README.md

RISC-V instructions used in real-world binaries

• target: 856 binaries in /bin

name	frequency (%)	cumulative (%)
ld	14.101200487156945	14.101200487156945
mv	10.010240605850349	24.111441093007294
addi	9.590532965163105	33.7019740581704
sd	8.400851481380371	42.10282553955077
li	8.125153393115912	50.22797893266669
jal	7.438872769547847	57.66685170221454
auipc	4.532127806046617	62.198979508261154
j	4.015768635289742	66.21474814355089
beqz	3.152443028634279	69.36719117218517
lw	2.5248689806228177	71.89206015280799
lbu	2.1917389047239393	74.08379905753192
add	1.9912981522204478	76.07509720975237
lui	1.9702957756804564	78.04539298543283
bnez	1.9151769290900005	79.96056991452282
slli	1.8403066904016967	81.80087660492453
beq	1.4300734866026883	83.23095009152722
sw	1.351695770780374	84.5826458623076
bne	1.2768774597481594	85.85952332205576
sb	1.2157397257473888	87.07526304780315
ret	1.1996138281837543	88.2748768759869
jalr	1.0494343260783947	89.3243112020653
addiw	1.0396908095176973	90.364002011583
or	0.9887544995904088	91.35275651117341
andi	0.8893272002680223	92.24208371144144
sext.w	0.7243199920012526	92.9664037034427
srli	0.6562145106925417	93.62261821413523
bltu	0.5048029069290707	94.1274211210643
bgeu	0.47227259073732414	94.59969371180162
nop	0.3855817292447637	94.98527544104638
addw	0.38516630799605184	95.37044174904244
and	0.3162205427979008	95.68666229184035
sub	0.30403642640102113	95.99069871824138
bltz	0.2792622137505661	96.26996093199195
srliw	0.27517881170356845	96.54513974369551
slliw	0.24515046439610944	96.79029020809162
bge	0.23518507512303197	97.02547528321465
xor	0.22665949813242184	97.25213478134707
subw	0.22046594496980806	97.47260072631687
blt	0.20866420494958368	97.68126493126645
lhu	0.2025603450111236	97.88382527627758
bgez	0.17207409019087994	98.05589936646845
ori	0.13585691041681527	98.19175627688527
blez	0.13491277121519732	98.32666904810047
fld	0.13308586186006657	98.45975490996054
sh	0.12211496433726597	98.58186987429781
jr	0.08760195582212173	98.66947183011993
seqz	0.081568906323783	98.75104073644371
sraiw	0.07992610411296777	98.83096684055668
not	0.07841548139037904	98.90938232194706
lwu	0.07510155279270002	98.98448387473977
snez	0.07374199234237018	99.05822586708214

show_freq_asm_inst.py

import sys
from typing import List
from collections import defaultdict


def show_frequently_used_insn(hist_sorted):
    upper_bound = 99.0
    cumu = 0
    result = list()
    for insn, value in hist_sorted:
        cumu += value
        result.append((insn, value))
        if cumu >= upper_bound:
            break

    cumu = 0
    for insn, value in result:
        cumu += value
        print(f"|{insn}|{value}|{cumu}|")


def _main() -> None:
    with open("concat.asm") as fin:
        asm_lines: List[str] = [l for l in fin.readlines() if (not l.strip().endswith(":"))]

    hist = defaultdict(int)
    n_ops = 0

    for asm_line in asm_lines:
        op_and_oprands = asm_line.split()
        if len(op_and_oprands) < 3:
            continue
        op_name = op_and_oprands[2]
        hist[op_name] += 1
        n_ops += 1

    for k, v in hist.items():
        hist[k] = (v / n_ops) * 100.0

    hist_sorted = sorted(hist.items(), key=lambda k_v: k_v[1], reverse=True)
    show_frequently_used_insn(hist_sorted)


if __name__ == "__main__":
    _main()