MacBook Air flash storage torture tests

Today's Best Tech Deals

Picked by Macworld's Editors

Top Deals On Great Products

Picked by Techconnect's Editors

As you probably know, the MacBook Air uses flash storage instead of the hard drives found in every other Mac. And while the initial results using our Speedmark 6.5 test suite showed that the Air’s flash storage helped provide a performance boost, we were concerned about whether that performance could be maintained over time.

Solid state drives (SSDs) and flash storage (Apple doesn’t call the Air’s flash-based storage an SSD; it doesn’t use a 2.5-inch hard drive housing like a conventional SSD) offer a number of advantages. Compared to hard drives, flash-based storage performs faster. And since it lacks moving parts, flash-based storage is more shock resistant and not prone to the mechanical failings that hard drives are susceptible to.

The obvious downsides are that flash-based storage devices are much more expensive from a price per gigabyte perspective, and they offer lower storage capacities.

Not so obvious, however, are the performance issues. The individual data cells on an SSD have a finite number of times that they can be written to, so manufacturers often include wear-leveling technologies to increase the life of the flash memory. Also, write performance can suffer when SSDs begin to fill up over time. Some SSD manufacturers use over-provisioning tactics—setting aside some capacity away from the user to swap for bad blocks and background tasks. Some SSDs offer TRIM, a way of handling some of the trash management duties that can also slow down write speeds, but TRIM isn’t supported by Mac OS X.

When shopping for an SSD, you can usually read about the kinds of technologies the manufacturer uses. Apple, on the other hand, doesn’t really talk in detail about its behind-the-scenes strategies for dealing with flash memory’s shortcomings. That can make people like me nervous— especially in a system like the MacBook Air, where the flash memory is not user replaceable. (Sites like iFixit show how to open up a MacBook Air and access the flash storage, but such a task voids your warranty.)

Flash storage torture tests

To find out how Apple’s flash storage performs over the long haul, we put three Macs that use flash memory for primary storage through a torturous set of tests. We ran the AJA System Test, our Speedmark 6.5 disk tests, and diglloydTools’ DiskTester Fill Volume tests on a 2008 2.13GHz MacBook Air with a 128GB SSD, a built-to-order 2010 27-inch 2.93GHz quad-core Core i7 iMac with a 128GB SSD, and a 2010 13-inch 1.86GHz MacBook Air with 256GB flash storage.

The baseline results for the iMac and 2010 Air were similar. The average results from three trials are listed below.

AJA System Tests: Read

  • iMac: 204MBps
  • 2010 MacBook Air: 201MBps
  • 2008 MacBook Air: 104MBps

AJA System Tests: Write

  • iMac: 177MBps
  • 2010 MacBook Air: 187MBps
  • 2008 MacBook Air: 67MBps

DiskTester Fill Volume: Read

  • iMac: 213MBps
  • 2010 MacBook Air: 210MBps
  • 2008 MacBook Air: 104MBps

DiskTester Fill Volume: Write

  • iMac: 172MBps
  • 2010 MacBook Air: 172MBps
  • 2008 MacBook Air: 60MBps

Duplicate a 1GB file

  • iMac: 14 seconds
  • 2010 MacBook Air: 13 seconds
  • 2008 MacBook Air 74 seconds

After getting those baseline results, we erased and reinstalled OS X and migrated our data from a Time Machine backup. We then used DiskTester to write 2,000,000 8K files, erased the files, and ran the test again. Then we erased and reinstalled OS X and data from a Time Machine backup, and twice ran another torture test, diglloydTools’ MemoryTester. After that, we erased the flash storage and reinstalled OS X and restore via Time Machine.

We ran our benchmark tests again and found that, unlike some of the SSDs we’ve tested in the past, performance was unchanged. Despite all of the wiping, filling, stressing, and imaging we’d done on these poor flash storage devices, they performed identically to the way the had before being tortured.

We took the testing a little further to see how performance was affected when a storage device was near full capacity. I filled the 256GB flash storage in our 2010 1.86GHz 13-inch MacBook Air until there was just 5GB of remaining space available and ran the AJA System Test and file duplication tests. Impressively, the results did not show any performance degradation.

One hiccup

The flash storage in one of our 2010 Airs, a 13-inch model with 128GB, stopped working properly before we were able to finish the tests. It failed three-quarters of the way through our final OS X installation task. We tried restoring from the included USB flash drive and from a retail Snow Leopard disc. No go. I booted from an external drive and used Super Duper to clone a working Air image, but it still failed to finish. I ran Disk Utility’s Repair Disk on the flash storage, but it didn’t find any problem with the drive. Other times, the drive would fail to unmount. Many hours were spent, but in the end we decided to risk the health of another new Air, the 256GB 1.86GHz 13-inch model referenced in the results above, which had no problem completing these same tests.

Seeing how three of the Mac flash drives made it through the testing with flying colors, I’ll assume that the one that crapped out was a fluke. We’ll keep testing and will let you know if this turns out to be anything more than that.

[James Galbraith is Macworld’s lab director.]

Note: When you purchase something after clicking links in our articles, we may earn a small commission. Read our affiliate link policy for more details.
At a Glance
  • Apple MacBook Air (Late 2010) family

Shop Tech Products at Amazon