[rescue] SS10/20 death

Dave McGuire mcguire at neurotica.com
Wed Feb 6 21:21:30 CST 2002


On February 6, Francisco Javier Mesa-Martinez wrote:
> > > Didn't one cray (Cray-2?) have two seperate memory systems? (And I'm
> > > not talking about an SSD.)
> > 
> >   Not that I'm aware of, at least not one of the PVP machines, which are
> > the ones I'm familiar with.  If one did, I don't think it could have
> > been the Cray-2.
> 
> That is correct only for the CRAY-1 (and maybe some of the later machines,
> XMP/YMP?) had a single memory system. No memory hierarchies for these

  What do you mean here?  "YMP" sums up pretty much all of the machines
between the XMP and the current SV1/SV2 machines.  The original name
for the J90, for example, was to be "YMP-J90"...it's a
YMP-architecture, YMP-binary-compatible machine.  So after the XMP
came the YMP, and then various other YMP models up to (but not
including) the SV1 and SV2.  Architecturally speaking, the T90 differs
from the other YMP-architecture machines only by its support of IEEE
floating point

> puppies, i.e. no cache and no virtual memory. Well not 100% correct if we
> think of the vector registers as a level of storage hierarchy, then you
> have 2 memory levels in an original  Cray, the registers and the main
> memory.

  By modern definitions, the vector registers wouldn't be considered a
1st-level memory any more than the scalar registers in a pentium would
be.

> However the Cray-2 had 2 levels of memory (3 if we include the vector
> registers), each vector processor had a local memory (16KWords) which was
> not accessible to the user unless you used some specific library calls and 
> it can be thought of as some sort of cache, but I think you had to
> specify the calls in your code in order to use such memory. All 4 processors 
> shared a 512MWord memory (around 4GB which was very large back then, hell
> even today it is pretty large... so imagine 15 years ago!). The cray-2 did
> also lack VM system. Memory access was pretty shitty for the Cray-2
> anyhow.. but it was relatively uniform.

  The processor-private memory in the -2 was not easily addressable and
not commonly used.  But then there weren't many Cray-2s to begin with.

> I think the XMP and later introduced the SSD, which were solid state disk
> storage units, but I do not think they were part of the directly
> addressable memory (perhaps scratch or pseudo-swap state). But I have
> never dealt with these machines...

  SSDs were introduced with the XMP.  They connect to the processor
through the same type of I/O channel as an IOS, and as such are VERY
fast.  An SSD appears to the OS as a disk, and was usually (on Unicos
systems) mounted as /tmp.  On Unicos systems, by convention, pretty
much all actual "work" is done in /tmp...home directories are used to
store stuff, but not run stuff...and so /tmp was made to be as fast as
you could make it.

         -Dave

-- 
Dave McGuire
St. Petersburg, FL         "Less talk.  More synthohol." --Lt. Worf



More information about the rescue mailing list