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

madeirabhoy

macrumors 68000
Original poster
Oct 26, 2012
1,681
663
short version of a long story, had a 2011 imac, bought an external 256gb thunderbolt SSD, the mac died. I cried. bought a 2017 one which has a fusion drive.

so ive got this SSD that's brand new. I also have a brand new thunderbolt to USB adaptor but clearly that's no use so will go on ebay.

however I can get a thunderbolt 3 to 1 adaptor for €50, and it says up to 10Gbps throughput.

so im thinking, am I best to keep it, buy the adaptor, and use it either as my startup disc, though I guess the fusion drive wouldn't be much different speed, or for bootcamp as clearly that will be slower from the fusion drive as it just used the HDD bit.

alternatively I could flog it for a loss on ebay, and buy a usb-c or thunderbolt 3, or even usb 3 SSD.

but in real terms would any of them be better?
 
Last edited:
Short answer: No, neither TB bus (2 or 3) will be slower than the internal Fusion Drive, and No, for a single SSD, neither TB bus (2 or 3) would limit performance.

Yes, a TB SSD will be faster than the FD overall. The FD will be as fast (roughly) for some common things (likely booting, opening apps, etc.), but it can't be the same speed as the SSD for most things.

If it were me, I would buy the USB C and not look back...unless you think you will ever need to connect to an older Mac. Either OS (Mac or bootcamp) will perform better overall on the TB SSD than on the FD. Your choice....which one you want to have the best performance.

-----

Keep in mind that the bus speed (USB, Firewire, Thunderbolt) rating is the maximum speed limit for things on that bus.

Drive speeds vary widely by brand and model, as well as by age and even how the drive is formatted, and size of files be read or written....but generally a ball park for transfer speed would be:

HD: 150 MB per second
SSD: 500 MB per second

Best to think of the bus—in this case TB—as a possible bottleneck. Can the drive(s) or any other connected device move data faster than the bus?

If YES: The bus is the limiting factor, and the data can only move at the maximum speed of the bus. Common problem with older USB, especially 2 and back. Nearly all drives, even back then, were faster than the bus. Bottleneck.

If NO: The drive is free to run (transfer data) at maximum speed. No bottleneck.

Also keep in mind that you you have at least 3 things that determine overall performance:

  • Origin (transferring from)
  • Bus (transferring through)
  • Destination (transferring to)

Whichever is the slowest of the 3 will determine the transfer rate...and become the bottleneck.
 
Last edited:
  • Like
Reactions: madeirabhoy
I think you can install windows to a thunderbolt ssd, but microsoft makes it difficult to install it on a usb3 ssd... (some sort of antipiracy/ enterprise feature)
 
  • Like
Reactions: madeirabhoy
Short answer: No, neither TB bus (2 or 3) will be slower than the internal Fusion Drive, and No, for a single SSD, neither TB bus (2 or 3) would limit performance.

Yes, a TB SSD will be faster than the FD overall. The FD will be as fast (roughly) for some common things (likely booting, opening apps, etc.), but it can't be the same speed as the SSD for most things.

If it were me, I would buy the USB C and not look back...unless you think you will ever need to connect to an older Mac. Either OS (Mac or bootcamp) will perform better overall on the TB SSD than on the FD. Your choice....which one you want to have the best performance.

-----

Keep in mind that the bus speed (USB, Firewire, Thunderbolt) rating is the maximum speed limit for things on that bus.

Drive speeds vary widely by brand and model, as well as by age and even how the drive is formatted, and size of files be read or written....but generally a ball park for transfer speed would be:

HD: 150 MB per second
SSD: 500 MB per second

Best to think of the bus—in this case TB—as a possible bottleneck. Can the drive(s) or any other connected device move data faster than the bus?

If YES: The bus is the limiting factor, and the data can only move at the maximum speed of the bus. Common problem with older USB, especially 2 and back. Nearly all drives, even back then, were faster than the bus. Bottleneck.

If NO: The drive is free to run (transfer data) at maximum speed. No bottleneck.

Also keep in mind that you you have at least 3 things that determine overall performance:

  • Origin (transferring from)
  • Bus (transferring through)
  • Destination (transferring to)

Whichever is the slowest of the 3 will determine the transfer rate...and become the bottleneck.


thanks, much appreciated. will probably go with the adaptor first and see how I get on. I suppose one issue of going straight with the thunderbolt 3 one is I haven't actually got any way to wipe the thunderbolt one and so I cant really sell it on ebay as it is. so will get the adaptor and if its not fast enough then I will sell the adaptor and the ssd on ebay and buy a thunderbolt 3 or usb-c one.
 
OP:

WHICH fusion drive did you get in the new iMac? 1tb, 2tb, or 3tb?

I doubt that ANY thunderbolt or USB3 drive you connect is going to be as fast as the fusion drive, IF the fusion drive is a 2tb or 3tb.

Please do this for us:
Download BlackMagic Speed Test.
Then, run it on your current fusion drive.
Post the read/write speeds here for us to see.
 
OP:

WHICH fusion drive did you get in the new iMac? 1tb, 2tb, or 3tb?

I doubt that ANY thunderbolt or USB3 drive you connect is going to be as fast as the fusion drive, IF the fusion drive is a 2tb or 3tb.

Please do this for us:
Download BlackMagic Speed Test.
Then, run it on your current fusion drive.
Post the read/write speeds here for us to see.

alas its a 1tb.

plus the fusion drive won't be fusion under bootcamp.

running disk speed test as we speak, will post after
 
right call me thick but I can't understand how disk speed test works as (on mine) it never stops, the amount just goes up and down whilst the start button is red. I left it running for hours as I went back to work and came in and they are still going up and down.
 
yeah, it keeps on going. Is it fluctuating wildly?If not, report the values (or screenshot it).
 
yeah, it keeps on going. Is it fluctuating wildly?If not, report the values (or screenshot it).
yes it bounces up and down from 0 to 900mb and I couldn't state an average as its so up and down.
[doublepost=1528844674][/doublepost]the first time the write went as high as 845mb but sat at 519mb when it started the read, which went up to 1049 but dropped to 683 and sat there.
 
OP:

WHICH fusion drive did you get in the new iMac? 1tb, 2tb, or 3tb?

I doubt that ANY thunderbolt or USB3 drive you connect is going to be as fast as the fusion drive, IF the fusion drive is a 2tb or 3tb.

How so? Are you suggesting that a hybrid drive, with most of the data living on a spinning HD is faster than an SSD?

Sorry, have to disagree.
 
Keep in mind there are many variables to HD speed tests. Format, block size, file size, file fragmentation, disk fragmentation, seek times, area of disk tested..and more. SSDs suffer much fewer limitations, in general, especially regarding disk location, fragmentation, and seek time.

On very small files, commonly used files....you may only be seeing the cache being tested; no the HD at all. On a FD, yes, some common functions will show native PCI bus SSD speeds, anywhere from 500 up to approaching 1000.

But what about the rest of your data? Hello HD limitations.

Other commercial grade benchmark options:

ATTO
HELIOS LAN Test (can test local and networked shares)
AJA
 
OP:

Download BlackMagic Speed Test and use that instead.


that's what im using. but why doesn't it run the test, stop and tell you the results? instead it just runs the test. then again. then again. and the dials are bouncing up and down. I can easily take a screen shot at the top of the bouncing to make it look great, or at the bottom of the bouncing to make it look crap. I am sure I am missing something as everyone uses this app but to me its useless as you can get virtually any reading you want simply by taking the snapshot at the right moment.
 
now im really confused. bought the apple thunderbolt 3 to 2 adaptor. the SSD seems really slow.

5ma4va.png

[doublepost=1528994568][/doublepost]from another app thanks to hobowankenobi

20hr4wh.png
 
the best app for me seems to be Aja system test lite. black magic bounces up and down. helios wouldn't run on the startup disk.

Fusion drive

2rro18g.png


thunderbolt SSD

2ag4z1u.png


bootcamp (using the windows version of the app)

23rtqtj.jpg


USB 3 external HD

2vuy7n5.png

[doublepost=1529008881][/doublepost]so using thunderbolt and the adaptor, the SSD is actually slower overall than bootcamp using the non HD part of the fusion drive, and not a lot more than the usb 3 4tb hard drive.
 
Wow, that's quite a hit.

Would be interesting to see how the TB SSD benches on a Mac with TB 2, no adapter.....just to verify your (reasonable) assumption the adapter is the huge bottleneck.

Yeah....AJA has been a go-to for a least a decade.

Helio is different animal, good for measuring real world performance of creating and moving files, not necessarily max drive performance or ability.

I am impressed how fast the FD is, even though that must be best possible performance, while using the solid state storage. The HD part of it likely maxes out under 200, maybe under 150.

One of the things that make any hybrid drive hard to accurately bench, troubleshoot, or even predict....in daily use, we as users get no say in how or what uses SSD, and how or what uses HD.
 
Referring to the speeds posted in #17 above:

You can clearly see here that the fusion drive of the 2017 is SUPERIOR to any other drive scenario that you have tried.

There's no point in trying to boot from the SSD.
It won't be faster if it's installed internally.
It won't be faster running from thunderbolt.
It won't be faster running from USB3.

Just use the fusion drive "as is".

I'd suggest a USB3 connection for the SSD.
An adapter like this will do the job well:
https://www.amazon.com/Sabrent-2-5-...478&sr=1-2-spell&keywords=sabremt+usb3+to+ssd
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Referring to the speeds posted in #17 above:

You can clearly see here that the fusion drive of the 2017 is SUPERIOR to any other drive scenario that you have tried.

There's no point in trying to boot from the SSD.
It won't be faster if it's installed internally.
It won't be faster running from thunderbolt.
It won't be faster running from USB3.

Just use the fusion drive "as is".

I'd suggest a USB3 connection for the SSD.
An adapter like this will do the job well:
https://www.amazon.com/Sabrent-2-5-...478&sr=1-2-spell&keywords=sabremt+usb3+to+ssd

Huh?

That FD number is just the SSD. Not real world for all uses. Once a user is into data that is not cached on the SSD section of an hybrid, the performance will drop to HD speeds.

Yes, the FD will boot fast, and do common things such as open applications fast. What about rendering 4K video, opening/saving/adding hundreds of high res photos, or other disk intensive activities? HD speeds.

Move a 5GB file and time it if you need proof.

Yes, an SSD on TB will be faster than any HD, internal or not.

The question is, why so slow now? It's not the TB bus, and it should not be the SSD (depending on brand/model/age). The adapter is the most likely culprit.
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Huh?

That FD number is just the SSD. Not real world for all uses. Once a user is into data that is not cached on the SSD section of an hybrid, the performance will drop to HD speeds.

Yes, the FD will boot fast, and do common things such as open applications fast. What about rendering 4K video, opening/saving/adding hundreds of high res photos, or other disk intensive activities? HD speeds.

Move a 5GB file and time it if you need proof.

Yes, an SSD on TB will be faster than any HD, internal or not.

The question is, why so slow now? It's not the TB bus, and it should not be the SSD (depending on brand/model/age). The adapter is the most likely culprit.


agreed. the problem isn't the fast fusion drive numbers, its the slow SSD numbers.

the SSD is silicon Power 240gb Thunderbolt T11, which I didn't test speed wise, but which made my 2011 iMac feel like a new machine.

but then the adaptor is apple, so it shouldn't be slower than what an original thunderbolt port was.
 
how weird. just ran Aja system test again.

same drive. same settings. twice the results.

ehi1jm.png
 
That is odd. But seems like more reasonable numbers.

Even with those, which seems a bit low my modern SSD standards, don't forget that is no seek time and no file fragmentation issues, so typically SSDs "feel" faster than they bench.

BTW, you should be able to run Helios on the internal drive. I have perhaps a hundred times on both HD and SSD....though never on a FD.
 
That is because sense you ran the test before your Mac cached the first test and added onto it! So always take the first test as the better tests!


that doesn't make sense though, surely. if that was the case then the writers of AJA would either run a script to delete the cache, or have a warning that once you run their app once on a machine theres no point in EVER running it again.

my machine has been restarted at least twice since those first scores.


plus on thinking about it, I ran it a couple of times on the same day as the first score and they were all similarly low.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.