# RV32I With Long Double Support
## Why Doing This
The default riscv-none-elf-gcc-xpack build of Newlib does not enable long double support. I want to enable long double support for RV32I.
The [ieee754.c](https://github.com/sysprog21/rv32emu/blob/master/tests/ieee754.c) file shipped with RV32EMU was originally written by [Dr. Vincent Lefèvre](https://www.vinc17.net/research/index.en.html) .
I found a FIXME and tried to fix it:
```cpp
#if 0 /* FIXME: make it work */
printf("LDBL_MAX = %La%s\n", lm, ERRSTR(lm != ldbl_max));
#endif
```
## How to Do
1. Rebuild Newlib with long double support
2. Compare RV32I long double support before and after
3. Compare calculation results between PC and RV32I emulator
4. Verify the results (Draft)
## Re-Build the Newlib
### Build Newlib Only
```bash
riscv-none-elf-gcc -v
Using built-in specs.
COLLECT_GCC=riscv-none-elf-gcc
COLLECT_LTO_WRAPPER=/home/eastwillow/.local/xPacks/@xpack-dev-tools/riscv-none-elf-gcc/14.2.0-3.1/.content/bin/../libexec/gcc/riscv-none-elf/14.2.0/lto-wrapper
...
```
The newlib used in riscv-none-elf-gcc is version 14.2.0-3.1
https://xpack-dev-tools.github.io/riscv-none-elf-gcc-xpack/blog/2024/12/06/riscv-none-elf-gcc-v14-2-0-3-released/
Newlib version: 4.4.0
Download from https://sourceware.org/pub/newlib/
```
wget ftp://sourceware.org/pub/newlib/newlib-4.4.0.20231231.tar.gz
tar xzf newlib-4.4.0.20231231.tar.gz
cd newlib-4.4.0.20231231
```
#### Configure:
```shell
../configure \
--target=riscv-none-elf \
--prefix=$HOME/opt/riscv-newlib-debug \
--enable-newlib-io-long-long \
--enable-newlib-io-c99-formats \
--enable-newlib-register-fini \
--enable-newlib-retargetable-locking \
--enable-newlib-mb \
--enable-newlib-reent-check-verify \
--enable-newlib-io-long-double
```
#### Build:
```shell
make CC_FOR_TARGET=riscv-none-elf-gcc CFLAGS="-g -O0" all
```
**Newlib configure reference**
From xPack Dev Tools [github action](https://github.com/xpack-dev-tools/riscv-none-elf-gcc-xpack/actions/runs/18744298753/job/53467796369)
See also:
* [newlib-cross.sh](
https://github.com/xpack-dev-tools/xbb-helper-xpack/blob/xpack/dependencies/newlib-cross.sh)
* [archlinux pkg-build](https://gitlab.archlinux.org/archlinux/packaging/packages/riscv32-elf-newlib/-/blob/main/PKGBUILD)
### Build the Entire riscv-none-elf-gcc-xpack
Clone and add --enable-newlib-io-long-double and debug flags:
```shell
riscv-none-elf-gcc-xpack.git/build-assets/xpacks/@xpack-dev-tools/xbb-helper/dependencies/newlib-cross.sh
```
Then build:
https://xpack-dev-tools.github.io/riscv-none-elf-gcc-xpack/docs/developer/
```shell
xpm run install -C ~/Work/xpack-dev-tools/riscv-none-elf-gcc-xpack.git/build-assets && \
xpm run docker-prepare --config linux-x64 -C ~/Work/xpack-dev-tools/riscv-none-elf-gcc-xpack.git/build-assets
xpm run build-development-debug --config linux-x64 -C ~/Work/xpack-dev-tools/riscv-none-elf-gcc-xpack.git/build-assets
```
Manual Installation globally (now version is `15.2.0-1`)
```bash
mkdir -p ~/.local/xPacks/riscv-none-elf-gcc
cd ~/.local/xPacks/riscv-none-elf-gcc
tar xvf ~/Work/xpack-dev-tools/riscv-none-elf-gcc-xpack.git/build-assets/build/linux-x64/deploy/xpack-riscv-none-elf-gcc-15.2.0-1-linux-x64.tar.gz
```
PATH setup in ~/.bashrc
```
export PATH=$HOME/.local/xPacks/riscv-none-elf-gcc/xpack-riscv-none-elf-gcc-15.2.0-1/bin:$PATH
```
## Long double has multiple implementation variants.
In glibc on x86_64, the default `LDBL_MANT_DIG` is 64.
When Newlib enables long double on RV32I, `LDBL_MANT_DIG` becomes 113.
According to the [GNU manual](https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html
), you need to add the flag `-mlong-double-128`
## PC Build Command with 128-bit Long Double
```shell
gcc -Wall -O2 -std=c99 -o tests/ieee754_pc tests/ieee754.c -lm -mlong-double-128 && \
tests/ieee754_pc > pc_output.txt
```
## RV32I Command (Using Custom-Built Newlib)
```shell
riscv-none-elf-gcc -I$HOME/opt/riscv-newlib-debug/riscv-none-elf/include -L$HOME/opt/riscv-newlib-debug/riscv-none-elf/lib -Wall -O2 -std=c99 -march=rv32if -mabi=ilp32 -o tests/ieee754_after tests/ieee754.c -lm && \
build/rv32emu tests/ieee754_after > rv32_after.txt
```
## RV32I Command (Using Full riscv-none-elf-gcc-xpack Build)
```shell
riscv-none-elf-gcc -Wall -O2 -std=c99 -march=rv32if -mabi=ilp32 -o tests/ieee754_after tests/ieee754.c -lm && \
build/rv32emu tests/ieee754_after > rv32_after.txt
```
## RV32I Supported Long Double Before & After Diff
```shell
git diff --no-index rv32_before.txt rv32_after.txt > rv32_before_after.diff
```
```diff
diff --git a/rv32_before.txt b/rv32_after.txt
index ec918d1..89ee1d5 100644
--- a/rv32_before.txt
+++ b/rv32_after.txt
@@ -1,6 +1,6 @@
-23:26:25 [32mINFO [0m [90msrc/riscv.c:552:[0m Log level: TRACE
-23:26:25 [32mINFO [0m [90msrc/riscv.c:565:[0m tests/ieee754_before ELF loaded
-23:26:25 [32mINFO [0m [90msrc/main.c:315:[0m RISC-V emulator is created and ready to run
+23:58:51 [32mINFO [0m [90msrc/riscv.c:552:[0m Log level: TRACE
+23:58:51 [32mINFO [0m [90msrc/riscv.c:565:[0m tests/ieee754_after ELF loaded
+23:58:51 [32mINFO [0m [90msrc/main.c:315:[0m RISC-V emulator is created and ready to run
FLT_RADIX = 2
FLT_MANT_DIG = 24
DBL_MANT_DIG = 53
@@ -20,11 +20,11 @@ LDBL_EPSILON = 1.92593e-34 = 0x1p-112
FLT_MIN = 1.17549e-38 = 0x1p-126
DBL_MIN = 2.22507e-308 = 0x1p-1022
-LDBL_MIN = 0 = 0x0p+0
+LDBL_MIN = 3.3621e-4932 = 0x1p-16382
FLT_MAX = 3.40282e+38 = 0x1.fffffep+127
DBL_MAX = 1.79769e+308 = 0x1.fffffffffffffp+1023
-LDBL_MAX = inf = inf
+LDBL_MAX = 1.18973e+4932 = 0x1.ffffffffffffffffffffffffffffp+16383
sizeof(float) = 4
sizeof(double) = 8
@@ -81,11 +81,11 @@ Constant expression 1 - DBL_MIN = 1
Variable expression 1 - DBL_MIN = 1
1/3 in float : 0x1.555556p-2
1/3 in double : 0x1.5555555555555p-2
-1/3 in long double : 0x1.5555555555555p-2
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = inf
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
LDBL_EPSILON = 0x1p-112
@@ -242,11 +242,11 @@ Constant expression 1 - DBL_MIN = 1
Variable expression 1 - DBL_MIN = 1
1/3 in float : 0x1.555556p-2
1/3 in double : 0x1.5555555555555p-2
-1/3 in long double : 0x1.5555555555555p-2
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = inf
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
LDBL_EPSILON = 0x1p-112
@@ -403,11 +403,11 @@ Constant expression 1 - DBL_MIN = 1
Variable expression 1 - DBL_MIN = 1
1/3 in float : 0x1.555556p-2
1/3 in double : 0x1.5555555555555p-2
-1/3 in long double : 0x1.5555555555555p-2
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = inf
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
LDBL_EPSILON = 0x1p-112
@@ -564,11 +564,11 @@ Constant expression 1 - DBL_MIN = 1
Variable expression 1 - DBL_MIN = 1
1/3 in float : 0x1.555556p-2
1/3 in double : 0x1.5555555555555p-2
-1/3 in long double : 0x1.5555555555555p-2
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = inf
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
LDBL_EPSILON = 0x1p-112
@@ -693,4 +693,4 @@ pow(-2, 1) = -2
pow(-2, -1) = -0.5
pow(-2, 2) = 4
pow(-2, -2) = 0.25
-23:26:25 [32mINFO [0m [90msrc/main.c:338:[0m RISC-V emulator is destroyed
+23:58:51 [32mINFO [0m [90msrc/main.c:338:[0m RISC-V emulator is destroyed
```
## PC Compare with RV32I Result (Draft)
```shell
git diff --no-index pc_output.txt rv32_after.txt > pc_rv32_after.diff
```
```shell
gcc -v
gcc version 13.3.0 (Ubuntu 13.3.0-6ubuntu2~24.04)
```
```diff
diff --git a/pc_output.txt b/rv32_after.txt
index 67d186e..617e449 100644
--- a/pc_output.txt
+++ b/rv32_after.txt
@@ -1,3 +1,6 @@
+23:54:32 [32mINFO [0m [90msrc/riscv.c:552:[0m Log level: TRACE
+23:54:32 [32mINFO [0m [90msrc/riscv.c:565:[0m tests/ieee754_after ELF loaded
+23:54:32 [32mINFO [0m [90msrc/main.c:315:[0m RISC-V emulator is created and ready to run
FLT_RADIX = 2
FLT_MANT_DIG = 24
DBL_MANT_DIG = 53
@@ -13,24 +16,20 @@ LDBL_MAX_EXP = 16384
FLT_EPSILON = 1.19209e-07 = 0x1p-23
DBL_EPSILON = 2.22045e-16 = 0x1p-52
-LDBL_EPSILON = nan = 0x3.38p-16385
+LDBL_EPSILON = 1.92593e-34 = 0x1p-112
FLT_MIN = 1.17549e-38 = 0x1p-126
DBL_MIN = 2.22507e-308 = 0x1p-1022
-LDBL_MIN = nan = 0x3.38p-16385
+LDBL_MIN = 3.3621e-4932 = 0x1p-16382
FLT_MAX = 3.40282e+38 = 0x1.fffffep+127
DBL_MAX = 1.79769e+308 = 0x1.fffffffffffffp+1023
-LDBL_MAX = nan = 0x3.38p-16385
+LDBL_MAX = 1.18973e+4932 = 0x1.ffffffffffffffffffffffffffffp+16383
sizeof(float) = 4
sizeof(double) = 8
sizeof(long double) = 16
-__STDC_IEC_559__ defined:
-The implementation shall conform to the IEEE-754 standard.
-FLT_EVAL_METHOD is 0 (see ISO/IEC 9899, 5.2.4.2.2#8).
-
x + y, with x = 9007199254740994.0 and y = 1.0 - 1/65536.0 (type double).
The IEEE-754 result is 9007199254740994 with double precision.
The IEEE-754 result is 9007199254740996 with extended precision.
@@ -82,14 +81,14 @@ Constant expression 1 - DBL_MIN = 1
Variable expression 1 - DBL_MIN = 1
1/3 in float : 0x1.555556p-2
1/3 in double : 0x1.5555555555555p-2
-1/3 in long double : nan
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = nan
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
-LDBL_EPSILON = nan
+LDBL_EPSILON = 0x1p-112
pow(nan, nan) = nan
pow(nan, inf) = nan
pow(nan, -inf) = nan
@@ -161,8 +160,8 @@ pow(-0.5, inf) = 0
pow(-0.5, -inf) = inf
pow(-0.5, 0) = 1
pow(-0.5, -0) = 1
-pow(-0.5, 0.5) = -nan
-pow(-0.5, -0.5) = -nan
+pow(-0.5, 0.5) = nan
+pow(-0.5, -0.5) = nan
pow(-0.5, 1) = -0.5
pow(-0.5, -1) = -2
pow(-0.5, 2) = 0.25
@@ -183,8 +182,8 @@ pow(-1, inf) = 1
pow(-1, -inf) = 1
pow(-1, 0) = 1
pow(-1, -0) = 1
-pow(-1, 0.5) = -nan
-pow(-1, -0.5) = -nan
+pow(-1, 0.5) = nan
+pow(-1, -0.5) = nan
pow(-1, 1) = -1
pow(-1, -1) = -1
pow(-1, 2) = 1
@@ -205,15 +204,15 @@ pow(-2, inf) = inf
pow(-2, -inf) = 0
pow(-2, 0) = 1
pow(-2, -0) = 1
-pow(-2, 0.5) = -nan
-pow(-2, -0.5) = -nan
+pow(-2, 0.5) = nan
+pow(-2, -0.5) = nan
pow(-2, 1) = -2
pow(-2, -1) = -0.5
pow(-2, 2) = 4
pow(-2, -2) = 0.25
Rounding toward 0
--------> toward 0 (to -oo / to nearest)
+-------> to nearest (to nearest / to nearest)
(double) 0 = 0
Signed zero tests (x is 0.0 and y is -0.0):
Test 1.0 / x != 1.0 / y returns 1 (should be 1).
@@ -240,17 +239,17 @@ Signed zero tests (x is 0.0 and y is -0.0):
Constant expression 1 + DBL_MIN = 1
Variable expression 1 + DBL_MIN = 1
Constant expression 1 - DBL_MIN = 1
-Variable expression 1 - DBL_MIN = 0.99999999999999988897
-1/3 in float : 0x1.555554p-2
+Variable expression 1 - DBL_MIN = 1
+1/3 in float : 0x1.555556p-2
1/3 in double : 0x1.5555555555555p-2
-1/3 in long double : nan
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = nan
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
-LDBL_EPSILON = nan
+LDBL_EPSILON = 0x1p-112
pow(nan, nan) = nan
pow(nan, inf) = nan
pow(nan, -inf) = nan
@@ -311,23 +310,23 @@ pow(0.5, inf) = 0
pow(0.5, -inf) = inf
pow(0.5, 0) = 1
pow(0.5, -0) = 1
-pow(0.5, 0.5) = 0.707106
+pow(0.5, 0.5) = 0.707107
pow(0.5, -0.5) = 1.41421
pow(0.5, 1) = 0.5
-pow(0.5, -1) = 1.99999
+pow(0.5, -1) = 2
pow(0.5, 2) = 0.25
-pow(0.5, -2) = 3.99999
+pow(0.5, -2) = 4
pow(-0.5, nan) = nan
pow(-0.5, inf) = 0
pow(-0.5, -inf) = inf
pow(-0.5, 0) = 1
pow(-0.5, -0) = 1
-pow(-0.5, 0.5) = -nan
-pow(-0.5, -0.5) = -nan
+pow(-0.5, 0.5) = nan
+pow(-0.5, -0.5) = nan
pow(-0.5, 1) = -0.5
-pow(-0.5, -1) = -1.99999
+pow(-0.5, -1) = -2
pow(-0.5, 2) = 0.25
-pow(-0.5, -2) = 3.99999
+pow(-0.5, -2) = 4
pow(1, nan) = 1
pow(1, inf) = 1
pow(1, -inf) = 1
@@ -344,8 +343,8 @@ pow(-1, inf) = 1
pow(-1, -inf) = 1
pow(-1, 0) = 1
pow(-1, -0) = 1
-pow(-1, 0.5) = -nan
-pow(-1, -0.5) = -nan
+pow(-1, 0.5) = nan
+pow(-1, -0.5) = nan
pow(-1, 1) = -1
pow(-1, -1) = -1
pow(-1, 2) = 1
@@ -356,25 +355,25 @@ pow(2, -inf) = 0
pow(2, 0) = 1
pow(2, -0) = 1
pow(2, 0.5) = 1.41421
-pow(2, -0.5) = 0.707106
-pow(2, 1) = 1.99999
+pow(2, -0.5) = 0.707107
+pow(2, 1) = 2
pow(2, -1) = 0.5
-pow(2, 2) = 3.99999
+pow(2, 2) = 4
pow(2, -2) = 0.25
pow(-2, nan) = nan
pow(-2, inf) = inf
pow(-2, -inf) = 0
pow(-2, 0) = 1
pow(-2, -0) = 1
-pow(-2, 0.5) = -nan
-pow(-2, -0.5) = -nan
-pow(-2, 1) = -1.99999
+pow(-2, 0.5) = nan
+pow(-2, -0.5) = nan
+pow(-2, 1) = -2
pow(-2, -1) = -0.5
-pow(-2, 2) = 3.99999
+pow(-2, 2) = 4
pow(-2, -2) = 0.25
Rounding to -oo
--------> to -oo (to -oo / to nearest)
+-------> to nearest (to nearest / to nearest)
(double) 0 = 0
Signed zero tests (x is 0.0 and y is -0.0):
Test 1.0 / x != 1.0 / y returns 1 (should be 1).
@@ -383,16 +382,16 @@ Signed zero tests (x is 0.0 and y is -0.0):
Test 1.0 / x != 1.0 / -x returns 1 (should be 1).
Test 1.0 / x == 1.0 / -y returns 1 (should be 1).
0 + 0 = 0
-0 - 0 = -0
-0 + -0 = -0
+0 - 0 = 0
+0 + -0 = 0
0 - -0 = 0
--0 + 0 = -0
+-0 + 0 = 0
-0 - 0 = -0
-0 + -0 = -0
--0 - -0 = -0
+-0 - -0 = 0
1 + 1 = 2
-1 - 1 = -0
-1 + -1 = -0
+1 - 1 = 0
+1 + -1 = 0
1 - -1 = 2
0 * 0 = 0
0 * -0 = -0
@@ -401,17 +400,17 @@ Signed zero tests (x is 0.0 and y is -0.0):
Constant expression 1 + DBL_MIN = 1
Variable expression 1 + DBL_MIN = 1
Constant expression 1 - DBL_MIN = 1
-Variable expression 1 - DBL_MIN = 0.99999999999999988897
-1/3 in float : 0x1.555554p-2
+Variable expression 1 - DBL_MIN = 1
+1/3 in float : 0x1.555556p-2
1/3 in double : 0x1.5555555555555p-2
-1/3 in long double : nan
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = nan
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
-LDBL_EPSILON = nan
+LDBL_EPSILON = 0x1p-112
pow(nan, nan) = nan
pow(nan, inf) = nan
pow(nan, -inf) = nan
@@ -472,23 +471,23 @@ pow(0.5, inf) = 0
pow(0.5, -inf) = inf
pow(0.5, 0) = 1
pow(0.5, -0) = 1
-pow(0.5, 0.5) = 0.707106
+pow(0.5, 0.5) = 0.707107
pow(0.5, -0.5) = 1.41421
-pow(0.5, 1) = 0.499999
-pow(0.5, -1) = 1.99999
-pow(0.5, 2) = 0.249999
-pow(0.5, -2) = 3.99999
+pow(0.5, 1) = 0.5
+pow(0.5, -1) = 2
+pow(0.5, 2) = 0.25
+pow(0.5, -2) = 4
pow(-0.5, nan) = nan
pow(-0.5, inf) = 0
pow(-0.5, -inf) = inf
pow(-0.5, 0) = 1
pow(-0.5, -0) = 1
-pow(-0.5, 0.5) = -nan
-pow(-0.5, -0.5) = -nan
+pow(-0.5, 0.5) = nan
+pow(-0.5, -0.5) = nan
pow(-0.5, 1) = -0.5
pow(-0.5, -1) = -2
-pow(-0.5, 2) = 0.249999
-pow(-0.5, -2) = 3.99999
+pow(-0.5, 2) = 0.25
+pow(-0.5, -2) = 4
pow(1, nan) = 1
pow(1, inf) = 1
pow(1, -inf) = 1
@@ -505,8 +504,8 @@ pow(-1, inf) = 1
pow(-1, -inf) = 1
pow(-1, 0) = 1
pow(-1, -0) = 1
-pow(-1, 0.5) = -nan
-pow(-1, -0.5) = -nan
+pow(-1, 0.5) = nan
+pow(-1, -0.5) = nan
pow(-1, 1) = -1
pow(-1, -1) = -1
pow(-1, 2) = 1
@@ -517,25 +516,25 @@ pow(2, -inf) = 0
pow(2, 0) = 1
pow(2, -0) = 1
pow(2, 0.5) = 1.41421
-pow(2, -0.5) = 0.707106
-pow(2, 1) = 1.99999
+pow(2, -0.5) = 0.707107
+pow(2, 1) = 2
pow(2, -1) = 0.5
-pow(2, 2) = 3.99999
+pow(2, 2) = 4
pow(2, -2) = 0.25
pow(-2, nan) = nan
pow(-2, inf) = inf
pow(-2, -inf) = 0
pow(-2, 0) = 1
pow(-2, -0) = 1
-pow(-2, 0.5) = -nan
-pow(-2, -0.5) = -nan
+pow(-2, 0.5) = nan
+pow(-2, -0.5) = nan
pow(-2, 1) = -2
-pow(-2, -1) = -0.500001
-pow(-2, 2) = 3.99999
+pow(-2, -1) = -0.5
+pow(-2, 2) = 4
pow(-2, -2) = 0.25
Rounding to +oo
--------> to +oo (to nearest / to nearest)
+-------> to nearest (to nearest / to nearest)
(double) 0 = 0
Signed zero tests (x is 0.0 and y is -0.0):
Test 1.0 / x != 1.0 / y returns 1 (should be 1).
@@ -560,19 +559,19 @@ Signed zero tests (x is 0.0 and y is -0.0):
-0 * 0 = -0
-0 * -0 = 0
Constant expression 1 + DBL_MIN = 1
-Variable expression 1 + DBL_MIN = 1.0000000000000002221
+Variable expression 1 + DBL_MIN = 1
Constant expression 1 - DBL_MIN = 1
Variable expression 1 - DBL_MIN = 1
1/3 in float : 0x1.555556p-2
-1/3 in double : 0x1.5555555555556p-2
-1/3 in long double : nan
+1/3 in double : 0x1.5555555555555p-2
+1/3 in long double : 0x1.5555555555555555555555555555p-2
Dec 1.1 = 0x1.199999999999ap+0
FLT_MAX = 0x1.fffffep+127
DBL_MAX = 0x1.fffffffffffffp+1023
-LDBL_MAX = nan
+LDBL_MAX = 0x1.ffffffffffffffffffffffffffffp+16383
FLT_EPSILON = 0x1p-23
DBL_EPSILON = 0x1p-52
-LDBL_EPSILON = nan
+LDBL_EPSILON = 0x1p-112
pow(nan, nan) = nan
pow(nan, inf) = nan
pow(nan, -inf) = nan
@@ -634,7 +633,7 @@ pow(0.5, -inf) = inf
pow(0.5, 0) = 1
pow(0.5, -0) = 1
pow(0.5, 0.5) = 0.707107
-pow(0.5, -0.5) = 1.41422
+pow(0.5, -0.5) = 1.41421
pow(0.5, 1) = 0.5
pow(0.5, -1) = 2
pow(0.5, 2) = 0.25
@@ -644,10 +643,10 @@ pow(-0.5, inf) = 0
pow(-0.5, -inf) = inf
pow(-0.5, 0) = 1
pow(-0.5, -0) = 1
-pow(-0.5, 0.5) = -nan
-pow(-0.5, -0.5) = -nan
-pow(-0.5, 1) = -0.499999
-pow(-0.5, -1) = -1.99999
+pow(-0.5, 0.5) = nan
+pow(-0.5, -0.5) = nan
+pow(-0.5, 1) = -0.5
+pow(-0.5, -1) = -2
pow(-0.5, 2) = 0.25
pow(-0.5, -2) = 4
pow(1, nan) = 1
@@ -666,8 +665,8 @@ pow(-1, inf) = 1
pow(-1, -inf) = 1
pow(-1, 0) = 1
pow(-1, -0) = 1
-pow(-1, 0.5) = -nan
-pow(-1, -0.5) = -nan
+pow(-1, 0.5) = nan
+pow(-1, -0.5) = nan
pow(-1, 1) = -1
pow(-1, -1) = -1
pow(-1, 2) = 1
@@ -677,20 +676,21 @@ pow(2, inf) = inf
pow(2, -inf) = 0
pow(2, 0) = 1
pow(2, -0) = 1
-pow(2, 0.5) = 1.41422
+pow(2, 0.5) = 1.41421
pow(2, -0.5) = 0.707107
pow(2, 1) = 2
-pow(2, -1) = 0.500001
+pow(2, -1) = 0.5
pow(2, 2) = 4
-pow(2, -2) = 0.250001
+pow(2, -2) = 0.25
pow(-2, nan) = nan
pow(-2, inf) = inf
pow(-2, -inf) = 0
pow(-2, 0) = 1
pow(-2, -0) = 1
-pow(-2, 0.5) = -nan
-pow(-2, -0.5) = -nan
-pow(-2, 1) = -1.99999
+pow(-2, 0.5) = nan
+pow(-2, -0.5) = nan
+pow(-2, 1) = -2
pow(-2, -1) = -0.5
pow(-2, 2) = 4
-pow(-2, -2) = 0.250001
+pow(-2, -2) = 0.25
+23:54:32 [32mINFO [0m [90msrc/main.c:338:[0m RISC-V emulator is destroyed
```
## Reference
https://zh.wikipedia.org/zh-tw/IEEE_754
https://en.wikipedia.org/wiki/Quadruple-precision_floating-point_format
https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html
Sign in with Wallet
Connect another wallet