It is faster - and I did buy those FW800 enclosures in PPC times.
:clears throat:
IEEE1394’s FireWire was still the fastest,
effective, device-based, external serial bus found on all Macs cranked out between 2006 and early 2012. (I emphasize
effective since data transfers with FW400’s 400Mbps typically outpaced USB 2.0’s greater packet overhead, thus undermining the negligible advantage of that standard’s 480Mbps specification.)
It’s noteworthy to mark these years because they cover virtually every model we discuss on this forum. [MacBook5,1 was the notable outlier for lacking FireWire altogether.]
Given the way Jobs declared FireWire to be “dead” in 2008 (as rebuttal to quick complaints how the aforementioned, unibody MacBook dropped FireWire was “intended” to be enough of a telegraphing of his sentiment before he had to do PR damage control), there was still a window of three years wherein FireWire was
still the fastest external bus port shipped on all Macs, save the MB5,1.
When set in this light, the demise of FireWire, as with the demise of ZFS in 2007–08, was spearheaded by Jobs.
In both, heavy engineering might and financial resources then got involved to step up, to find technological workarounds to emphasize the D.O.A./deadness of Jobs’s capricious PR decrees: respectively, working with Intel to develop Light Peak (Thunderbolt); and snubbing Sun CEO’s faux pas of scooping Jobs’s announcement for ZFS
by one day, to devote a nine-year development of APFS to replace HFS+ (something Jobs didn’t live to witness).
In short, it’s to say Jobs never did anything by half-measures: either things he championed, proactively, ended up successful (or laid groundwork for future success, like Lisa becoming Macintosh; the G4 Cube paving way for Mac mini; or iPod paving ground for iPhone); or, initiatives he pursued reactively and/or defensively ended up as remedial kludges which, eventually, were brought to force through compulsory inclusion on hardware and system software.
Sure, Thunderbolt is fast
today. But consider the billions in R&D involved for a clean-sheet, proprietary serial bus standard-over-PCIe, when a related standard was already in place.
That existing standard, IEEE1394, already played nice with PCIe. It was still undergoing standards updates and conveyance improvements. Multiple revisions were already deployed and in the pipeline. (There’s no reason to discount an IEEE1394 revision to have been also been able technically to carry digital display signal alongside other data.) IEEE1394’s data-transfer-over-optical fibre and Ethernet were already in place before optical-fibre transfer arrived to Thunderbolt).
One would be remiss if they didn’t consider the white whale aspect of tossing IEEE1394 for Light Peak/TB, or eschewing ZFS for APFS, all because the show man got showed up.
Also, FW80 wasn't faster anymore when the Thunderbolt to FireWire adapter was released in 2012 - cause Macs were updated to include USB3 at about the same time.
Again, three years passed between Jobs’s faux pas in 2008 and the arrival of Thunderbolt (in 2011) and USB 3.0 (over a year after Thunderbolt). That’s when a
lot of Mac models came through, were sold, and dominated the circulation of what end users were using, from consumer level Mac mini and MacBooks to server-level Mac Pros and Xserves. FireWire 800, throughout, was still de facto the fastest offering across all of them (or, with a cheap adapter, could be adapted to FW400 on the remote side).
During those three, long years of Jobs doing damage control and the arrival of Thunderbolt, FW1600 had been out there and
was quite ready to step up. But for Apple to adopt FW1600 for their Macs at that point would have made Jobs look, uh, doubly worse for having prematurely delcaring its “death” in the first place. If anything, that death was anything but organic the way, say, SCSI was.
EDIT to add: After posting this, I
found another post on the forums with a straightforward breakdown of Thunderbolt 1’s effective data transfer speeds. We depend upon GT/s —
gigatransfers-per-second — ratings when discussing Thunderbolt and PCIe, but this is the
physical signal speed of the wire/cable medium, not the
data speed. With header/footer subtracted from a 5GT/s Thunderbolt 1 packet — 20 per cent of that data is header/footer overhead — the effective,
total data transfer rate is only 4000Mbps; but duplexed, that’s 2000MB/s.
2Gbps is barely quicker than than 1.6GBps data speed/transfer rate of FW1600, which was two years old and in place before Intel’s very first Light Peak/TB demonstrations rolled out in 2009. Optical cable transfer for IEEE1394b-2002 was
seven years ahead of that first Light Peak demonstration, which relied on modified USB plugs paired with an optical cable. FW1600 was also to be succeeded in 2012 by FW3200, making it, again, significantly faster than TB1 just one year later.
My hot take, given all this: Light Peak/Thunderbolt was an engineering make-work project. It was designed to improve revenue on
licensing the use of a new serial standard at an order much steeper than either the licensing of IEEE1394 (which never gained ubiquitous traction on Windows-ready Intel boxes), or either of the unlicensed USB or direct-PCIe. It may have been simpler for a non-proprietary standard, not unlike adopting USB in 1996, to engineer a non-proprietary, direct PCIe-over-external-cable standard. But we got the Thunderbolt we deserve, and we also pay dearly for that embedded cost of its licensing. 🙃
xairy.io
