These speeds are actually very surprising to me. I just recently tried a USB 3.0 drive on a PC recently and i only got firewire 800 speeds. I guess it could have been the actual speed of the drive in the enclosure, but I still don't think it's all that.
Yes, for the most part, what you were observing was that your bandwidth bottleneck was the physical (and single spindle) HDD's limits, not the protocol that was transferring it (SATA, FW, USB3, etc).
Firewire 1600 and 3200 should be promising, and USB 3.0 still relies on the CPU host. Firewire has it's own, so professionals should still keep it in mind.
It has mostly been the advent of higher bandwidth capabilities via RAID-0 and solid state (ie, SSDs) that have increased the appetite for higher I/O interfaces...along with higher bandwidth media applications (HD video, audio) which have time latency factors that bring the requirement to a "Real Time" objective. It has been interesting (and IMO, disappointing) to observe how Firewire's performance has been allowed to languish...not sure really why this is.
Firewire 3200 might theoretically be slower than usb 3.0, but in actuality, it might be faster than usb 3.0. The theoretical speeds are always exaggerated.
This is more a case of product marketing: the theoretical speed is what is marketed, but it includes all of the overhead costs, which decrements the real world pragmatic I/O performance. The classical example of FW400 vs USB2 is that the "theoretical" comparison is 400 vs 480 (~20% advantage to USB2), but after accounting for their respective differences in their overhead budgets, the "reality" comparison is closer to 350 vs 200 (~75% advantage to FW400).
Really, that's the best USB 3.0 can do? Pretty lame - only 1.75x to 2.4x faster than FW800 by those numbers. FW3200 is 4x faster. So much for the claims that USB 3.0 is a worthy competitor to FW3200.
Actually, until there's real world (and populist/mainstream) products really shipping, neither has "won". Similarly, some of this is like debating over the merits of SATA 3GB vs SATA 6GB: it is only relevant for those (currently very few) applications where the current bandwidth being consumed is greater than 3GB. If your application is <3GB, having the "higher spec" interface isn't accomplishing anything for you today, other than incurring some cost for some future-proofing of unknown eventual merit.
Shame on Apple for not having put FW3200 in all their products by now, but holding off on USB 3.0 looks like a pretty wise move by these numbers. On the other hand, if Light Peak really is just around the corner, then no point in wasting time with either.
Agreed, except that there is the merit of being able to exploit a capability more so in the near term than in the long term. Personally, I suspect that the dilemma that Apple (& others) face is that buffing out the I/O raises some system integration and system optimization factors: the historical parallel can be seen from the original PowerMac G5, which raised the bar on alleviating bottlenecks in system bandwidth ... now consider the following: how much have we improved these capabilities? It comes back to the question of identifying precisely where the current system's performance is limited (bottlenecked) and what can be done (hopefully cheaply) to alleviate these ... as such, the potential is that adding a FW3200 port could very well simply magnify an existing system design shortcoming, so strategically, Apple does not wish to highlight this system design weakness in their products (and ditto for Intel, too).
Maybe it's approaching the limits of the tested hard drive? If the drive is the weakest link, then SATA and FW 3200 isn't going to show massive gains, maybe a few percent, not a few times.
Exactly. The last time that I looked at detailed technical specs, the bandwidth of current state-of-the-shelf commodity HDDs wasn't all that far above SATA-I (1.5GB) specs. As such, a SATA II (3GB) was more than enough capabilty (& capability growth) and SATA III (6GB) is for most consumer instances an utter waste of money. The same business pragmatism will apply to I/O connections outside of the box, be they USB3, FW3200, eSATA, or Lightpeak.
It's been about two and a half years and no one has produced any FW3200 devices that I've seen. FW3200 devices were originally projected to ship Oct 2008. Where are they?
Great question. IMO, the answer is going to get into 'Market Forces', probably mostly on the supplier side.
I don't think it's going to take long to get established. Newegg has more than 60 USB 3 hard drive enclosures available for sale right now...
Yet similarly, NewEgg has had eSATA enclosures now for how many years, and how has that done in
taking off?
Yeah like FW took off? We had toslink optical for audio. It's not longer terribly relevant. You need HDMI cable to handle more bandwidth for the latest BD based Dolby and DTS formats. Optical isn't everything, particularly over short runs. You also have to consider cost. Lightpeak is far from a sure sell. Its bandwidth just isn't THAT much more than USB 3.0 to overtake it when it's going to be running 1-3 years behind. They should be looking ahead to something to beat USB4 to the market. They've lost already to USB 3.0 in a general sense. Lightpeak appears to be more of a mobile orientated device anyway (fast connections between iOS devices, for example), not a general purpose replacement for all peripherals like USB traditionally has been.
Good points, although I suspect that for the consumer-centric marketplace, the first gate is cost, but there's potentially other factors of significance to consider, such as the ease-of-cabling (if the consumer actually gets the opportunity to voice a real preference in the marketplace). Historically, think about our physical I/O cable connections and see if you can think of any that aren't sensitive to being installed "upside down" (yes, there's a few). USB doesn't pass this ease-of-use test. Nor does FW. Nor eSATA. Nor HDMI. For LightPeak, I've not seen a proposed plug design yet...
-hh
