The integer emulation of the round-to-nearest-even float-to-half
conversion is exact but costly: a compute shader with about 3600
conversions never finished compiling on NVIDIA. When the device enables
shaderFloat16 and guarantees shaderRoundingModeRTEFloat16,
shaderDenormPreserveFloat16 and shaderSignedZeroInfNanPreserveFloat16,
it advertises Float16, RoundingModeRTE and DenormPreserve, and the
recompiler emits OpFConvert under the 16-bit RoundingModeRTE,
DenormPreserve and SignedZeroInfNanPreserve execution modes instead.
NaNs keep the emulated quiet NaN with the high payload bits. Devices
without these guarantees keep the emulation.
Graphics validation accepts the RoundingModeRTE, DenormPreserve and
SignedZeroInfNanPreserve execution modes for 16-, 32- and 64-bit floats,
so vertex and pixel shaders with a native conversion still draw.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
The smallest mip levels of a tiled surface are packed together into one
block, the mip tail. A size limit decides which level is the first to
go into the tail.
For SW_64KB_Z_X, AnyPS5 used the same limit as the other 64 KiB modes.
In addrlib, that is only the case on Navi10. On every other GFX10 chip,
including the PS5's gfx1013, addrlib halves the limit for 8 and 16 bpp
Z_X surfaces: the width at both sizes, and also the height at 8 bpp.
The setting behind this is dsMipmapHtileFix (see HwlConvertChipFamily
and ComputeSurfaceInfoMacroTiled in gfx10addrlib.cpp).
With the larger limit, the tail started too early. For example, a
256x256 R8 surface with 9 levels started its tail at mip 1 instead of
mip 2, and took 131072 bytes instead of 196608. Every level was then
read and written back at the wrong offset.
GetMipTailLayout now halves the limits the same way. Only the level
where the tail starts changes; the tail slot positions stay the same.
This follows addrlib's setting for gfx1013. The console's own layout
has not been measured.
Fixes#1590
Co-Authored-By: Claude Opus 5.5 <[email protected]>
v_cvt_pknorm_i16/u16_f32 rounded |x| * 32767 or 65535 to nearest even
with a 64-bit integer product, variable 64-bit shifts and a sticky bit.
In f32, p = |x| * scale is the f32 nearest to the exact product, so
roundEven(p) is the answer unless p is exactly k + 0.5, where the sign of
the exact error fma(|x|, scale, -p) decides. Lanes that select a
constant (NaN, |x| >= 1) multiply 0 instead.
v_cvt_pk_u8_f32 truncated with the saturating conversion and rounded the
fraction by hand; it is now roundEven(x) between the 0 and 255 clamps,
NaN giving 0, read without the f32 denormal flush.
New test agc_driver_f32_pack_rounding: 563,232 inputs around every
rounding tie, checked against the exact product in double.
When f32 denormals are flushed, v_cmp_*_f32 compares flush(a) with
flush(b). With one operand a constant, the flush of the other operand
can go into the compare: a denormal and its flushed zero compare the same
way with a NaN, an infinity or a normal constant, and comparing flush(x)
with a zero is comparing x with +-2^-126 (flush(x) < 0 is x <= -2^-126,
flush(x) == 0 is |x| < 2^-126, and so on, the unordered forms alike).
emitFloatCompare reads both operands unflushed, finds a constant operand
after its abs/neg modifiers, mirrors the predicate when the constant is
first, and emits one compare. Two registers, f16 compares and modes that
keep f32 denormals are unchanged. New test agc_driver_f32_compare_constant
checks all 12 predicates against flush(a) op flush(b).
The translation of v_cvt_i32/u32/flr_i32/rpi_i32_f32 and v_cvt_i16/u16_f16
already maps NaN to 0 and clamps the value into range before
ConvertS32F32/ConvertU32F32, and the backend truncated it again and
repeated the NaN and range selects on the result. Those IR ops have no
other producers, so the backend now emits one OpConvertFToS or
OpConvertFToU, and EmitF32ToU32 and EmitTruncF32Value, which had no
other users, are removed.
The translation drops its FPTrunc32, as the conversions round toward
zero, and the selects whose constant the conversion of the clamped value
already gives. v_cvt_i32_f32 and v_cvt_u32_f32 read their source without
the f32 denormal flush: a denormal truncates to 0 either way.
The guest process is assumed to run as root. Under FreeBSD's
sys_seteuid and sys_setegid, id 0 is then accepted without a privilege
check and changes no credential, so both return 0. Any other id would
change the effective credential, which is not modelled, and throws.
Part of #1578.
When the embedded vertex fetch adds the base vertex or start instance
user SGPR to the index VGPR, the driver folds that SGPR into the draw's
first vertex and first instance. v5/v8 were gl_VertexIndex and
gl_InstanceIndex, which already include it, so the translated add
counted it twice, on direct draws and CPU-path indirect records alike.
The entry of a vertex or local program now subtracts the folded SGPR's
user data word from the builtin, which gives the zero-based indices the
hardware has. A record drawn in place on the GPU keeps the CPU's word,
so the fetch analyzer counts the SGPR as shared unless the fetch add is
the only reader of v5/v8 and the SGPR (no earlier access or EXEC write,
no carry, clamp, second add or other write of the SGPR); such records
take the CPU path. The fold moves into Driver::FoldDrawOffsets,
ClassifyIndirectDraw becomes public for the test, and the env switches
that skipped the fold go.
A program with one GLC or DLC buffer load or store decorated the whole
guest buffer array Coherent, so its other loads were coherent too. On
NVIDIA that keeps them out of the L1 cache: one GLC store made the
re-reads of a 4 KiB window about 2.7 times slower.
RDNA keeps those loads incoherent. A load without GLC or DLC can hit the
per-CU L0 and per-SA L1 caches, which are not coherent between
work-groups, and code that needs another work-group's data reads it
with GLC/DLC or invalidates the caches first (LLVM AMDGPU memory model,
GFX10: an acquiring load is followed by buffer_gl1_inv; buffer_gl0_inv).
The program metadata now records whether a program invalidates a cache
(buffer_gl0_inv, buffer_gl1_inv, s_gl1_inv, s_dcache_inv). Without one,
a program with GLC/DLC accesses gets a second array on the same binding,
buffers_coherent, decorated Coherent, and both arrays are Aliased as
buffers_u64 already is. Loads without GLC or DLC read buffers; GLC/DLC
loads, stores and atomics use buffers_coherent. A program that
invalidates a cache keeps the whole array Coherent.
gh label create exits non-zero for a label that already exists, and the
step runs with bash -e, so the first existing unforced label stopped the
job. Skip labels that already exist, keeping hand-set colors.
Fixes#2320
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceKernelAprSubmitCommandBuffer rejected a buffer object with EINVAL
unless sceAmprAprCommandBufferConstructor had tagged it. The tag is
AnyPS5's own: in the retail layout the type word only carries state
bits and the constructors zero the object, so a zero-filled object
bound with SetBuffer is a valid APR buffer.
CRI's file system in Sonic Origins (PPSA05324) lays out its
AprCommandBuffer objects in its own zeroed work memory and never calls
the constructors. Its first read submit failed with 0x80020016, and the
CRI file system stopped there.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sceNpEAAccessInitialize and sceNpEAAccessTerminate resolve the NIDs
7CxI50-xlCk and pMxXhNozUX8, which shadPS4 names in np_partner.cpp. The
stubs still throw; only the names and no-argument signatures change.
FreeBSD's unlink ignores the file's own permission bits, so remove()
succeeds on a file the guest made read-only with chmod(0400). On Windows
the guest chmod sets FILE_ATTRIBUTE_READONLY, and DeleteFileW then fails
with ERROR_ACCESS_DENIED, which remove() reported as EACCES. Clear the
attribute and retry, restoring it if the retry still fails.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Use INFINITY and NAN from <cmath> instead of a new <limits> include, and
list the TechnicalDebt entry beside the other NGS2 voice entries, so the
change merges cleanly with the pan, channel map and reverb changes.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The template allocated a fixed 4 KiB of LDS, so kernels with immediate
offsets past it could not run. --lds (Python lds) sets the group segment
size in bytes, up to 64 KiB, with 4096 as the default.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
The guest passes and expects UTF-8 names, but on Windows opendir and
readdir used MinGW's narrow dirent API, which works in the ANSI code
page: opendir failed for a directory whose name is outside the code page,
and readdir returned names as ANSI bytes, with "?" for characters the code
page lacks. Use _wopendir and _wreaddir there and convert each name to
UTF-8.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A reverb_size other than the default was accepted and ignored, and below
about 7.5 kHz an allpass length went negative and wrapped to a 16 GB buffer.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
WindowsMappings::Collect resets each part of the range as it walks it:
it marks shared views seen and resets private pages' write-watch bits.
When it then met a NOACCESS view or memory that is not committed
private memory, it returned false with the earlier parts already reset,
so their writes were lost to the caller's fallback. With reset, the
whole range is now checked first and nothing is reset when it fails.
AI-assisted (Claude Code).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Parse legacy decimal, octal and hexadecimal forms before host lookup. Preserve FreeBSD shorthand, whitespace and 64-bit accumulation, clear the SDK resolver error on numeric success, and add regression coverage.
getcwd(NULL, size) returned memory from the guest heap directly, while
the caller frees it with free(), which goes to the application heap.
With a title's malloc replacement registered, that free reached the
replacement with a pointer it never returned. Allocate from the
application heap like strdup, and report ENOMEM when it returns null.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
