the desired output, they were happy.
As a guy who's bee writing software for 37 years, I disagree.

Netadmins crack me up.
 
Last edited:
  • Like
Reactions: 09872738 and nvmls
It makes the Macbook Pro 13" more compelling. Of course, I would wait for the new keyboard design at this point.

It should be pointed out that this as always been the case with the Air though. The Pro 13" has always been the better deal when performance is factored in.
 
  • Like
Reactions: albebaubles
It makes the Macbook Pro 13" more compelling. Of course, I would wait for the new keyboard design at this point.

It should be pointed out that this as always been the case with the Air though. The Pro 13" has always been the better deal when performance is factored in.
Spot on. I bought an Air once (the small one, 11") years back out of curiosity -- knew it wouldn't handle my coding needs but was curious if I would prefer over my iPad -- was was severely disappointed - gave it to my sister (a teacher) who has been using it with no complaints for 4 years.

Everyone's mileage varies based on needs.
 
I wonder if they went with cheaper SSD - TLC?

Eliminate the 128GB SSD altogether, doubling the size gives higher write speeds and much more endurance. win for apple, win for customers, it is illogical for apple to keep the 128GB SSD

SSD capacity almost follows moore's law, the size doubles for the same price every 18 months or so. It has been long apple is keeping the capacities at the same price. (on the other hand, RAM has difference pricing strategy, where it is expensive when the tech (DDR4 and such) is new as well as when the tech is old (parts are hard to come by, where manufacturer stops producing them DDR2).

the real MBA to buy is mostly the one with 10th Gen Intel CPU, hopefully that time it starts with 256GB/512GB capacities.
 
Last edited:
Not so, "Big Boy"

Have you ever heard of "virtual memory" and "paging"?

The speeds shown above are aligned to today's ultra, low-cost SATA 3.0 SSDs and not PCMCIA SSDs.

The internal SATA SSDs with 250GB capacity are now retailed for less the $40. (Check NewEgg, et al.)

The amount of swapping a system will perform depends on the amount of AVAILABLE MEMORY.
AVAILABLE MEMORY::= FREE + CACHED MEMORY
It is not only for "large files".

And, if you use, for example, a Windows VM (to access corporate and educational programs not available under MacOS), then you are going to swap like crazy, when you only have 8GB of memory.

No offense dude. I think you need to be (re)educated on virtual memory.
BlackMagic test: 5GB file
VM page: 4k

This benchmark is indeed only for large files.

Also, PCMCIA?! These transfer speeds would have been miraculous on PCMCIA cards... Did you benchmark a SunDisk card?
Not necessarily. All OS's will do some paging even if you have sufficient RAM. A lot of software also uses temp files on disk that can benefit from speed.

There are things that matter with NVME. yes, if you're average use is just firing up notepad, word, chrome. you'll probably not notice the difference between SATAIII SSD and NVME SSD. But start doing more complex tasks that do a lot of IO and that difference starts to be noticable. In addition, larger programs that need to load more into memory will benefit greatly from NVME.

Example in point: Games can be quite tremendously large with lots of content streamed to or from disk. Something as simple as, Cities Skylines for example has benefitted from moving from SATA based SSD to NVME for me in the way of load times that are significantly quicker.

in addition, many of us do numerous things at the same time. Yesterday I built 5 Windows 10 VM's and a PFSENSE VM and ran and installed them all at the same time. My disk IO usage was up in the 2,000 to 3,000 range.

Or the Database backup for the terrible engine I use, which peaked at 3,900Mbit/s when it was running.

the point is, Not all workloads are the same. Advertising "fastest" speeds, then cheaping out and using a set of storage that is slower than the cheap options available retail today is just more of the same of Apple cheaping out where they think they can hide it in order to maximize profiteering.
Incorrect, and IOPS is also a misleading measurement that can overstate performance depending on the platform.

Anecdote Time: I had nutanix sales people try and push that the Nutanix platforms 4million or so IOPS meant that no matter what, it was the correct choice for my databases. So I had them provision a top of the line cluster for me to test... Despite having millions of IOPS performance was less than our current spinning rust based server.

The reason is a bit more complicated than average user probably knows or cares, but it evidences that raw benchmark numbers do not always accurate predict performance outcomes

As well, you're partially write about the 4k read/writes. But most day to day computer users are not doing bulk of 4k read/writes, but typically somewhere between larger and smaller hybrid. For example, reading an OS from disk to memory is going to do more consecutive, and sequential reading. Same with the bulk of larger programs loaded off disk. if you swap a lot, you'll get a lot more 4k read/writes. If you do a lot of database work, you'll get that read/writes of 4k.

it always comes down to use cases. but IMHO, if you're paying for top of the line, you best be getting top of the line. Something Apple doesn't seem to be trying to deliver.

SO while you think this was clickbaity (it was), your comments are also fairly innacurate and generalizing of the impact of the slower storage.

but you do bring up one valid point. Sequential read/write isn't the whole story and would love to see if someone could run and post a crystal disk mark run or two of this computer to see what all the reported metrics are.
The article provides a single specification: large file throughput. There's no basis to talk about virtual memory performance of these drives. BlackMagic writes files that are typically 5GB in size. Virtual Memory pages are typically 4k in size. This is also why IOPS are sized at 4k. So you can't, on one hand, dismiss the argument that these BlackMagic benchmarks are only relevant if you're doing massive reads and writes (which is true, that's why all the check boxes relate to video formats), then say that virtual memory paging is critical, but then dismiss 4k accesses because sometimes you write something bigger.

If you're building out 5 Windows VMs simultaneously, the Air is probably not your preferred device. Also not for database backups. If your machine is slow because it's thrashing the swapfiles, then the problem isn't your SSD performance, it's your RAM. If the OS is paging out while there's still sufficient memory, then it's all hidden in a background operation and has little if any impact on performance. If your app is writing temp files for anything but the most trivial calculations and then ingesting them again without skipping a beat, they are likely buffered and not the slow part of your operation. And none of this, none, is informed by the article we've been presented with.

It is, however, informed by the people you're arguing with: these benchmarks don't give much insight into real world performance of these drives for most workflows and the Air in particular is not the machine that benchmark is tuned for.
if you're average use is just firing up notepad, word, chrome. you'll probably not notice the difference
In other words, if you're a typical Macbook Air user, you won't notice a difference-- which is, after all, what we're talking about.
 
Last edited:
  • Like
Reactions: brentsg
The read and write speeds of SSD's are overrated. What matters a lot more in real world use is the access times.

Pretty much ANY SSD has access speeds 100 times that of a spinning hard drive, and that's where the real payoff is.

Only if you're copying super large files might you start to see the benefit of crazy fast read/write speeds. And even then, only if you're copying that super large file from and equally fast drive (otherwise the source drive will be the bottleneck).

Sure because nobody on MacBook Pros edits video… oh, wait…
 
considering the slim price difference... I am sure more will purchase the MBP who were maybe on the fence? ... I know I was....

I've been on the fence for a while, and cant make a decision. I'm currently using a 2011 iMac 27", which runs fine, but i couldnt update to Mojave, and in September will fall even further behind. But i agree with you that the MBP seems more attractive.
 
  • Like
Reactions: kp98077
In a previous job I worked in a software development firm. I can safely claim that 90% of the developers might be good at writing code, but almost all of them couldn't tell how a computer actually works, or what each component of a computer is for. as long as the software had the desired output, they were happy.

Nahhhhhhhh, I'm having a very hard time believing that.

Anyways, 2018 MBA or 2019 MBA choices choices choices.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.