#IDE – SCSI
27 messages in this thread
I find it incredibly hard to believe that IDE will surpass SCSI in performance
by the end of the year. There are so many high performance workstations out
there that have faster busses, need for higher throughput than 99.9% PCs, and
use SCSI that I just can't imagine HD manufacturers shooting themselves in the
foot by developing faster technology for a system that is used in one platform
(IDE in PCs). While there are more PCs out there than just about everything
else, SCSI is a standard for PCs, and also the standard for just about every
other platform from workstation on down. Just seems nuts.
John Tissavary (LUNA cie)
>> I find it incredibly hard to believe that IDE will surpass SCSI in
performance by the end of the year. << I agree. Its appears that all the R&D
for HD's is going into SCSI and IDE is getting the spinoff, not to make a bad
pun. <g>
On a similar but not directly related note, I think its interesting that both
the engineers for the Targa 2000 and the MAX chose the EISA approach. Haven't
heard a thing about either Truevision or Matrox adopting PCI for this
technology, have you?
-JE
Might that not be because EISA is open and MCA and PCI are closed
architectures? ie. Go with the bigger market. Might be different in another
year though.
>> Might that be … go with the bigger market. << I imagine that was the
thought. But I suspect PCI will dominate, although there are several companies
like Compaq who successfully built P5's on EISA architecture. Compaq is beating
IBM in PC sales, which I find interesting just because of their history. I'm
hearing more horror stories about DEC. These high tech firms are taking a
beating. I wonder how well the Intel board is going to be recieved.
-JE
I think they chose EISA because of its buss throuput (33mb per second)
capabilities, and because it is the most proven, solid PC architecture out
there. My guess is they were more confident in the short term with EISA. I'll
bet when PCI gets a track record, and the spec tightens up or whatever these
developers feel they need, both companies will release PCI versions or new
products based on PCI.
John Tissavary (LUNA cie)
>> I think they chose EISA because of its bus throughput… <<
Yes but they could have chosen to hang the drive directly off their card as DPS
did. In that regard, in the short term at least, the potential market will be
much larger for the DPS than for either the MAX or the Targa 2000. We all seem
to be in agreement that PCI is probably going to be the next standard. I've
been looking at that new Intel P90. Their board, their BIOS and their Chip.
(Although they may have subcontracted the mother board.) And the price is
right. Of course the $24 question is, how well does it run 3DS?
-JE
Rumblings in Truevision indicate that a PCI T2000 is in the works. It only
makes sense since that's where the market will be in 6 months (maybe sooner).
Personally I'd have a hard time recommending someone reconstruct an existing
system just to to go T2000/EISA, much less purcase new. The immediate payback
would have to be significant.
Ken Burke
Designlab
>> Personally I'd have a hard time recommending someone recostruct an existing
system just to go T2000/EISA. <<
You've got a good point, but lets say someone needs to do desktop video for
multi-media or game developement _now_. The fact that Compaq and AMI are
continuing to support EISA with P5's means its not dead yet. And as is
apparent, the technology rush is making it such that we're upgrading everything
every 12 months anyway. I suspect that the PAR and MAX technology will become
dated shortly (18 months?) not that there useful life will not extend beyond
that. (Some would argue its already dated as the price of Accoms and Abekas
continue to drop.) Abekas has already come out with a 10 bit desktop version.
Look for more of the same, which raises the question why go JPEG except for
CD-ROM release or archiving, which will forstall the demise of JPEG. Mastering
for video is another story.
-JE
How long ago did they announce the MAX? I guess it's been over two years. There
were no PCI then. VESA local bus was just starting to break grounds. I would
think that they though EISA would be a better bet. Now they're screwed with a
dead beat bus that nobody uses. Well, maybe it was a better decision than going
with Micro Channel! The war now will be PCI x VESA. With all these "big" guys
coming out with Multi Pentiums, MIPS 4400, Alpha, what have you, with PCI
buses, it's going to be pretty obvious that EISA has joined the Kaypro club.
>> Now they're screwed with a dead beat bus that nobody uses. <<
That's if you discount all the Compaq and AMI boards out there among others,
both of whom have P5 boards in the EISA standard. So its obviously not dead
yet. We've been hearing the prediction for the death of DOS for a long time
also. Since the orginal release of Windows as a matter of fact, but it probably
won't happen until Windows NT takes a strong footing. PCI may be the death
knell for EISA and VESA but its obvious we are in a period of radical
transition. My predicition is that Windows NT and PCI will be the ultimate
victors but as they say, "it ain't over, till its over"… and it sure ain't
over, although it does look like the writing is on the wall.
FWIW I learned my lesson as a member of the Osborne club, which was decimated
by Kaypro, only to be undone by Compaq… and the battle continues to rage. <g>
Ahh yes, the Kaypro. I'm still the proud owner of a Kaypro II… it's all
pretty amazing how far things have gone in 10-12 years. And they said we'd
never need all 64k of RAM!
Ken Burke
Designlab
>> I'm still the proud owner of a Kaypro II… <<
Well I'm the owner (proud? <g>) of an Osborne I in the original tan case. It
was my first and I'm hanging on to it to remind Microsoft what can be done with
a 2k operating system. <G>
-JE
John
Did you hear that Gary Kildall, founder of Digital Research, died this
weekend. End of an era?
Ted(greycase)B
!^NavFont01F0007MGHHG61F7E2
Ted,
>> Did you hear that Gary Kildall…. died this weekend. <<
That's a little reminiscent of Mark Twain, who was born and died on the day of
Halley's comet. In this case comets impacting Jupiter. End of an era? Most
probably, the PC is no longer an adolescent but growing up. But at least the
Rolling Stones are doing _another_ World Tour and I don't feel quite so old.
<g>
I'd try to get the old dude (tan case) fired up but the modem died along time
ago. Its limping along on one 360k drive (a $450 upgrade btw). Curiously
enough, there's still an Osborne User Group in San Diego, but I've never been.
I should see about getting another drive for it. What kind of shape is your's
in? 🙂
-JE
John
My Osborne is in pretty good shape. I had paid for the disk upgrade just when
they went out of business. FWIW, my first copy of Autocad was for CPM, not on
the Osborne but an NEC APC.
TedB
!^NavFont01F0007MGHHG9BA928
>> my first copy of Autocad was for CPM… <<
Jeez Ted, you've got everyone I know beat including a guy down here who teaches
at Palomar college who I could swear was originally constructed in AutoCAD. The
famous D.L. <g> He built his house with ACAD on the coast of Del Mar. Very
remarkable design, especially the roof. I thought the original AutoCad was
written for IBM DOS for the XT, but that was before my involvement with ADESK.
Can you elaborate on that?
Thanks,
-JE
John
D.L. certainly has me beat in the bad joke department though. My Autocad for
CPM was an early 1.4 and shortly after I got the machine an MS-DOS copy arrived
so I switched and never turned back<g>. remember, you had to crtl-C every time
you changed a diskette (8") and it got confusing with Acad.
I've seen pictures of Dick's mansion, tell him I say hi.
TedB
!^NavFont01F0006MGHHHoFA40
John,
It's really pretty simple – EISA works, but PCI is still undergoing
revision, and there are implementation problems. PCI should be shaping up this
fall, and you should expect more manufacturers, such as Truevision, to migrate
to it.
Kevin Krell – Computer Support Associates
This is my take on the issue exactly. I'm quite sure EISA will be yesterday's
news very soon (pretty much already is…), but it's still a useful spec for
the time being.
John Tissavary (LUNA cie)
>> I'm quite sure EISA will be yesterday's news very soon… <<
I'm not so sure, I also see the possiblity that the market is fragmenting;
Alpha, Power-PC, MCA, EISA, VESA, PCI. We may not end up with one standard but
specialty markets with possibly one or two predominant platforms. Intel may not
hold on to its lead in which case we're in for even more confusion.
-JE
The problem with this fragmentation (and I agree with you on that) is that it
gets harder for hardware manufacturers to select a bus to support. It ain't
cheap designing boards for ISA, EISA, VLB, and PCI all at the same time. It's
even more expensive to design mother boards. The manufacturers are going to
have to "select" one and see hoppe for the best. The bus who gets the greatest
number of manufacturers wins. EISA bus is mostly being used for network servers
as both SCSI and NIC support is strong in EISA. Machines designed around EISA
today are thought out as straight file servers. You won't find a EISA "desktop"
machine that easily. ISA is the common denominator. All of the above support
it. VLB got a lot of attention in the past year. Just about every video board
out there has a VLB version (just a very few EISA and that's for the reason
stated above, EISA = file server). PCI started sorta screwed up but gained a
lot of momentum and seems to be the "hot" thing on the market today (it doesn't
mean it will last longer than tomorrow as this market is more volatile than
processed chili beans)…
I guess we all agree with this in one way or another. The point I was trying to
make was that EISA was the "obvious" choice two years ago as it could not be
any other way. It's just that it took them too long to have the product out and
they may now be facing a situation where they are already late and to consider
a bus change would make it even later.
Technically speaking, ISA is limited to 8Mhz and 16 bit transfer. EISA is
limited to 16MHz and 32 bit transfer. Both VLB and PCI are limited to 33MHz and
32 bit transfer. That in itself shows the route of things. Note that to break
that, they will have to "invent" something. Increasing the buss width will make
the connector too long (say, a 64 bit bus). Increasing the bandwidth will make
it almost impossible to avoid leaks (connector to connector signal leak) and
interference (you will be able to "tune" your computer on channel 2).
Your technical arguments are sound and it not only makes it more difficult for
hardware manufactures but consumers who get confused by the proliferation of
different standards. As a whole it hurts the industry, but as we've seen with
IBM and DEC they stubbornly persist in going their own way and as a result have
been clobbered, but they just don't get it. So they cut back on staff and
plunge headlong into it again. The industry needs to rally and accept PCI as
the standard and form a committee <horrors> to agree on the development and
continuation of that standard in a practical way. Of course if that happens,
they could get hit with the "Sherman Anti-Trust" act. Maybe what's need is a
committee like "NTSC" but then the technology might stay dated for 40 years. I
don't know what the answer is, but I think we're all agreed, there should be a
standard platform and now is a critical time as NT starts to take hold. My vote
as I've said is with PCI and NT and here's hoping both can deliver as promised.
If not you can be sure other platforms will spring up, not to mention other OS.
BTW I wouldn't mind tuning in to Channel 2 if it was devoted to Computer
Related stuff. Don't you think its about time computers had their own channel?
<g>
-JE
>> Don't you think its about time computers had their own channel? …
There are a couple of them. All through satellite only though. Galaxy 5
transponder 21 is the "biggest" one.
When I was in Tucson I had a dish in the backyard, but here in the city… well
I miss it, especially for these space related shots. I'm sure you're having a
blast… (bad pun <g>).
-JE
Just quoting the pundits. I'm no expert but with thruputs reaching the
purported 15-17 MB/sec susteined near the first of the year it may just happen.
Nothing in this computer industry surprises me anymore. A friend of mine
working as a network consultant showed me some prerelease spec sheets from Fuji
that were pretty amazing. Still its vaporware though.
You're right, there's no telling what's going to come… but FUJI makes SCSI
drives, too. And they all sell to all markets (PC, MAC, SGI, SUN, etc…), so
from a marketing $$$ standpoint it seems the high performance stuff wouldn't
want to exclude SCSI.
John Tissavary (LUNA cie)
Oh I do agree but it does seem to me more performance developments have
surfaced in the IDE technology recently.