"As part two (see previous attempt) of my ongoing series in ‘computationanecromancy,â I’ spent the last year and a half or so constructing my own 1/10-scale, binary-compatible, cycle-accurate Cray-1. This project falls purely into the “becausI can!â category â€I was poking around the internet one day looking for a Cray emulator and came up dry, so I decided to do something about it. Luckily, the Cray-1 hardware reference manual turned out to be useful enough that implementing most of this was pretty straightforward. The Cray-1 is one of those iconic machines that just makes you say “Nothat‘a super computer!†Sure, your iPhone is 10X faster, and it†completely useless to own one, but admit it . . you really want one, don†you?"
But apparently, finding actual Cray-1 software is tricky ...
> "As part two (see previous attempt) of my ongoing series in > ‘computationanecromancy,â I’ spent the last year and a half or > so constructing my own 1/10-scale, binary-compatible, cycle-accurate > Cray-1. This project falls purely into the “becausI can!â category > â€I was poking around the internet one day looking for a Cray emulator > and came up dry, so I decided to do something about it. Luckily, the > Cray-1 hardware reference manual turned out to be useful enough that > implementing most of this was pretty straightforward. The Cray-1 is one of > those iconic machines that just makes you say “Nothat‘a super > computer!†Sure, your iPhone is 10X faster, and it†completely > useless to own one, but admit it . . you really want one, don†you?"
> But apparently, finding actual Cray-1 software is tricky ...
> // Christian
Sure, modern desktops are theoretically faster, but they're running with
the boat anchor of bloat from Windows or some other GUI. I bet the Cray
would still be faster for the sort of jobs it was designed for. Maybe
if you just ran BSD with no X-windows the modern computer would be
competitive.
Software? You've got a Fortran compiler and an assembler, right? What
more do you want?
In article <86ehr141b5....@zipcon.net>,
Patrick Scheible <k...@zipcon.net> wrote:
> Sure, modern desktops are theoretically faster, but they're running with
> the boat anchor of bloat from Windows or some other GUI. I bet the Cray
> would still be faster for the sort of jobs it was designed for. Maybe
> if you just ran BSD with no X-windows the modern computer would be
> competitive.
> Software? You've got a Fortran compiler and an assembler, right? What
> more do you want?
Back in the 1980's that may have been the case.
Today no way the Gooey Interfaces don't much interfere with numerical computation and you can play with your output in graphical format in real time, instead of plotting it by Calcomp plotter or some drudge manually plotting points. I've looked at Activity Monitor and if I'm not doing heavy graphics or sound or movies the confuser cpus are over 90% idle and that's using a laptop early 2010 model. Converting OTA TV to Quicktime turns on the fan.
Now if you're doing word processing in Weird, maybe so, but you couldn't do that on a Cray anyway.
> "As part two (see previous attempt) of my ongoing series in
> ‘computationanecromancy,â I’ spent the last year and a half or
> so constructing my own 1/10-scale, binary-compatible, cycle-accurate
> Cray-1. This project falls purely into the “becausI can!â category
> â€I was poking around the internet one day looking for a Cray emulator
> and came up dry, so I decided to do something about it. Luckily, the
> Cray-1 hardware reference manual turned out to be useful enough that
> implementing most of this was pretty straightforward. The Cray-1 is one
> of
> those iconic machines that just makes you say “Nothat‘a super
> computer!†Sure, your iPhone is 10X faster, and it†completely
> useless to own one, but admit it . . you really want one, don†you?"
> But apparently, finding actual Cray-1 software is tricky ...
>> "As part two (see previous attempt) of my ongoing series in
>> ‘computationanecromancy,â I’ spent the last year and a half or
>> so constructing my own 1/10-scale, binary-compatible, cycle-accurate
>> Cray-1. This project falls purely into the “becausI can!â category
>> â€I was poking around the internet one day looking for a Cray emulator
>> and came up dry, so I decided to do something about it. Luckily, the
>> Cray-1 hardware reference manual turned out to be useful enough that
>> implementing most of this was pretty straightforward. The Cray-1 is one
>> of
>> those iconic machines that just makes you say “Nothat‘a super
>> computer!� Sure, your iPhone is 10X faster, and it†completely
>> useless to own one, but admit it . . you really want one, don†you?"
>> But apparently, finding actual Cray-1 software is tricky ...
>> "As part two (see previous attempt) of my ongoing series in
>> ‘computationanecromancy,â I’ spent the last year and a half or
>> so constructing my own 1/10-scale, binary-compatible, cycle-accurate
>> Cray-1. This project falls purely into the “becausI can!â category
>> â€I was poking around the internet one day looking for a Cray emulator
>> and came up dry, so I decided to do something about it. Luckily, the
>> Cray-1 hardware reference manual turned out to be useful enough that
>> implementing most of this was pretty straightforward. The Cray-1 is
>> one of
>> those iconic machines that just makes you say “Nothat‘a super
>> computer!†Sure, your iPhone is 10X faster, and it†completely
>> useless to own one, but admit it . . you really want one, don†you?"
>> But apparently, finding actual Cray-1 software is tricky ...
> "As part two (see previous attempt) of my ongoing series in
> ‘computationanecromancy,â I’ spent the last year and a half or
> so constructing my own 1/10-scale, binary-compatible, cycle-accurate
> Cray-1. This project falls purely into the “becausI can!â category
> â€I was poking around the internet one day looking for a Cray emulator
> and came up dry, so I decided to do something about it. Luckily, the
> Cray-1 hardware reference manual turned out to be useful enough that
> implementing most of this was pretty straightforward. The Cray-1 is one of
> those iconic machines that just makes you say “Nothat‘a super
> computer!†Sure, your iPhone is 10X faster, and it†completely
> useless to own one, but admit it . . you really want one, don†you?"
> But apparently, finding actual Cray-1 software is tricky ...
> // Christian
Great memories.
I have a friend who was on the design team for the Cray 1
He gave me a tour of the plant.
So glad I got to see a part of history.
>> "As part two (see previous attempt) of my ongoing series in
>> ‘computationanecromancy,â I’ spent the last year and a half or
>> so constructing my own 1/10-scale, binary-compatible, cycle-accurate
>> Cray-1. This project falls purely into the “becausI can!â category
>> â€I was poking around the internet one day looking for a Cray emulator
>> and came up dry, so I decided to do something about it. Luckily, the
>> Cray-1 hardware reference manual turned out to be useful enough that
>> implementing most of this was pretty straightforward. The Cray-1 is
>> one of
>> those iconic machines that just makes you say “Nothat‘a super
>> computer!†Sure, your iPhone is 10X faster, and it†completely
>> useless to own one, but admit it . . you really want one, don†you?"
>> But apparently, finding actual Cray-1 software is tricky ...
>> // Christian
> Great memories.
> I have a friend who was on the design team for the Cray 1
> He gave me a tour of the plant.
> So glad I got to see a part of history.
Did any big organisation, such as the Meteorological Office, keep archives of Cray software?
>>> "As part two (see previous attempt) of my ongoing series in
>>> b computationanecromancy,? Ib spent the last year and a half or
>>> so constructing my own 1/10-scale, binary-compatible, cycle-accurate
>>> Cray-1. This project falls purely into the b becausI can!? category
>>> ??I was poking around the internet one day looking for a Cray emulator
>>> and came up dry, so I decided to do something about it. Luckily, the
>>> Cray-1 hardware reference manual turned out to be useful enough that
>>> implementing most of this was pretty straightforward. The Cray-1 is
>>> one of
>>> those iconic machines that just makes you say b Nothatb a super
>>> computer!b Sure, your iPhone is 10X faster, and it?? completely
>>> useless to own one, but admit it . . you really want one, don?? you?"
>>> But apparently, finding actual Cray-1 software is tricky ...
>>> // Christian
>> Great memories.
>> I have a friend who was on the design team for the Cray 1
>> He gave me a tour of the plant.
>> So glad I got to see a part of history.
> Did any big organisation, such as the Meteorological Office, keep
> archives of Cray software?
Or ask a university who had/has a Cray if they have any archives.
On May 3, 1:16 pm, Al Kossow <a...@bitsavers.org> wrote:
> On 5/3/12 10:33 AM, Patrick Scheible wrote:
> > Software? You've got a Fortran compiler and an assembler, right?
> Wrong
> He doesn't even have a complete copy of the OS
According to the web site, though, he _does_ have an assembler - not
the original one from Cray, but one he wrote himself.
There's always the option of porting Linux - that is, just
a) retarget gcc for the target architecture, and
b) rewrite the assembler-language portions of the kernel
and one has software to run on it... and, gradually, one could, say,
take one of the GPL Fortran 95 compilers out there, and make it use
the underlying Cray-1 with more efficiency.
So turning the thing into a real platform, while a lot of work, is
probably less work than has been done already.
But it may be someone else's turn to do that work.
"Quadibloc" <jsav...@ecn.ab.ca> wrote:
>There's always the option of porting Linux - that is, just
>a) retarget gcc for the target architecture, [...]
A sort-of similar solution was used at Oak Ridge National Labs (aka "X-10") when it installed a 360/75 (running an early version of OS/360).
The /75 replaced a CDC 1604, on which the FORTRAN compiler was considered to be critically deficient by the researchers, in response to which Jerry Sullivan wrote a very good, efficient, and feature-rich. Over time the compiler was rewritten in its own language (much like FORTRAN H).
Porting Jerry's compiler to the /75 was relatively straight-forward: the 1604 compiler was modified to emit S/360 object decks, and was then fed its own source. The output was the compiler for OS/360 which used the native FORTRAN library. (There may have been a couple of specialized assembly language programs that had to be recoded from scratch.)
The ported compiler remained the language translator of choice for a long time, well after FORT-H became available. I don't know if it was still in use when the optimizing FORTRAN compiler program product showed up.
If anyone here has ever worked with a FORTRAN program that came from ORNL and it refers to FORTRAN unit numbers starting at 50, it's likely that the program was originally intended for use with Jerry's FORTRAN compiler.
> Is the model truly faithful to the complete Cray architecture? That
> is, does it have vector registers and do vector pipeline
> floating-point arithmetic? If so, seems like porting some Fortran
> compiler to use that hardware is rather more involved. Especially
> optimizing code for vector arithmetic and memory vector read/write
> like the Cray did.
Haven't looked at it myself, but I've heard it consistently described as
"cycle-accurate", which had better mean it has all those features.
I can see the appeal of getting gcc and linux working on it, but doing
it "right" would mean the original OS and software tools.
>>> "As part two (see previous attempt) of my ongoing series in
>>> ‘computationanecromancy,â I’ spent the last year and a half or
>>> so constructing my own 1/10-scale, binary-compatible, cycle-accurate
>>> Cray-1. This project falls purely into the “becausI can!â category
>>> â€I was poking around the internet one day looking for a Cray emulator
>>> and came up dry, so I decided to do something about it. Luckily, the
>>> Cray-1 hardware reference manual turned out to be useful enough that
>>> implementing most of this was pretty straightforward. The Cray-1 is
>>> one of
>>> those iconic machines that just makes you say “Nothat‘a super
>>> computer!†Sure, your iPhone is 10X faster, and it†completely
>>> useless to own one, but admit it . . you really want one, don†you?"
>>> But apparently, finding actual Cray-1 software is tricky ...
>> <Quote>
>> Intel’s working on an 80-core CPU that will break the Teraflop
>> barrier for PCs. This may ship by 2011 or 2012
>> </Quote>
>> Anyone seen one yet?
> I'm not sure what I'd do with it. I can't even keep four cores busy.
Each of your four cores are several times faster than a Cray-1. Even so simulating even a simple electronic circuit (I used an valve audio amp and simulated one second of operation) with LTSpice which is multi-core aware will keep all four happy for a few seconds. Just the sort of work the Cray was designed for...
The Cray Fortran had extensions so that array manipulation could be run in the pipeline. However keeping on busy was tricky. Where I worked one of our Ocean Modellers who produced mathematical models of tides and currents used to get an hour of Cray time per week, and most weeks he didn't use it all.
Given the preponderance of multi-cpu PCs I wonder of we will ever see a parallel Fortran (or C or Algol) for Linux of Windows.
>Given the preponderance of multi-cpu PCs I wonder of we will ever see a >parallel Fortran (or C or Algol) for Linux of Windows.
OpenMP is a standard that allows programmers to annotate their otherwise-sequential code, so that compilers to parallelize that code, in Fortran, C and C++. Compilers:
* GCC (the Gnu Compiler Collection) supports OpenMP 3.1, including on Linux and Windows. * IBM's XL C/C++ and Fortran compilers also support OpenMP, including on Linux.
* Intel have compilers that support OpenMP for Linux and Windows.
While this is programmer-guided parallelization as opposed to fully-automatic, it still allows you to compile the same program and have the compiler figure out how that best maps to the CPU cores you have available.
