I'm into Black Magic

I've been using this BlackMagic drive speed test program for OS/X lately. It's free on the App Store and quite handy. Missing some tests I'd like to see but still for free who can complain.

I thought I'd share some read/write drive speeds for various hard drives, USB devices, SSD drives, etc that I had. Note sure if anyone will find them useful but here they are regardless.

Lexar 8 gb USB thumb drive:

I had a cheap USB 2.0 thumb drive cost was $9.99 on sale, it was formatted with Mac OS Extended (Journaled).

I tried plugging it into my MacBook directly, and then into my wire Apple keyboard which is plugged into a USB hub in turn plugged into my Thunderbolt Cinema display that then connects via Thunderbolt to my MacBook. I thought it would be interesting to see the difference a 3 hop daisy chain made. I also tried plugging in a old thumb drive beside it I had that I believed was USB 1.1 to see if it would bring down the overall speed of the hub.

Test Performed	Write	Read
8GB drive (plugged direct into MacBook)	5.4 MB/s	20.6 MB/s
8GB drive (3 USB hops thru various hubs)	4.6 MB/s	11.9 MB/s
8GB drive (3 USB hops thru various hubs plugged in beside a suspected USB 1.1 device)	4.5 MB/s	11.9 MB/s

Conclusion

Well there was a difference, especially in read performance, this was expected but I was surprised plugging in the old USB 1.1 thumb drive beside it on the Apple keyboard didn't drag it down further. Perhaps that old drive was actually USB 2.0. It was too small to run the Black Magic speed test utility on to get numbers for it.

Seagate Momentous Hybrid vs stock Apple hard drive:

I bought a Seagate Momentous 500gb Hybrid hard drive. It featured a SATA II interface and used a 7200rpm hard drive married with a 4gb SLC Nand chip to promise SSD performance at HDD prices. I'd had one before in my old white 2007 MacBook and liked it so figured it would be a suitable replacement for the stock Apple drive. I particularly wanted a SATA II interface for the drive in the optical bay of my MacBook because there's stories of unreliability using SATA III devices in this bay, Apple's own Superdrive negotiates a plain old SATA 1.0 link so Apple's never claimed even SATA II support for that bay. 

I compared the Seagate Momentous to the stock Apple 500gb 5400rpm drive my MacBook came with.

Test Performed	Write	Read
Apple stock 500gb 5400rpm	53.1 MB/s	52.9 MB/s
Seagate 500b Momentous hybrid	99.2 MB/s	103.6 MB/s

Conclusion

Huge difference! These compare with the numbers I used to get on my white 2007 MacBook, the Seagate is double the Apple 5400rpm drive. I believe too the Seagate's NAND memory isn't having a huge impact on these numbers. It's a sequential write followed by a sequential read. Seagate doesn't write  the data to the NAND as per their spec sheets so the write is just the 7400rpm drive and the read only happens once in the BlackMagic test so it's coming from disk and also measures real disk performance. I believe real world use of the Seagate drive is actually going to be much faster since Seagate's adaptive technology can keep the most used content in that 4gb NAND.

SSD drive vs Seagate vs Stock 5400rpm

As you can expect the SSD is going to win. The SSD is SATA III with MLC NAND chips in the main bay that fully takes advantage of the SATA III 6gb/s link, the other two drives are SATA II devices and made of spinning platters of molasses.

Test Performed	Write	Read
Apple stock 500gb 5400rpm	53.1 MB/s	52.9 MB/s
Seagate 500b Momentous hybrid	99.2 MB/s	103.6 MB/s
OCZ Agility 3 SSD drive	139.2 MB/s	202.2 MB/s

Conclusion

Obviously the SSD wins but on write speed it would seem to not be leaps and bounds ahead. On read speed it clearly dominates. But the trouble is the test is showing the mechanical drives off at their best and the SSD at its worst. Mechanicals love sequential read/write tests, the data is lined up nicely on the platters and they can just scoop it up, they also don't care if the data is compressible or not, they write each byte 1:1 to the platters. The SSD though hates sequential writes and reads; due to how NAND works they much prefer random read/writes and with the SandForce controller in the Agility 3 if the data can be compressed they use that to their advantage and gobble it up even faster and write fewer bytes to the NAND as a consequence. So real world; reading data the stock drive is ~50MB/s best case, the Seagate is likely a fair bit better than 100MB/s and the Agility should be managing somewhere close to double the 200MB/s reported. I just wish I had a test to prove that.

Firewire 800 vs USB 2.0

I have a Firewire 800 and USB 2.0 ports on my Thunderbolt Cinema display so thought a test of Firewire 800 vs USB 2.0 would be fun. The Firewire 800 drive I have is made by Seagate but probably a nothing fancy drive, no specs are listed on the box for it. For USB I put the SSD drive into a USB 2.0 enclosure (of dubious Chinese manufacture). I tried various combinations of Thunderbolt/MacBook direct and even distinguished between having multiple USB devices plugged in vs a lone USB 2.0 device.

Test Performed	Write	Read
Firewire 800 with unknown drive (plugged into Thunderbolt Cinema display)	41.8 MB/s	25.5 MB/s
Firewire 800 with unknown drive (plugged into Macbook directly)	42.8 MB/s	26.7 MB/s
Firewire 800 with Seagate Hybrid (plugged into Thunderbolt Cinema display)	41.2 MB/s	33.0 MB/s
USB 2.0 with Agility 3 SSD (plugged into Thunderbolt Cinema display)	14.0 MB/s	31.7 MB/s
USB 2.0 with Agility 3 SSD (plugged into Thunderbolt Cinema display with a USB 2.0 hub plugged in beside it)	7.4 MB/s	12.5 MB/s
USB 2.0 with Agility 3 SSD (plugged into MacBook directly)	27.5 MB/s	33.7 MB/s

Conclusion

Wow! Thunderbolt does a good job offloading ports from the MacBook without losing speed. Unfortunately USB 2.0 shows it has a well known but fatal flaw that performance depends a lot on what's plugged in around it. Alone USB 2.0 and Firewire 800 weren't that far apart but plugging a USB 2.0 hub in beside the USB 2.0 drive murdered its performance. I even tried pulling everything from the USB 2.0 hub so it was JUST the hub and still got the same numbers.

I am glad to have the Firewire drive as because it's always isolated from the USB 2.0 stuff I have attached and can maintain it's peak performance. I can also see my Firewire 800 drive had better reads with the Seagate Hybrid which suggests to me my numbers for Firewire were not yet at their peak, I wish I'd tried the SSD in the Firewire 800 enclosure to really remove any chance of the hard drive being a limiting factor. 

Anyway it was fun and interesting to see what performance all these drives and access methods afford me.