Text preview for : 19770919_Simonyi_Progress_Report.pdf part of xerox 19770919 Simonyi Progress Report xerox sdd memos_1977 19770919_Simonyi_Progress_Report.pdf

Back to : 19770919_Simonyi_Progress | Home

I ntcl'-()ffice I'v1enlorandu m

To DJyid Liddle Date September 19. 1977

From Charles Simonyi location Palo Alto

SUbject Progress Report Org:mization SOD/SD

A' ~\

Filed on: (Simonyi)sddsep.memo

Prototype Software (Charles Si monyi)

Programming of the Diamond Recompilation, Service, and Test Modules has been-
progressing well. The total size of these programs is now 29500 lines. All of the code has
been converted to Mesa 3. The Recompilation Module is nearing completion. There were
some difficulties in binding the Service l'w1odules. A minor simplification has been defined
to solve these problems. The Test fvlodule is working well. but it still needs to be integrated
with the Service Modules.

The Tvlesa group has been briefed about the details of these activities. A demo of the Test
Module was given as part of this briefing. It is hoped that improved communication about
our problems and about our approaches to solving them will lead to increased coordination
and cooperalion.

A valuable opportunity was exploited by a two man-week study of the Mesa microcode to
find out if Butler Lampson's original implementation ideas, which had the potential of
doubling the current execution speed, can- be made to fit the available space. The original
sketches were microcoded and a number of space-saving ideas were contributed. The results
are not only that the fast microcode can be fitted. but also a fair amount (about 50-80 words
out of lK) of microcode space will be left for future developments. These results- were
handed over to Roy Levin of CSL for further optimization anel implementation. -\Vith the
new microcode. Diamond will have the potentia) of outperforming Bravo even on the Alto.

Copies to: Bergsteinsson, Heini-ich, Irby, Lynch, Metcalfe. Shultz, Szelong, Stottlemyre,-
Townsend, vVeaver