8ef7c31089
No change to linuxppc and qemuppc. They continue to run ego without any descr file. I copied m68020.descr to powerpc.descr and changed some numbers. My numbers are guesses; I know little about PowerPC cycle counts, and almost nothing about ego. This powerpc.descr causes most of the example programs to shrink in size (without descr -> with descr): 65429 -> 57237 hilo_b.osxppc -8192 36516 -> 32420 hilo_c.osxppc -4096 55782 -> 51686 hilo_mod.osxppc -4096 20096 -> 20096 hilo_p.osxppc 0 8813 -> 8813 mandelbrot_c.osxppc 0 93355 -> 89259 paranoia_c.osxppc -4096 92751 -> 84559 startrek_c.osxppc -8192 (Each file has 2 Mach segments, then a symbol table. Each segment takes a multiple of 4096 bytes. When the code shrinks, we lose a multiple of 4096 bytes.) I used "ack -mosxppc -O6 -c.so" to examine the assembly code for hilo.mod and mandelbrot.c, both without and with descr. This reveals optimizations made only with descr, from 2 ego phases: SP (stack pollution) and RA (register allocation). In hilo.mod, SP deletes some instructions that remove items from the stack. These items get removed when the function returns. In both hilo.mod and mandelbrot.c, RA moves some values into local variables, so ncg can make them into register variables. This shrinks code size, probably because register variables get preserved across function calls. More values stay in registers, and ncg emits shorter code. I believe that the ego descr file uses (time,space) tuples but the ncg table uses (space,time) tuples. This is confusing. Perhaps I am wrong, and some or all tuples are backwards. My time values are the cycle counts in latency from the MPC7450 Reference Manual (but not including complications like "store serialization"). In powerpc.descr, I give the cost for saving and restoring registers as if I was using chains of stw and lwz instructions. Actually ncg uses single stmw and lmw instructions with at least 2 instructions. The (time,space) for stmw and lmw would be much less than the (time,space) for chains of stw and lwz. But this ignores the pipeline of the MPC7450. The chains of stw and lwz may run faster than stmw and lmw in the pipeline, because the throughput may be better than the latency. By using the wrong values for (time,space), I'm trying to tell ego that stmw and lmw are not better than chains of stw and lwz.
119 lines
2.2 KiB
Plaintext
119 lines
2.2 KiB
Plaintext
wordsize: 4
|
|
pointersize: 4
|
|
%%RA
|
|
general registers: 19
|
|
address registers: 0
|
|
floating point registers: 0
|
|
use general as pointer: yes
|
|
|
|
register score parameters:
|
|
local variable:
|
|
(2 cases)
|
|
pointer,general
|
|
(1 size)
|
|
default -> (3,4)
|
|
general,general
|
|
(1 size)
|
|
default -> (3,4)
|
|
address of local variable:
|
|
(2 cases)
|
|
pointer,general
|
|
(1 size)
|
|
default -> (0,0)
|
|
general,general
|
|
(1 size)
|
|
default -> (0,0)
|
|
constant:
|
|
(2 sizes)
|
|
fitbyte -> (-1,-1)
|
|
default -> (-1,-1)
|
|
double constant:
|
|
(1 size)
|
|
default -> (-1,-1)
|
|
address of global variable:
|
|
(1 size)
|
|
default -> (2,8)
|
|
address of procedure:
|
|
(1 size)
|
|
default -> (-1,-1)
|
|
|
|
opening cost parameters:
|
|
local variable:
|
|
(2 cases)
|
|
pointer
|
|
(1 size)
|
|
default -> (3,4)
|
|
general
|
|
(1 size)
|
|
default -> (3,4)
|
|
address of local variable:
|
|
(2 cases)
|
|
pointer
|
|
(1 size)
|
|
default -> (1,4)
|
|
general
|
|
(1 size)
|
|
general -> (1,4)
|
|
constant:
|
|
(2 sizes)
|
|
fitbyte -> (1000,1000)
|
|
default -> (1000,1000)
|
|
double constant:
|
|
(1 size)
|
|
default -> (1000,1000)
|
|
address of global variable:
|
|
(1 size)
|
|
default -> (2,8)
|
|
address of procedure:
|
|
(1 size)
|
|
default -> (1000,1000)
|
|
|
|
register save costs:
|
|
(21 cases)
|
|
0 -> (0,0)
|
|
1 -> (6,8)
|
|
2 -> (12,16)
|
|
3 -> (18,24)
|
|
4 -> (24,32)
|
|
5 -> (30,40)
|
|
6 -> (36,48)
|
|
7 -> (42,56)
|
|
8 -> (48,64)
|
|
9 -> (54,72)
|
|
10 -> (60,80)
|
|
11 -> (66,88)
|
|
12 -> (72,96)
|
|
13 -> (78,104)
|
|
14 -> (84,112)
|
|
15 -> (90,120)
|
|
16 -> (96,128)
|
|
17 -> (102,136)
|
|
18 -> (108,144)
|
|
19 -> (114,152)
|
|
0 -> (0,0)
|
|
%%UD
|
|
access costs of global variables:
|
|
(1 size)
|
|
default -> (5,12)
|
|
access costs of local variables:
|
|
(1 size)
|
|
default -> (3,4)
|
|
%%SR
|
|
overflow harmful?: no
|
|
array bound harmful?: yes
|
|
reduce sli if shift count larger than: 0
|
|
%%CS
|
|
#include "em_mnem.h"
|
|
first time then space:
|
|
addressing modes: op_ads op_adp op_lof op_ldf op_loi op_dch op_lpb -1
|
|
op_ads op_adp op_lof op_ldf op_loi op_dch op_lpb -1
|
|
cheap operations: op_cii op_ciu op_cui op_cuu op_cmi op_cmu op_cmp -1
|
|
op_cii op_ciu op_cui op_cuu op_cmi op_cmu op_cmp -1
|
|
lexical tresholds: 1 1
|
|
indirection limit: 8
|
|
do not eliminate sli if index on shiftcounts: -1
|
|
-1
|
|
forbidden operators: -1 -1
|
|
%%SP
|
|
global stack pollution allowed?: yes
|