Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Hellhammer

Moderator emeritus
Dec 10, 2008
22,164
582
Finland
Is it even possible to identify one? :(

Most likely no, unless you open it in-store to check the SSD model.

EDIT: Anyone with those newer SSDs willing to open the bottom cover to see is it a Samsung SSD? No need to take anything out, just remove the bottom cover and take a pic of the SSD, shouldn't void your warranty ;)
 

gdeputy

macrumors 6502a
Jul 23, 2008
838
84
New York
What exactly is Native Command Queuing?

My Model btw is CJAA0201 and I do have a TS128C.

Sort of a letdown that there is such a marked difference in read/write considering the short timespan between them.
 

KPOM

macrumors P6
Original poster
Oct 23, 2010
18,019
7,862
If I had to guess, the SSD in newer MBAs is Samsung 470-series as that has speeds of 250/220 MB/s, which makes sense when looking at the benchmarks. Apart from the speed differences, it looks like the newer ones support Native Command Queuing while older ones don't.

It could be. Perhaps someone running a newer MacBook Air could run Crystal Disk Info from within Windows to see if NCQ is supported? It isn't on the Toshiba drives.
 

marzxbarz

macrumors member
Apr 23, 2009
43
0
Looks like Apple used a completely different SSD board in the 256 GB models, as all of us with the Ultimates have the slower SSD while everyone with the 128 GB have the faster ones... Ack!
 

KPOM

macrumors P6
Original poster
Oct 23, 2010
18,019
7,862
Ultimate MacBook Air 13 (bought yesterday):

So it turns out I was sold an older revision, right? That's why it's slower, right? I'm mad!!! :mad::mad::mad:

Should I have asked the sales person to pull out the latest revision? Is it even possible to identify one? :(

In reality, those speeds are still considerably faster than a hard drive. It's a bit like the different screens that are out there. Sure there are some differences, but I doubt they will be particularly noticeable.

Looks like Apple used a completely different SSD board in the 256 GB models, as all of us with the Ultimates have the slower SSD while everyone with the 128 GB have the faster ones... Ack!

It may be a supply issue. That said, I'm a bit surprised that didn't turn up in the AnandTech review. The 11" Ultimate ought to have had the Samsung if you are correct.
 
Last edited by a moderator:

Hellhammer

Moderator emeritus
Dec 10, 2008
22,164
582
Finland
It could be. Perhaps someone running a newer MacBook Air could run Crystal Disk Info from within Windows to see if NCQ is supported? It isn't on the Toshiba drives.

According to the screenshot posted in earlier this thread, it does.

screenshot20110206at849.png


Looks like Apple used a completely different SSD board in the 256 GB models, as all of us with the Ultimates have the slower SSD while everyone with the 128 GB have the faster ones... Ack!

I have 128GB in my 13" MBA and it is the slower Toshiba.
 

sniffies

macrumors 603
Jul 31, 2005
5,644
14,608
somewhere warm, dark, and cozy
Ok, I've just got off the phone with the store I bought my machine from. Basically, what I was told is that the most expensive model of MacBook Air ($1799) comes with revision CJAA0201 (slower SSD).

Could someone else confirm this. Thanks.
 

zenio

macrumors 6502
Feb 2, 2011
472
0
I ran it yesterday and got similar results to yours. The SSD is the same revision as yours. Got it in late November.

Mine must have come from the same production run, as my experience mirrors yours.

I'm really enjoying this machine. 100% satisfied.
 

fibrizo

macrumors 6502
Jan 23, 2009
411
5
A
However, this does not apply to SSDs since they don't spin. 40GB SSD is as fast as 480GB one with the same controller. Amount of NANDs may affect the speed but usually they don't.

This can be true for the most part if lower density nands are used instead of less nands.

ie the intel SSDs show this. the 160 is faster than the 80 which is faster than the 40, simply because they didn't fill the nand slots and left them empty. I'm not sure how the mba ones work, as I think the 256 would have more nand modules than the 128 and perhaps the 64, but we'd have to tear it open to see...

Then again the nand modules could be slower at higher capacities etc, so it is difficult to say for sure, unless we test.

So far it seem the AX ssds have higher read and write speeds than the CJ models.

But I'm actually curious about the random read and write 4k and such blocks. While numbers for continuous read and write are nice, the random read and write seem to contribute more to my day to day usage patterns. ie I have an intel 80gb x-25m G2, which read: 250mb/sec and write 70mb/sec advertised... and an OCZ agility 2 sandforce drive with read: 285mb/sec write 275mb/sec... in theory the sandforce SSD should blow away the intel drive. what I've actually seen is on the same computer, the intel drive boots faster, and gets a higher windows disk score (7.7) where the sandforce drive gets a 7.4 and boots about 1 second slower. Crystal disk benchmarks also show while the sequential speeds are faster on the sandforce drive, the intel drive is faster in random reads/writes. Don't get me wrong. the sandforce drive is still really fast, but I think that read/write numbers don't always mean much if we don't look at random small file read/writes.

Does any one have benchmarks for either CJ or AX ssds? My 128gb ultimate 11 inch uses the CJ ssd.

Also the native command queuing might play a role... here's something from wiki


NCQ is also used in newer solid-state drives where the drive encounters latency on the host, rather than the other way around. For example, Intel's X25-E Extreme solid-state drive uses NCQ to ensure that the drive has commands to process while the host system is busy processing CPU tasks. [3]

NCQ also enables the SSD controller to complete commands concurrently (or partly concurrently, for example using pipelines) where the internal organisation of the device enables such processing.

For example, the SandForce 1200[4] based OCZ Vertex II 50GB drive running on a Dell Perc 5i (which doesn't support SATA NCQ) delivers about 7,000 4k IOPS (50% write) at a controller queue depth of 32 IOs. Moving the drive to the similar Dell Perc 6i increases this to over 14,000 IOPS on the same basis
 

elfxmilhouse

macrumors 6502a
Oct 15, 2008
602
144
Northeast USA
just for comparison:
OWC 240GB for MBA 2010
4.0 GB 1920x1080 10-bit RGB

Read: 274.5 MB/s
Write: 218.4 MB/s

Didn't know about the AJA system test program before I did the upgrade so i dont have any stock numbers for comparison.
 

torbjoern

macrumors 65816
Jun 9, 2009
1,204
6
The Black Lodge
Ok, I've just got off the phone with the store I bought my machine from. Basically, what I was told is that the most expensive model of MacBook Air ($1799) comes with revision CJAA0201 (slower SSD).

Could someone else confirm this. Thanks.

Did they say anything about why?
 

Hellhammer

Moderator emeritus
Dec 10, 2008
22,164
582
Finland
This can be true for the most part if lower density nands are used instead of less nands.

ie the intel SSDs show this. the 160 is faster than the 80 which is faster than the 40, simply because they didn't fill the nand slots and left them empty. I'm not sure how the mba ones work, as I think the 256 would have more nand modules than the 128 and perhaps the 64, but we'd have to tear it open to see...

Then again the nand modules could be slower at higher capacities etc, so it is difficult to say for sure, unless we test.

According to Intel, the read speed is the same for all, 250MB/s. 80GB has write speed of 80MB/s while higher capacities have write speed of 100MB/s. This is supported by AT's bench.

This shows nearly no difference in sustained speeds between 40GB and 120GB SSD with the same controller.

FYI, 64GB and 128GB SSDs in MBA consist of four NANDs (4x16GB and 4x32GB) and the 256GB is made of eight 32GB NANDs.
 

fibrizo

macrumors 6502
Jan 23, 2009
411
5
FYI, 64GB and 128GB SSDs in MBA consist of four NANDs (4x16GB and 4x32GB) and the 256GB is made of eight 32GB NANDs.

Thanks! The random IO read seems to be drastically different in the 40 vs 120gb comparison you linked to, and per my post I am rather curious about the random io performance rather than the sequential performance. I wonder if the 256 will show a higher random io vs 128 and 64 given the benchmarks you provided show a 3.5 fold random 4k read increase. Also I would like to see if the newer ssds are faster in this regard as well so I can feel thoroughly bad about not having the faster drive lol.
 

torbjoern

macrumors 65816
Jun 9, 2009
1,204
6
The Black Lodge
If those who paid more for their SSDs get the slower ones, it sucks. But that's mainly on a general principle, and will have no practical consequences. If it takes 2.5 seconds to open Photoshop instead of 2.0, I can live with that.
 

JPizzzle

macrumors 6502
Oct 30, 2008
315
46
CJAA0201, 185write 205 read. 128gb 11.6in. bought 2/1/11. A little dissapointed. I'm still within my 14 days, is this something that I should return it over??? Are any of the people who are getting the faster speeds getting them with the 11.6 in? I wonder if model size has an affect of SSD allocation.
 

KPOM

macrumors P6
Original poster
Oct 23, 2010
18,019
7,862
just for comparison:
OWC 240GB for MBA 2010
4.0 GB 1920x1080 10-bit RGB

Read: 274.5 MB/s
Write: 218.4 MB/s

Didn't know about the AJA system test program before I did the upgrade so i dont have any stock numbers for comparison.

Hmm. OWC might need to change its advertising about being 22% faster than factory, then.
 

DeBilbao

macrumors member
Jan 13, 2008
53
6
I include my results on the survey:

13" MacBook Air manufactured January, 19th 2011.
Read: 258.4 MB/s
Write: 207.2 MB/s
Model: SM128C (supporting NCQ)
Revision: AXM09A1Q
Capacity: 128 GB
Occupied: 75 GB (58%)

30t13zm.jpg


b706j6.jpg
 

DeBilbao

macrumors member
Jan 13, 2008
53
6
BTW, I have run the Crystal DiskMark in my virtualized Windows 7 and the results are just incredible :D

1hc9he.jpg

Have some fun and try on yours!
 

whuybr0

macrumors newbie
Mar 11, 2009
21
0
Belgium
Ok, I've just got off the phone with the store I bought my machine from. Basically, what I was told is that the most expensive model of MacBook Air ($1799) comes with revision CJAA0201 (slower SSD).

Could someone else confirm this. Thanks.

I bought my "ultimate MBA" yesterday from an official Mac reseller (location : Antwerp, Belgium) and have the 'faster' SSD (see my previous post). As the MBA is BTO and was in stock, I presume they already had it ordered a while ago. So I don't know why in the 'Big Apple' they're stating the ultimate MBA comes with slower SSD, while in a tiny country as Belgium, we get faster specs.

PS : ultimate = 2,13 GHz, 4GB, 256SSD
 

Mr. Zarniwoop

macrumors 6502a
Jun 9, 2005
751
139
MacBookAir3,2 (Late 2010) 2.13GHz/13.3"/4GB/256GB

Read: 207.4 MB/s
Write: 181.0 MB/s

Purchased: October 2010

Revision: CJAA0201
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.