main
Create tests/func-tests
func-tests: enable all float16 for x86
Fix the acos function, make it faster and improve tests
Add functional tests
tests: skip unsupported atomic tests on Xe targets
Fix backward tests
Add Linux ppc64le (PowerPC64) architecture support using generic targets (#3735)
Added unsigned versions for broadcast/rotate
Skip cbrt-double.ispc on platforms without double support
Update error handling in cbrt-float.ispc
Fix all tests of the recent stdlib-math-function PRs
Extended int16 dot product functions
Added generic implementations of dot4add_u8u8packed and dot4add_i8i8packed
Add exclusive_scan functions for int8/int16 types
Add extract/insert tests for unsigned types
Do not use rcp in fmod tests to decrease stack usage
Split fmod-[double|float]-varying.ispc tests
Split fmod-double-varying-2.ispc
Split isfinite/isinf tests to fp16 and fp/fp64 types
Skip ldexp-double-zero.ispc on targets without double support
Fix ldexp zero handling
ldexp-float16 func-test: reduce input range to avoid float16 overflow
Add functional tests for log1p
Allow skip func tests in jit mode
Extend binary operators support
Update couple of operators tests
Add support of Unary operators
Add support of Assign operators
Add missing types for packed_load/packed_store
Add support of func tests in JIT mode
Add reduce_equal for int8/int16 types
Add reduce_min/max int16 for wasm/xe
Add generic implementations for reduce_min/max for int8
Fix skipping tests
Skip round-double tests on PPC64 due to LLVM 19 v2f64 froundeven ISel bug
Skip tests with round functions for generic targets on windows
Skip round-float16-varying test on PPC64 due to missing half-precision support
Skip select-short-vec-func.ispc on tgllp/dg2
Skip select-short-vec.ispc on some Xe targets
Added unsigned versions for shifts/shuffles
Legalize soa<> members in structs
Split srgb8.ispc test to two (uniform/varying)