Why don't we have internal firewire?

[McCarthy]

If we do have internal firewire anywhere, please excuse my mistake and tell me more about it...

Not having seen any myself I'm curious though why it's not being mentioned for internal use. For those who have read the posts about legacy free boards you've no doubt seen people giving grudging support for keeping the floppy around a bit longer. But if the floppy became a firewire piece, voila, there goes the legacy problem and chipsets can be cleaned up some more. Firewire CAN BE bootable, right?

I used to ponder this about USB with floppies, but not being a fan of the reliability of USB and having only that one device in mind it wasn't much of an idea. Where firewire could easy support the (99.99999999% of the time idle) floppy, CDroms and even extra HD's what's the downside?

400Mbps, around 40mb/s, right? While ATA has gotten faster recently, that's still a good speed for internal devices for the majority of us. And it would seem to be easier to implement more firewire ports than more IDE. I've even seen little adapters you can use to run an IDE drive off firewire, though I don't know that they'd be very popular for internal use except for those out of IDE ports and wanting to cram more drives in.

Anyway, where's the fatal flaw in this idea? I'm not saying replace ATA, not at all. I can certainly see where someone inputing video from their digital camera over firewire to their firewire drive would run into ploblems. But as we don't record directly from the camera (or scanner, or other devices) to CD-R or DVD-R then the bus wouldn't be saturated, it'd be going firewire to IDE, edit, then IDE to firewire later when burned.

You'd have a motherboard in the end with:
4 IDE channels as usual *8 for RAID boards
No floppy controller
No more ISA bus in the chipset (giving up my PS/2 ports in this as well)
No more parallel/seriel, so lots of USB/USB2.0 / Firewire ports out back
And a couple firewire ports on the board inside, maybe where the floppy controller was

The internal firewire (throw in a USB too while we're at it) which aren't used for internal devices are then free to be hooked up to front panel ports. Of course if you don't have any need for internal use and don't want front panel ports then just hook them up to run out the back, nothing lost.

Good idea or rotten one?

--Mc

Edited to fix a Mbps/mb typo

 

[pm]

It all dates back to this: "Apple Ignites Firewire Fury" (link to 1999 EETimes article). Consider also that chipsets have an 18+ month design cycle.

For other reading (still at EETimes): Sour Apple, and Turning the 1394 goose into a goose egg.

IIRC, Apple and the rest of the world resolved most of the difference of opinion over royalties within a year from these stories in early 1999. But with the extended design cycles, it will take a while for the designs to appear. And now USB 2.0 may start to take some steam away from 1394 as well.

As far as the idea of a motherboard with no parallel, serial, floppy, or PS/2 ports. That's definitely where the PC is headed. It will just take a while to get there. Someone (Abit?) just released a motherboard similar to your idea, IIRC.

Edit: took me a while to find it. Here's a review of the Abit AT7. I thought I remembered Abit did it. Here's a review of a motherboard with on-board Firewire a ton of USB and no legacy ports. It does have a floppy though.

 

[McCarthy]

Yeah, the AT7 I'm familiar with. Rather dislike it, but that's a whole different thread.

Where firewire is peer to peer like scsi and USB isn't and has more overhead I don't rate them as comparable. Yeah, USB 2.0 is a bit faster, 480Mbps vs 400Mbps, but the overhead and real world performance will still favor firewire from everything I've heard and read. USB was a good idea, but came about 10 years too late to really be handy. Still, since it's supposed to be cheaper I guess there's a place for both.

I shy away from USB floppy for a few reasons, even though there are already Mac ones around. First, I'm not sure if they're bootable on the PC, just plain don't know. Second there's the overhead and the reliability problems I personally have had with USB 1.0 devices so far, personal bias, but there it is. And finally, and most importantly, USB wouldnt' be appropriate for internal CDroms (and DVD of course) so why mix in two when I can just mention firewire as a replacement for the traditional FDC.

Maybe a better question for me to pose would be "If you could get a firefire floppy and internal CDroms for about the same price would you choose them over a traditional floppy controller and IDE CDs."?

I would for my replacement drives. Unless someone shows me a technical reason why it'd be worse.

--Mc

Again, same typos fixed

 

[Barnaby W. Füi]

i like the legacy-free thing. i dont use floppies anymore, i dont use any serial/parallel devices, however i do use ps/2 for my keyboards and one of my mice, but i suppose there are adapters.

out with the old, in with the new. if someone needs a parallel or serial port that bad, they can buy a 15 dollar or whatever PCI card. get those dumbass ports off my motherboard

its about time we stop the insanity of infinite backwards compatability. apple has always been good at that. i mean, they have usb/firewire/gigabit ethernet on their machines, thats pretty nice. of course you pay for it...but its cool nonetheless

 

[spanner]

Parralel, serial ports etc are far from legacy, most proffesional equipment i.e multi thousand dollar electronic sensors still use floppies, serial and parralel ports. Sure you could eliminate them and save a couple of cents on the home computers but then they would need separate ones for the big engineering companies and universities. Face it people, parralel ports and serial ports are gonna be around for a very long time, floppies can possibly be done away with in the future, but only when another suitable, cheap universal standard comes along that can do everything a floppy drive can do.

 

[nortexoid]

uh, isn't it 400Mbit/sec, NOT 400MB/sec? - huge difference...anyway...

USB is utter poo compared to firewire, and the topic's been beaten so bad we won't even bring it up here, again...

spanner, that wasn't his point - i.e. - that we should get rid of floppy drives...rather, we should get rid of the legacy floppy 34-pin interface and use firewire instead...that is, firewire interface floppy drives...in fact, u could run everything off native firewire and ditch all legacy ports, except parallel ATA...

AFAIK, all firewire ATAPI devices still use parallel ATA interfaces deep down inside...somebody correct me if this has changed.

i agree w/ this proposal, and apple's already done it...they're the only intelligent ones with the ability to totally dispose of legacy interfaces on their computres since they lack any competitors...far harder to do in the PC industry when most people adore cheapness..

 

[Eug]

<< Why don't we have internal firewire? >>



1) 1394a is only 50 MB/s max. Too slow by today's standards for the future of hard drive technology. It would really have to be 1394b, which isn't out yet.
2) The role will be taken by serial ATA, which similarly has nice small round cables, and which has the benefit of backwards compatibility with standard IDE drives.
3) As far as I know, Firewire is not currently bootable on a Windows PC, but I'm not sure about that. I do know that a Mac can boot from Firewire though.
4) Licensing fees. They were initially US$1 per port, but I think it's now $0.25 per device. (Or is that per port?)

By the way, I DO have internal Firewire, but right now that port is unused. I cannot use it for a front port, because my port replicator drive bay actually only works with external ports.

I could stick a Firewire drive on it, but all an internal Firewire drive is at the moment is an IDE drive with an (expensive) IDE->Firewire bridge board. The prices I've seen for the bare bridge boards are as high (~$70) as complete external enclosures which include a bridge board, so I have only been buying external enclosures, which gives me the added benefit of using my drives with Mac as well.

My setup:

PC:
Two hard drives on IDE
One DVD-ROM on IDE
One CD-ROM on IDE
Two external Firewire ports (one on back and one on front)
One internal Firewire port

iBook:
One hard drive on IDE
One CD-RW/DVD on IDE
One external Firewire port

External Firewire:
Panasonic DVD-RAM/-R with Initio chipset enclosure
Teac 24X CD-RW waiting to be put into an Oxford 911 chipset enclosure I have on order.
IBM 75GXP 60 GB hard drive in Oxford 911 chipset enclosure
Lexar CompactFlash reader
6-port powered Firewire hub on order.

 

[nortexoid]

oh right...that's another big issue....firewire devices aren't bootable on PC..that's a problem.

i don't think 1394a being limited to 400MBit is a problem tho...especially with b just around the corner...especially with its advantages over parallel ATA

i still think USB blows...its power output is too low - it's sad...who wants to connect every peripheral to an external AC outlet?...

 

[Eug]

<< i don't think 1394a being limited to 400MBit is a problem tho...especially with b just around the corner...especially with its advantages over parallel ATA

i still think USB blows...its power output is too low - it's sad...who wants to connect every peripheral to an external AC outlet?... >>


Well, b is only 100 MB/s. Not bad, but serial ATA is also RoundTheCorner (TM).

Yeah, I agree that USB blows in many ways. Buggy drivers, higher CPU utilitilization, and lower power (at I think 500mA x 5V = 2.5 Watts max). Firewire is much more stable and provides up to 15 Watts (1.25A x 12V), but what the PC makers have to do is have better support for power over Firewire. I don't know of a single PC laptop that provides a powered Firewire port, whether thru a dedicated port or over PCMCIA. OTOH, if I want I can power a whole portable hard drive like an iPod or a portable 2.5" notebook drive, using only my iBook's Firewire power. My Firewire CF reader doesn't even have an external power option, so I don't think it works on PCs without a PCI card or powered hub for a laptop.

 

[NukemAll]

<< Parralel, serial ports etc are far from legacy, most proffesional equipment i.e multi thousand dollar electronic sensors still use floppies, serial and parralel ports. Sure you could eliminate them and save a couple of cents on the home computers but then they would need separate ones for the big engineering companies and universities. Face it people, parralel ports and serial ports are gonna be around for a very long time, floppies can possibly be done away with in the future, but only when another suitable, cheap universal standard comes along that can do everything a floppy drive can do. >>



You're funny. Obviously you do not know what legacy means in the context of PC hardware. Everything you say here that is far from legacy, is 100% legacy. Legacy is defined by these types of devices and interfaces. Don't bet that they will be around for a long while - already you would be very hard pressed to find a new motherboard with ISA slots. Abit has laready announced the release of 100% legacy free motherboards. If you can find a brand new pc or motherboard not custom built nor refurbished in 3 years with PS/2, 9-pin serial, Standard floppy or Parallel printer port, they would be amazing. You can be sure that by the end of this year, any new PC hardware coming out(as in componants) will be legacy free. A couple of years after that existing non-legacy free producst will have all exited the marketplace. Sure, if you are willing to pay a premuim, you will find non legacy hardware, you can still buy 386 processors today. But most will be inventory built up by large manufacturr for the industrial sector for warranty and service replacements. They got to support there products if they want to keep their customers. Why the hell anyone now would want to pay 5 or 10 grand for a 386 for use at home or as a workstation is beyond me(I am guessing you would). Likewise, 5 or 10 years from now why anyone would want to pay 10-20 thousand Dollars for a PIII 1GHz with onboard floppy, PS/2, serial and parallel ports is your guess.

 

[McCarthy]

nortexoid - sorry about the typos, I get all confuzzled sometimes

BingBongWongFooey - yeah, I know at least Belkin makes a PS/2 to USB adapter. $50, but I'd probably spend that to keep my beloved old IBM tanks of keyboards. After all it'd be a one time cost.

Spanner - parallel, serial, PS/2, floppy controller are all legacy ports. I agree, they still have usefulness, are even required, for some. But unlike the scenerio painted, there's no reason to have to keep around a 10 year old computer to have them. The parallel and serial ports are easily reintroduced via a PCI expansion card for those who do need them. Which is an ever shrinking group and the time to remove them from the mainstream has arrived or shortly will.

Eug - If you noticed I said I'm not saying replace ATA, not at all. (Current parallel or future serial) I'm talking about slower devices here, especially the floppy which is going to require us to hang onto the ISA bus longer than anything else, and throwing in CDroms and DVD drives simply because 50mb/s is fast enough for them right now, and would benefit all the ATA devices in use at the same time by freeing up their resources. At least for current CD/DVD drives, by the time they're exceeding 50mb/s then 1394b should be around. Not only backwards compatable, but plug compatable as well like USB 1.0 to USB 2.0. Someone's going to make a mint selling clunky dongles for awhile so people can use their current parallel ATA drives in new computers and vica versa.

NukemAll - the Abit isn't 100% legacy free. Still has the floppy controller. And as I've recently come to understand, as long as it has that it still has to have all the same legacy ISA bus built into the chipset that it would if it had the parallel/serial/PS/2 ports in place. Just that the physical connectors have been taken off the board. A chop shop job to try to make a Buick look like a Porsche if you will.

The bootable problem - If Apple is already doing it I'd going to go out on a limb here and guess PCs could boot off 1394 if there was a reason to set it up that way.

The other nice benefit of putting the floppy on firewire, aside from cleaning up chipsets of ISA legacy, and getting rid of a big connector on the motherboard thereby freeing space, and getting a cluttering ribbon cable out of the case....is that it could be powered right from the 1394 bus. Some more clutter gone.

I'm not quiet sure how serial ATA found it's way into this. Unless the idea was to go to serial ATA floppy drives. Way I understood serial ATA we're still going to be limited in the number of channels available, actually 1 channel per controller with it isn't it? Seems a pretty big waste to give one to a floppy when 1394 can support so many devices.

Inevitably this thread has become in part what it had to, a commentary on legacy hardware and it's replacement. Though it may seem like a reversal on my stance taken against the Abit AT7 in a previous thread, it isn't. I'm for removing legacy devices if there's a benefit to doing it. Simply removing the physical connectors and editing them out of the bios like the AT7 did is little more than a gimmick. Given current available chipsets that's all they could do though, so as a 'trial balloon' I guess what they had available to try. As long as the floppy is there the idea of legacy free is somewhat moot though, just putting up a trial balloon of my own here to see if there's hope for true legacy free in the near future.

Please keep adding thoughts, this is fun.

--Mc

 

[nortexoid]

this is how i see things, hopefully soon:

all legacy interfaces gone, replaced by USB, firewire and serial ATA all natively supported by modern chipsets.
boards will start sporting good onboard audio and video (courtesy of, say, SIS' 330 graphics core or the like) thereby slowing phasing out the fullATX spec in favor of flex/mATX...

full size roms will no longer be needed and should be replaced with half-height (i.e. laptop) drives...
cases will obviously follow suit....slim-line will become popular finally
breakout boxes wiht a plethora of connectors might become more common as well
mediaGX (what the hell are they called?) will be standard for easy front porting of a flash memory reader.
flat panels will still cost a fortune...hence CRTs will look gigantic compared to modern PC cases
etc...

 

[sparks]

I converted my Plextor 40/12/48 to an internal 1394 device. I already have the 1394 for DV editing and I needed the IDE port for a new HD, so I bought an IDE to Firewire bridge and connected it. It works great, no complaints.

 

[dullard]

McCarthy what do you gain from doing this? I can see one drawback: everyone that uses these devices must go out of their way to add in cards. But what is the benefit? The cost for firewire far outweight the cost for the legacy parts - so your proposal will make motherboards more expensive and less capable. The fact is a large number of people still use these legacy parts - so why get rid of them? What is the incentive?

The fatal flaw as I see it is that you want to make a motherboard less capable and have a higher price.

 

[Eug]

<< Eug - If you noticed I said I'm not saying replace ATA, not at all. (Current parallel or future serial) I'm talking about slower devices here, especially the floppy which is going to require us to hang onto the ISA bus longer than anything else, and throwing in CDroms and DVD drives simply because 50mb/s is fast enough for them right now, and would benefit all the ATA devices in use at the same time by freeing up their resources. At least for current CD/DVD drives, by the time they're exceeding 50mb/s then 1394b should be around. Not only backwards compatable, but plug compatable as well like USB 1.0 to USB 2.0. Someone's going to make a mint selling clunky dongles for awhile so people can use their current parallel ATA drives in new computers and vica versa. >>

That's precisely my point. Why change DVD/CD to Firewire, when we already have IDE. IDE works great for DVD. Most home users don't have 4 drives in their system, so their IDE channels can easily accomodate their needs. Firewire would only add to the cost. I do recommend Firewire to most people I know who are buying a new computer, but that's only because I like Firewire and if I ever need to do stuff on their computer my external devices would work on their machines. However, when it comes to Joe Schmoe and DVD, their IDE bus is fine. In fact, in some ways it's superior to Firewire 1394a: It's cheaper for one thing, and for another it's faster. For floppy, the controllers are cheap as borscht and it works fine. Going Firewire certainly isn't make a floppy drive run at 5 MB/s or anything. That said, I'd like to see the floppy stuff gone too. I wonder if it'd be easier to implement IDE floppy though on a large scale than floppy Firewire.

Now, external devices are a different story. USB 2 does offer backwards compatibility to USB 1.1, but there have been some problems, and currently for high bandwidth devices the implementation of it is nowhere near as good as Firewire. Thus I do agree with the inclusion of Firewire on computers for external devices.

P.S. I have two 5.25" floppy drives sitting on my floor. Anyone in Toronto want one?


<< I converted my Plextor 40/12/48 to an internal 1394 device. I already have the 1394 for DV editing and I needed the IDE port for a new HD, so I bought an IDE to Firewire bridge and connected it. It works great, no complaints. >>

Well, the only problem here is that it still isn't really a 1394 device. It's an IDE device with a 1394 bridge. Also, a lot of the time it's actually cheaper to buy a PCI IDE card, that would have provided you with space for 4 more IDE devices. (But if you already have the port, why not use it and save a PCI slot of course. )

 

[McCarthy]

dullard - in a couple threads about the Abit AT7 and legacy free as a concept I expressed support for keeping the PS/2 ports around as long as possible. As long as the ISA bus is still present in the southbridge the benefit of removing them is minimized, just saving a bit of space on the backplane.

I found out somewhat to my suprise that those interested enough to comment were mostly for doing away with them.

Now when you consider the serial ports you find a physical connector requiring about the same physical space as the PS/2 port. Per serial port. You also find fewer people still using them than PS/2 ports. (few external modem users, few older digital camera users, and very small groups of other hardware users). That market no longer demands these, most people leave them unused and have for several years.

With the parallel port you find more still in use currently than serial, but it's also the biggest consumer of space on that backplane. Even with the parallel port the number of users is falling fast; what percentage of new printers don't have USB or Firewire? What percentage of scanners? What percentage of the market still uses laplink through the parallel to share files between their home computer and laptop? All very very small numbers.

Now you could argue what percentage of people have firewire devices, and that too is a very small number currently. However it's an increasing number, not a decreasing one.

What's the incentive? Incentive is to A) clean up board designs B) free board and backplane space C) reduce cost of board by eliminating legacy hardware D) eventually remove the ISA bus from the southbridge, reducing elements from it's design, increasing efficiency.

Or more to the point, why should 100% of the market have to pay for the inclusion legacy hardware with each motherboard purchase and suffer any drawback from it when the 10% of the market still using any of the above parts can replace them with a one time purchase of a parallel/serial expansion board?

Anyway, enough of that. It's going to happen, my problem is if it happens for A and B, but not C or D because the southbridge still requires the ISA bus to support a friggin floppy device.

Eug - Most home users only have 2 IDE devices, the HD and CDrom or DVD drive that came in their Dell. Do you therefore support reducing the number of IDE channels available to 2? Which may well become a concern for some boards. From a short commentary at Hardwarecentral:


<< Another area I find troubling is the Serial ATA 1.0 limitation of only one device per connector channel. I admit to going a bit over the top with my home PC, but I currently have eight parallel ATA devices installed (four hard disks plus CD-ROM, CD-RW, DVD-ROM, and floppy drives) that could potentially be replaced by Serial ATA.
I don't imagine I'm alone in this respect, as hard disks and CD-RWs are among the most popular upgrades for home and business users alike. The Serial ATA cards and chipsets I've seen all support only two independent Serial ATA channels and connectors, which according to the specification, means only two Serial ATA devices (excluding RAID arrays). >>



Now unless the writer is misinformed it would seem to me that means putting in a new controller for each device for serial ATA which wouldn't be able to be used for front panel connectors or other purposes. This vs throwing some slower devices (floppy, Zip drives and yes perhaps even CDroms) to Firewire. As firewire works, is already available and I support it I'd rather go that route than the untested and questionable future of serial ATA for this purpose. But that's me.

Like I said in those previous threads, I'd actually prefer to have 10 PCI slots and nothing onboard, no audio, no video, no parallel, serial, USB, firewire and no, not even IDE channels. I actually liked the way things were with 486's where you had everything on a card: time for a new motherboard you swapped it and got to reuse your existing components. Legacy was up to you. But as long as we have one size fits all to a degree with current trends, this is the combo I'd choose.

--Mc

 

[Eug]

<< Another area I find troubling is the Serial ATA 1.0 limitation of only one device per connector channel. I admit to going a bit over the top with my home PC, but I currently have eight parallel ATA devices installed (four hard disks plus CD-ROM, CD-RW, DVD-ROM, and floppy drives) that could potentially be replaced by Serial ATA.
I don't imagine I'm alone in this respect, as hard disks and CD-RWs are among the most popular upgrades for home and business users alike. The Serial ATA cards and chipsets I've seen all support only two independent Serial ATA channels and connectors, which according to the specification, means only two Serial ATA devices (excluding RAID arrays). >>



I'm not sure how it is going to work but from my readings there seem to be two options:

1) 1 serial ATA controller per serial ATA device. In this scenario a dual controller set up would only allow 2 drives. Not ideal.
2) 1 serial ATA controller attached to a serial ATA switch. In this scenario a single controller would be hooked up to a smart switch allowing multiple drives, sharing the same overall bandwidth. Dunno if the switch is onboard or a separate internal board.

(I must say, I tend to keep it to 1 device per parallel ATA channel anyway, at least for hard drives. In the every least I try to keep my writer off the channels of the main hard drive and the source CD-ROM.)

I don't think daisychaining is a major goal in later iterations of serial ATA but I dunno for sure.