| Author |
Message |
kus@free.net
Guest
|
 Posted:
Tue Aug 09, 2005 10:06 pm Post subject:
What was new&important in computer architecture 10 years ago |
|
|
I'm looking what was especially "new & important" in computer
architecture (I don't say about "pure researches") 10 years ago, in
1995.
1) Introducing of Pentium Pro : it was begin of wide using of x86 chips
for serious server applications (and x86 Beowulf clusters later).
2) Clear win of 64-bit RISC chips:
to Alpha 21164 it was added SGI/MIPS R8K/90 Mhz (75 Mhz started at
1994?)
and PA-8000
3) Start of win of RISC SMP-servers over vector mini-supercomputers
(Cray J90) or even over large vector systems (as Cray C90 - taking into
account price/performance ratio).
This win was based on SMPs like SGI Power Challenge, DEC 8200/8400
and Convex SPP servers (btw, ccNUMA in large configurations;
Sequent, if I'm correct, proposed ccNUMA on quad Pentium Pro boards).
What is your opinion - is it correct ? What should be added else ?
Mikhail Kuzminsky
Computer Assistance to Chemical Research Center
Zelinsky Institute of Organic Chemistry
Moscow |
|
| Back to top |
|
 |
Nick Maclaren
Guest
|
| Posted:
Wed Aug 10, 2005 12:15 am Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
In article <1123607191.752418.154800@g49g2000cwa.googlegroups.com>,
kus@free.net <kus@free.net> wrote:
| Quote: | I'm looking what was especially "new & important" in computer
architecture (I don't say about "pure researches") 10 years ago, in
1995.
1) Introducing of Pentium Pro : it was begin of wide using of x86 chips
for serious server applications (and x86 Beowulf clusters later).
|
Not really. That was the 80386. The increase in the use of IA-32
for serious server applications has been more-or-less monotonic
since then.
| Quote: | 2) Clear win of 64-bit RISC chips:
|
OK. I'll pass that.
| Quote: | 3) Start of win of RISC SMP-servers over vector mini-supercomputers
(Cray J90) or even over large vector systems (as Cray C90 - taking into
account price/performance ratio).
|
Not the start - more-or-less the end. The start was much earlier,
perhaps even 20 years ago.
Regards,
Nick Maclaren. |
|
| Back to top |
|
|
rhnlogic
Guest
|
| Posted:
Wed Aug 10, 2005 12:15 am Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
Nick Maclaren wrote:
| Quote: | In article <1123607191.752418.154800@g49g2000cwa.googlegroups.com>,
kus@free.net <kus@free.net> wrote:
....
3) Start of win of RISC SMP-servers over vector mini-supercomputers
(Cray J90) or even over large vector systems (as Cray C90 - taking into
account price/performance ratio).
Not the start - more-or-less the end. The start was much earlier,
perhaps even 20 years ago.
|
Who was shipping vector mini-supers or RISC SMP boxen in 1985?
The "attack of the killer micro's" paper was circa 1989, IIRC, a few
year later.
--
rhn A.T. nicholson d.O.t C-o-m |
|
| Back to top |
|
|
Del Cecchi
Guest
|
| Posted:
Wed Aug 10, 2005 6:22 am Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
<kus@free.net> wrote in message
news:1123607191.752418.154800@g49g2000cwa.googlegroups.com...
| Quote: | I'm looking what was especially "new & important" in computer
architecture (I don't say about "pure researches") 10 years ago, in
1995.
1) Introducing of Pentium Pro : it was begin of wide using of x86 chips
for serious server applications (and x86 Beowulf clusters later).
2) Clear win of 64-bit RISC chips:
to Alpha 21164 it was added SGI/MIPS R8K/90 Mhz (75 Mhz started at
1994?)
and PA-8000
3) Start of win of RISC SMP-servers over vector mini-supercomputers
(Cray J90) or even over large vector systems (as Cray C90 - taking into
account price/performance ratio).
This win was based on SMPs like SGI Power Challenge, DEC 8200/8400
and Convex SPP servers (btw, ccNUMA in large configurations;
Sequent, if I'm correct, proposed ccNUMA on quad Pentium Pro boards).
What is your opinion - is it correct ? What should be added else ?
Mikhail Kuzminsky
Computer Assistance to Chemical Research Center
Zelinsky Institute of Organic Chemistry
Moscow
|
NUMA and clusters. That is about the time frame they became commercial.
Sysplex. etc.
> |
|
| Back to top |
|
|
Ketil Malde
Guest
|
| Posted:
Wed Aug 10, 2005 8:15 am Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
"kus@free.net" <kus@free.net> writes:
| Quote: | I'm looking what was especially "new & important" in computer
architecture (I don't say about "pure researches") 10 years ago, in
1995.
|
So you're looking for "new on the market"? (The problem with anything
"new" is that it has done before in some more or less obscure design -
I sometimes suspect there is a prize to the c.a. poster digging up the
oldest reference :-)
| Quote: | 1) Introducing of Pentium Pro : it was begin of wide using of x86 chips
for serious server applications (and x86 Beowulf clusters later).
|
The fun thing about PPro was that it was hampered by a perception of
being too slow on 16-bit code, and consequently on the current
Windows - and it wasn't tremendously successful. (It was a great CPU
to run Linux on, though.) IIRC, one of the improvements in the PII
was faster 16-bit execution.
| Quote: | 2) Clear win of 64-bit RISC chips:
3) Start of win of RISC SMP-servers over vector mini-supercomputers
What is your opinion - is it correct ? What should be added else ?
|
4) Wasn't this about the time SGI was producing (beefy) workstations
with real-time 3D-graphics? The RealityEngine was early nineties,
IIRC.
Prior to this, I think 3D tended to be either simplistic (flat-shaded
polygons in PC games), or not RT (rendering of Tron, Last Starfighter,
etc).
(Of course, if I'm off by a few years here, 1995 marks the start of
the RT 3D era for the PC instead :-)
-k
--
If I haven't seen further, it is by standing in the footprints of giants |
|
| Back to top |
|
|
Nick Maclaren
Guest
|
| Posted:
Wed Aug 10, 2005 8:15 am Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
In article <1123623487.100371.305810@o13g2000cwo.googlegroups.com>,
rhnlogic <rhnlogic@yahoo.com> wrote:
| Quote: | Nick Maclaren wrote:
In article <1123607191.752418.154800@g49g2000cwa.googlegroups.com>,
kus@free.net <kus@free.net> wrote:
...
3) Start of win of RISC SMP-servers over vector mini-supercomputers
(Cray J90) or even over large vector systems (as Cray C90 - taking into
account price/performance ratio).
Not the start - more-or-less the end. The start was much earlier,
perhaps even 20 years ago.
Who was shipping vector mini-supers or RISC SMP boxen in 1985?
|
FPS and others. No, people weren't shipping RISC SMP to any
significant extent, but people were abandoning vector systems in
favour of clusters. And the FIRST step to FPS's demise was the
80386/80387. I was referring to your "START of win".
| Quote: | The "attack of the killer micro's" paper was circa 1989, IIRC, a few
year later.
|
The paper may have been. The event was earlier, for those of us
who were observant in the field.
Regards,
Nick Maclaren. |
|
| Back to top |
|
|
Jason Lee Eckhardt
Guest
|
| Posted:
Wed Aug 10, 2005 2:39 pm Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
In article <ddcc53$kog$1@gemini.csx.cam.ac.uk>,
Nick Maclaren <nmm1@cus.cam.ac.uk> wrote:
| Quote: | In article <1123623487.100371.305810@o13g2000cwo.googlegroups.com>,
rhnlogic <rhnlogic@yahoo.com> wrote:
Nick Maclaren wrote:
In article <1123607191.752418.154800@g49g2000cwa.googlegroups.com>,
kus@free.net <kus@free.net> wrote:
...
3) Start of win of RISC SMP-servers over vector mini-supercomputers
(Cray J90) or even over large vector systems (as Cray C90 - taking into
account price/performance ratio).
Not the start - more-or-less the end. The start was much earlier,
perhaps even 20 years ago.
Who was shipping vector mini-supers or RISC SMP boxen in 1985?
FPS and others.
|
The Convex C1 probably also qualifies as a vector mini-super.
It started shipping sometime in 1985, IIRC. Though it (and
the follow-on, the C2) somehow managed to survive until the
early 1990's.
By the time I arrived at Convex in late 1992, C-series sales
were definitely drying up (if not completely evaporated), and we
began doing our own "RISC SMP" :)
| Quote: | No, people weren't shipping RISC SMP to any
significant extent, but people were abandoning vector systems in
favour of clusters. And the FIRST step to FPS's demise was the
80386/80387. I was referring to your "START of win".
The "attack of the killer micro's" paper was circa 1989, IIRC, a few
year later.
The paper may have been. The event was earlier, for those of us
who were observant in the field.
Regards,
Nick Maclaren. |
|
|
| Back to top |
|
|
Nick Maclaren
Guest
|
| Posted:
Wed Aug 10, 2005 4:15 pm Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
In article <ddd3ih$sjm$1@joe.rice.edu>,
jle@ural.owlnet.rice.edu (Jason Lee Eckhardt) writes:
|>
|> The Convex C1 probably also qualifies as a vector mini-super.
|> It started shipping sometime in 1985, IIRC. Though it (and
|> the follow-on, the C2) somehow managed to survive until the
|> early 1990's.
|>
|> By the time I arrived at Convex in late 1992, C-series sales
|> were definitely drying up (if not completely evaporated), and we
|> began doing our own "RISC SMP" :)
Yes. The success of FPS etc. led some people to think that the
future of HPC was vector mini-supers - because it assuredly wasn't
traditional mainframes or vector supercomputers. However, by the
time that well-engineered products were produced, it was clear
that they were going to have much of their market cut from under
them by the killer micros.
At the time, there just wasn't enough of a market for mid-range
HPC systems to make it economic to develop product lines for that
alone. The killer fact was that it was cheaper for each worker
to have his own workstation than to share out a vector mini-super,
so the market was limited to people with job sizes too large for
a workstation but not needing a high-end supercomputer.
It was sad, because machines like the Convex were very good of
their kind, and met a real requirement. But the writing was on
the wall on the day of its birth ....
Regards,
Nick Maclaren. |
|
| Back to top |
|
|
kus@free.net
Guest
|
| Posted:
Wed Aug 10, 2005 4:15 pm Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
| Quote: | So you're looking for "new on the market"?
|
Yes, you are absolutely right ;-) !
Yours
Mikhail
| Quote: | (The problem with anything
"new" is that it has done before in some more or less obscure design -
I sometimes suspect there is a prize to the c.a. poster digging up the
oldest reference :-) |
|
|
| Back to top |
|
|
Anne & Lynn Wheeler
Guest
|
| Posted:
Wed Aug 10, 2005 9:41 pm Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
"kus@free.net" <kus@free.net> writes:
| Quote: | I'm looking what was especially "new & important" in computer
architecture (I don't say about "pure researches") 10 years ago, in
1995.
1) Introducing of Pentium Pro : it was begin of wide using of x86 chips
for serious server applications (and x86 Beowulf clusters later).
2) Clear win of 64-bit RISC chips:
to Alpha 21164 it was added SGI/MIPS R8K/90 Mhz (75 Mhz started at
1994?)
and PA-8000
3) Start of win of RISC SMP-servers over vector mini-supercomputers
(Cray J90) or even over large vector systems (as Cray C90 - taking into
account price/performance ratio).
This win was based on SMPs like SGI Power Challenge, DEC 8200/8400
and Convex SPP servers (btw, ccNUMA in large configurations;
Sequent, if I'm correct, proposed ccNUMA on quad Pentium Pro boards).
|
... slightly earlier ...
here is reposting of Summary of Spang RObinson Report on Sumpercomputing
and Parallel Processing (from jan. 88)
http://www.garlic.com/~lynn/2001b.html#56 Why SMP at all anymore?
another post on some market segment overviews from 92
http://www.garlic.com/~lynn/2001n.html#83 CM-5 Thinking Machines, Supercomputers
note that SCI was picking up in the early 90s ... with 64-port memory
interface ... convex did exemplar with 64 two-way HP-RISC processor
boards and both sequent and DG doing 64 four-way intel processor
boards.
http://www.scizzl.com/
--
Anne & Lynn Wheeler | http://www.garlic.com/~lynn/ |
|
| Back to top |
|
|
Anne & Lynn Wheeler
Guest
|
|
| Back to top |
|
|
Andrew Reilly
Guest
|
| Posted:
Thu Aug 11, 2005 8:15 am Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
On Wed, 10 Aug 2005 16:11:36 +0000, Nick Maclaren wrote:
| Quote: | Yes. The success of FPS etc. led some people to think that the
future of HPC was vector mini-supers - because it assuredly wasn't
traditional mainframes or vector supercomputers. However, by the
time that well-engineered products were produced, it was clear
that they were going to have much of their market cut from under
them by the killer micros.
|
Speaking of vector mini-supers, what did they use for memory? If
off-the-shelf DRAM, how did they deal with non-unit stride accesses?
Cheers,
--
Andrew |
|
| Back to top |
|
|
Guest
|
| Posted:
Fri Aug 12, 2005 12:15 am Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
"Del Cecchi" <dcecchi.nospam@att.net> writes:
| Quote: | NUMA and clusters. That is about the time frame they became commercial.
Sysplex. etc.
|
Clusters had been around for over 10 years then. Both Vaxen with VMS
and Tops-20 CFS.
--
Paul Repacholi 1 Crescent Rd.,
+61 (08) 9257-1001 Kalamunda.
West Australia 6076
comp.os.vms,- The Older, Grumpier Slashdot
Raw, Cooked or Well-done, it's all half baked.
EPIC, The Architecture of the future, always has been, always will be. |
|
| Back to top |
|
|
Del Cecchi
Guest
|
| Posted:
Fri Aug 12, 2005 4:15 pm Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
prep@prep.synonet.com wrote:
| Quote: | "Del Cecchi" <dcecchi.nospam@att.net> writes:
NUMA and clusters. That is about the time frame they became commercial.
Sysplex. etc.
Clusters had been around for over 10 years then. Both Vaxen with VMS
and Tops-20 CFS.
OK, just NUMA then. And give or take a few years. |
--
Del Cecchi
"This post is my own and doesn’t necessarily represent IBM’s positions,
strategies or opinions.” |
|
| Back to top |
|
|
Nick Maclaren
Guest
|
| Posted:
Fri Aug 12, 2005 4:15 pm Post subject:
Re: What was new&important in computer architecture 10 years |
|
|
In article <3m3krfF158slqU1@individual.net>,
Del Cecchi <cecchinospam@us.ibm.com> writes:
|> >
|> >>NUMA and clusters. That is about the time frame they became commercial.
|> >>Sysplex. etc.
|> >
|> > Clusters had been around for over 10 years then. Both Vaxen with VMS
|> > and Tops-20 CFS.
|> >
|> OK, just NUMA then. And give or take a few years.
Well, only if you disregard companies like Kendall Square Research
as non-commercial.
Regards,
Nick Maclaren. |
|
| Back to top |
|
|
|
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in