

8/23/85
Rob Horning



I ran the full spice models.  The slow model looked very close to the
abbreviated model.  The fast model looked close in the critical timing
areas but had some ringing in a non critical area that the abbreviated
model did not have.  I will continue to develop with the abreviated
models but I will always check my final results with the full model.

There has been some concern with the feedback that the Cheetah has been
getting from the MTC on Hitachi.  The concern was that they are using
a different process than everyone else.  We were assuming that Sang Park
was concerned with the reliability of the process.  What his real concern
is that we will qualify there parts and then end up not using Cypress.
There is pressure for them to try to buy more RAM's from U.S. vendors.
We will continue to work closely with Hitachi and MTC will help us
evaluate the parts, but we will also make an effort to work with Cypress
ands IDT even if we determine that we do not need them.  Sang feels that
there will be many vendors with 16K x 4 24 ns SRAM's by the time we are
in production.

The CCU/TCU I to L will be in about four weeks.  It will take this long for
the Cheetah guys to give feed-back on the results of there timing simulations.
It looks like there is plenty of margin on Firefox but the budgets have not
been finalized.

There was an IO_ACD review this week.  There were no big problems.  I send
Dave James some comments on areas in the memory module document that were
not clear.  This was in the area of  error logging and mapping out hard
errors.

I spent yesterday with the NIO people and have a feel for what NIO involves.
I let them know that we wanted the clock to be the same as MID_BUS.  I am
not sure what synchronizing will do to complexity, but I want to make sure
that there are real advantages to adding this complexity.  They would like
to see a general purpose set of chips developed to interface to NIO.  I am
fairly sure that our section does not have the resources to do this.  I
don't think that we want to wait for someone else to develop the chips
either.

