-march=x86-64-v3 kept the port and the installer to CPUs with AVX2 and FMA. --cpu v2 compiles the generated code and the runtime for SSE4.2 instead. Without FMA3 the FP lowering calls fma/fmaf, so runtime/fma.c provides correctly rounded ones (musl, MIT) that honour MXCSR's rounding mode, FTZ and DAZ; the UCRT's software fallback rounds incorrectly by up to half an ulp. simd.h's fast paths keep their assumptions.
The installer's host check asks for x86-64-v2 and selects libtotkrt.a or libtotkrt-v2.a; both release scripts build and ship the pair. The differential tests compile fma.c as well, so a --cflags=-march=x86-64-v2 run tests it.
Tested on an i7-3770K (no AVX2, FMA or BMI) with an RTX 3060 Ti, game 1.4.3: the installer ran end to end, --selftest passed and play holds 40-60 fps at 1080p. fma and fmaf matched 152,288 exact vectors over two seeds, including ties, subnormals and every rounding mode, FTZ and DAZ combination. The v3 build is unchanged: fma.c is an empty translation unit there.
The generator emits a guest poll such as `ldr w9, [x19]; cmp w9, #1; b.eq` as a plain C load, and clang hoists it out
of the loop: the lock wait in f_7001fa0e2c compiled to `cmp; jne; jmp .`, so a thread that found the lock held spun
forever after its holder released it. A fight run hung this way in 3 of 13 launches on an RX 9070, with the UBO
snapshot on and off; gdb showed the thread on that `jmp` in every sample. The gate is an opaque call on every
back-edge of the 868 loops tools/spin_waits.py lists, so each pass reloads memory: the exe goes from 182 self-jumps to
148, and 144 of those are the game's own `b .` instructions. 8 of 8 fight runs then passed; tests/diff/func.py with
--patches passed 5470 cases on 878 functions, and the four probes passed on 1.4.3.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The FP helpers compared c->fpcr with a cached copy before every operation to resync MXCSR lazily. MXCSR now
follows FPCR.FZ at once: MSR FPCR calls rt_fpcr_sync, and so do rt_call (thread, fiber and host entry into guest
code), fiber resume (FIBER_FLAG_FLOAT_SWITCH fibers keep their own MXCSR) and the differential test harness per
case. op and vop only test FPCR.RMode for the slow path, and their redo check is one ordered-not-equal compare of
|r| with the smallest normal, the same NaN-or-smallest-normal set as before.
Hot guest code shrinks by about a quarter (132 functions holding half the guest cycles, 1.4.3); fight frame time
moves 1-3%, inside the run-to-run spread. lift, simd, func, atomic and flags_check pass on 1.4.3; simd.py fails
without the harness sync.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every flag-setting instruction packed N, Z, C and V into c->nzcv and every reader unpacked them. recomp/flags.py
now runs a reaching-definitions pass per function: a compare stores its operands in a local Flags F, and a reader
whose definitions are all of one kind computes its condition from them, so cmp and b.cond become one host compare.
CCMP, FCMP, MSR NZCV and ADCS write the packed form, which readers reached by several kinds also use. Calls and
returns publish nothing (AAPCS64 neither passes nor preserves NZCV); tail branches, falls into the next function and
br through rt_dispatch publish c->nzcv, since discovery can start a function inside another one. Functions with an
unresolved computed br keep flags in c->nzcv.
1.4.3 main: 492k direct and 89k packed flag reads, .text 342 -> 299 MiB. Fight bench (bench-fight2-2026-10-27b,
frames 5400-7500, uncapped) 13.15 and 13.63 ms vs 13.9 before. tests/flags_check.py checks the new helpers against
the packed ones; tests/diff/func.py no longer compares NZCV, which generated code leaves untouched.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CHEN Feng