Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Annotation of /sml/branches/SMLNJ/src/MLRISC/hppa/README.hppa
ViewVC logotype

Annotation of /sml/branches/SMLNJ/src/MLRISC/hppa/README.hppa

Parent Directory Parent Directory | Revision Log Revision Log


Revision 410 - (view) (download)

1 : monnier 409 Additions to the instruction set
2 :     =====================================
3 :     1. SHnADD[O] instructions, which performs a*(2^n)+b
4 :     2. Branch on bit instruction (BB)
5 :     3. Single precision floating points + lots of other conversion ops
6 :     4. Loads and stores with scaled indexed and auto increment addressing modes
7 :    
8 :     New optimizations in the HPPA backend
9 :     =====================================
10 :    
11 :     1. An expressions such as a*n+b where n = 2, 4, and 8 now emits SHnADD[O]
12 :     instructions.
13 :    
14 :     2. An expression such as
15 :    
16 :     BCC(T.CMP(T.NE,0,T.ANDB(mask,e))
17 :    
18 :     where mask is a power of 2 now emits BB instructions.
19 :    
20 :     SML/NJ: this improves boxity testing code by replacing 3 instructions with 1
21 :     instruction.
22 :    
23 :     3. Strength reduction for multiplication with small constants.
24 :    
25 :     SML/NJ: benefits array indexing and multiway branches
26 :    
27 :     4 Addressing modes computation now takes into account of scaled
28 :     indexed mode. Unfortunately, due to tagging this optimizations
29 :     is not used in SML/NJ.
30 :    
31 :     5 Propagation of annotations
32 :    
33 :     6 Delay slot filling has been added
34 :    
35 :     7 NOTB folding has been implemented. This combines NOTB with other
36 :     bit ops such as ANDB, ORB, XORB.
37 :    
38 :     8. COND expressions now generate the composite instruction sequence:
39 :    
40 :     COMCLR,cc r1, r2, t
41 :     LDO i(b), t

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0