Fastest file system on MacOS?

Jakob H
Jakob H used Ask the Experts™
As I have copied files on a USB 3.0 drive between two SSD based Macs (10.12 Sierra and 10.13 High Sierra), I have noted that there is a significant difference in performance depending on what file system I have chosen for the USB Drive.

I have read some performance comparisons, but they are either outdated or Windows-based and they appear inconsistent with my findings from the file copying.

Hence, could someone share some insights on what file system to use for the purpose of copying files most efficiently between SSD based Macs using a USB 3.0 Drive? (The performance of the drive will also factor in, of course, but in this question I am interested in the effects of the file system as they can potentially reduce the effective performance of a USB Drive.)

The file systems that I have available to choose from in my MacOS setup:

- ExFat
- MS-DOS (FAT)   [Disregarding the 4GB file size limit to focus on performance]
- Mac OS Extended (Journaled)

(I am disregarding the Encrypted and Case-sensitive alternatives of Mac OS Extended assuming that they will not perform better.)

What I am doing is
- copying to and from a USB 3.0 Drive (meaning that both writes and reads will factor in)
- approximately 300 GB of data consisting of tens of thousands of files with a file size of a few KB's up to hundreds of MB's. (There are some files in sizes of GB as well, but they are exceptions.)

If I left out something significant from the question, let me know and I am prepared to update the question as needed.
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
There are a heap of other variables at play here as well.

The performance of USB keys varies, particularly when you look closely. One device may be very fast at sustained, sequential access, but run like  a gravid pig when accessing blocks randomly. Some are fast at reading data, but slow at writing it.

These performance differences could easily mean that the optimal file system may differ from device to device.

Realistically, you will probably have to run some "real world" tests to answer this question.
I'd agree with Mal .. there are a number of factors apart from the filesystem which may have a bigger impact than just the file format.  Data fragmentation or the overall size of the external HD are also big factors.

I'd suggest you use a proper tool which can read/write a consistent dataset to compare properly. Here's a free one which works

If you've only Macs there is NO reason to use or advantage in either of the FAT formats (FAT32 or exFAT).  Remember that the OSX native formats are always the best to use and required for TimeMachine, Drive cloning software etc.  FAT formats are Windows-based and simply not as compatible with OSX.

Remember also that FAT32 doesn't support files larger than a couple of GB and hard drives larger than 2TB by default which may rule it out for certain tests.

The OSX format HFS+ (aka Extended Journaled) writes a more detailed file index than the FAT formats so is better suited to data recovery in the event of drive issues or if the computer fails while writing data the risk of corruption is slightly less.
FAT is not a "Windows based" format, it is universal, and supported by most systems.  TVs, routers, Linux boxes AmigaDOS whatever. Nearly all Windows PCs run NTFS these days.   FAT was originally used in the 70s for 8" floppy disks, long before Windows was even thought of.

USB keys are usually preformatted with FAT as this will be recognised by nearly all contemporary devices.
OWASP: Avoiding Hacker Tricks

Learn to build secure applications from the mindset of the hacker and avoid being exploited.

@Mal - I accept your clarification.  I had meant that FAT formats are more relevant if you're working with windows-based computers as well as OSX.  

The failure by Apple to natively support writing to NTFS formats leaves the FAT formats as the only universal formats for OSX to/from Windows for reliable file exchange.


Thank you for your comments!

I agree that other factors play a role in the overall performance, which is why I formulated the question to acknowledge that but to focus on the performance effects of the file systems mentioned and disregard also functional differences such as the 4 GB file size limit.

As per my own limited testing, my perception is that large (say >= 1 GB), sequential writes is close to identical between the file systems, but that the overhead of the file table metadata makes all the difference when there are many smaller files.

Is that a fair statement and, if so, what does the expertise say about the comparative overhead for FAT, ExFAT and Mac OS Extended (Journaled) respectively?

Am I mistaken in assuming that such comparison of overhead could provide a generic ballpark answer that helps determining the recommended filesystem for a file copy scenario such as the one described?

Although I didn't specify this in the question initially, I also meant to include if there is a potential impact of the Partition Map as well (GUID, MBR or Apple).

(Regarding performance testing using Blackmagic Disk Speed Test, this seems helpful when looking at large, sequential writes and reads, but not so for smaller writes and reads where file table metadata overhead of different file systems plays a decisive role.)
Someone's run the tests a few years back, but you'll need a Microsoft account to fully access the data on his Onedrive account.

I'd just use HFS+ for Mac sharing and NTFS for windows sharing.  I stopped using FAT and exFAT a while back.

FAT was for DOS based systems originally.  Windows NT uses the more robust NTFS and supported FAT as a backward compatible feature, not because FAT is used for Windows.  Any support of FAT in modern systems is due to backwards compatibility, not because it's designed as a universal format.  It just turned out to be a common format that many systems had to support for the Migration from old DOS based systems and was kept because of legacy support.

FAT, including FAT32 and ExFAT really should be deprecated.  They're so inefficient and wasteful.  ExFAT is only supported on newer Windows systems and was designed to support the newer USB memory sticks that have larger capacities.  It's not really meant for spinning disks.

If you're not sharing with windows, then use HFS+ with a journal.  It's more efficient than FAT or ExFAT on a Mac.  FAT is going to be slower for access on larger storage with numerous small files, because of the way it indices files.  Larger files don't differ as much because they don't access the allocation table so frequently so they're closer in copy speeds.  If you share with Windows, it's much better to just enable NTFS write or buy Paragon NTFS.  NTFS is a much better file system than FAT for storing data.  You can already read NTFS on a Mac by default, just not write by default.

The main reason we still have FAT based disks is because IBM created it and allowed people to use it and there's still sufficient numbers of legacy systems around.  NTFS is proprietary from Microsoft.  HFS+ is proprietary from Apple.  Apple should just make OS X fully read and write NTFS properly.  Since they don't, you have to hack a setting in OS X to enable NTFS write or you should buy something like Paragon NTFS.  FAT and DOS are obsolete.  It's time to stop using them.


Thanks for the reference to the comparison, @serialband! Unfortunately, the OneDrive file seems to be gone. (I did access using OneDrive credentials.)
The forum post you refer to is still available though, although the x/horisontal axis legend/description is missing so it isn't clear if it measures time (less is better) or throughput (more is better). Assuming that NTFS (Win) is faster than FAT32 in the lower of the two images, it indicates that HFS+ is actually slower than FAT32! However, is your recollection that the full PDF provided another conclusion?

About the partition map, what is the potential effect of switching between GUID, MBR and Apple?

On newer systems, there is also another choice of format: APFS. Any known comparisons on how that stacks up?
David AndersTechnician

I believe that using Target Mode and terminal copy commands would provide the fastest speeds. Depending on the exact models and ports available on your two Macs.
Thank you for all the comments.

In summary, the comments provided indicates that it is possible to benchmark the file systems to give generic advice on performance, such as the benchmarking that serialband referred to that is however no longer available.

Until such benchmarks are served, it is an open question in which scenarios ExFAT, FAT, HFS+ and APFS (as well as choice of partition map; GUID, MBR or Apple) would be fastest.

(While there are also other factors at play and while there are also other methods to copy files; the question in this case is specifically about the performance of the Mac file systems.)


My solution gives the most helpful summary of the comments provided:

"In summary, the comments provided indicates that it is possible to benchmark the file systems to give generic advice on performance, such as the benchmarking that serialband referred to that is however no longer available.

Until such benchmarks are served, it is an open question in which scenarios ExFAT, FAT, HFS+ and APFS (as well as choice of partition map; GUID MBR or Apple) would be fastest."

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial