vf2h truncates the mantissa (no rounding), gives a signed zero below 2^-14
and a signed inf from 65536 up, and keeps the low ten mantissa bits of a
NaN, so one with those clear becomes inf. vh2f flushes a subnormal half to
a signed zero and keeps inf/NaN mantissa bits unshifted. The x86 JIT's own
vh2f gets the subnormal flush; the other backends go through the
interpreter. Recorded in cpu/vfpu/specials.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
mtvc to RCX0-7 keeps the low 20 bits of the value and forces the top
twelve to 0x3F8 (cpu/vfpu/vrnd), which is also the form vrnds and vrndi
leave them in. We masked with 0x3FFFFFFF. The generator itself matches
hardware for every seed and state in that test.
Interpreter, IR and the arm64 JIT. The x86 and ARM32 JITs don't mask
control register writes at all and are left as they were.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
WriteMMIO_U32 was missing the return after the kernel-mode check, so a
user-mode write raised the exception and then went through anyway. The
other five MMIO accessors already return here; this one lost it when the
GPIO/syscon branches were added.
Int_Vrot scanned all four entries of dregs, but GetVectorRegs only fills
the first n - so a vrot with vs == 0 (S000) matched a lane that isn't
there and took the cosine from it. The IR backend gets this right via
IsOverlapSafe, so the two disagreed.
The breakpoint checks in MIPSInterpret and RunUntilDowncountZeroWithChecks
read instr->flags without checking for null, which MIPSGetInstruction
returns for the eight primary opcodes (and many subops) that don't decode.
With a memcheck or register breakpoint active, landing on one of those
crashed instead of raising ILLEGAL.
Also: GetVectorOverlap decoded the second vector with size1 (no callers
today), and RegisterFunction left most of its AnalyzedFunction
uninitialized, including the size that ends up in knownfuncs.ini.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Was looking over VFPU stuff, and noticed this.
Presumably, on x86 the code was already doing exactly this, but still, ouch.
Did retest the code on all relevant (|x|>2^32) available inputs, all matches (except NaN payloads, as usual, but that is unrelated).
* Rename LogType to Log
* Explicitly use the Log:: enum when logging. Allows for autocomplete when editing.
* Mac/ARM64 buildfix
* Do the same with the hle result log macros
* Rename the log names to mixed case while at it.
* iOS buildfix
* Qt buildfix attempt, ARM32 buildfix
Drops these functions down the ranking of top functions by quite a bit in GTA,
speedup at most 0.5% though. But enough of these small ones and they
start adding up.
Not sure why GTA falls back to the interpreter for these so much though.
I guess some "uneaten" prefix..
See #18249. Speedup for this function ranges 10%..100%,
depending on system. Updated verification and speed measurements:
https://godbolt.org/z/W1z3sj6hz
PR #16984 added more accurate versions of these functions, but they require
large lookup tables stored in assets/.
If these files are missing, PPSSPP would simply crash, which isn't good.
We should probably try to warn the user somehow that these files are
missing, though...
This makes the results much more accurate to the PSP's results.
Could narrow a bit further swapping sin/cos/neg, which might be what the
hardware does given vrot.
