Home
last modified time | relevance | path

Searched refs:ulp (Results 1 – 10 of 10) sorted by relevance

/llvm-project-15.0.7/llvm/test/CodeGen/AMDGPU/
H A Damdgpu-codegenprepare-fdiv.ll6 ; NOOP: %md.25ulp = fdiv float %a, %b, !fpmath !0
8 %md.25ulp = fdiv float %a, %b, !fpmath !0
15 ; CHECK: %md.half.ulp = fdiv float %a, %b
16 ; CHECK: %md.1ulp = fdiv float %a, %b
30 %md.1ulp = fdiv float %a, %b, !fpmath !2
33 %md.25ulp = fdiv float %a, %b, !fpmath !0
36 %md.3ulp = fdiv float %a, %b, !fpmath !3
60 %md.25ulp = fdiv float 1.0, %x, !fpmath !0
308 ; CHECK: %md.1ulp = fdiv float %a, %b
309 ; CHECK: %md.25ulp = fdiv float %a, %b
[all …]
H A Dfdiv.f16.ll90 ; We could not do 1/b -> rcp_f16(b) under !fpmath < 1ulp.
/llvm-project-15.0.7/libc/AOR_v20.02/math/
H A DDir.mk14 $(S)/test/ulp.c \
29 build/bin/ulp \
55 $(B)/test/ulp.o: $(S)/test/ulp.h
81 build/bin/ulp: $(B)/test/ulp.o build/lib/libmathlib.a
99 check-math-ulp: $(math-tools)
102 check-math: check-math-test check-math-rtest check-math-ulp
111 .PHONY: all-math check-math-test check-math-rtest check-math-ulp check-math install-math clean-math
/llvm-project-15.0.7/libc/utils/MPFRWrapper/
H A DMPFRUtils.cpp660 << std::to_string(mpfr_result.ulp(libc_result.f)) << '\n'; in explain_unary_operation_two_outputs_error()
804 double ulp = mpfr_result.ulp(libc_result); in compare_unary_operation_single_output() local
805 return (ulp <= ulp_tolerance); in compare_unary_operation_single_output()
826 double ulp = mpfr_result.ulp(libc_result.f); in compare_unary_operation_two_outputs() local
831 return (ulp <= ulp_tolerance); in compare_unary_operation_two_outputs()
852 double ulp = mpfr_result.ulp(libc_result.f); in compare_binary_operation_two_outputs() local
863 return (ulp <= ulp_tolerance); in compare_binary_operation_two_outputs()
884 double ulp = mpfr_result.ulp(libc_result); in compare_binary_operation_one_output() local
886 return (ulp <= ulp_tolerance); in compare_binary_operation_one_output()
905 double ulp = mpfr_result.ulp(libc_result); in compare_ternary_operation_one_output() local
[all …]
/llvm-project-15.0.7/libc/AOR_v20.02/math/test/
H A Drunulp.sh25 $emu ./ulp -r $r -e $Lt $flags "$@" && PASS=$((PASS+1)) || FAIL=$((FAIL+1))
29 $emu ./ulp -f -q "$@" >/dev/null
/llvm-project-15.0.7/libc/AOR_v20.02/
H A Dconfig.mk.dist46 # Use if mpfr is available on the target for ulp error checking.
/llvm-project-15.0.7/libclc/generic/lib/math/
H A Derf.cl217 * guarantee the error is less than one ulp for erf.
H A Derfc.cl218 * guarantee the error is less than one ulp for erf.
/llvm-project-15.0.7/llvm/include/llvm/IR/
H A DIntrinsicsAMDGPU.td2428 // Emit 2.5 ulp, no denormal division. Should only be inserted by
/llvm-project-15.0.7/llvm/docs/
H A DLangRef.rst6232 of them, then ``ulp(x) = |b - a|``, otherwise ``ulp(x)`` is the
6234 nearest ``x``. Moreover, ``ulp(NaN)`` is ``NaN``.