[rescue] IBM :: Different
George Adkins
george at webbastard.org
Wed Feb 13 17:38:59 CST 2002
On Wednesday 13 February 2002 05:17 pm, you wrote:
> On Wed, 13 Feb 2002, George Adkins wrote:
> > > These aren't mainframes. These are RS/6000's. This is *really* a
> > > 128MB SRAM cache. Big ballsy systems. Microprocessor-based.
> >
> > Yes, Yes, I apologize for deigning to write the word 'mainframe' would
> > you be offended or otherwise manage to find some fault if I said 'system'
> > instead? (Of course you would, please note in advance the lack of
> > capitalization when using the word system here, I meant in no way to
> > imply System, as in System 360 or System 390, it was intended to mean
> > system as in the generic term referring to 'computer system'.)
>
> You still didn't get what I'm trying to say. This is not a multi-tiered
> main memory architecture. That is a 128MB transparent Static RAM *cache*.
> A *real cache*. With a hit-detect algorithm and everything. This is a
> UNIX box running on microprocessors.
>
Yes, I do. I did before. nothing has changed. I UNDERSTAND.
Look, I wrote:
> Yes, but remember, for IBM Mainframes compared to everyone else, this is
> like having a 2 or 4 processor server with 512K or 1 meg cache each, with
> 8 megs of unified L2 cache for the processors and 128 megs of buffer RAM
> between the Proc and a 2GB main RAM. (no doubt all kickass fast...)
Which part of 'buffer RAM between the Proc and a 2 gig main RAM' implied that
I was saying this was a multi-tiered memory archetecture?
and no, this is not an invitation to debate the subtle differences between a
buffer and a cache, I didn't say it was *a* buffer, only that it buffered
access to the 'main RAM'.
Please, stop attempting to nitpick anything I post to death...
More information about the rescue
mailing list