arm64: support signed-128
View Changes

Potential life-time issues
View Changes

x86-64: add unboxed signed 128 bit registers
View Changes

Fix an eql+abs transform
View Changes

Add a type deriver for SB-SYS:WITH-INTERRUPTS
View Changes

Turn (+ unsigned-word -constant) into (- unsigned-word constant)
View Changes

Typo.
View Changes

Fix some avx512 k* instructions encodings
View Changes

Add sb-impl::read-line-buffers to ignore-symbol-value-change
View Changes

arm64: better single-float-bits
View Changes

Revert "x86-64: implement disassembler support for Intel APX EVEX instructions"
View Changes

Revert "x86-64: implement Intel APX (Advanced Performance Extensions) and hardware fuzz suite"
View Changes

life: Remove FASTP kludge.
View Changes

Better diagnostics for make-host-1 warnings
View Changes

arm64: more compact double-float-high-bits
View Changes

Simplify single-float handling in struct by value on x86-64
View Changes

arm64: faster imagpart
View Changes

arm64: improve complex float multiplication vops
View Changes

x86-64: use SSE3 for (* complex-float complex-float)
View Changes

arm64: fix direct call vops not saving NFP
View Changes

Fix tests on -x86-64
View Changes

tests(x86-64): add bit-exact encoding assertions for vbroadcastss and vbroadcastsd
View Changes

feat(sb-simd): support Intel AVX10.1 and AVX10.2 instruction sets
View Changes

x86-64: implement Intel AVX10.1 and AVX10.2 converged vector instructions and CPUID
View Changes

x86-64: implement Intel APX (Advanced Performance Extensions) and hardware fuzz suite
View Changes

x86-64: implement disassembler support for Intel APX EVEX instructions
View Changes

tests(sb-simd): add regression tests for sap-ref bindings, CPUID bounds, and missing instructions
View Changes