forked from psoberoi/stepanov-conversations-course
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
added lec02 notes; fixed types on lec01
- Loading branch information
Showing
2 changed files
with
141 additions
and
8 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,133 @@ | ||
| Lecture 2 | ||
|
|
||
| Last lecture: Programming is about bits | ||
|
|
||
| This was not always self evident | ||
| - bits were invisible for many years | ||
| - bytes for decades | ||
| - the abstract machine model took along time to emerge | ||
|
|
||
| Bits not self-evident | ||
| - it was clear that computers must operate on numbers | ||
| - but how should numbers be represented? | ||
| - Knuth, 1968 - support both binary and decimal | ||
| - many architectures supported BCD, including x86 | ||
| - word sizes were not powers of two - all kinds | ||
| - cray: 60 bit words | ||
|
|
||
| Popular word sizes: 36 bits and 60 bits | ||
| - divisible by 6 | ||
| - 6 bits enough to encode English characters | ||
| - but 6-bit "bytes" were not addressable | ||
| - IBM STRETCH: bit addressable | ||
|
|
||
| Mid-60s: IBM 360 | ||
| - first architecture that spanned a family of machines | ||
| - great team of people | ||
| - c machine model a result of IBM 360 | ||
| - byte addressable | ||
| - full-blown pointers | ||
| - IBM Principles of Operation | ||
| - Brooks and Blaauw | ||
|
|
||
| Revolutionary | ||
| - lots of clones | ||
| - copied widely internationally | ||
| - including in the USSR (allegedly aided by the CIA) | ||
| - slowly the "exotic" word sizes disappeared | ||
|
|
||
| However, no access to this low level abstract machine | ||
| - all serious programming done in assembly | ||
| - need access to bits | ||
| - performance | ||
|
|
||
| 1953: John Backus: Formula Translation | ||
| - Wrote memo to mgmt, asked for large team (10 best people) | ||
| - Due diligence: mgmt asked von Neumann | ||
| - terrible idea | ||
| - why waste valuable computer time on this? | ||
| - just use "girls" | ||
| - IBM goes ahead anyway | ||
| - History very well documented by Paul McJones | ||
|
|
||
| FORTRAN compiler | ||
| - First | ||
| - They had to invent all the basic ideas | ||
| - "compiler", "basic block", "optimization", "frequency" hints | ||
| - conditional statement, do statement, arrays | ||
| - primitives derived from hardware instructions | ||
| - Goal: as fast as hand-written code | ||
| - took 4 years (released in 1957) | ||
| - very efficient code, right from the start | ||
| - very successful from the start - physicists embraced it | ||
|
|
||
| FORTRAN's success led to the notion of "portability" | ||
| - Initially only ran on IBM 704 | ||
| - Other companies realized that the language was not machine specific | ||
| - They implemented their own implementations to attract physicists | ||
| - Physicists could freely move code from machine to machine | ||
| - They were in a state of bliss | ||
| - No other portable software for quite a while | ||
|
|
||
| But FORTRAN was widely maligned | ||
| - Computer scientists thought it was a horrible language | ||
| - Dijkstra, Hoare - wrote nasty things | ||
| - Physicists used it anyway, and still do | ||
| - It was, and remains, a great achievement | ||
|
|
||
| FORTRAN v/s Smalltalk style | ||
| - FORTRAN - arrays | ||
| - Smalltalk - lots of small objects, all alike | ||
| - Both styles possible in most languages | ||
| - with the resulting performance characteristics | ||
|
|
||
| FORTRAN supported both ints and floating point | ||
| - von Neumann - no need for floating point - any mathematician should be able to normalize the result in their head | ||
| - but very few mathematicians were like von Neumann | ||
| - hardware manufacturers supported floating point | ||
|
|
||
| Integer representation | ||
| - decimal or binary | ||
| - even after binary became common, how to represent sign | ||
| - ints were always signed | ||
| - Two's Complement | ||
| - Signed Magnitude | ||
| - One's Complement | ||
| - signed magnitude and 1's complement had two zeros | ||
| - many violent arguments about zeros | ||
| - integer semantics still not quite understood, even today | ||
| - integer sizes were different across machines, of course | ||
| - floating point eventually standardized | ||
| - Kahan | ||
| - no such luck for integers... | ||
|
|
||
| Languages | ||
| - FORTRAN -> II -> IV -> 77 -> 90 -> ... | ||
| - Algol 60 | ||
| - Int, Float | ||
| - PL/1 | ||
| - universal language | ||
| - systems programming needs, in a HLL | ||
| - large and unwieldy | ||
| - Multics, big failure | ||
| - but, led to Ken and Dennis | ||
| - some access to bits | ||
| - PL/S | ||
| - BLISS | ||
|
|
||
| C | ||
| - UNIX initial version: assembly | ||
| - not portable... | ||
| - BCPL, B, C | ||
| - Dennis was a genius | ||
| - C ugly, not easy to parse | ||
| - unlike Wirth languages (Pascal, Modula-2) | ||
| - But: came up with a machine model roughly matching IBM 360 | ||
|
|
||
| Revolutionary | ||
| - At that time, designers were convinced that the job of the language was to hide the machine | ||
| - as late as ~1985, in a a Wirth/Knuth interview: | ||
| - Wirth: A pointer type should not have < defined on it | ||
| - obviously | ||
| - Knuth: So if you have a list of pointers to records, you cannot use binary search? | ||
|
|