DaniPopes:interp-perfs - Branch - bluealloy/revm

perf: interpreter improvements

#2760Merged

Comparing

DaniPopes:interp-perfs

(

f47751d

) with

main

(

a5dd047

)

+33%

Improvements: 132

Untouched: 39

Ignored: 6

Benchmarks

Improved

burntpix

bins/revme/benches/evm.rs::benches::evm

+33%

614.6 ms460.8 ms

snailtracer

bins/revme/benches/evm.rs::benches::evm

+22%

212.5 ms173.5 ms

JUMPI_50

bins/revme/benches/evm.rs::benches::evm

+12%

25.8 µs23.1 µs

NOT_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23 µs

BYTE_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.1 µs23.5 µs

AND_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.8 µs23.2 µs

DUP8_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.8 µs23.2 µs

DUP14_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.1 µs

XOR_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.8 µs23.2 µs

DUP13_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.1 µs

DUP4_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.1 µs

DUP16_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.1 µs23.4 µs

OR_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.8 µs23.2 µs

SLT_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.8 µs24.1 µs

DUP5_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.6 µs23 µs

analysis

bins/revme/benches/evm.rs::benches::evm

+11%

49 µs44.1 µs

DUP11_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.9 µs23.3 µs

DUP15_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.6 µs23.1 µs

DUP10_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.1 µs

DUP6_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.1 µs

DUP12_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.1 µs

EQ_50

bins/revme/benches/evm.rs::benches::evm

+11%

26 µs23.4 µs

ISZERO_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.8 µs23.2 µs

SGT_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.7 µs24.1 µs

ADD_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.2 µs

DUP9_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.6 µs23.1 µs

DUP1_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.6 µs23.1 µs

DUP2_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.6 µs23.1 µs

SIGNEXTEND_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.8 µs24.2 µs

LT_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.2 µs23.6 µs

GT_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.2 µs23.6 µs

DUP7_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.7 µs23.2 µs

DUP3_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.6 µs23.1 µs

SUB_50

bins/revme/benches/evm.rs::benches::evm

+11%

25.9 µs23.3 µs

SHR_50

bins/revme/benches/evm.rs::benches::evm

+11%

27.1 µs24.4 µs

SHL_50

bins/revme/benches/evm.rs::benches::evm

+11%

26.9 µs24.3 µs

SAR_50

bins/revme/benches/evm.rs::benches::evm

+10%

27.1 µs24.6 µs

analysis-inspector

bins/revme/benches/evm.rs::benches::evm

+10%

433.8 ms394.9 ms

DELEGATECALL_50

bins/revme/benches/evm.rs::benches::evm

+10%

99.6 µs90.7 µs

MUL_50

bins/revme/benches/evm.rs::benches::evm

+10%

26.7 µs24.3 µs

CALL_50

bins/revme/benches/evm.rs::benches::evm

+10%

105.5 µs96.2 µs

STATICCALL_50

bins/revme/benches/evm.rs::benches::evm

+9%

104.2 µs95.4 µs

SWAP16_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.9 µs21 µs

SDIV_50

bins/revme/benches/evm.rs::benches::evm

+9%

29.6 µs27.1 µs

MOD_50

bins/revme/benches/evm.rs::benches::evm

+9%

28.2 µs25.9 µs

SWAP10_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.9 µs21 µs

MCOPY_COLD_50

bins/revme/benches/evm.rs::benches::evm

+9%

26.4 µs24.2 µs

MCOPY_50

bins/revme/benches/evm.rs::benches::evm

+9%

22 µs20.2 µs

DIV_50

bins/revme/benches/evm.rs::benches::evm

+9%

28.2 µs25.9 µs

SWAP6_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.7 µs20.9 µs

ADDMOD_50

bins/revme/benches/evm.rs::benches::evm

+9%

33.5 µs30.8 µs

SWAP12_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.7 µs20.9 µs

SWAP9_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.6 µs20.8 µs

SWAP4_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.7 µs20.8 µs

SMOD_50

bins/revme/benches/evm.rs::benches::evm

+9%

29.3 µs27 µs

SWAP11_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.6 µs20.8 µs

SWAP5_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.6 µs20.8 µs

SWAP15_50

bins/revme/benches/evm.rs::benches::evm

+9%

23 µs21.2 µs

SWAP8_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.6 µs20.8 µs

SWAP14_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.6 µs20.8 µs

POP_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.3 µs20.6 µs

PUSH2_50

bins/revme/benches/evm.rs::benches::evm

+9%

19.6 µs18.1 µs

SWAP7_50

bins/revme/benches/evm.rs::benches::evm

+9%

22.7 µs21 µs

SWAP13_50

bins/revme/benches/evm.rs::benches::evm

+8%

22.6 µs20.9 µs

SWAP1_50

bins/revme/benches/evm.rs::benches::evm

+8%

22.6 µs20.8 µs

SWAP3_50

bins/revme/benches/evm.rs::benches::evm

+8%

22.6 µs20.8 µs

SWAP2_50

bins/revme/benches/evm.rs::benches::evm

+8%

22.6 µs20.8 µs

PUSH18_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.1 µs18.5 µs

PUSH9_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.8 µs18.2 µs

MULMOD_50

bins/revme/benches/evm.rs::benches::evm

+8%

31.7 µs29.3 µs

PUSH8_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.7 µs18.2 µs

PUSH16_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.9 µs18.5 µs

PUSH21_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.2 µs18.7 µs

PUSH5_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.7 µs18.2 µs

CLZ_50

bins/revme/benches/evm.rs::benches::evm

+8%

24.3 µs22.5 µs

PUSH28_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.3 µs18.9 µs

REVERT_50

bins/revme/benches/evm.rs::benches::evm

+8%

120.7 µs111.8 µs

PUSH10_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.8 µs18.3 µs

PUSH7_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.8 µs18.4 µs

RETURN_50

bins/revme/benches/evm.rs::benches::evm

+8%

119.3 µs110.6 µs

PUSH11_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.9 µs18.4 µs

PUSH4_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.6 µs18.2 µs

PUSH14_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.9 µs18.5 µs

PUSH13_50

bins/revme/benches/evm.rs::benches::evm

+8%

20 µs18.5 µs

PUSH24_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.2 µs18.7 µs

PUSH15_50

bins/revme/benches/evm.rs::benches::evm

+8%

20 µs18.6 µs

PUSH26_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.3 µs18.8 µs

PUSH17_50

bins/revme/benches/evm.rs::benches::evm

+8%

20 µs18.5 µs

PUSH32_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.5 µs19 µs

PUSH19_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.1 µs18.7 µs

PUSH3_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.6 µs18.2 µs

PUSH0_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.3 µs17.9 µs

PUSH25_50

bins/revme/benches/evm.rs::benches::evm

+8%

20.2 µs18.8 µs

GASLIMIT_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.3 µs18 µs

PUSH6_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.7 µs18.3 µs

PUSH1_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.3 µs18 µs

PUSH12_50

bins/revme/benches/evm.rs::benches::evm

+8%

19.8 µs18.4 µs

EXP_50

bins/revme/benches/evm.rs::benches::evm

+8%

33.3 µs30.9 µs

EXTCODECOPY_50

bins/revme/benches/evm.rs::benches::evm

+7%

51 µs47.4 µs

PUSH30_50

bins/revme/benches/evm.rs::benches::evm

+7%

20.4 µs19 µs

PUSH31_50

bins/revme/benches/evm.rs::benches::evm

+7%

20.5 µs19.1 µs

PUSH23_50

bins/revme/benches/evm.rs::benches::evm

+7%

20.2 µs18.8 µs

TIMESTAMP_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs17.9 µs

NUMBER_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18 µs

PUSH20_50

bins/revme/benches/evm.rs::benches::evm

+7%

20 µs18.6 µs

GASPRICE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.5 µs18.2 µs

PUSH22_50

bins/revme/benches/evm.rs::benches::evm

+7%

20.1 µs18.7 µs

CHAINID_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18 µs

PUSH29_50

bins/revme/benches/evm.rs::benches::evm

+7%

20.3 µs19 µs

ORIGIN_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18 µs

CODESIZE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18.1 µs

COINBASE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.4 µs18.1 µs

PC_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.4 µs18.1 µs

MSTORE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.8 µs18.5 µs

CALLVALUE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.2 µs18 µs

PUSH27_50

bins/revme/benches/evm.rs::benches::evm

+7%

20.2 µs18.9 µs

GAS_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.2 µs18 µs

RETURNDATASIZE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18 µs

CALLDATASIZE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18.1 µs

MSIZE_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.4 µs18.1 µs

DIFFICULTY_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.5 µs18.3 µs

CALLER_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18.1 µs

ADDRESS_50

bins/revme/benches/evm.rs::benches::evm

+7%

19.3 µs18.1 µs

EXTCODESIZE_50

bins/revme/benches/evm.rs::benches::evm

+6%

39.1 µs36.7 µs

EXTCODEHASH_50

bins/revme/benches/evm.rs::benches::evm

+6%

39.1 µs36.8 µs

JUMP_50

bins/revme/benches/evm.rs::benches::evm

+6%

18.8 µs17.7 µs

SELFBALANCE_50

bins/revme/benches/evm.rs::benches::evm

+6%

21.8 µs20.6 µs

MSTORE_COLD_50

bins/revme/benches/evm.rs::benches::evm

+6%

24 µs22.7 µs

MLOAD_50

bins/revme/benches/evm.rs::benches::evm

+5%

60.3 µs57.6 µs

CALLDATALOAD_50

bins/revme/benches/evm.rs::benches::evm

+5%

60.2 µs57.5 µs

CALLDATACOPY_50

bins/revme/benches/evm.rs::benches::evm

+5%

59.1 µs56.5 µs

MSTORE8_50

bins/revme/benches/evm.rs::benches::evm

+4%

56.9 µs55 µs

Passed

RETURNDATACOPY_50

bins/revme/benches/evm.rs::benches::evm

+3%

109.1 µs106 µs

JUMPDEST_50

bins/revme/benches/evm.rs::benches::evm

+3%

15.9 µs15.4 µs

KECCAK256_50

bins/revme/benches/evm.rs::benches::evm

+2%

146.3 µs143.9 µs

CREATE_50

bins/revme/benches/evm.rs::benches::evm

+2%

418.7 µs412.4 µs

transact_1000tx_commit_inner_every_40

bins/revme/benches/evm.rs::benches::evm

+1%

2 ms2 ms

transfer_finalize

bins/revme/benches/evm.rs::benches::evm

+1%

16.3 µs16.1 µs

blake2/2_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

+1%

3.1 µs3 µs

blake2/10_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

+1%

4.3 µs4.3 µs

ecrecover precompile

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

+1%

203.3 µs202 µs

ecpairing precompile

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

+1%

4.7 ms4.7 ms

blake2/64_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

9.4 µs9.4 µs

blake2/512_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

51.7 µs51.7 µs

g2_msm (size 2)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

2.7 ms2.7 ms

blake2/1024_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

100.1 µs100 µs

bn128 mul precompile

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

335.3 µs335.3 µs

map_fp_to_g1

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

243.7 µs243.6 µs

g1_msm (size 1)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

714 µs714 µs

g1_msm (size 128)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

47 ms47 ms

g2_msm (size 256)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

162.3 ms162.3 ms

pairing (16 pairs)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

36.5 ms36.5 ms

blake2/200K_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

18.9 ms18.9 ms

blake2/100K_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

9.4 ms9.4 ms

blake2/4_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

3.4 µs3.4 µs

blake2/12_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

4.5 µs4.5 µs

kzg precompile

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

7.5 ms7.5 ms

pairing (8 pairs)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

19.4 ms19.4 ms

pairing (2 pairs)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

6.5 ms6.5 ms

pairing (1 pairs)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

4.4 ms4.4 ms

g2_msm (size 1)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

1.3 ms1.3 ms

g1_msm (size 256)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

89.2 ms89.2 ms

map_fp2_to_g2

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

860.7 µs860.7 µs

g2_msm (size 128)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

88.2 ms88.2 ms

g2_add

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

42.7 µs42.7 µs

g1_msm (size 2)

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

1.4 ms1.4 ms

g1_add

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

28.1 µs28.1 µs

bn128 add precompile

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

21.4 µs21.5 µs

transact_commit_1000txs

bins/revme/benches/evm.rs::benches::evm

2.7 ms2.7 ms

blake2/compress_12_rounds

crates/precompile/bench/main.rs::benches::benchmark_crypto_precompiles::Crypto Precompile benchmarks

-1%

3.3 µs3.3 µs

transfer

bins/revme/benches/evm.rs::benches::evm

-1%

12.6 µs12.7 µs

Ignored

evm-build

bins/revme/benches/evm.rs::benches::evm

Ignored

-1%

10.6 µs10.7 µs

LOG4_50

bins/revme/benches/evm.rs::benches::evm

Ignored

-2%

61.2 µs62.5 µs

LOG0_50

bins/revme/benches/evm.rs::benches::evm

Ignored

+5%

35.5 µs33.7 µs

LOG2_50

bins/revme/benches/evm.rs::benches::evm

Ignored

+3%

57 µs55.3 µs

LOG3_50

bins/revme/benches/evm.rs::benches::evm

Ignored

+4%

61 µs58.3 µs

LOG1_50

bins/revme/benches/evm.rs::benches::evm

Ignored

+25%

53.8 µs43.1 µs

Commits

Click on a commit to change the comparison range

Base

main

a5dd047

-0.45%

perf: move interpreter to first field in InstructionContext It's more accessed than host, and when calling a function that takes interpreter as the first argument, no registers have to change.

fa89252

5 months ago

by DaniPopes

+25.98%

perf: mark halt as cold This function massively bloats all callsites because it can drop the previous action.

865b66a

5 months ago

by DaniPopes

+4.1%

perf: use push_slice again We want to write directly into the maybe-uninitialized memory of the stack. We cannot do this using existing APIs, so we must add `push_slice` to `StackTr`.

6898a9f

5 months ago

by DaniPopes

+2.33%

perf: reintroduce resize_memory fn Reintroduce https://github.com/bluealloy/revm/pull/1361.

b4f6a94

5 months ago

by DaniPopes

+0.54%

perf: workaround rust#144329 The compiler does not elide the "not Some" check after inlining `popn_top`. See https://github.com/rust-lang/rust/issues/144329.

b4e4d46

5 months ago

by DaniPopes

+0.89%

perf: remove RefCell overhead in memory Move runtime checks to debug mode only.

f47751d

5 months ago

by DaniPopes

Home Terms Privacy Docs