Is FireWire 800 really faster than SATA II for HDDs/SSDs?

[retta283]

Just trying to wrap my head around this and be sure I'm really correct with the math here. It looks to me that SATA speeds are measured in bits not bytes, but the 400/800 for FireWire is bytes. This would mean FireWire 800 is about 6GB/s versus SATA II being only 3GB/s. I understand some of this may be sucked up by "overhead" so you might be losing 10-30 Megabytes/s but FireWire on paper seems a lot faster than SATA II.

I ask mostly because my iMac's SATA is no good at all anymore and I really don't want to bother cracking into it now that I have my SSD armada if a FireWire 800 external enclosure is going to be faster or even the same. I do not know what my enclosure's ratings are though other than it has a 8MB cache so I'm not sure what effect this will have on the speed. It is also a bus powered enclosure with option for DC mains power.

 

[Amethyst1]

the 400/800 for FireWire is bytes.

It's bits - megabits to be exact, making FW800 slower than SATA II's 3Gbps.

 

[iluvmacs99]

Just trying to wrap my head around this and be sure I'm really correct with the math here. It looks to me that SATA speeds are measured in bits not bytes, but the 400/800 for FireWire is bytes. This would mean FireWire 800 is about 6GB/s versus SATA II being only 3GB/s. I understand some of this may be sucked up by "overhead" so you might be losing 10-30 Megabytes/s but FireWire on paper seems a lot faster than SATA II.

I ask mostly because my iMac's SATA is no good at all anymore and I really don't want to bother cracking into it now that I have my SSD armada if a FireWire 800 external enclosure is going to be faster or even the same. I do not know what my enclosure's ratings are though other than it has a 8MB cache so I'm not sure what effect this will have on the speed. It is also a bus powered enclosure with option for DC mains power.

Firewire 800 is 800Mbps; megabits per second or 100MB/s bandwidth throughput, so it's 3x slower than SATA 2 which is 3Gbps (3 gigabits/sec) or roughly 300MB/s bandwidth throughput.

 

[Jack Neill]

I was in the same boat, tried a OWC FW800 enclosure with a EVO 860 and I got 50-60 R/W's I just bit the bullet and cracked it open and put a MX500 in it. I get 240's now.

 

[Raging Dufus]

In my experience FireWire 800 was a lot of marketing hype. It was billed at twice the speed of FW400, but I have never once seen that achieved on any of my FW800 equipment.

FW400 typically reaches speeds of ~35-40 MB/s in most of my use cases, and while I've seen FW800 exceed 70 MB/s a few times, it's rare. FW800 reliably moves at ~50-60 MB/s IME. So, it's better than 400, but certainly not twice as fast in the real world. Same with FW800 vs ATA: it's marginally faster than ATA-66, and roughly equal to ATA-100 in my real world uses.

Although, I have to admit I've never tried a FW800 RAID, so that might actually achieve its potential - but I doubt it would exceed ATA-100 RAID capability. Of course, it's not easy to run an ATA-100 RAID setup externally, whereas FW800 was built for that. Seems a lot of people used to use it for that purpose before eSATA and then USB3 came along, so maybe I just missed the boat.

 

[Juicy Box]

As many have mentioned, FW800 is only 800Mbps versus the 3000Mbps of SATAII.

The difference between the two is night and day.

That said, I have used external FW800 boot drive before many times, and it is perfectly fine depending on what you are doing with your iMac.

Actually, for overall experience, I would much rather use an external FW800 SSD over an internal HDD as my boot drive.

If I was doing long sustained writes or reads, then an internal SATAII HDD would be a better choice, but I think for most people, the experience of an external SSD, even at a slower potential speed, would be better than an internal HDD.

 

[Juicy Box]

It was billed at twice the speed of FW400, but I have never once seen that achieved on any of my FW800 equipment.

FW400 typically reaches speeds of ~35-40 MB/s in most of my use cases, and while I've seen FW800 exceed 70 MB/s a few times, it's rare. FW800 reliably moves at ~50-60 MB/s IME.

I don't think it is just hype nor just marketing.

I personally have seen FW800 SSDs consistently get over 80MBps for sequential read speeds.

Much faster than FW400 and USB2.

 

[Amethyst1]

I personally have seen FW800 SSDs consistently get over 80MBps for sequential read speeds.

It also depends on the SATA to Firewire bridge chip used in the enclosure. Would be interesting to compare a few and find out which is fastest.

 

[Juicy Box]

I decided to run some tests using FW800, USB2, SATAII SSD, and SATAII HDD disk speed tests to compare for the thread.

I also tested FW800, USB2, and SATAII HDD in a SW RAID0 just for comparison.

All the tests were ran on the default settings of AmorphousDiskMark. Here are the results.

Internal SATAII HDD:


This is what I would expect for the sequential reads and writes, it is actually pretty fast for a HDD. But, the random results, which is what people mostly experience in real world general usage, shows the weakness of using the HDD.



Here is FW800 SSD:


While the sequential is much lower than the HDD, it isn't that bad. Many people might be surprised at how well it performed with the random when compared to the internal HDD.



Here is the USB2 SSD:


Depending on the test, it performed about half of what the FW800 did. Just like the FW800, the USB2 SSD did better than the HDD at the random.



Internal SATAII SDD:


Not really a surprised that this one performed the best out of all of the single drive options. What is surprising is how close the FW800 got to the internal SSD at the random writes.



Here is the Internal SATAII HDD in a SW RAID0:


No surprised that this performed the best overall for sequential. I was surprised at how poorly it performed at the random, but maybe this was do to the block size?



Here is FW800 SSD in a SW RAID0:


The results of this was very surprising overall. I was expecting about twice the performance of the single FW800, but it was only a little bit faster.


Here is USB2 SSD in a SW RAID0:


Similarly to the FW800 RAID0, the results were not a huge improvement over the single USB SSD. Although, the performance gain of using a RAID0 drive over a single drive was better for the USB2 than FW800.


This has gotten me curious, so maybe I might do some tests on other computers and some other FW800 enclosures.

It also depends on the SATA to Firewire bridge chip used in the enclosure. Would be interesting to compare a few and find out which is fastest.

This is true.

I have a few different FW800 enclosures and the older ones tend to not perform as well.

 

[retta283]

I misunderstood the meaning of the 800 MBPS of FireWire then, I thought it was 800 megabytes which seemed awfully high considering SATA II came out around the same time if not a bit later.

I have been using a 5400rpm 2.5" drive in a FW800 enclosure to power this computer and it's been generally okay, although for some reason I get major lag on YouTube that I never did before which doesn't quite add up to me. It's not particularly fast, but it doesn't seem much slower to me than it was with the original drive. The original install I was using seemed to bog down too so I redid it a few months ago and we're running pretty smooth now. This is my girlfriend's TV substitute and I sometimes use it when I need to access certain apps so I'm thinking putting an SSD in that enclosure should set it up for quite some time to come. The SATA connector in the iMac does not seem to function anymore and I'm not sure how hard it is to replace that hardware.

 

[Amethyst1]

The SATA connector in the iMac does not seem to function anymore and I'm not sure how hard it is to replace that hardware.

If you don't need the optical drive and it's connected via SATA, you could get one of those adapters/caddies that allows putting a HDD/SSD in place of it.

 

[Raging Dufus]

I decided to run some tests using FW800, USB2, SATAII SSD, and SATAII HDD disk speed tests to compare for the thread.

I also tested FW800, USB2, and SATAII HDD in a SW RAID0 just for comparison.

All the tests were ran on the default settings of AmorphousDiskMark. Here are the results.

Internal SATAII HDD:
View attachment 1748791

This is what I would expect for the sequential reads and writes, it is actually pretty fast for a HDD. But, the random results, which is what people mostly experience in real world general usage, shows the weakness of using the HDD.



Here is FW800 SSD:
View attachment 1748792

While the sequential is much lower than the HDD, it isn't that bad. Many people might be surprised at how well it performed with the random when compared to the internal HDD.



Here is the USB2 SSD:
View attachment 1748793

Depending on the test, it performed about half of what the FW800 did. Just like the FW800, the USB2 SSD did better than the HDD at the random.



Internal SATAII SDD:
View attachment 1748794

Not really a surprised that this one performed the best out of all of the single drive options. What is surprising is how close the FW800 got to the internal SSD at the random writes.



Here is the Internal SATAII HDD in a SW RAID0:
View attachment 1748797

No surprised that this performed the best overall for sequential. I was surprised at how poorly it performed at the random, but maybe this was do to the block size?



Here is FW800 SSD in a SW RAID0:
View attachment 1748799

The results of this was very surprising overall. I was expecting about twice the performance of the single FW800, but it was only a little bit faster.


Here is USB2 SSD in a SW RAID0:
View attachment 1748800

Similarly to the FW800 RAID0, the results were not a huge improvement over the single USB SSD. Although, the performance gain of using a RAID0 drive over a single drive was better for the USB2 than FW800.


This has gotten me curious, so maybe I might do some tests on other computers and some other FW800 enclosures.


This is true.

I have a few different FW800 enclosures and the older ones tend to not perform as well.

Thanks, that was an eye-opener. You've clearly got some good equipment, I was surprised how well even USB2 performed there.

 

[EugW]

In my experience FireWire 800 was a lot of marketing hype. It was billed at twice the speed of FW400, but I have never once seen that achieved on any of my FW800 equipment.

FW400 typically reaches speeds of ~35-40 MB/s in most of my use cases, and while I've seen FW800 exceed 70 MB/s a few times, it's rare. FW800 reliably moves at ~50-60 MB/s IME. So, it's better than 400, but certainly not twice as fast in the real world. Same with FW800 vs ATA: it's marginally faster than ATA-66, and roughly equal to ATA-100 in my real world uses.

Although, I have to admit I've never tried a FW800 RAID, so that might actually achieve its potential - but I doubt it would exceed ATA-100 RAID capability. Of course, it's not easy to run an ATA-100 RAID setup externally, whereas FW800 was built for that. Seems a lot of people used to use it for that purpose before eSATA and then USB3 came along, so maybe I just missed the boat.

86 MB/s for me with FW800 hard drives, and 70 MB/s on FW800 Compact Flash cards.

Here was my post on this from 12 (!) years ago:

Kramer VS-30FW FireWire 800 Hub | Seagate FreeAgent Desk for Mac | SanDisk Extreme FireWire CompactFlash Reader | Transcend 600X CompactFlash

Because current iMacs only come with a single FireWire 800 port, a hub is often necessary if you have more than one FireWire device. Unfortu...

everythingapple.blogspot.com

EDIT:

BTW, I ran an SSD in a FW800 enclosure for a while, and it worked fine. It's nowhere near as fast as internal SSD, but it's not bad, and way faster than USB 2. The main problem is not the speed though. It's the fact that there is no TRIM. You gotta keep a fair amount of the drive empty or else it slows right down with large file transfers.

But like I said, overall it was fine. That was the reason I didn't bother putting an internal SSD in my 2010 iMac.

 

[Raging Dufus]

86 MB/s for me with FW800 hard drives, and 70 MB/s on FW800 Compact Flash cards.

Here was my post on this from 12 (!) years ago:

Kramer VS-30FW FireWire 800 Hub | Seagate FreeAgent Desk for Mac | SanDisk Extreme FireWire CompactFlash Reader | Transcend 600X CompactFlash

Because current iMacs only come with a single FireWire 800 port, a hub is often necessary if you have more than one FireWire device. Unfortu...

everythingapple.blogspot.com

EDIT:

BTW, I ran an SSD in a FW800 enclosure for a while, and it worked fine. It's nowhere near as fast as internal SSD, but it's not bad, and way faster than USB 2. The main problem is not the speed though. It's the fact that there is no TRIM. You gotta keep a fair amount of the drive empty or else it slows right down with large file transfers.

But like I said, overall it was fine. That was the reason I didn't bother putting an internal SSD in my 2010 iMac.

Well, you guys are just piling on now - but it's good to know.

Note to self, buy better stuff!

 

[Amethyst1]

@vertical smile -

Is this on a Mac Pro? If so, were the two FW 800 SSDs connected to different ports for the RAID? Just wondering whether that theoretical 800Mbps is per port or shared among all devices.

 

[Juicy Box]

Is this on a Mac Pro?

Yes.

If so, were the two FW 800 SSDs connected to different ports for the RAID?

Yes.

Just wondering whether that theoretical 800Mbps is per port or shared among all devices.

I am unsure about this. I would think that it would be 800Mbps per port, but the results show that this may not be the case.

Maybe I might test this theory by testing the two FW800 ports at the same time.

If they each have decent speeds, then I would think that the 800Mbps is not being shared between the ports.

 

[joevt]

@vertical smile -

Is this on a Mac Pro? If so, were the two FW 800 SSDs connected to different ports for the RAID? Just wondering whether that theoretical 800Mbps is per port or shared among all devices.

My MacPro3,1 only has one Fire Wire controller. It's a PCI device connected to a PCIe bridge all in one.

├┬00:1c.2-[24-25]       # g1x1           [8086:2694] [0604] (rev 09) PCI bridge                : Intel Corporation 631xESB/632xESB/3100 Chipset PCI Express Root Port 3
│└┬24:00.0-[25]         # g1x1           [104c:823e] [0604]          PCI bridge                : Texas Instruments XIO2213A/B/XIO2221 PCI Express to PCI Bridge [Cheetah Express]
│ └─25:00.0             #                [104c:823f] [0c00]          FireWire (IEEE 1394)      : Texas Instruments XIO2213A/B/XIO2221 IEEE-1394b OHCI Controller [Cheetah Express]


25:00.0 FireWire (IEEE 1394) [0c00]: Texas Instruments XIO2213A/B/XIO2221 IEEE-1394b OHCI Controller [Cheetah Express] [104c:823f] (prog-if 10 [OHCI])
	Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
	Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
	Latency: 64 (500ns min, 1000ns max), Cache Line Size: 64 bytes
	Interrupt: pin A routed to IRQ 18
	Region 0: Memory at a8204000 (32-bit, non-prefetchable)
	Region 1: Memory at a8200000 (32-bit, non-prefetchable)
	Capabilities: [44] Power Management version 3
		Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1+,D2+,D3hot+,D3cold-)
		Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME+

The PCI bridge is 250 MB/s max.
The FireWire device is 66MHz * 32 bit = 266 MB/s.
So one would expect something like 160 MB/s for a FireWire RAID.
Does the MacPro1,1 use the same FireWire controller?

I think all the ports are on the same bus so they need to share the 800 Mb/s and you can't get much better than 80 MB/s for a RAID (98.304 MB/s theoretical max without protocol overhead). This allows devices connected to the same controller to communicate with each other. Wikipedia says this:

FireWire can connect up to 63 peripherals in a tree or daisy-chain topology (as opposed to Parallel SCSI's electrical bus topology). It allows peer-to-peer device communication — such as communication between a scanner and a printer — to take place without using system memory or the CPU. FireWire also supports multiple hosts per bus.

The XIO2213B data sheet says this:

During packet reception, the serial data bits are split into 2-, 4-, or 8-bit parallel streams by the PHY section and sent to the link-layer controller (LLC) section. The received data is also transmitted (repeated) on the other connected and active cable ports.

The XIO2213A/B datasheets say they have 3 ports but the MacPro3,1 has four ports. Does that mean there's a FireWire hub somewhere? The FireWire section of System Information.app doesn't indicate topology. I don't think topology can be programatically determined (it's like an ADB bus or I2C bus or whatever - you can physically connect them in a tree but they are all equal peers).

 

[Juicy Box]

I think all the ports are on the same bus so they need to share the 800 Mb/s

I am thinking this is the case for my 1,1.

What I don't understand is that if all the ports share the 800Mbps, why the random speeds were not improved, with the exception for the RND4k QD64 Read test, which was doubled. You would think that the bus speed wouldn't be limiting this test.


Either way, I think all the tests show that a SSD on FW800 is not only an okay alternative to an internal HDD, but it is actually a much better boot drive when it comes to the random reads and writes.

 

[EugW]

I am thinking this is the case for my 1,1.

What I don't understand is that if all the ports share the 800Mbps, why the random speeds were not improved, with the exception for the RND4k QD64 Read test, which was doubled. You would think that the bus speed wouldn't be limiting this test.


Either way, I think all the tests show that a SSD on FW800 is not only an okay alternative to an internal HDD, but it is actually a much better boot drive when it comes to the random reads and writes.

Just remember there is no TRIM over FireWire. However it should usually be OK if you oversize the drive.

 

[Project Alice]

In my experience FireWire 800 was a lot of marketing hype. It was billed at twice the speed of FW400, but I have never once seen that achieved on any of my FW800 equipment.

FW400 typically reaches speeds of ~35-40 MB/s in most of my use cases, and while I've seen FW800 exceed 70 MB/s a few times, it's rare. FW800 reliably moves at ~50-60 MB/s IME. So, it's better than 400, but certainly not twice as fast in the real world. Same with FW800 vs ATA: it's marginally faster than ATA-66, and roughly equal to ATA-100 in my real world uses.

Although, I have to admit I've never tried a FW800 RAID, so that might actually achieve its potential - but I doubt it would exceed ATA-100 RAID capability. Of course, it's not easy to run an ATA-100 RAID setup externally, whereas FW800 was built for that. Seems a lot of people used to use it for that purpose before eSATA and then USB3 came along, so maybe I just missed the boat.

FW800 is absolutely twice the speed of 400. When I'm using a HDD (HDD, not SSD) FW800 is honestly not much slower in my experience than a USB3 HDD. I have yet to see anything I own connected over USB3 to achieve higher than 75MB/s, usually no more than 60MB/s. FW800 runs at it's top speed consistently.
This could be due to a limitation of the hard disk, but one would think a modern SATA 3 HDD would be faster...

 

[Amethyst1]

I have yet to see anything I own connected over USB3 to achieve higher than 75MB/s, usually no more than 60MB/s.

My Samsung T5 SSD does >400MB/s via USB3. Try that via Firewire.

 

[EugW]

My Samsung T5 SSD does >400MB/s via USB3. Try that via Firewire.

Indeed.

My Samsung T7 over USB-C:

 

[Project Alice]

My Samsung T5 SSD does >400MB/s via USB3. Try that via Firewire.

Indeed.

My Samsung T7 over USB-C:

Yeah, that’s why I specified that this has all been with HDDs, not SSDs lol. it’s possible my enclosure is also not up to par, it’s just a $20 2.5” enclosure that I use as a ghetto dock for my 2.5” SATA drives.
Every USB 3.0 flash drive I have is also pretty slow though.. I have a PNY one that gets 125MB/s. But all the other ones I have top out at like 60MB/s tops.
I’m too poor to afford putting SSDs in enclosures especially when they’re pretty much only being attached to PowerPC Macs so there’s not really any speed benefit when the SATA hard disk pretty much utilizes the FW bus anyways.

 

[EugW]

Yeah, that’s why I specified that this has all been with HDDs, not SSDs lol. it’s possible my enclosure is also not up to par, it’s just a $20 2.5” enclosure that I use as a ghetto dock for my 2.5” SATA drives.
Every USB 3.0 flash drive I have is also pretty slow though.. I have a PNY one that gets 125MB/s. But all the other ones I have top out at like 60MB/s tops.

Just FYI, I wouldn't get too hung up on max sequential transfer speeds when it comes to USB flash drives. I've noticed quite a few drives that have fast sequential transfer speeds actually are moderately slow when it comes to small file random transfer speeds.

For a lot of people, the small file random speeds are much more important than the sequential speeds.

I’m too poor to afford putting SSDs in enclosures especially when they’re pretty much only being attached to PowerPC Macs so there’s not really any speed benefit when the SATA hard disk pretty much utilizes the FW bus anyways.

The Samsung T5 and T7 are dedicated USB-C (or USB 3) SSDs, designed specifically to be used this way, and consequently never run into power usage issues over USB. (One issue I've run into before with SSDs in third party USB 3 or FireWire enclosures is flakiness when run using only bus power, that was fixed when external power sources were added.) The other thing about these drives is they have excellent random and sequential transfer speeds.

And yeah, my USB-C speed tests were on a 2017 Core i5 iMac. For my one PowerPC iBook still in service I just use a hard drive, cuz it's only running 10.4 and is surprisingly peppy with that for OS navigation on an old laptop HD. I just use it to configure my 802.11 AirPort Expresses. My other PowerPC machines have been retired in my little personal Mac museum.

As for Mac-ing on a budget, I did pick up a US$80 2008 Mac Pro 2,1 with 8-core 3.0 GHz Xeons and upgraded it for about another US$150 with a new GPU, RAM, and SSD among other things, and it's actually quite speedy running 10.11. But of course, it doesn't need to run the SSD over FW.

 

[m1maverick]

Just FYI, I wouldn't get too hung up on max sequential transfer speeds when it comes to USB flash drives. I've noticed quite a few drives that have fast sequential transfer speeds actually are moderately slow when it comes to small file random transfer speeds.

For a lot of people, the small file random speeds are much more important than the sequential speeds.


The Samsung T5 and T7 are dedicated USB-C (or USB 3) SSDs, designed specifically to be used this way, and consequently never run into power usage issues over USB. (One issue I've run into before with SSDs in third party USB 3 or FireWire enclosures is flakiness when run using only bus power, that was fixed when external power sources were added.) The other thing about these drives is they have excellent random and sequential transfer speeds.

And yeah, my USB-C speed tests were on a 2017 Core i5 iMac. For my one PowerPC iBook still in service I just use a hard drive, cuz it's only running 10.4 and is surprisingly peppy with that for OS navigation on an old laptop HD. I just use it to configure my 802.11 AirPort Expresses. My other PowerPC machines have been retired in my little personal Mac museum.

As for Mac-ing on a budget, I did pick up a US$80 2008 Mac Pro 2,1 with 8-core 3.0 GHz Xeons and upgraded it for about another US$150 with a new GPU, RAM, and SSD among other things, and it's actually quite speedy running 10.11. But of course, it doesn't need to run the SSD over FW.

Isn't the 8-core 3.0GHz Mac Pro 2,1 a 2007 model? I ask because the 2008 Mac Pro is a different model than the 2006 and 2007 models (the 2007 being really a firmware upgraded 2006).