TDEF fuses TCLASS+METHOD+DEF for normal method definitions.
SDEF fuses SCLASS+METHOD+DEF for singleton method definitions.
Saves 4 bytes per method definition (8 bytes -> 4 bytes).
Falls back to unfused instructions if irep index exceeds 255.
Co-authored-by: Claude <noreply@anthropic.com>
Fuses MOVE+LOADI_0+GETIDX pattern into single instruction.
Saves 4 bytes per arr[0] access (7 bytes -> 3 bytes).
Co-authored-by: Claude <noreply@anthropic.com>
fuse MOVE+ADDI+MOVE and MOVE+SUBI+MOVE patterns into single instructions.
ADDILV/SUBILV add/subtract an immediate to a local variable in-place.
BBB format: a=local, b=working space for method call, c=immediate.
saves 5 bytes per instance (9->4 bytes), 40 occurrences in stdlib.
Co-authored-by: Claude <noreply@anthropic.com>
Change OP_MATCHERR from Z format (unconditional) to B format
(conditional on register). This allows fusing JMPIF + MATCHERR
sequence into a single MATCHERR instruction for simple patterns.
Before: JMPIF R2 target (4 bytes) + MATCHERR (1 byte) = 5 bytes
After: MATCHERR R2 (2 bytes)
Saves 3 bytes per pattern match with raise_on_fail.
Co-authored-by: Claude <noreply@anthropic.com>
Replace 4-instruction sequence (GETCONST + STRING + SEND + RAISEIF)
with single OP_MATCHERR instruction that raises NoMatchingPatternError
with "pattern not matched" message.
Bump RITE binary format version from 0300 to 0400 due to opcode
number shift.
Co-authored-by: Claude <noreply@anthropic.com>
Added error handling for file descriptors larger than FD_SETSIZE in mrb_hal_io_fdset_set and mrb_hal_io_fdset_isset functions, for posix hal.
I actually don't know how to fix this on windows, or if it needs fixing.
The purpose is to avoid using the `MRB_TT_CPTR` object.
The reasons are as follows:
- The `MRB_WORD_BOXING` setting involves object creation.
- If object creation fails, the `ary_set_t` data leaks memory.
wrap hash and eql callbacks with mrb_protect_error() to catch exceptions
during khash table rebuild. when an exception occurs (e.g., SystemStackError
from infinite recursion), return a safe default value and store the exception
in mrb->exc for later processing. this prevents memory leaks from orphaned
allocations when exceptions propagate through khash rebuild.
Co-authored-by: Claude <noreply@anthropic.com>
Function was only called once and contained just 2 lines of code.
Inlining directly reduces code size and improves clarity.
Co-authored-by: Claude <noreply@anthropic.com>
Add CPython-style parsing for base-10 string to integer conversion:
- Parse 9 digits at a time into decimal-base array
- Convert decimal-base to binary in single pass
- Use memory pool for temporary decimal buffer
- Use realloc for result buffer to reduce allocations
Also add digit_pairs lookup table for faster to_s output.
Performance: 2-5x faster for to_i, 60% fewer allocations.
Co-authored-by: Claude <noreply@anthropic.com>
- Add mrb_gc_protect() after arena_restore to prevent result from being collected before returning to caller
- Add comment to suspend_task_internal explaining why WAITING and DORMANT tasks can also be suspended
- Move argc/argv cast at the beginning of function with comment
This improves to_s performance for medium-sized bigints (40-50 limbs,
~800-1000 digits) by approximately 5x by enabling the divide-and-conquer
algorithm earlier.
Benchmark results:
40 limbs (772 digits): 88 us -> 18 us (5x faster)
50 limbs (964 digits): 134 us -> 25 us (5.4x faster)
Co-authored-by: Claude <noreply@anthropic.com>
Benchmarks show the previous threshold of 50 was too low, causing
Toom-3's setup overhead to outweigh its asymptotic benefits for
medium-sized numbers. Raising to 100 limbs provides:
- 2x faster at 300 limbs (192 -> 96 us)
- 2.5x faster at 120 limbs (42 -> 17 us)
- 2.6x faster at 80 limbs (31 -> 12 us)
Co-authored-by: Claude <noreply@anthropic.com>
When multiplying numbers where one is significantly larger than the other
(at least 2x size difference), split the larger number into chunks matching
the smaller number's size, multiply each chunk, and combine results. This
avoids pathological performance when Toom-3 pads asymmetric operands with
zeros.
Benchmarks show 6-16x speedup for size ratios from 10:1 to 40:1, with no
regression for symmetric cases.
Co-authored-by: Claude <noreply@anthropic.com>
mpz_init_heap() already allocates the requested size, so immediately
calling mpz_realloc() with the same size is a no-op. Remove these
redundant calls from mpz_and, mpz_or, mpz_xor, mpz_mod_2exp, and
mpz_abs.
Co-authored-by: Claude <noreply@anthropic.com>
When the output and input are the same variable, avoid unnecessary
heap allocations by modifying in place:
- mpz_neg: just flip the sign
- mpz_abs: just make sign positive
- ulshift: use mpn_lshift in-place (safe since it processes high-to-low)
Co-authored-by: Claude <noreply@anthropic.com>
Add mpz_sqr_toom3() that performs Toom-3 squaring with reduced memory
and computation:
- Only evaluates x (not y), reducing evaluation buffer from 6 to 3
- Uses recursive squaring instead of multiplication for all 5 products
- Simplifies interpolation since squared values are always positive
The specialized squaring uses the same Toom-3 structure but avoids
redundant computation when both operands are the same number.
Co-authored-by: Claude <noreply@anthropic.com>
The mpn_divexact_3 function used ap[i] in the borrow computation after
writing to rp[i]. When rp == ap (in-place operation), this read the
modified value instead of the original input, causing incorrect borrow
propagation.
This bug caused Toom-3 multiplication to produce wrong results for
certain input patterns where t6 - t5 had non-zero values followed by
zeros. The corrupted r3 coefficient then propagated errors to the final
result.
Fix by saving the original ap[i] value before writing rp[i].
Co-authored-by: Claude <noreply@anthropic.com>
move in-place optimization directly into mpz_sub() so callers
just use mpz_sub(ctx, x, x, y) and get automatic optimization.
remove separate mpz_sub_inplace() function.
Co-authored-by: Claude <noreply@anthropic.com>
when destination equals source in mpz_div_2exp(), use memmove
and mpn_rshift in-place instead of allocating a temporary.
reduces sqrt allocations by 49% since Newton iteration uses
in-place division by 2 on each iteration.
Co-authored-by: Claude <noreply@anthropic.com>
add usub_inplace() and mpz_sub_inplace() for allocation-free
subtraction when the minuend is larger than the subtrahend.
apply to Mersenne multiplication which reduces allocations by
17% and improves performance by 7-9% for small numbers.
Co-authored-by: Claude <noreply@anthropic.com>
skip two's complement conversion in mpz_and, mpz_or, mpz_xor when both
operands are positive. this avoids the per-limb make_2comp overhead and
provides up to 1.6x speedup for large bigints.
Co-authored-by: Claude <noreply@anthropic.com>
convert mpz_init_heap to mpz_init_temp for temporary quotient and
remainder variables in div_limb. these variables are now allocated
from the memory pool when possible, reducing heap allocation overhead.
the div_limb function already uses pool_save/pool_restore, so these
temporary variables are proper candidates for pool allocation.
Co-authored-by: Claude <noreply@anthropic.com>
Replace Karatsuba multiplication (O(n^1.585)) with Toom-3 (O(n^1.465))
for large number multiplication. Toom-3 splits numbers into thirds and
evaluates at 5 points, providing better asymptotic performance.
Threshold is 50 limbs (~1600 bits). For operands below threshold or
highly asymmetric sizes, schoolbook multiplication is used.
Co-authored-by: Claude <noreply@anthropic.com>
Remove automatic downgrade to 16-bit limbs on 32-bit Windows.
Modern compilers (including MSVC) have supported uint64_t for decades.
MRB_NO_MPZ64BIT remains available for constrained platforms.
Adjust BATCH_DIVISOR and BATCH_DIGITS for 16-bit limb compatibility:
- 32-bit limbs: 10^9 (9 digits per batch)
- 16-bit limbs: 10^4 (4 digits per batch)
Co-authored-by: Claude <noreply@anthropic.com>
Add trim() after mpz_set in early return paths to prevent propagation
of inflated sz values. When an mpz_t has sz larger than actual allocated
limbs, copying it without trim causes subsequent operations to read
beyond allocated memory.
Fixed functions:
- mpz_add: when one operand is zero
- mpz_neg: when copying operand
- mpz_mod_2exp: when x < 2^e
Co-authored-by: Claude <noreply@anthropic.com>
Use mrb_protect_error API instead of direct MRB_TRY/MRB_CATCH to handle
exceptions in mpz_mul_all_ones and mpz_to_s_dc. This maintains C++
compatibility (issue #6702) while ensuring temporary mpz_t allocations
are properly freed even when exceptions occur.
Co-authored-by: Claude <noreply@anthropic.com>
Remove MRB_TRY/MRB_CATCH exception handling from bigint.c to fix
compilation errors when using mruby-bigint in C++ projects with
MRB_USE_CXX_EXCEPTION enabled.
The exception handling was added for cleanup on error, but it requires
throw.h which doesn't work when a C file is compiled in a C++ context
with C++ exceptions enabled. Accepting potential memory leaks on
exception (rare) is preferable to breaking C++ builds.
Fixes#6702
Co-authored-by: Claude <noreply@anthropic.com>
Yukihiro "Matz" Matsumoto