0000 **/**/** **:**:** * CM4NEWS CLASSIC/CLASSIQUE *
---- -------- -------- -------------------------------------------------
************************************************************************
****************** PRELIMINARY CMOS4 DEVELOPMENT NEWS ******************
************************************************************************
- Circulation: 50 (September 10, 1985)
- Editor-in-Chief: ?????????????????
Editor's note:
- Please do not copy "CM4NEWS" to your A-disk because it might contain
BNR/NT private/restricted/proprietary information. Besides, "CM4NEWS"
on your A-disk will screw up "CM4NEWS WRITE".
------------------------------------------------------------------------
CM4NEWSMANIA BUILDING UP
========================
= The situation is similar to the one in early 1980. The retired
CM4NEWS editor-in-chief is carefully assessing the possibility of
forming a majority editorship.
NEWS ITEM FOR CM4NEWS EDITOR
============================
WED 11SEP85 12:36:15
After reading CM4NEWS this morning Jeff Bennett stormed into
the hyper nest rabbit hutch that ************* calls home
and demanded the instant resignation of the CM4NEWS editor.
No one was present - no such action was taken.
BENNETT ROYAL COMMISSION RECOMMENDS ABOLITION OF CM4NEWS
========================================================
= Our $8-a-day man Jeff Bennett handed down a 1-sentence recommendation
September 7 on the future of CM4NEWS. In his long-awaited report,
Jeff Bennett, who was once touted as the next CM4NEWS editor, said
"Since the small-team effort of the CMOS4 development is over, the
communication should return to the normal channel." (meaning COCOS)
= Unlike Brian Mulroney, the CM4NEWS editor-in-chief will take an
immediate action. CM4NEWS will be phased out September 11, 1985.
JOHN ROTH ANNOUNCES HIRING FREEZE (MITEL-INSPIRED LAYOFF COMING UP?)
====================================================================
| INQUIRE CCCMINS Updated September 5th, 1985 |
John Roth's Closing Comments
----------------------------
In 1984, R&D was $430M. In 1985, we have an R&D of $580M. Given
our current expenditure rate we will need an R&D of $730M for
1986. However, we will probably be receiving R&D of around
$640M.
Therefore, we do have a problem. The problem is in terms of
headcount. The consequence is that we will stop hiring out of
Ottawa. Regional Labs have been notified that they are not to
increase their headcount. Effective immediately, there is a hir-
ing freeze in Ottawa. All VP's as well as HR people have been
informed of the situation.
CM4NEWS CIRCULATION REACHES 50
==============================
= Despite the threat from the reconfigured competitor 5SNEWS (issue 3),
CM4NEWS is spreading like an epidemic. (Like AIDS, one might say)
Vacationing CM4NEWS editor-in-chief, who is analyzing the similarities
between his CM4NEWS and Pierre E. Trudeau's Cite Libre (the absolute
value of the individual, the supremacy of rationality, the constant
struggle between totalitarian and democratic tendencies at BNR, and
the obligation of every individual to involve himself/herself in the
engineering process) is so pleased that he wants to form a separate
company called "BNR CM4NEWS Limited/Limitee", and self-appoint
President and Chief Executive Officer of the small business. The
editor-in-chief is dreaming of the day when Roly Reader, Ian McWalter,
Terry Curtis, Cesar Cesaratto, John Roth and Edmund Fitzgerald (all
featured in CM4NEWS) will contract the disease of reading CM4NEWS.
= The CM4NEWS editor-in-chief was the only one with a high-tech look
(meaning without a tie) at the first Canadian Club Luncheon Meeting
September 4 at Chateau Laurier featuring Archbishop of Canterbury.
The Archbishop said "People in Great Britain watch television to see
if the person in the House of Lords is still alive or not." Well, at
BNR one can always type "CM4NEWS" to see if the person is still active
in the CMOS4 development. Although the editor-in-chief didn't speak
up this time, he might have appeared on the local TV news again.
SOME MISUNDERSTANDING OF A SIMPLE JOKE.....
===========================================
THUR 05SEP85 17:16:41
YOU KNOW WHAT, **? I THINK YOU'VE GOT A CHIP ON YOUR SHOULDER.
THANK GOD THAT MOST PEOPLE WHO WORK HERE HAVE BRAINS AND CAN SEE B.S.
WHEN THEY READ IT! ON YOUR RISE TO THE TOP YOU MAY WANT TO READ THE
FOLLOWING PHRASE (PERHAPS USE IT AS YOUR LOGO). IT'S A CARTOON OF A
PERSONNEL OFFICER TALKING TO A "PROSPECTIVE" EMPLOYEE: "THERE'S NO
ROOM AT THE TOP, AND THERE'S NO ROOM IN THE MIDDLE, BUT THERE IS A
LITTLE ROOM AT THE BOTTOM."
ROB GIBBINS FINDS LOOPHOLES IN DE007 (RELEASE 1W04)
===================================================
THUR 05SEP85 11:03:21
In case anybody still reads CM4NEWS, below is a COCOS I sent
to all relevant satellite centre librarians:
Freshly back from holidays, I just finished perusing the
1W04 release of DE007, and uncovered two minor errors.
D. Smith informs me that the next release may not be
for some time; consequently, here are the errors found:
1) contact window There are two (1.6 x 1.6) headings
rules page near the top of the page. The
first (left) one should be preceded by
the symbol for "less than or equal to",
while the second should be preceded by
the "greater than" symbol.
2) corner-to-corner The last line on this page:
rules page
metal 2: x = 1.5 um, y = 1.6 um
should read:
metal 2: x = 1.7 um, y = 1.6 um
Please alter your document accordingly.
Thank you.
THE CM4NEWS EDITOR-IN-CHIEF ON VACATION
=======================================
= Now that the half-grid issue is finally resolved for good and Y13 is
gone for 16 weeks or so, the CM4NEWS editor-in-chief is taking a one-
week vacation starting September 3. Please use the "CM4NEWS WRITE"
facility to keep CM4NEWS exciting. Any revenge is especially welcome!
The editor-in-chief is anxiously hoping that his successor be chosen
when he returns.
= Mysteriously, Go S. Sunatori is going on holidays at the same time.
Unlike that of the editor-in-chief, Go's vacation is indefinite.
The sign says "Go S. Sunatori is on vacation till God knows when".
If a newpaper article "Vacations Stressful for True Workaholic" were
true, he would come back in 2 days. If not, he would be away for 2
months and he might find himself unemployed (again) or demoted to a
janitor when he returns. In any case, Go says "Don't bother me during
my vacation. I am no longer on call!"
= August 30 was the last day for the two key test chip workers, Greg
Ward and John Kester (the ones who did the real work). Jean Wills
is also leaving next week. Lucky Greg and John were treated in an
unprecedented 3-hour lunch August 29 at Peter's Pantry. They would
shortly find out that very hard work at the university is much better
than being abused to hell at BNR.
FULL-COLOUR DE007 (ISSUE 1W04) RELEASED
=======================================
= Thanks to Dave Smith, issue 1W04 of DE007 was released August 30.
This issue features amazing full-colour pictures and a handy list
called "CMOS4 - MAJOR DESIGN RULE SUMMARY", which should be especially
useful for high-level managers.
5SNEWS STRIKES BACK (WITH ONLY 473 LINES)
=========================================
1985 Aug 29 at 14:13 EDT
| bulletin: NEWSLETTER
|
| from: Terry Curtis (BNR) Dept 5A10 CRK
|
| subject: LSI COMMUNICATOR FOR 5S00
|
| sent by: Carol Edwards (BNR) Dept 5S60-S CRK
|
| Please read 5S00 News, Issue 3. You will find it
| interesting and most informative.
|
| You can do this by typing 5SNEWS. Enjoy!
HALF-GRID CASE FINALLY CLOSED
=============================
= The final "Half-Grid Meeting" was held August 28 to resolve the most
controversial issue in the CMOS4 development once and for all. Our
very big boss Terry Curtis was cordially invited as a special guest
to this climax of the hottest controversy, and even uninvited guest
Ian McWalter showed up. Terry, who had previously been briefed
that this issue was a real chaos, was pleasantly surprised to see a
consensus among most attendants for the full-grid system with half-
grid offset. It immediately wiped out one of Terry's three major
headaches, and Terry was stressing "We have already 6 CMOS4 codes
waiting." Roly Reader, who had to miss the very exciting meeting,
was relieved of strain, too.
= Last week's PRELIMINARY stamp on the ultimate full-grid proposal has
been upgraded to BNR RESTRICTED. Steve Sunter, who was all dressed up
for this big occasion, addressed a number of key convincing arguments
for the half-grid offset. In spite of a desperate attempt by Jeff
Bennett to keep the complete full-grid system alive, OMNI/HORIZON Cell
Library was permanently recalled. In reply to Len McCready's question
"If you had another chance, would you do it again?", confident Go S.
Sunatori, whose main objective has been to build the best cell library
in the world, simply said "Yes!"
= After the meeting, jubilant CM4NEWS editor-in-chief was asking Len
McCready "Is there going to be a Half-Grid Victory Lunch?", while
exhausted Go S. Sunatori sighed and said "Good grief! At last, I can
go on vacation which I had postponed for 2 weeks because of the half-
grid issue. But I will definitely miss it since it was sort of my
raison d'etre!"
= Now that the half-grid issue is gone, the only major remaining CAD
issue seems to be the MWIN connection between METAL1 and METAL2 in
the integrated OMNI package supported by Rick Epplett. Robert Hum
estimates the work to be 6 to 8 person-weeks total. However, Grant
Martin, who had initially estimated his work on TIARA to be 2 weeks
to change the handling of cell origins, revealed that it took him only
2 hours to modify 50 lines of code in his program.
SOME MISCELLANEOUS NEWS
=======================
= Rob Koziy is back from an exciting 4-day trip to HONEYWELL. He said
he was especially impressed by a large individual office with a door
and a window for every employee at HONEYWELL. (Better than that of a
typical C-level manager at BNR!)
= Bruce McCracken experienced the ultimate entertainment of reading
CM4NEWS August 27. Bruce was heard bitching "I'm not doing any more
work on CMOS4 because I wasn't invited to the Y13 Celebration Lunch!"
= Dave Gibson devised a term "Sea of Pins" August 27 while looking at
Go S. Sunatori's plot of the Scannable 4-bit Binary Counter, to be
presented at the half-grid meeting August 28.
= Go S. Sunatori, meanwhile, worked only 5 hours August 27. Go was
witnessed by Dave Gibson when he was sneaking out at 15h13. Go's
excuse this time was "I have a night shift tonight to prepare for the
extremely important half-grid meeting tomorrow with special guest
Terry Curtis."
= John Hogeboom is on vacation this week. Although he had logged this
vacation a long time ago, he did not remind anybody of his plan, and
everybody was wondering where he was.
NEXT: SWEEPING CELL NAME CHANGES, NEW INVC AND BUFC
===================================================
= The exciting CMOS4 cell development still goes on..... Our Auditor
General Jeff Bennett pointed out a serious inconsistency between the
present cell names and a clause in Steve Sunter's CMOS4 Cell Library
Philosophy Document which reads:
| All cell names which are the same as used in CMOS3 are functionally
| the same - this ensures rapid conversion of CMOS3 designs into CMOS4.
= Following is a proposal for the cell name changes:
- CM4RDTPS -----------------> CM4RDFFS
- CM4RDTP -----------------> CM4RDFF
- CM4RLAT -----------------> CM4SLAT
- CM4TGIP -----------------> CM4TGPI
= John Hogeboom recommends some changes in INVC and BUFC cells with much
less area at the expense of about 10% drive capability:
NEW OLD
--- ---
- CM4INVC (10 x 10 grids) ----- CM4INVC (14 x 10 grids)
- CM4BUFC ( 8 x 10 grids) ----- CM4BUFC (12 x 10 grids)
WHERE IS THE BEEF?
==================
= We ask Jeff Bennett "Where is the beef in the complete full-grid
system?" Here we find some reasons why OMNI/HORISON Cell Library is
not good enough compared with CADILLAC Cell Library:
It is too full-grid.
--------------------
- One has a good old problem of identifying coincident cell origins.
It is not full-grid enough.
---------------------------
- One has to use half-grid for the wide power/ground bus.
WHATPROC EXECDATA ON A-DISK SCREWS UP JEFF BENNETT
==================================================
= Jeff Bennett could not run SCAMPER on NAND4 August 21 because of his
own WHATPROC EXECDATA file on his A-disk. Go S. Sunatori ended up
trouble-shooting Jeff's problem for half an hour, and found out that
Jeff had copied the file on December 4, 1984. One has to remember
that any modern CAD program goes through the WHATPROC routine to find
out which technology disk you are linked, and he/she should not copy
WHATPROC EXECDATA file to his/her A-disk. Use your own common sense!
= Jeff also had CM4 PROCESS file on his A-disk. Please note that any
file on the CM4 Library Disk should not be copied to your A-disk. Go
also discovered such plain junk files as LOAD MAP, RESTORE LOADMEMO,
CRM178 NETLOG along with JEFFB NETLOG on JEFFB 191 disk. What's more,
Jeff's PROFILE EXEC did not contain the two critical CP/CMS settings
advertized in ICNEWS 94 (84/08/01) and ICNEWS 100 (84/08/17):
- CP SET TIMER OFF
- SET BLIP OFF
DEX FOR CMOS4
=============
= Thanks to Bill Gunn, DEX for CMOS4 is now up on LDISK. It can be
invoked by typing "CMOS4", and all the rock-solid standard cells on
Y13 will be captured on DEX soon, enabling us to resume the cell
development. For more information, please contact Go S. Sunatori.
BNR PATENT OFFICER EXAMINES OUR CMOS4 COMPOSIT CELL LIBRARY
===========================================================
= BNR patent officer Stuart Wilkinson examined our CMOS4 Composit
Cell Library August 20 with curiosity. Not knowing the hottest
controversy in the CMOS4 development, Stuart suggested that we should
apply for another patent on the half-grid strategy! Steve Sunter
honestly admitted that we stole the idea from HONEYWELL.
DAVE SMITH CELEBRATES 32ND BIRTHDAY AUGUST 20, LOSES DE007 (1W04)
=================================================================
= Claudine Simson hosted a birthday party for Dave Smith August 20 in
her office. There were actually 32 candles on the delicious blueberry
cake, and the smoke was just about to trigger the smoke detectors.
Dave wants to buy a house, get married, and have 3 children, all in
parallel.
= Is anybody wondering where 1W04 version of DE007 is? Happy Dave Smith
confessed after eating two pieces of cake that his only copy got lost
in the Documentation Department at Mallorn Centre.
JEFF BENNETT DECLINES MOST GENEROUS OFFER
=========================================
= Citing personal reasons, Jeff Bennett turned down the most generous
offer August 20 for the CM4NEWS editorship, which nobody else could
have possibly refused. (Maybe his wife objected.....)
= Accordingly, the CM4NEWS Editor Emeritus demoted himself to the much
less prestigious position of CM4NEWS editor-in-chief. He is so glad
that he never said never.
"CM4NEWS EDITOR EMERITUS" APPOINTED
===================================
= The original CM4NEWS editor appointed himself to the extremely
prestigious status of "CM4NEWS Editor Emeritus" August 19. He might
still make occasional appearances, just as retired Charles Lynch does
in The Ottawa Citizen.
= The CM4NEWS Editor Emeritus paid $26.00 and joined The Canadian Club
of Ottawa August 19, hoping that he would some day be invited to speak
on BNR/NT's CMOS4 development along with such celebrities as John Roth
(President, Bell-Northern Research), Marc Garneau (Commander, RCN),
Robert White (President, United Automobile Workers (Canada)),
Joe Schlesinger (CBC National TV News), and Maureen Forrester (The
Canada Council).
= Ontario NDP Leader Bob Rae preached August 19 exactly what the CM4NEWS
Editor Emeritus has felt for some time. Rae said "I say to you, I
cannot accept some people working 50 hours a week and other people not
working at all....." In our CMOS4 development, the situation was much
worse - some engineers were working 70 to 80 hours a week while some
managers were working only 10 hours a week, as professed below by one
of them. (Sorry, this is his last attempt to attack people.)
JEFF BENNETT: THE NEW CM4NEWS EDITOR
====================================
MON 19AUG85 10:46:23
Since it was getting boring only working 10 hrs/week I
considered putting in an Ottawa Journal type take-over
bid for CM4NEWS and the liquidating *************'s
assets. However for a capitalist (and part time scribe)
like myself CM4NEWS's free speech is too free to turn
a profit. Anyone join me for greenmail and a life of
grime?
.......Jeff
= Our famous Iacocca-class authority Jeff Bennett decided to get into
the press business as shown above, pending sanity clearance. Let us
all wish best of luck for Jeff's new and exciting career, which he is
determined to devote 27.5 hours/week.
CHALLENGING CAREER OPPORTUNITY IN CM4NEWS EDITORSHIP
====================================================
= Looking for the opportunity to realize your true potential? The
search for the replacement of the CM4NEWS editor has been extended.
The ideal candidate for the CM4NEWS editorship is:
- a strong supporter of the Canadian Charter of Rights and Freedoms.
- willing to respect and strictly conform to the BNR/NT corporate
policy in a capitalistic way.
- not afraid of management and/or peer pressure in challenging
injustice and wrongdoings.
- capable of converting facts into black or white humour.
- able to dedicate his/her time to CM4NEWS, 24 hours a day, 7 days a
week, and able to wake up at 3 a.m. to write stories.
- anxious to set standards and write history or even a bible on the
cutting edge of the state-of-the-art CMOS4 development.
- motivated enough to compete with 5SNEWS.
= This infinitely attractive, stimulating position requires no related
experience, and absolutely no extra compensation or comprehensive
benefits package is offered. Sanity clearance desired. Act now -
send your VMID or nominations through the CM4NEWS WRITE facility!
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+++++++++++++++++++++ Equal Opportunity - M/S/T/O. +++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
"OMNI/HORIZON CELL LIBRARY"
===========================
= Sparked by Go S. Sunatori's inaction on investigating the complete
full-grid approach, John Kester have built a set of cell library
composed of rather compact cells with FULL/FULL via sites. Although
solidly backed by our Lee Iacocca-class authority Jeff Bennett, the
OMNI/HORIZON-quality cells have encountered countless technical
difficulties.
= Meanwhile, Go S. Sunatori is gaining more confidence in his ultimate
full-grid plus half-grid offset proposal with Grant Martin's assurance
that 3.03 TIARA could handle the proposed scheme without any major
problems. Gary Sakauye now calls the scheme "It's the best so far!"
THE CM4NEWS EDITOR SPEAKS IN BROADCASTING POLICY PUBLIC HEARING
===============================================================
= The CM4NEWS editor, who boasts of a 100% Canadian content in his
CM4NEWS, went to Skyline Hotel August 14 to attend The Federal Task
Force on Broadcasting Policy Public Meeting. Although his original
intention was to study the impact of the federal broadcasting policy
on CM4NEWS, he could not help standing up to make a very patriotic
speech before the 7-member Task Force appointed by Minister of
Communications Marcel Masse. Still under the influence of the zombi
he drank at lunch, he testified "I watch an American program STAR TREK
Sunday morning on CBC because the hero of the show Captain Kirk is a
proud Canadian!" He was filmed by numerous TV cameras, and his voice
was broadcasted on the CBC AM-radio network.
Y13 PARTY IS OVER
=================
= In a very surprising move, Len McCready suddenly decided to hold the
Y13 Celebration Lunch August 14. Vacationing Rob Gibbins received a
phone call at home, so he did not miss the lunch. Happy faces were
everywhere with lots of booze, and especially happy was Len McCready
who found his expected expenses well overestimated.
= Erratic behaviours and negative productivity were observed everywhere
in the afternoon. No wonder the company doesn't offer free lunches
very often.
THE FULL-GRID SYSTEM: AN ULTIMATE SOLUTION TO THE HALF-GRID ISSUE
=================================================================
= Go S. Sunatori, whose controversial half-grid system was found by
Gary Sakauye to contain a fatal flaw (an inconsistency between cell
placements and macro placements), was sympathized by the two verses
"remember you almost had it all" and "remember you almost made it"
in Melissa Manchester's "Don't Cry Out Loud" August 13. Go is now
proposing a total full-grid system with a half-grid offset, originally
conceived by Steve Sunter:
----------------------------------------------------------------------
- All cell/macro placements are on FULL/FULL grid points
- All interconnect lines can be on FULL/FULL grid points
----------------------------------------------------------------------
= With this novel scheme, almost everything will be consistent at
FULL/FULL grid throughout the entire hierarchy in both automatic and
manual layout, and a designer would have to issue "SET GRID INP 66
DIS 66" in LAYED only once during his/her entire session except for
join cell placement and wide power/ground bus routing.
= The initial response from the CAD group is encouraging. Rick Epplett
doesn't seem to have any problems with his OMNI package (combined
WIDEN, MCCHECK and VERPLACE). As usual Gary Sakauye doesn't guarantee
anything at this stage, but his reaction was "It looks much better."
Jamie Twidale was a little upset because he was all set to go for
the original half-grid approach.
= Now, the action items for each key person to achieve this system are:
- Dave Gibson: To change his great CM4CLIO EXEC.
- Go S. Sunatori: To change the origin of all cells to (x,y)=(1/2,1/2)
in CM4 CAL and re-produce CM4 CELLIO.
- Len McCready: Not to worry a bit because the physical levels of
the cells do not change (still rock-solid).
A PATENT APPLICATION ON CMOS4 COMPOSIT CELL LIBRARY
===================================================
= Steve Sunter and Go S. Sunatori are making a patent application on
the most basic CMOS4 Standard Cell Library philosophy - a minimum set
of primitive primary cells to construct composit primary cells using
various join cells. Steve's analogy is that reduced-instruction-set
computers (RISCs) can have much better performance than conventional
computers. Go's best analogy is that only 26 Roman alphabets in
English et francais are far more efficient and flexible in word-
processing than say, thousands of Chinese ideographic characters.
A FAREWELL LUNCH FOR JIM SERACK AUGUST 16
=========================================
= Jim Serack is finally leaving us to do his Ph.D. work in Scotland.
A farewell lunch will be held Friday, August 16 at the March House.
Therefore, the Y13 Celebration Lunch will be re-scheduled again in
spite of Steve Sunter's comment "The Y13 Lunch is much more important
because we will still have to work together, while we won't have to
work with Jim any more."
A QUOTE FROM JOHN ROTH'S SPEECH
===============================
= Here is a quote from John Roth's speech in BNR INSIGHT, vol.5, no.6:
"So no matter how great the pressure is to meet a schedule, it cannot
be done at the expense of product quality."
TERRY CURTIS DEVELOPING KEEN INTEREST IN CMOS4
==============================================
= Our very big boss Terry Curtis was given a tour of the CMOS4 Cell
Library August 12 for the first time. According to Terry, the
official $4M contract with HONEYWELL is nowhere near the sight.
The CM4NEWS editor was still wondering whatever happened to Terry's
5SNEWS, the only competitor of any significance.
= Terry was entertained by the now-popular "Design Systems" which Steve
Sunter presented at a conference in Kingston:
----------------------------------------------------------------------
- DAD : Dinner-Assisted Design
- BDD : Bug-Driven Design
- CDD : Complaint-Driven Design
- CDD : Crisis-Driven Design
- FOD : Frustration-Oriented Design
- POD : Panic-Oriented Design
- UOD : Underground-Oriented Design
----------------------------------------------------------------------
Y13 COSTS $86.1K (US)/LAWYERS INVOLVED IN THE NON-DISCLOSURE AGREEMENT
======================================================================
= Len McCready received the final quote from HONEYWELL August 12:
$86.1K (US) which translates to just over $100K (CDN). This figure
includes PG to the final wafer production. The CM4NEWS editor agrees
with Len who thinks the deal reasonable considering the estimate of
as much as $60K (CDN) for an E-beam mask-set alone at NTE.
= Corporate lawyers are involved in interpreting the complicated non-
disclosure agreement with HONEYWELL, according to Len. The agreement
was originally signed between NTE and HONEYWELL, and we would like it
to cover between BNR and HONEYWELL. No wonder some corporate lawyers
like John N. Turner and John B. Aird can make half a million bucks a
year by changing just a few letters (NTE to BNR, in this case) with
their signatures!
BRUCE MCCRACKEN PREPARING GDS-II TAPE/BUREAUCRACY MAY DELAY Y13 SHIPMENT
========================================================================
= The CM4NEWS editor's favourite song these days is "My Way". Y13 is
now on the final stretch, and as soon as you hear the name Bruce
McCracken, you would know "the end is near". The DRC error-free
super CALMA file "Y13 CALMA" was tossed to Bruce August 9, and it
is now up to him to produce two CALMA GDS-II tapes to be shipped to
HONEYWELL August 12.
= Meanwhile, Len McCready is still struggling with corporate red tape
to analyze the detailed quote from HONEYWELL, as well as to clear up
the purchase order and the complicated non-disclosure agreement with
HONEYWELL.
= Incidentally, Len invited his boss's boss Terry Curtis to the Y13
Celebration Lunch just to let Terry pick up the expected $600 bill.
The CM4NEWS editor thinks that Cesar Cesaratto should be invited
because of his tremendous influence on the half-grid issue.
COLOUR VERSATEC PLOTTER SABOTAGED
=================================
= Somebody cut out his/her own plot from the Colour Versatec plotter,
and did not put the roll of paper back on about 20h00, August 8.
This affected two of the Y13 plots, and the queue was longer than 18
August 9. Enraged Bruce McCracken started a witch hunt asking every-
body around, and he even went down to the security and checked the
sign-up sheets to find out who were in the building Thursday night.
There was a rumour of a $50.00 reward for catching the vandal.
DAVE SMITH DOES IT AGAIN
========================
= Rob Koziy, who ran DRC on Dave Smith's Y13TCT section August 9 as a
courtesy and got "NO ERRORS FOUND", was shocked to hear that Dave had
put level 29 (exclude check layer) around his entire section. Rob was
absolutely speechless when he heard one of Dave's structures had a
0.7um metal2 gap! Poor Go S. Sunatori, whose response to Dave's goof
was "Oh my God!", got stuck with fixing Dave's mistake.
= Another surprise struck Bob Penchuk, who was singing blues after
finding absolutely no metal1 level on his final Y13MCT quadrant plot
produced by LIBSTAT. Bob figured PENFILE got screwed up somehow.
"Y13 CALMA" ON CM4 LIBRARY DISK
===============================
= The integrated "Y13 CALMA" file with all 4 quadrants plus Dave Smith's
test structure was placed on CM4 Library Disk August 8. The top-level
cells are named as follows:
----------------------------------------------------------------------
- Y13CCT: Macro quadrant
- Y13GCT: Gate-array quadrant
- Y13MCT: Memory quadrant
- Y13SCT: Standard-cell quadrant
- Y13TCT: Test transistors
----------------------------------------------------------------------
CMOS4 CELL LIBRARY FROZEN AUGUST 8, 1985
========================================
= Go S. Sunatori was under fire August 7 for the very minor cell changes
he had made. Though the changes were suggested by Dave Smith, poor Go
was getting all the blame from everyone. Among the most upset, John
Kester was pointing a shotgun at Go because of an extra DRC check he
had to run. When Go said "Nobody told me to stop the cell design!",
Len McCready immediately corrected the situation by ordering Go to
completely freeze the CMOS4 cells. So the CM4 Library Disk is now
rock-solid as of August 8, and it will never change until Y13 goes
out and all the cells get captured on DEX. This means absolutely
no more suggestions/comments/complaints against any cell will be
accepted! Consequently, the daily CM4 Library Disk build process will
come to an end August 9.
PAD MACROS ADDED TO CMOS4 CELL LIBRARY
======================================
= The following PAD macros designed by Perry Pavlovic and Rob Gibbins
passed the tough DRC check and they were added to CM4 CAL August 7,
and to CM4 CELLIO August 8:
----------------------------------------------------------------------
- CM4OP: Output PAD macro
- CM4IP: Input PAD macro
- CM4VDD: VDD PAD macro
- CM4VSS: VSS PAD macro
- CM4VAA: Analog PAD macro
----------------------------------------------------------------------
A MESS ON CM4 PROCESS FILE
==========================
From CRM389A sent on 85/08/08 at 09:23:54 :
go: the process file that gibbins used was a quick and dirty
'slow' column for output drivers only. it should not be
used for general users. i will create a full process
file for the next release of software. (david smith)
= Because of the note above, both "SLOW" and "FAST" process columns in
the CM4 PROCESS file on the CM4 Library Disk were taken out August 8.
DAVE GIBSON ON CMOS4 TESTCHIP (Y13) STATISTICS
==============================================
1985 Aug 8 at 10:28 EDT
+--------------+-------------------+---------------+
| Quadrant | Size | Transistors |
+--------------+-------------------+---------------+
| Standard | 128 x 120 mil | 10767 |
| cell | (15360 mil2) | |
+--------------+-------------------+---------------+
| Gate Array | 123 x 141 mil | 3912 |
| | (17343 mil2) | |
+--------------+-------------------+---------------+
| Macros | 146 x 146 mil | 8858 |
| | (21316 mil2) | |
+--------------+-------------------+---------------+
| Memory | 128 x 126 mil | 14623 |
| (256x8 bit) | (16128 mil2) | |
+--------------+-------------------+---------------+
+--------------+-------------------+---------------+
| Totals | (70147 mils2) | 38160 |
+--------------+-------------------+---------------+
This is truly amazing since the entire testchip has been produced
in only 9 weeks. If you were to consider the number of working days
(ie. the mythical 5 day work week with only 7.5 hours per day) then
the productivity is a staggering 848 transistors per day or 113
transitors per hour or 2 transistors per minute. However, these
figures are a bit misleading since most of the designers did a lot
more than the recommended mythical work week.
ROB KOZIY WINS A TRIP TO HONEYWELL
==================================
= The CM4NEWS editor congratulates Rob Koziy for winning a trip to
Minneapolis, Minnesota, to deliver our CALMA GDS-II tape to HONEYWELL.
= In the final test chip meeting August 7, Dave Gibson presented the
latest stats for Y13, which shows close to 40 thousand transistors in
4 quadrants, much more than W82 has. This incredible Y13 was designed
in record 2 months, and Len McCready calls it "a schedule-driven
mammoth chip".
= It was decided in the same meeting that "Y13 CALMA" file be placed on
CM4 Library Disk as a master CALMA file for duplicate cellname check,
and more importantly for the final DRC run and the single-layer plots
on all 4 quadrants to be done August 8.
= Hojjat Salemi, Bob Penchuk and John Kester all reported "NO ERRORS
FOUND" in their DRC run, while Rob Koziy reported a few explainable
DRC errors in his gate-array border macros. The only worry for Go
S. Sunatori now is the narrow doping mask at the butted cell edges,
commonly seen in CMOS3 design.
= The Y13 Celebration Lunch was re-scheduled to Friday, August 16.
VERY MINOR CELL CHANGES
=======================
= Some metal lines in NAND3, NAND4, NOR3 and NOR4 were tidied August 6
for less extra capacitance and for better yield. At the same time,
the passivation opening in the PADT cell was given what Dave Smith
calls "a more professional look".
WORST CASE SCAMPER SIMULATION APPLICATION NOTE
==============================================
= Thanks to Rob Gibbins, the CM4 PROCESS file on the CM4 Library Disk
now has both "SLOW" and "FAST" process columns.
= Following is a set of parameters to be used for CMOS4 worst case
SCAMPER simulation:
----------------------------------------------------------------------
' CMOS4 WORST CASE ANALYSIS:
' VDD = 4.5 V
' TEMP = 70.0 C
' PROCESS = SLOW
' GATE LENGTH = 1.75 um
'
*VDD = 4.5
*VSS = 0.0
*VRF = 2.42
**TEMP = 343.0
'
$CHANGE PROC 1 3
$CHANGE POLY 3 0.25
$ANALYZE TIME
----------------------------------------------------------------------
CMOS4 CUSTOM DESIGN APPLICATION NOTE
====================================
= If one understood the basic CMOS4 half-grid philosophy, he/she would
not have too much frustration in digitization in LAYED. Following is
a set of guidelines to layout CMOS4 circuits manually:
----------------------------------------------------------------------
1) All cell placements to be on HALF(X)/HALF(Y) grid points
---> "set grid inp 33 dis 66" in LAYED
2) All interconnect lines to be on FULL(X)/FULL(Y) grid points
---> "set grid inp 66 dis 66" in LAYED
----------------------------------------------------------------------
RDTP ADDED TO CMOS4 CELL LIBRARY/INVZ REDESIGNED
================================================
= The latest addition to the CMOS4 standard-cell family is an RDTP cell
(RDTPS without a built-in MUX), born July 31. Since there is no
scan-path feature in the RDTP cell, the cell should be used only in
shift-registers or in bus structures where 100% testability is ensured
without the scan-path technique. Exultant Dave Gibson exclaimed in
COCOS: "I am extremely impressed that you made an RDTP."
= The troble-making INVZ cell, which was generating tons of DRC errors
at the side edges when butted with other cells in the standard-cell
quadrant, was totally redesigned with a switching point of 2.32 volts,
and the latest DRC run July 31 did not show any errors around INVZ.
PAD MACRO CONTROVERSY (BDN AND BDP CELLS FIGHT FOR SURVIVAL)
============================================================
= The question is "How much drive should the output PAD macro have?"
The design target for everyone was to drive off chip at 50MHz under
the worst-case condition (VDD = 4.5 volts, TEMP = 70 degrees Celsius,
and "slow process"). But the real question is "Into what load?"
Steve Sunter quotes 20pF, Len McCready says 10pF, and Eric Hall wants
even 50pF! It was sort of decided that at this stage of the game,
anything between 10pF and 20pF is sufficient for a regular PAD macro
CM4OP which has two BDNs and two BDPs. A high-power PAD macro will
eventually be designed with more BDNs and BDPs. Especially happy was
Go S. Sunatori who does not have to redesign the border cells BDN and
BDP which are now "version 7".
= Rob Gibbins and Go S. Sunatori were puzzled to see identical SCAMPER
results July 29 for both "nominal process" and "slow process". It
turned out that Dave Smith had not implemented the "slow process" and
"fast process" in his CM4 PROCESS file.
= Bob Penchuk had not realized the INVD cell size change until as late
as July 26, and he had to re-route the signals in his memory block at
8 o'clock Friday night. Bob also joined the team of assassins to get
Dave Smith.
= One of the border macro rules "maximum device well overlap of butt
contact" was changed from 2.4 um to 1.6 um. Both Rob Koziy and Go S.
Sunatori got upset about this change for the sake of DRC.
"THE HALF-GRID TASK FORCE"
==========================
= Dave Gibson, Hojjat Salemi, Jamie Twidale and Go S. Sunatori held a
day-long meeting July 29 to discuss the half-grid issue. Dave has
built a benchmark Cruise Control using the half-grid approach to
demonstrate that the CAD tools work. Although he had to live with a
slight inconvenience in digitization and a lack of schematic, he felt
that there is no real problem. All the advantages and disadvantages
of the half-grid system were documented, and it was finally approved
in a meeting July 30 in Len McCready's office with about a douzen
people. (Everybody was so excited in the meeting that somebody had to
open the door of Len's office.) A list of required tool enhancements
will be presented to the CAD group for the integrated CMOS4 Design
System in SDS release 3.03. (January, 1986)
CMOS4 CELL TYPES
================
= Here is a summary of the CMOS4 cell types. The complete information
can be browsed on-line by typing "CM4INFO".
----------------------------------------------------------------------
- Type 1: Standard Height (no metal2)
- Type 2: Standard Height
- Type 3: Internal (no metal2)
- Type 4: Internal
- Type 5: Utility (no metal2)
- Type 6: Utility
- Type 7: Macro-block (no metal2)
- Type 8: Macro-block
- Type 9: Boundary (no metal2)
- Type 10: Boundary
----------------------------------------------------------------------
= PADT is now a type 1 cell. That means a LOGED symbol exists for PADT
for FUNSIM and even AUTOLAY. All the credits should go to John Kester
who insisted on making PADT a non-utility cell.
THE FATE OF GATE-ARRAY (AND ROB KOZIY)
======================================
= It is anybody's guess at present. The betting circle is growing very
rapidly.
ROB GIBBINS ON CHANGES TO THE CMOS4 DRC
=======================================
MON 29JUL85 15:26:44
Changes to the CMOS4 DRC
Soon to be forthcoming, the next version of the rule document
will contain comprehensive sections on:
CAP POLY structures
BORDER region rules
CORNER to CORNER, CORNER to EDGE rules for metal, metal2, poly
SCRIBE CHANNEL rules
RULES of THUMB for current handling of metals, contacts, vias
Also, for ease of understanding, in the next version, design
rule numbers will be correlated to CALMA (oops, LAYED) layers,
eg. a contact error will be 9 point something.
In the interim, new rules have been encoded with the old
numbering system, as of Sat., July 27. These rules are:
1) minimum nwell spacing: 5 um (still 11 um for different
potential nwells)
2) minimum pguard spacing: 3 um
3) minimum pguard width : 10 um
As of 2pm today, the pad rules have been updated. For all
you worried souls, the internal test pad rules are:
Test pad spacing: set by metal2 spacing
Test pad passivation to via spacing: anything you want
Test pad passivation to metal1 spacing: anything you want
Metal2 overlap of passivation opening: min 6 um
Minimum passivation opening: 9 um by 9 um
As usual, please report any DRC funnies to me posthaste.
A special nod to Hojjat for a couple of spurious errors
found, and to Rob Penchuk for finding yet another problem
with rule 5.7b.
ERIC LIVERMORE ON CMOS4 SUDS
============================
FRI 26JUL85 08:12:09
Concerning future (near term) CM4 products, we have as
you may have heard but forgotten the following candidates
1) R10
2) R30 an ADN chip
3) R?? the 'syntran' chip - initially for DMT-400 now
put back to 'exploratory'
4) Rxx maybe some of the DMS-100E would be redone
for a CR
5) plus a whole host of others
**** Caution though***** Roly will restrict the initial
number of codes in 1986 to some 'manageable ' number
such as one or two ---- seems like the right thing to do.
WELCOME BACK, "CM4NEWS II" (PLUS "CM4INFO")
===========================================
= Just a few minutes after the CM4NEWS editor finished writing his own
obituary below, he received a personal visit by his boss's boss, and
the editor was finally persuaded that he should continue CM4NEWS
before John Roth or even Edmund Fitzgerald begs him to do so.
= The rejuvinated "CM4NEWS II" comes with a new feature "CM4INFO" which
contains the on-line CMOS4 Cell Library Philosophy document maintained
by Steve Sunter. Please direct your comments/complaints about the
document to Steve.
AU REVOIR, CM4NEWS (EPILOGUE)
=============================
= Although the CM4NEWS editor attracted more popular support than Joe
Clark did in Winnipeg where he drew 66% of the conservative vote in
the 1983 PC leadership review, the support for the CM4NEWS editor fell
short of his psychological target of 80%, and the editor's decision to
quit is now as firm as Pierre Trudeau's on February 29, 1984. The
CM4NEWS editor feels that now is the right time to go, just as the
popular BUILDING file died before the official opening of Lab-3 at
BNR Carling in spite of an enormous support. After seeing hot-air
balloons in "Destination Canada '85" July 27, the editor now wants to
pursue his career as a balloonist. Incidentally, the CM4NEWS editor
will be desperately seeking your support in the 2004 Federal Liberal
leadership convention.
= So far we seem to have survived the two major battles:
- The 6.6 um/grid battle
- The half-grid battle
The CM4NEWS editor believes that we have already achieved a major
kilometer-post in the CMOS4 development, and only history will tell
whether the incredible Y13 becomes as legendary as "a small step for
Dept. 5K, but a giant step for BNR/NT". And remember, Y13 is just a
beginning of the unlimited potential of the CMOS4 Technology which is
scheduled to peak in 1992.
= The hype may be over now, and the CM4NEWS editor wishes to thank every
reader and contributer for sharing the ups and downs of the early
CMOS4 development as well as joys, laughs, angers, sorrows, agonies,
crises, panics, long days, late nights and deadlines. Special thanks
are due to his managers for their continuous support and encourage-
ment. Thank you all again for staying tune for the CM4NEWS editor's
personal diary, et merci beaucoup!
<<<<<<<<<<<<<<<<<<<<<< ------------------------ >>>>>>>>>>>>>>>>>>>>>>>
<<<<<<<<<<<<<<<<<<<<<< See you again in CM5NEWS >>>>>>>>>>>>>>>>>>>>>>>
<<<<<<<<<<<<<<<<<<<<<< ------------------------ >>>>>>>>>>>>>>>>>>>>>>>
TUES 16JUL85 12:53:07
########################### IMPORTANT NOTICE ###########################
############# TO ALL DESIGNERS INVOLVED WITH THE TESTCHIP ##############
########################################################################
PLEASE NOTE THAT YOU WILL HAVE TO PUT LDISK 3.01 IN YOUR PROFILE
SO THAT YOU WILL OVER-RIDE THE DEFAULT 3.02 SOFTWARE. WE ARE STICKING
TO 3.01 SOFTWARE UNTIL THIS CHIP IS COMPLETE.
************************************************************************
****************** PRELIMINARY CMOS4 DEVELOPMENT NEWS ******************
************************************************************************
Editor's note:
- Please do not copy "CM4NEWS" to your A-disk since it will screw up
"CM4NEWS WRITE".
- Extensive personal references are made not to offend/embarrass/upset/
insult/humiliate/defame/mortify somebody but to highlight his/her
activity. If you feel offended, please file your complaint to the
CM4NEWS editor through "CM4NEWS WRITE" facility. For example:
------------------------------------------------------------------------
TUES 09JUL85 02:10:23
My favourite show "The Flintstones", mentioned in your article
"ANYBODY ELSE ON CALL?", was replaced by "Spider-Man" and is no
longer shown on CTV at noon! I watch Les Nouvelles instead. (GSS)
------------------------------------------------------------------------
WED 26JUN85 21:45:04
I work only 70 hours a week now, not 80 hours as stated in your
article "HALF-GRID IMPLICATION..... "! (GSS)
------------------------------------------------------------------------
CM4 CAL NOW CONTAINS CM4CPYRT AND CM4NTL
========================================
= Thanks to Bob Penchuk, the error-free version of the following two AID
cells have been incorporated in CM4 CAL file on the CM4 Library Disk.
- CM4CPYRT (copyright symbol)
- CM4NTL (NT symbol)
"DECISION CMOS4": AN ELECTRONIC POLL TO DECIDE THE FATE OF CM4NEWS
==================================================================
= The CM4NEWS editor went for a lonely walk in a rainstorm July 22 and
decided that it was time for him to step down as editor. The present
CM4NEWS editor has been criticized for:
- Attacking people
- Singing too much
- Invading people's privacy
- Making false accusations
- Conveying inaccurate information
- Distorting some facts
- and much, much more.....
= Now, here is your once-in-a-lifetime opportunity to cast a democratic
vote. Starting July 24, as soon as you leave this CM4NEWS panel, an
electronic voting system will be activated to decide the future of
CM4NEWS, despite a very strong opposition by Steve Sunter who said
"Can't we make an engineering decision?" You are all on the eligible
voters' list, and you vote by one secret ballot. Following are the
three options:
**********************************************************************
OPTION 1: Phase out CM4NEWS entirely
OPTION 2: Continue under new management (nominate somebody)
OPTION 3: Continue under present management
**********************************************************************
Please type in your choice, general comments on CM4NEWS if any, and
also nomination for a successor if you should choose OPTION 2. You
may have to bribe the present CM4NEWS editor if you were to choose
OPTION 3.
= An advance poll was conducted July 23 in the SDS Hotline area. Please
type "CM4POLL" to see the results. Remember, every vote counts!
CESAR CESARATTO INVOLVED IN THE HALF-GRID ISSUE/MORE CMOS4 CUSTOMERS
====================================================================
= An ultimate lineup of Cesar Cesaratto, Roly Reader, Eric Livermore,
Steve Sunter, Robert Hum and Jamie Twidale got together July 23 for
yet another meeting to discuss the half-grid issue. Cesar read the
affidavit regarding AUTOLAY2 with interest, and the outcome of the
meeting was a highest-priority action item (over W82 according to
Cesar, believe it or not) for Dave Gibson and Go S. Sunatori to try
out a real test circuit like Cruise Control System to demonstrate the
area advantage of the half-grid system. It was advised that Y13 not
go out until the half-grid issue is resolved. Poor Go, who had been
in a somewhat relaxed mood these days with the return of the Coke
Classic, cried out "I have to give up my daily 2 o'clock sunbathing
for the next 2 weeks!" The CM4NEWS editor wonders how long it would
be before John Roth or even Edmund Fitzgerald intervenes with this
issue.
= Dave Smith pointed out more facts July 23 regarding the cell size.
A 20% cell area increase will lead to:
- 20% decrease in yield
- 20% increase in interconnect length, leading to 20% increase in
interconnect capacitance which means 20% increase in RC delay time
= Cesar Cesaratto was seen selling CMOS4 Cell Library himself July 23
along with Roly Reader to yet another potential customer.
"CADILLAC LIBRARY" VERSUS "LADA LIBRARY"
========================================
= The half-grid issue is still controversial. Ian McWalter was viewing
the CMOS4 library cells July 19 and he said "We could put all via
sites on full grid points by expanding the cells half grid in each
side, if necessary." Et Voila, we now have two sets of cell library:
- CADILLAC-grade cell library with HALF/HALF via sites
- LADA -grade cell library with FULL/FULL via sites
Both versions are now on display in the Cell Library Group area.
The CADILLAC library is what Go S. Sunatori calls "the best cell
library in the world", and it is 20 to 30% more area efficient than
the LADA library. The performance characteristics of the LADA library
cells are beyond anybody's imagination. (Ask Leo Potvin who recently
had to scrap his 1982 LADA.)
Y13 DESIGN REVIEW MEETING TO BE HELD JULY 23
============================================
= The CM4NEWS editor welcomes Len McCready and Dave Gibson back on board
into the reality of the July 26 deadline, replacing Jeff Bennett who
is now serving his wife for the next two weeks. Len and Dave would
be very pleased to know, despite Steve Sunter's pessimistic forcast
"It will never be done." based on his own numerous bitter experiences,
that Y13 is still on track aiming for PG on July 26, 1985. They might
also notice that while they were away CM4NEWS, which started as a
source of serious information, has transformed itself into a major
source of entertainment.
JULY 20 BNRLSI SHUTDOWN FROM 07H30 THROUGH MIDNIGHT
===================================================
= Outraged by the inconsistency between NEWS 173 and the logon message:
| 173 85/07/11 11:16:02 *BNRECAD and BNRLSI unavailable July 20/85 *
| ----------------------------------------------------------------------
| On Saturday, July 20, BNRECAD and BNRLSI will be down from 07:30
| until 19:30 due to electrical work being done at Corkstown.
| * SYSTEM SHUTDOWN JULY20/85 07:30-->24:00 DUE TO BUILDING POWER DOWN *
the CM4NEWS editor phoned the Corkstown operator, who confirmed that
the shutdown is until midnight, not till 19h30 as stated in the news
item above. Representing the entire CMOS4 design community, the
CM4NEWS editor filed an official complaint to USER SERVICES, who is
responsible for NEWS, for misleading us at this critical phase of Y13.
There is a potential of wasting up to 50 minutes of each designer's
precious time as well as gas/energy to drive/cycle into work after
19h30. In response to the protest, USER SERVICES immediately issued
NEWS 183 with the correct shutdown period.
= One of the non-stop crazy holiday workers, Bob Penchuk was planning
to get up at 5h00 Saturday morning to catch two-and-half hours of
work before the shutdown. What a spirit! Meanwhile, Rob Koziy was
planning to enjoy a taste of manangement Saturday doing long-overdue
paperwork.
CM4WIDE EXEC PLACED ON LIBSTAT 194
==================================
= Due to popular demand (Rob Gibbins, Hojjat Salemi and Perry Pavlovic),
a new WIDEN utility exec CM4WIDE was added to LIBSTAT 194 disk. The
exec widens all centre-line levels 21, 22 and 23, places the expanded
levels 6, 10 and 12 back onto the original cell, and creates "FINAL
CALMA" and "FINAL ICDATA" files.
= A fatal bug was found in CM4WIDE EXEC July 25. The "FINAL CALMA" file
was erased from Bob Penchuk's disk when he tried to run the exec on
"FINAL CALMA" file, and he had to get it back from the backup system.
One is not supposed to run this utility exec on "FINAL CALMA" file,
and a protection mechanism was added immediately to the exec.
CMOS4 WAFER-SCALE INTEGRATION/BOOMING BUSINESS FOR CMOS4
========================================================
= No kidding! The CM4NEWS editor was astonished to learn that Peter
Schvan of P986 (NTE) has a plan to develop Wafer-Scale Integration
using our CMOS4 Standard Cell Library.
= The CM4NEWS editor has also learned that Colin Rowe, now a moderately
big boss in 5S, wants to do at least two CMOS4 chips in 1985.
Y13 CELEBRATION LUNCH AUGUST 9
==============================
= According to Jeff Bennett's minutes of the July 16 CMOS4 test chip
meeting, the BNR-funded Y13 Celebration Lunch will be held August 9.
Too bad for those of you who are planning a vacation on that day.
The CM4NEWS editor asks "Only one lunch? We have 4 (four) quadrants,
which are closely related with one another."
= As Steve Sunter told Cesar Cesaratto in the last SDT Review Meeting,
we are setting two company records with Y13:
- Chip PG in 2 months from scratch
- Cell Library development in 2 months
GO S. SUNATORI F---S UP CM4 LIBRARY DISK (AGAIN)
================================================
= Again? Yes, and it was worse this time affecting the entire CM4
Library Disk. The daily 6 o'clock CM4 Library Disk build process,
which had never failed since it started June 17 with a big fanfare,
did not happen July 18. It turned out that Go S. Sunatori had erased
CMOS4 CELLIO file on SUNATORI 193 disk by mistake when he used ERASE
and a number of ='s in FLIST while re-redesigning the border cells.
Confronted with infuriated Jean Wills, John Kester, Rob Gibbins and
Perry Pavlovic, Go quickly started CM4BUILD EXEC at 12h20, but it was
not until 14h35 when the whole CM4 Library Disk was rebuilt.
= Despite this emergency, Go managed to escape from work for his daily
2 o'clock sunbathing with beer. But even the sun was turning its cold
eyes to him, hiding behind thick clouds from time to time. Go was
bitching about CM4NEWS saying "I was probably just a little drunk when
I goofed last night. It is not fair that every time I make a minor
mistake, it becomes a CM4NEWS headline! Besides, I cannot even have
my own privacy!" Go made a fatal error making everybody upset, and
he has to realize that it is totally unacceptable to screw up the CM4
Library Disk at this stage of the game!
A TASTE OF NIGHT SHIFT/RE-REDESIGNED BORDER CELLS
=================================================
= Now Dave Smith is wanted by 2 people! Rob Koziy, plagued by numerous
design spec changes to his own border macros requested by Dave Smith,
was heard uttering "Christ!" many times during what Steve Sunter calls
"the best consultation on border cells". Rob immediately joined the
team of assassins to get Dave Smith.
= In the mean time, Go S. Sunatori, fatigued after non-stop 15-hour day
Tuesday, decided to go home early to have a much-needed rest. With
his favourite Lowenbrau in his hand, he was sunbathing and feeling
Rocky-mountain high with John Denver's "Sunshine on My Shoulders".
When his CD player hit "Fly Away" and he just felt like flying away
from his 15th-floor balcony, he suddenly recalled his promise to Perry
Pavlovic to deliver a set of border cells by Thursday morning. So he
returned to his night shift and completed re-redesign of the border
cells which now should satisfy all the requirements by Steve Sunter,
Dave Smith and Rob Gibbins.
= The following border cells have been re-redesigned to comply with
the new border macro rules, latch-up protection and current handling
capability. Please note the change in the cell size from yesterday's
announcement.
NEW OLD
--- ---
- CM4BDN ( 7 x 10 grids) ----- CM4NNN ( 6 x 10 grids)
- CM4BDP (10 x 10 grids) ----- CM4PPP ( 6 x 10 grids)
- CM4BDI (17 x 10 grids) ----- CM4III (18 x 10 grids)
- CM4INVD (23 x 10 grids) ----- CM4INVD (24 x 10 grids)
- CM4BUFD (21 x 10 grids) ----- CM4BUFD (22 x 10 grids)
* Metal2 connection is mandatory if used in the border region.
* A one-grid gap is required between BDN and BDP in the border region.
* CM4NNN, CM4PPP and CM4III are to be obsolete after the test chip.
DAVE SMITH TO BE ASSASINATED
============================
= Dave Smith handed out a new set of pad macro rules July 16 to Rob
Gibbins. Unfortunately it was too late for poor Go S. Sunatori who
had already finished re-design of BDN, BDP, BDI, INVD and BUFD cells
with Rob Gibbins' version of the pad macro rules. It was even after
Go sent out a COCOS message to his manager when he saw Dave's version.
"Give me a f---ing break! I'm gonna kill Dave!", screamed Go, who
ended up missing his supper and staying till 23h52 to fix up all the
cells.
ROB GIBBINS' YET ANOTHER NOTE ON PDRC
=====================================
WED 17JUL85 17:57:04
Due to popular demand (Hojjat asked) yet another plotfile
is sent to you when the DRC is run: filename PRTPLTC (sends
plot to the colour VERSATEC).
This file produces a "functional" plot of the cell, which means
it is half way between a latchup plot and an all-layer
plot. It is useful in detecting poor design practises, such
as current constriction points, unnecessary capacitance,
inadequate conductor widths, etc.
The doping mask information is logically ANDED with the
device well after poly has been removed to delineate
diffusion regions, as in a latchup plot. In addition, all
diffusions, poly, contacts, vias and metals are shown.
No modification of this file is necessary in order to generate a
functional plot. Simply use the px exec as before, except specify
"C" as the letter. Example, for an analyzed cell in CALMA
file TEST:
px test 9 c 40 (cr) (40 inch plot)
or
px test 9 c s (cr) (scaled plot; further questions
asked to determine scale factor)
This note will be appended to the PLTGUIDE TEXT file on the DRCDSK.
GO S. SUNATORI MAKES A BRIEF TV APPEARANCE JULY 15
==================================================
= In an unrelated move, Go S. Sunatori took Monday morning and a part
of afternoon off just to attend the official opening of the Sparks
Street Mall Speakers' Corner at noon. There he said hello to
Commons speaker John Bosley, saw journalist Charles Lynch speak, and
was filmed when a heckler disrupted Lynch's speech. A part of his
face was shown for quarter of a second in the CBC/Radio-Canada news
(channel 4, 5 and 11/canal 9), and his back was shown for half a
second on CJOH (channel 13) and CHOT (canal 40). This marks his
second TV appearance in 4 years. His first TV appearance was on
Canada Day in 1981 when he appeared in the local CHCH news singing
O CANADA in Gage Park in Hamilton, Ontario. Go said "I would rather
be talking about CMOS4 Cell Library on TV."
MANAGING CMOS4 TECHNOLOGY
=========================
= Jeff Bennett brought an interesting article "Cut Product Development
Time" to the CM4NEWS editor's attention. Here is a summary of the
article found in the July 1985 issue of IEEE Spectrum (vol.22, no.7):
- Avoid setting too many objectives for a program.
- Wherever possible, use technologies developed externally.
- Break the design and development process into small steps that can
each be tackled by a team of no more than five people.
- Perform as many design and development steps as possible in
parallel.
- Test new technologies at an early stage, as well as the marketing
strategy and manufacturing approach.
- Keep program reviews relatively informal.
- Choose team leaders who are experienced in all aspects of
development, who can anticipate logjams and find ways around them.
- Make sure that support groups understand project priorities so that
they will make themselves available if a project team has to work
nights or weekends.
PERRY PAVLOVIC WORKING ON PAD MACROS
====================================
= Fresh from a one-week vacation, Perry Pavlovic is now in a critical
path building PAD macros under Rob Gibbins' supervision.
AN ANONYMOUS BIG BROTHER
========================
MON 15JUL85 09:08:27
SUNDAY WASN'T THE ONLY DAY THAT GO PLAYED COMPUTER GAMES.
ALTHOUGH HE DID IN FACT CYCLE INTO 'WORK' ON SATURDAY, I
KNOW FOR A FACT THAT HE SPENT HIS TIME PLAYING ON THE
MAINFRAME AND WITH THE VERSATEC. ACTUALLY HE WAS PLOTTING
THE NUMBER OF HOURS PER WEEK HE HAS WORKED SINCE JOINING
BNR. LATELY THE PLOT HAS BEEN SHOOTING UPWARD, SO I GUESS
HE DESERVES A BREAK TO FOOL AROUND.
LET'S SEE IF THIS MAKES IT INTO CM4NEWS!
- ANONYMOUS
LIMITED GUARANTEE FOR ALL CMOS4 STANDARD CELLS (JULY 14)
========================================================
= The following 44 CMOS4 standard cells have been fully simulated with
SCAMPER. FUNSIM MACRO file has been created as well as FUNSIM LOGIC
and SUMMARY (SCAMPER output) files, and a limited guarantee has been
issued. Please ask Steve Sunter for a full guarantee.
OR2, OR3, OR4, OR5, OR6, OR7, OR8, ORAND,
AND2, AND3, AND4, AND5, AND6, AND7, AND8, ANDOR,
NAND2, NAND3, NAND4, NOR2, NOR3, NOR4, EXOR,
INV, INVB, INVC, INVD, BUF, BUFB, BUFC, BUFD, INVL,
BDN, BDP, BDI, TGP, TGIP, MUX, MUX2, RLAT, RDTPS,
BUSC, BUSD, BUSE
* LIMITED Guarantee: Go S. Sunatori will resign if the cell doesn't work
* FULL Guarantee: Steve Sunter will resign if the cell doesn't work
= Finally, Go S. Sunatori does not have to bear starvation or dirty
clothes any more. He did both grocery shopping and laundry Saturday,
July 13, phoned 1-(900)-200-COKE to vote "YES" to the return of the
classic original taste of Coca-Cola (and 5SNEWS, too?), and cycled to
work. He said "Crossing the creek at Corkstown by bike wasn't very
easy." After seeing an impressive lineup of crazy holiday workers,
he was singing Simon and Garfunkel's "Mrs. Robinson" with a special
emphasis on the "Where have you gone, Joe DiMaggio?" part.
Incidentally, Go played 20 video games Sunday without spending a cent,
thanks to some coupons.
= While Go is taking a brief break and fooling around, Jean Wills is
now working day and night creating LOGED symbols for all the CMOS4
standard cells.
COMING UP: ROB KOZIY'S MAJOR ANNOUNCEMENT RE PREFIXES
=====================================================
= In a brief and productive meeting July 11 without Steve Sunter in
Steve Sunter's office, Rob Koziy, Jeff Bennett and Go S. Sunatori
came up with a reasonable prefix strategy for Y13. Rob is now
preparing a COCOS message to all the quadrant primes re prefixes.
A PROPOSAL FOR CMOS4 PREFIX
===========================
= Here is a preliminary proposal for the CMOS4 prefixes:
Standard Cells Gate Array Cells
-------------- ----------------
CM4$INV CM4®INV
CM4$NOR3 CM4®NOR3
= It surely reflects the fact that a gate array chip is less expensive
than a standard cell chip, n'est-ce pas? If there is any opposition
to the suggestion above, please file a complaint to Go S. Sunatori.
PENCHUK MAKES GIBBINS' DAY
==========================
= After clearing up all the Phoenix errors in his SRAM memory block,
Bob Penchuk was more than confident of getting "NO ERRORS FOUND" in
the latest Phoenix run. Instead, he found just one tiny corner-to-
corner violation July 12. He said "I am very frustrated!", but it
certainly made Gibbins' day all right.
ROB GIBBINS ON PDRC CORNER CHECKING (PLUS AN APOLOGY)
=====================================================
FRI 12JUL85 10:46:04
Ladies and Germs,
Rules for corner-to-corner and corner-to-edge spacings for
poly-to-poly, metal1-to-metal1, and metal2-to-metal2 are
now in place in the CMOS4 DRC. These rules have been
validated with test structures, and will be included in the
next DE007 document (promised within 2 weeks). An advance
copy of these rules, along with rule-of-thumb design
guidelines, may be obtained from me.
Much to by chagrin, the CMOS4 library ran clean of these
errors. Please make my day by rulechecking your structures
and showing me some CC or CE errors.
SAT 13JUL85 14:07:54
RUNNING THE DRC
Sorry, it had escaped my attention that the documentation
for running the DRC was not on the DRCDSK. It is now.
The relevant file is:
DRC TEXT: how to run the DRC
Don't forget about:
PLTGUIDE TEXT: how to get DRERROR or LATCHUP plots
GO S. SUNATORI SCREWS UP CM4 CAL AND CM4 CELLIO
===============================================
= Go S. Sunatori, who applied for a MasterCard at Bank of Montreal July
11 just to collect a $5.00 credit, was in a very big trouble. First
of all Hojjat Salemi found out in the morning that Go missed out MWIN
and PWIN in CM4 CELLIO. Then in the afternoon, Rob Gibbins discovered
literally hundreds of Phoenix errors in Go's standard cells which Go
had not touched since they were declared Phoenix error free July 8.
A careful examination revealed that Go screwed up his CM4LEVEL EXEC
when he changed it to accomodate Rob Gibbins' PADs. After fixing both
bugs quickly, apologetic Go sighed saying "I am just too overworked",
but he needs a better explanation than that!
= Another blow to Go is a set of PAD rules Dave Smith is now writing.
That means Go will have to re-design BDN, BDP, BDI, INVD and BUFD.
To obviously dismayed Go, cool Dave just said "Life is tough".
So please be prepared for drastic changes in these border cells,
including the cell size. (Sorry, Terry Thomas who said to P861 in late
1983 quote "I don't care what you do, but never change the cell size!"
unquote.)
SOME ORGANIZATIONAL CHANGES IN 5L
=================================
= The CM4NEWS editor learned July 11 that Peter Benedek is about to be
promoted to the biggest boss in combined SDS/CBDS group. Mike Nakhla
and Robert Hum will be given more responsibilities. Congratulations!
BNRLSI SHUTDOWN JULY 20
=======================
= There is a scheduled BNRLSI shutdown from 7h30 to 19h30, Saturday,
July 20, according to NEWS 173. This last-minute shutdown one week
before the D-day would affect especially such a dedicated non-stop
holiday worker as Bob Penchuk, who doesn't even mind staying here at
Corkstown overnight provided that Bill Loughrin gives him a couch to
sleep on.
= Leo Potvin is back from a one-week vacation, making coffee again every
morning. This should please Go S. Sunatori who spent an average of
$1.60 a day last week for four large cups of coffee.
FIRST PROSPECTIVE CUSTOMER FOR CMOS4
====================================
= The CM4NEWS editor was astonished to learn July 10 that there is
already a potential customer for CMOS4 this year from Dept. 1C.
Roly Reader, Eric Livermore and Eric Hall are now negotiating a deal
for the CMOS4 SUDS chip to be done in 1985!
= Jeff Bennett also heard a rumour July 10 that somebody in Dept. 1E
wants to do SRAM in CMOS4 soon. Can anybody keep up with the 80-hour
weeks for the rest of the year? "No f---ing way!", Go S. Sunatori
exclaimed.
ANYBODY ELSE ON CALL?
=====================
= As the D-day July 26 approaches, the CM4NEWS editor invites every key
player of Y13 to join Go S. Sunatori, who is showing an early symptom
of breakdown, to be available for professional duty at any time for
emergency. (After long exhausting wee hours, Go may be watching "The
Flintstones" on TV or listening to his CD music at lunch time.)
- Go S. Sunatori: (613) 726-1693
ROB GIBBINS ON PDRC PLOTS
=========================
TUES 09JUL85 15:31:15
Ladies and Gentlemen,
The current version of the CMOS4 DRC now sends two plot control
files to the user:
1) filename PRTPLTB - used for error plotting (B/W plot)
2) filename PRTPLTL - used for latchup plots (colour plot)
where filename refers to the file name of the CALMA file containing
the checked cell. The user must perform some small modifications to
these plotfiles before use. A brief user's guide file is on the
DRCDSK: PLTGUIDE TEXT.
PAD AND PAD MACROS/CORNER RULES/LATCH-UP PLOT
=============================================
= Rob Gibbins almost finished his design of PAD, which is 30 x 30 grids.
He is now moving on to the design of PAD MACROs, another key item in
every quadrant. It was decided in the July 9 meeting that two non-
inverting PAD MACROs (input and output) be designed A.S.A.P.
= Dave Smith is moving along fast these days giving us a preliminary set
of corner-to-corner and corner-to-diagonal rules. The document is now
available from Rob Gibbins, who is encoding the rules for PDRC along
with the new integrated PAD rules (processing + manufacturing)
compiled by Steve Sunter.
= Another achievement of Rob Gibbins is to make the latch-up plot
facility available for CMOS4. Rob just produced a latch-up plot of
INVD cell, and he assures that INVD will never latch up!
SCOOP: STAN JEDRYSIAK SIGNS AN AFFIDAVIT! (AND ITS AFTERMATH)
=============================================================
= Our AUTOLAY expert Stan Jedrysiak signed an affidavit July 8, stating
that AUTOLAY2, which could be available in release 3.03 (January,
1986), could handle the proposed half-grid system without any problem.
In fact, he uses a phrase "absolutely no problem" in his affidavit.
The original affidavit was circulated in the test chip meeting July 9.
= There is absolutely no commitment at this point in time that the work
could be done, according to Gary Sakauye July 9. Gary says that Stan
was just assessing the technical feasibility in AUTOLAY2.
= The CM4NEWS editor deeply regrets if the previous statement has left
any false impression to the readers.
ANOTHER PREVIEW: CMOS4 LOGED SYMBOLS
====================================
= Whipped by Steve Sunter, Go S. Sunatori, who cannot even find the time
for laundry this week, is fighting back with a preliminary version of
CM4 LIBDAT file for the LOGED symbols. Please report any bugs to him.
(By the way, Go welcomes any volunteer to do his laundry and/or
grocery shopping.)
ALL CELLS IN HLPCOM INDEX ARE PHOENIX CLEAN
===========================================
= The latest Phoenix run at 8h30, July 8, 1985 on all the standard cells
in HLPCOM INDEX finally showed "NO ERRORS FOUND"! Special thanks are
due to John Kester and Rob Gibbins for their remarkable job on PDRC.
= Roly Reader was quite impressed in the July 9 test chip meeting to
hear that all standard cells are Phoenix error free.
NOW YOU CAN HAVE FUN WITH FUNSIM AND HLPCOM
===========================================
= The amazing story of CMOS4 development continues ..... (more amazing
than W82!?) The CM4FUNS MACLIB, which was previewed a couple of days
ago now has correct delay and LD values for the cells with limited
guarantee.
= The CM4SIM EXEC has been enhanced to include semi-automatic generation
of the deloading factors, and the new version of CM4FUNS MACLIB was
released July 8, 1985. C'est incroyable, n'est-ce pas? This means
Steve Sunter will not be constantly bothered by people like Phil
Wilcox and Ian Roane (remember that master of complaints?) regarding
inaccurate deloading values in FUNSIM MACROs.
= At the same time, more HLPCOM functions have been added:
- HLPCOM FUNS OR7 (FUNSIM MACRO file)
- HLPCOM SIML AND4 (FUNSIM LOGIC file)
- HLPCOM SIMI NOR2 (SCAMPER input file)
- HLPCOM SIMO INVL (SCAMPER output file)
= The following HLPCOM functions have been available for some time:
- HLPCOM INDEX (a list of semi-official cells)
- HLPCOM SUBC RDTPS (SCAMPER SUBCCT file)
- HLPCOM CAP MUX (capacitance file)
JOHN GORDON HOGEBOOM USING LAYED, CREATES SCHEMATIC SYMBOLS
===========================================================
= John Hogeboom, one of our managers, was seen using LAYED on a spectra-
graphics terminal July 4. He said it has been "quite a while" since
he used LAYED last time. The CM4NEWS editor congratulates John's deep
commitment to the CMOS4 development.
= John Hogeboom, who is about to go on vacation, created a set of
beautiful schematic symbols July 11 to be used on level 16. These
cells were promptly placed in CM4 ICDATA (not in CM4 CAL which
contains only semi-officially released cells) on CMOS4 Library disk.
VERSION 1W03 OF DE007 RELEASED
==============================
= At last, Dave Smith is back on track giving us the official version
1W03 of DE007 July 4. The CM4NEWS editor strongly advises everybody
to destroy any previous versions of DE007, official or pirated.
CM4NEWS EDITORIAL POLICY
========================
= The CM4NEWS editor, who is almost throwing up on his terminal after an
all-you-can-eat lunch (his only meal for the next two days), exercised
his power to censor Rob Gibbins' note below to include "Ladies and".
The CM4NEWS conforms to Section 15 of the Canadian Charter of Rights
and Freedoms, which has been in effect since April 17, 1985.
ROB GIBBINS ON PDRC
===================
THUR 04JUL85 11:19:21
Ladies and Gentlemen,
The CMOS4 DRC was updated around 5 p.m. July 2nd, so please
relink the DRCDSK disk by typing: DRCDISK (cr).
Pad rules have not yet been added to the DRC.
Large non-butt contacts must be inside or partially inside
a layer 28 polygon in order to be recognized by the DRC.
The DRC will generate the error message:
LENGTH SMALLCNT GT 1.6
for any large non-butt contacts not inside layer 28.
To facilitate the use of layer 28 with existing designs,
the user can run a special DRC check to quickly distinguish
large/small contacts. Type:
DRC filename cellname TESTC (cr)
The resulting file: filename PDRCERR A will contain a
summary and listing of all large non-butt contacts not
properly identified. The user can then copy over to
his A-disk the file: TESTC PRTPLTB from the DRCDSK disk,
rename it to: filename PRTPLTB and modify this file
to include his USERID and filename in the title line.
Then a plot of the incorrectly identified large contacts
can be generated by typing:
PX filename 9 B # (cr)
or
PX filename 9 B S (cr)
where # refers to the plot size in inches (40 maximum)
S flags to the plot routine that a scale size is
desired. For this case the user will then be
prompted for the scale size.
SNEAK PREVIEW: CM4FUNS MACLIB
=============================
= Finally, FUNSIM MACRO is coming! Inspired by Dave Gibson's semi-
automatic CELLIO generator, Go S. Sunatori is now giving a finishing
touch to his semi-automatic FUNSIM MACRO generator. After working
record 17 hours a day July 2, he almost completed the amazing
CM4SIM EXEC around 3 a.m., and a preliminary CM4FUNS MACLIB was
created in the daily build around 7 a.m., July 3.
CODE FOR THE JULY TEST CHIP
===========================
= Does anybody know the code for the July test chip? John Hogeboom
and Yachin Afek, both of whom are managers, said that nobody has
thought of it. Well, the most dedicated holiday worker CM4NEWS
editor hasn't forgotten! He suggests H series for Honeywell, and N
series for NTE for simplicity and consistency. Qu'est-ce que vous
pensez, Dave?
= Cheri Kathnelson gave us "Y13" for the Honeywell test chip and "Y14"
for the NTE test chip.
"CMOS4" OR "CMOS IV"?
=====================
= The CM4NEWS editor questioned Roly Reader after seeing "CMOS IV" in
Tim White's Technology Time presentation June 28. Roly insists we
use "CMOS4" although Dave Lawrence still prefers "CMOS IV". It makes
perfect sense to the CM4NEWS editor since "CMOS IV" could be mis-
interpreted as "CMOS one-V". So let's stick to "CMOS4"!
CM4NEWS GOES PUBLIC (PLUS II)
=============================
= CM4LIB EXEC, which links to LIBSTAT 104 disk, was released to LDISK
thanks to Jeff Bennett, Dave Gibson, and Ed Attfield. It means that
CM4NEWS is now available to every LDISK user on BNRLSI! Despite
this ever-increasing popularity of CM4NEWS, the editor is committed
to leave the facility as an informal and preliminary source of
information. (Try "CM4LIB ?" to find a typo bug in the EXEC)
= 5SNEWS, our major competitor, will be reconfigured as a Division 5
news facility and will still be maintained by Terry Curtis, according
to new 5S00 Director Ian McWalter (June 26). Remember many people
switched to PEPSI when COKE changed its formula?
= CM4NEWS may contain BNR/NT private/restricted/proprietary information,
and it should not be copied to your A-disk. Here is a quote from the
BNR corporate policy: "The use of electronic mail to transmit BNR
PRIVATE/RESTRICTED information is not allowed." (Section 14.2
HANDLING COMPANY INFORMATION, "A Guide to BNR Ottawa", P.55-57)
= Besides, "CM4NEWS WRITE" will not work properly if one has CM4NEWS
file on his/her A-disk. Jeff Bennett inadvertently destroyed CM4NEWS
June 27 because he had an earlier version of CM4NEWS on his A-disk.
JEFF BENNETT ON DE007
=====================
THUR 27JUN85 17:16:13
There is a new set of design rules! Until they are officially
released through Documentation use the special copy available
from Jeff Bennett. You can recognise it by the date June27/85
and the impressive line of signatures at the top. Dave Smith
has put his hand on his heart to say that these rules are
compatible with Honeywell processing (metal2 width of 3u
and 10% shrink possibly excepted). Jaan Kolk has convinced
us there is no problem in the rules for NTE maskmaking.
The rules preview copy will be missing introduction sheets.
These are unchanged from 1w01. Unfortunately DRC will need an
extra design layer to distinguish minimum and non-minimum
contacts- stand by for further details. Also stand by for new
bonding pad rules expected imminently from Steve.
May your graphics ever be righteous!
CM4NEWS EDITOR'S COMMENT
========================
= The CM4NEWS editor expresses deep condolence to poor Dave Smith and
Jaan Kolk who were confronted with a group of highly motivated
designers spearheaded by Jeff Bennett June 27 for not providing the
design community with the official 1W02 version of DE007.
= This is another example of poor version control. In response to Dave
Smith's complaint against the design community that the designers
want more stable rule document and more flexible rules at the same
time, Go S. Sunatori made a point saying that Dave should provide the
designers with more official versions (1W02, 1W03, etc.) so that we
do not have to be forced to rely on several pirated versions of DE007!
= Jaan Kolk promised us that corner-to-corner and corner-to-diagonal
rules will be included in the future version of DE007.
= Thanks to John Hogeboom's excellent imagination, it was decided to
encircle large non-butt contacts with a dummy rectangle on level 28.
This way, Rob Gibbins can code Phoenix rules to produce zero errors
(if your design is correct, of course) without any change in the
contact rules.
= Level 28 mentioned above will be called Large Contact Identification,
taking Rob Gibbins' suggestion, and he says butt contacts may or may
not be enclosed because it makes no difference to PDRC. (Steve Sunter
should include this in the Cell Library Philosophy document)
MORE CELLS ADDED TO CMOS4 CELL LIBRARY
======================================
= The following utility cells have been created due to popular request:
- DJOIN: Metal2 join cell (34 x 34 units)
- PWIN: Same as in CMOS3 (34 x 34 units, level 15 only)
- MWIN: Same as in CMOS3 (34 x 34 units, level 15 only)
= A new TTL-to-CMOS converter cell INVL was added to the CMOS4 Library.
It has a P/N ratio of 1:2 with a switching point of 1.7453 volts.
(Note that it is not called TMOS any more)
= Please note that the CM4 CAL file now contains only semi-officially
released cells listed in the HLPCM4 INDEX file.
HLPCOM INDEX REPLACES BNR/NT CMOS4 CELL LIBRARY STATUS
======================================================
= BNR/NT CMOS4 Cell Library Status will no longer be published in
CM4NEWS. (Watch out, Steve Sunter!) Instead, please type HLPCOM INDEX
for the most updated list of CMOS4 cells.
USE OF "PRELIMINARY"
====================
= Steve Sunter was in hot water recently for not using "PRELIMINARY"
in his Cell Library Philosophy document. The CM4NEWS editor
strongly urges everybody to use "PRELIMINARY" if it is not final.
A "PRELIMINARY" stamp is available from our secretary Nancy Porter.
= Dave Smith boasts of still having "PRELIMINARY" even in his 5-year old
NMOS PROCESS files.
GOSSIP: STAN CHOW IS FINALLY LEAVING/ZAHIR ADDETIA IS ALSO LEAVING
==================================================================
= The CM4NEWS editor has just learned that our DEX/PLS hero Stan Chow
is finally leaving Division 5 for BNR RTP. Zahir Addetia is
coordinating a farewell lunch for him.
= In another surprising move, Zahir Addetia is also leaving the DEX/PLS
group for DMS. The CM4NEWS editor wonders who's left?
DAVE SMITH IS BACK
==================
= After a long vacation/course/conference in Irland, Dave Smith is back!
= "Holy S---!" was his first reaction when he saw the plot of the
memory block with lots of 45-degree lines all over the place.
= "Holy Christ!" was his next reaction when he saw the plot of Bob
Penchuk's DRAM sense amplifier with circular poly gates with 45-degree
lines.
= According to Dave, an extra-large contact with the substrate contact
at the middle has to be 16 x 60 units.
HALF-GRID IMPLICATION ALMOST CLEARED
====================================
= In one of the most exciting meetings ever held in the CMOS4
development June 25, the half-grid implication was extensively
debated. After the meeting, our top brains came up with the
following proposal to the CAD group:
- All cell placements are on HALF grid points (both X and Y!).
- All interconnect lines are on FULL grid points (everywhere!).
- Cell VDD and VSS boundaries (top and bottom edges) are drawn
on FULL grid points on level 15.
- All side I/O points are on HALF grid points (in X direction)
except for VDD and VSS in FT cell which are on FULL grid points.
= The solution above requires major changes in CELLIO table, but not
the actual graphical data. That means Dave Gibson will have to modify
his great GETVIA EXEC and CM4CLIO EXEC before getting married.
= Go S. Sunatori, who was seriously considering taking an early
retirement from his position if he were to face re-design of all the
CMOS4 cells, is now back again to his 80-hour week business.
A MAJOR SETBACK EXPECTED TO ALL CMOS4 CELLS
===========================================
= Depending on the outcome of the next Tuesday's very important
meeting to discuss the half-grid implication, the announcement
*** ALL CELLS ARE READY (CAL/CELLIO/SUBCCT/simulation) AS SCHEDULED ***
would not be issued June 29 as planned, but would be delayed by
10 days to July 7 or so to shift the VIA locations on side edges to
full-grid points.
= Here is the action items for each of designer if it is decided to go
ahead with this major change:
- Dave Gibson: modify GETVIA EXEC and CM4CLIO EXEC to accomodate
half-grid VIA locations on top/bottom but full-grid VIA locations on
both sides. (1/2 day)
- Go S. Sunatori: change layout of all 47 cells in 48 hours in 4 days
(12 hours a day, that is), fix PRDC errors for all the cells,
re-generate CELLIO for all the cells, and re-simulate all the cells.
(10 days total)
- Greg Ward: change layout, run PDRC, and re-generate CM4CLIO for the
FADD cell. (1 day)
- John Kester: re-run PDRC for Go and Greg. (1/2 day)
WHERE IS DE007?
===============
= Roly Reader was astonished to learn that nobody has seen the official
1W02 version of DE007. Our boss Len McCready has made a top-priority
committment to get hold of a copy from the Documentation immediately.
= In another development, Jaan Kolk will notify Ian McWalter how much
corner-to-corner and corner-to-diagonal separation is required for
poly and metal in order not to be, quote "f---ed up again!" unquote.
= Rob Koziy just confirmed with Dave Smith verbally that the new minimum
via-to-poly spacing is 1.8 microns, not 2.5 microns as in the pirated
version (1W02) of DE007. Rob Gibbins also noted another major error.
So the CM4NEWS editor asks "Where is the official version?", "Whatever
happened to Len McCready's top-priority committment?".
ROB GIBBINS IN A CRITICAL PATH
==============================
= In the same meeting, Rob Gibbins was identified as being in a critical
path. His challenges for the next few weeks include:
- Re-coding of Phoenix DRC rules
- Design of pads and protection devices
= The former critical path Go S. Sunatori says he is very satisfied to
almost meet the major milestone of June 29 deadline for the design of
some 50 standard cells, although the half-grid implication is yet to
be solved. Best of luck to Rob from the CM4NEWS editor.
PLEASE STICK TO 3.01 SOFTWARE TOOLS
===================================
= Everybody agreed in the test-chip meeting June 25 that we should
stick to more reliable 3.01 software tools as much as possible.
MIMIMUM CMOS4 RESOLUTION IS 0.1 MICROMETER
==========================================
= Steve Sunter just corrected his typo on the preliminary cell design
philosophy document, and it now clearly states that the minimum
design resolution for CMOS4 is 0.1 micrometer.
IS RDTPS CELL FAST ENOUGH?
==========================
= Here is the preliminary results for RDTPS cell:
no load 0.1pF load CMOS3 (no load)
------- ---------- ---------------
- CK to Q (rise) ......... 1.3 ns 1.5 ns (3.4 ns)
- CK to Q (fall) ......... 1.4 ns 1.6 ns (2.7 ns)
- R to Q (fall) ......... 1.4 ns 1.6 ns (1.0 ns)
* All values were measured at VRF = 2.4 volts.
* CMOS3 shows the values for RDTP cell in CMOS3 (even without MUX!)
= If any of the CMOS4 cells do not work functionally, Go S. Sunatori
will be quietly asked to quit, or he would do an honorable thing
and submit a resignation from his position.
A STAR LOST
===========
= Our MGM star/director/producer Phil Nowe has left BNR/NT. The CM4NEWS
editor sends best wishes for Phil's new career.
CELL NAMES ARE STILL OPEN
=========================
= Any suggestions for the cell names should be directed to the CM4NEWS
editor through the CM4NEWS WRITE facility.
BUG IN SUBC FIXED
=================
= Thanks to Dave Gibson's remarkable contribution, a bug in SUBC file
mentioned below was fixed June 18, and included in the daily build
June 19. One should be able to run SCAMPER on INVD and BUFD now.
= BUFD and INVD cells cannot be simulated correctly due to bugs in
their SUBCCT files. The correct SUBCCT files will be provided in the
near future (God knows when).
JOHN KESTER WORKING ON PHOENIX DRC
==================================
= While Gerry Vandervalk is on vacation for 4 weeks, our hero John
Kester is responsible for checking the CMOS4 rule set and also
checking both standard cells and gate-array cells for any design
rule violation.
DAILY CM4 LIBRARY DISK BUILD BEGINS JUNE 17
===========================================
= Beginning June 17, the following 8 files on the CM4 Library Disk
(LIBSTAT 104) will be built daily at 6h00 in the morning
(BNRALOG REQUEST DAILY 0600 CM4BUILD):
- CM4 CAL (contains all levels)
- CM4 ICDATA (contains only design levels)
- CM4 CELLIO
- CM4SUBC MACLIB
- CM4SUBC LIBRARY (for SCAMPER)
- CM4CAP MACLIB
- HLPCM4 INDEX
- DEMO GGPLOT (contains all semi-officially released cells)
= This incredible build process is intended to prevent the CM4 Library
Disk users from being frustrated by too many disk updates in the
middle of the day. (You are out of luck if your day happens to start
at 6 a.m.) It also allows the users to see any changes next day,
not 2 months later in some other administration.
SUBCCT AND CAP FILES AVAILABLE
==============================
= Another giant step for the CMOS4 development is the availability of
SCAMPER SUBCCT files and CAP files on CM4 Library disk for all CMOS4
cells. (Oh boy, somebody must be working 80 hours a week to get
these things done!)
= In SCAMPER, just as in 3.01 version for CMOS3 simulation, one does
not have to copy SUBCCT files using HLPCOM SUBC and declare $SUBCCT
in SCAMPER file if he/she types:
- SCAMPER GSS CM4SUBC
-------
LIMITED HLPCOM FUNCTION OPERATIONAL
===================================
= Please try out HLPCOM for INDEX, SUBC and CAP functions:
- HLPCOM INDEX (for a list of semi-official cells)
- HLPCOM SUBC INVB
- HLPCOM CAP NAND2
= Because of Go S. Sunatori's oversight, HLPCOM was not operational
as announced for those who do not link to LIBSTAT 194 disk. The
situation was corrected June 18, and it should be working now for
everybody. CM4NEWS editor sends regret to Rob Koziy who could not
do HLPCOM SUBC BDI.
CM4 CELLIO COMPLETE
===================
= Preliminary CM4 CELLIO table for all 46 CMOS4 cells (including JOIN
and PJOIN cells) is now available on the CM4 Library disk.
UPGRADING THE EXTRACT PACKAGE
=============================
= In a heated discussion held in Ian McWalter's office June 12, 1985,
it was finally agreed that the present EXTRACT package be upgraded
to handle CMOS4. The critical item is the double-metal extraction.
Robert Hum and Jamie Twidale are negotiating the time frame for
for delivery. Special thanks to Perry Pavlovic and Dave Gibson for
addressing this very important issue to Peter Benedek!
CMOS4 IS NOW THE HOTTEST ITEM IN TOWN: DEADLINE IS JULY 26!
===========================================================
= Dave Gibson is now writing the minutes of the meeting held June 11,
1985. It calls for a committment from vertually everybody in 5K43,
5K44, 5K53 and 5K54 towards the first CMOS4 test chip by the end of
July. (1985 that is, believe it or not!)
= By the way, Dave Gibson was appointed the prime for the test chip.
Congratulations to Dave from the CM4NEWS editor!
= According to the minutes of the CMOS4 meeting held June 18, the
absolute deadline for the test chip is July 26. Good luck to all
involved in this incredible schedule from the CM4NEWS editor.
CMOS4 PARAMETERS
================
- Gate-Length = 1.5 um
- VDD = 5.0 V
- 1 unit = 0.1 um (minimum CMOS4 resolution)
- 1 grid = 6.6 um = 66 units
- Metal1/metal2/poly centre-line expansion = 34 units
- Standard cell height = 10 grids = 660 units
- Standard load = 0.1 pF
DESIGN LEVELS
=============
= Design levels have been defined as follows: LAYED VERPLOT
1 - N-Well ................................ green ?
2 - Device Well ........................... cyan red
* 3 - P-Guard Exclusion ..................... ? ?
6 - Polysilicon ........................... green yellow
7 - N+ Doping Exclusion ................... red ?
* 8 - P+ Doping ............................. blue ?
9 - Contact ............................... yellow black
10 - Metal1 ................................ white blue
11 - VIA ................................... blue black boundary
12 - Metal2 (not in standard cells!) ....... red green boundary
13 - Passivation Windows (PAD only) ........ white ?
15 - Boundary (should be closed polygon) ... green black boundary
16 - Schematic ............................. white black
(18)- P-Well for EXTRACT .................... blue ?
21 - Expandable Metal1 ..................... blue ?
22 - Expandable Poly ....................... red ?
23 - Expandable Metal2 ..................... ? ?
(28)- Large Contact Identification (PDRC) ... ? ?
(29)- Exclude Check Layer ................... ? ?
= Levels marked with * will be automatically generated by CM4LEVEL EXEC.
= Since CALMA is gone for good, the layers should never be called
CALMA LAYERS! Any suggestions for colours are welcome!
SOME USEFUL EXECS ON LIBSTAT 194 DISK
=====================================
------------------------------------------------------------------------
CM4GO
------------------------------------------------------------------------
SAY " THE CM4GO EXEC CREATES 2 GGPLOT, SUBCCT AND "
SAY " CAPEXT FILES FOR A CMOS4 CELL. "
SAY " * FILENAME CELLIO MUST EXIST! "
SAY " * CONVCLIO XEDIT MACRO MUST EXIST! "
SAY " "
SAY " FORMAT: CM4GO CELLNAME FILENAME ICDATA "
SAY " CM4GO CELLNAME FILENAME "
------------------------------------------------------------------------
CM4LEVEL
------------------------------------------------------------------------
SAY " THE CM4LEVEL EXEC PRODUCES "CM4 CALMA" FILE "
SAY " WITH THE FOLLOWING LAYERS: "
SAY " "
SAY " - LEVEL 1: N-WELL "
SAY " - LEVEL 2: DEVICE WELL "
SAY " * LEVEL 3: P-GUARD EXCLUSION "
SAY " - LEVEL 6: POLYSILICON "
SAY " - LEVEL 7: N+ DOPING EXCLUSION "
SAY " * LEVEL 8: P+ DOPING "
SAY " - LEVEL 9: CONTACT WINDOWS "
SAY " - LEVEL 10: METAL1 "
SAY " - LEVEL 11: VIA "
SAY " * LEVEL 12: METAL2 "
SAY " - LEVEL 13: PASSIVATION WINDOW "
SAY " "
SAY " FORMAT: CM4LEVEL CELLNAME FILENAME ICDATA "
------------------------------------------------------------------------
CM4SIM
------------------------------------------------------------------------
SAY " THE CM4SIM EXEC RUNS SCAMPER, CREATES A SUMMARY "
SAY " FILE WITH DELAY VALUES AND DELOADING FACTORS, "
SAY " AND CREATES MACRO IF TMACRO EXISTS. "
SAY " "
SAY " * CELL SCAMPER MUST EXIST! "
SAY " * RISE/FALL ORDER IS MANDATORY IN $MEASURE! "
SAY " * $CM4 MUST EXIST AT THE BOTTOM OF TMACRO FILE! "
SAY " * IF DELOAD EXISTS IN MACRO, AND CELL CAPEXT OR "
SAY " CM4CAP MACLIB EXISTS, DELOADING IS DONE. "
SAY " "
SAY " FORMAT: CM4SIM CELL CM4SUBC "
------------------------------------------------------------------------
CM4WIDE
------------------------------------------------------------------------
SAY " THE CM4WIDE EXEC WIDENS CENTRE-LINE. "
SAY " * INPUT: ICDATA, CELLIO "
SAY " * OUTPUT: FINAL CALMA/ICDATA "
SAY " "
SAY " FORMAT: CM4WIDE CELLNAME FILENAME "
------------------------------------------------------------------------
GETVIA
------------------------------------------------------------------------
say 'This EXEC calls the GETVIA program to determine cell size'
say ' and via locations. The output of this file can be '
say ' editted to include labels,iotypes,equiv,logical,'
say ' and also pins on other layers. This data can then'
say ' be used by CM4CLIO to create the CELLIO.'
say ' '
say 'Syntax: GETVIA cellname calma_fn calma_ft'
say ' where, cellname is the name of the CALMA cell.'
say ' The other argument define the library in which'
say ' the cell is contained.'
say ' '
say 'Output File: cellname VIALOC'
say ' which contains the locations of the'
say ' vias sorted by x co-ordinate and'
say ' the cell height and width.'
say ' '
say ' N.B. - lines beginning with an "*" are comments'
------------------------------------------------------------------------
CM4CLIO
------------------------------------------------------------------------
say 'This EXEC reads a file: cellname CM4CLIO and generates the'
say ' necessary commands for EDCEL to generate a cellio for'
say ' that cell.'
say ' '
say 'Syntax: CM4CLIO cellname cellio_filename'
say ' where, cellname is the name of the cell'
say ' cellio_filename = CELLIO library name.'
say ' (ft=CELLIO)'
say ' '
say 'Input File Secification:'
say ' '
say ' - file: cellname CM4CLIO'
say ' - first record that is not a comment contains the'
say ' width and height of the cell'
say ' - subsequent non-comment records contain i/o locations'
say ' - format for data is as follows:'
say ' '
say ' LABEL X Y IOTYPE LAYER EQUIV (LOGICAL)'
say ' '
say ' 4 grids INPUt Poly # optional'
say ' char INPOut Metal logical'
say ' Output Via pin equivalence'
say ' VDd number'
say ' VSs '
say ' Feedth'
say ' '
say ' N.B. - Capitilized letters indicate valid abbreviations'
say ' - lines beginning with an "*" are comments'
------------------------------------------------------------------------
PROCESS FILE
============
= Dave Smith gave us the new HONEYWL PROCESS file May 27, 1985.
The simulation results are dramatically different from those
with the old PROCESS file. So, watch out!!
= CM4 PROCESS file was created June 12, 1985. It is an exact copy
of HONEYWL PROCESS file mentioned above, but FILE HONEYWL is no
longer needed in your SCAMPER file!
DESIGN RULES (DE007)
====================
= Those who grabbed an un-official copy of DE007 are advised to destroy
it as soon as you get an official copy from the Documentation!
= Steve Sunter and John Hogeboom just signed off Dave Smith's DE007
(June 7, 1985) with some major changes.
EDGE RULES
==========
- Metal1 to edge: inset 10 units
- Poly to edge: inset 10 units
- Device well to edge: inset 16 units
- Doping mask to edge: On-line or inset 20 units
- N-well to edge: On-line or inset 54 units
......................
HONEYWELL
=========
= The deal is all set according to Roly Reader (June 3, 1985).
PERSONNEL
=========
= Yachin Afek is now manager in charge of CMOS4 MACRO design.
= Go S. Sunatori is too overloaded these days!
MEETINGS/PRESENTATIONS
======================
= The weekly CMOS4 test-chip meeting is held at 9h00, Tuesday.
[Go to]
[Mail any comments or suggestions to]
[Credit]
- Copyright © 2010 by HyperInfo Canada Inc.
- All rights reserved.
- Produced in Canada.
- ISBN 978-0-929105-40-6
- 2019-03-18T21:49:02,0Z