Merge pull request #1599 from gitlost/fract_trunc_classify_#1574

Fix for trunc_f16/32/64 (#1574)

core/math/math.odin

@@ -396,7 +396,7 @@ trunc_f16   :: proc "contextless" (x: f16) -> f16 {
 		e := (x >> shift) & mask - bias
 
 		if e < shift {
-			x &= ~(1 << (shift-e)) - 1
+			x &~= 1 << (shift-e) - 1
 		}
 		return transmute(f16)x
 	}
@@ -428,7 +428,7 @@ trunc_f32   :: proc "contextless" (x: f32) -> f32 {
 		e := (x >> shift) & mask - bias
 
 		if e < shift {
-			x &= ~(1 << (shift-e)) - 1
+			x &~= 1 << (shift-e) - 1
 		}
 		return transmute(f32)x
 	}
@@ -460,7 +460,7 @@ trunc_f64   :: proc "contextless" (x: f64) -> f64 {
 		e := (x >> shift) & mask - bias
 
 		if e < shift {
-			x &= ~(1 << (shift-e)) - 1
+			x &~= 1 << (shift-e) - 1
 		}
 		return transmute(f64)x
 	}
@@ -473,6 +473,7 @@ trunc_f64   :: proc "contextless" (x: f64) -> f64 {
 }
 trunc_f64le :: proc "contextless" (x: f64le) -> f64le { return #force_inline f64le(trunc_f64(f64(x))) }
 trunc_f64be :: proc "contextless" (x: f64be) -> f64be { return #force_inline f64be(trunc_f64(f64(x))) }
+// Removes the fractional part of the value, i.e. rounds towards zero.
 trunc       :: proc{
 	trunc_f16, trunc_f16le, trunc_f16be,
 	trunc_f32, trunc_f32le, trunc_f32be, 
@@ -958,7 +959,7 @@ classify_f16   :: proc "contextless" (x: f16)   -> Float_Class {
 			return .Neg_Zero
 		}
 		return .Zero
-	case x*0.5 == x:
+	case x*0.25 == x:
 		if x < 0 {
 			return .Neg_Inf
 		}
@@ -1027,6 +1028,8 @@ classify_f64   :: proc "contextless" (x: f64)   -> Float_Class {
 }
 classify_f64le :: proc "contextless" (x: f64le) -> Float_Class { return #force_inline classify_f64(f64(x)) }
 classify_f64be :: proc "contextless" (x: f64be) -> Float_Class { return #force_inline classify_f64(f64(x)) }
+// Returns the `Float_Class` of the value, i.e. whether normal, subnormal, zero, negative zero, NaN, infinity or
+// negative infinity.
 classify       :: proc{
 	classify_f16, classify_f16le, classify_f16be,
 	classify_f32, classify_f32le, classify_f32be,
@@ -1715,4 +1718,22 @@ F32_BIAS  :: 0x7f
 
 F64_MASK  :: 0x7ff
 F64_SHIFT :: 64 - 12
-F64_BIAS  :: 0x3ff
+F64_BIAS  :: 0x3ff
+
+INF_F16     :f16: 0h7C00
+NEG_INF_F16 :f16: 0hFC00
+
+SNAN_F16    :f16: 0h7C01
+QNAN_F16    :f16: 0h7E01
+
+INF_F32     :f32: 0h7F80_0000
+NEG_INF_F32 :f32: 0hFF80_0000
+
+SNAN_F32    :f32: 0hFF80_0001
+QNAN_F32    :f32: 0hFFC0_0001
+
+INF_F64     :f64: 0h7FF0_0000_0000_0000
+NEG_INF_F64 :f64: 0hFFF0_0000_0000_0000
+
+SNAN_F64    :f64: 0h7FF0_0000_0000_0001
+QNAN_F64    :f64: 0h7FF8_0000_0000_0001

examples/demo/demo.odin

@@ -94,6 +94,7 @@ the_basics :: proc() {
 		z: f64 // `z` is typed of type `f64` (64-bit floating point number)
 		z = 1  // `1` is an untyped integer literal which can be implicitly converted to `f64`
 				// No need for any suffixes or decimal places like in other languages
+				// (with the exception of negative zero, which must be given as `-0.0`)
 				// CONSTANTS JUST WORK!!!
 
 

tests/common/common.odin

@@ -0,0 +1,40 @@
+// Boilerplate for tests
+package common
+
+import "core:testing"
+import "core:fmt"
+import "core:os"
+
+TEST_count := 0
+TEST_fail  := 0
+
+when ODIN_TEST {
+	expect  :: testing.expect
+	log     :: testing.log
+} else {
+	expect  :: proc(t: ^testing.T, condition: bool, message: string, loc := #caller_location) {
+		TEST_count += 1
+		if !condition {
+			TEST_fail += 1
+			fmt.printf("[%v] FAIL %v\n", loc, message)
+			return
+		}
+	}
+	log     :: proc(t: ^testing.T, v: any, loc := #caller_location) {
+		fmt.printf("[%v] ", loc)
+		fmt.printf("log: %v\n", v)
+	}
+}
+
+report :: proc(t: ^testing.T) {
+	if TEST_fail > 0 {
+		if TEST_fail > 1 {
+			fmt.printf("%v/%v tests successful, %v tests failed.\n", TEST_count - TEST_fail, TEST_count, TEST_fail)
+		} else {
+			fmt.printf("%v/%v tests successful, 1 test failed.\n", TEST_count - TEST_fail, TEST_count)
+		}
+		os.exit(1)
+	} else {
+		fmt.printf("%v/%v tests successful.\n", TEST_count, TEST_count)
+	}
+}

tests/core/Makefile

@@ -1,7 +1,8 @@
 ODIN=../../odin
 PYTHON=$(shell which python3)
 
-all: download_test_assets image_test compress_test strings_test hash_test crypto_test noise_test encoding_test
+all: download_test_assets image_test compress_test strings_test hash_test crypto_test noise_test encoding_test \
+	 math_test linalg_glsl_math_test
 
 download_test_assets:
 	$(PYTHON) download_assets.py
@@ -26,4 +27,10 @@ noise_test:
 
 encoding_test:
 	$(ODIN) run encoding/json -out=test_json
-	$(ODIN) run encoding/varint -out=test_varint
+	$(ODIN) run encoding/varint -out=test_varint
+
+math_test:
+	$(ODIN) run math/test_core_math.odin -out=test_core_math -collection:tests=..
+
+linalg_glsl_math_test:
+	$(ODIN) run math/linalg/glsl/test_linalg_glsl_math.odin -out=test_linalg_glsl_math -collection:tests=..

tests/core/math/linalg/glsl/test_linalg_glsl_math.odin

@@ -0,0 +1,85 @@
+// Tests "linalg_glsl_math.odin" in "core:math/linalg/glsl".
+// Must be run with `-collection:tests=` flag, e.g.
+// ./odin run tests/core/math/linalg/glsl/test_linalg_glsl_math.odin -collection:tests=./tests
+package test_core_math_linalg_glsl_math
+
+import glsl "core:math/linalg/glsl"
+
+import "core:fmt"
+import "core:math"
+import "core:testing"
+import tc "tests:common"
+
+main :: proc() {
+
+    t := testing.T{}
+
+	test_fract_f32(&t)
+	test_fract_f64(&t)
+
+	tc.report(&t)
+}
+
+@test
+test_fract_f32 :: proc(t: ^testing.T) {
+
+	using math
+
+	r: f32
+
+	Datum :: struct {
+		i: int,
+		v: f32,
+		e: f32,
+	}
+	@static data := []Datum{
+		{ 0, 10.5, 0.5 }, // Issue #1574 fract in linalg/glm is broken
+		{ 1, -10.5, -0.5 },
+		{ 2, F32_MIN, F32_MIN }, // 0x1p-126
+		{ 3, -F32_MIN, -F32_MIN },
+		{ 4, 0.0, 0.0 },
+		{ 5, -0.0, -0.0 },
+		{ 6, 1, 0.0 },
+		{ 7, -1, -0.0 },
+		{ 8, 0h3F80_0001, 0h3400_0000 }, // 0x1.000002p+0, 0x1p-23
+		{ 9, -0h3F80_0001, -0h3400_0000 },
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = glsl.fract(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%v (%h)) -> %v (%h) != %v", i, #procedure, d.v, d.v, r, r, d.e))
+	}
+}
+
+@test
+test_fract_f64 :: proc(t: ^testing.T) {
+
+	using math
+
+	r: f64
+
+	Datum :: struct {
+		i: int,
+		v: f64,
+		e: f64,
+	}
+	@static data := []Datum{
+		{ 0, 10.5, 0.5 }, // Issue #1574 fract in linalg/glm is broken
+		{ 1, -10.5, -0.5 },
+		{ 2, F64_MIN, F64_MIN }, // 0x1p-1022
+		{ 3, -F64_MIN, -F64_MIN },
+		{ 4, 0.0, 0.0 },
+		{ 5, -0.0, -0.0 },
+		{ 6, 1, 0.0 },
+		{ 7, -1, -0.0 },
+		{ 8, 0h3FF0_0000_0000_0001, 0h3CB0_0000_0000_0000 }, // 0x1.0000000000001p+0, 0x1p-52
+		{ 9, -0h3FF0_0000_0000_0001, -0h3CB0_0000_0000_0000 },
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = glsl.fract(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%v (%h)) -> %v (%h) != %v", i, #procedure, d.v, d.v, r, r, d.e))
+	}
+}

tests/core/math/test_core_math.odin

@@ -0,0 +1,310 @@
+// Tests "math.odin" in "core:math".
+// Must be run with `-collection:tests=` flag, e.g.
+// ./odin run tests/core/math/test_core_math.odin -collection:tests=./tests
+package test_core_math
+
+import "core:fmt"
+import "core:math"
+import "core:testing"
+import tc "tests:common"
+
+main :: proc() {
+    t := testing.T{}
+
+	test_classify_f16(&t)
+	test_classify_f32(&t)
+	test_classify_f64(&t)
+
+	test_trunc_f16(&t)
+	test_trunc_f32(&t)
+	test_trunc_f64(&t)
+
+	tc.report(&t)
+}
+
+@test
+test_classify_f16 :: proc(t: ^testing.T) {
+
+	using math
+	using Float_Class
+
+	r: Float_Class
+
+	Datum :: struct {
+		i: int,
+		v: f16,
+		e: math.Float_Class,
+	}
+	@static data := []Datum{
+		{ 0, 1.2, Normal },
+		{ 1, 0h0001, Subnormal },
+		{ 2, 0.0, Zero },
+		{ 3, -0.0, Neg_Zero },
+		{ 4, SNAN_F16, NaN },
+		{ 5, QNAN_F16, NaN },
+		{ 6, INF_F16, Inf },
+		{ 7, NEG_INF_F16, Neg_Inf },
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = classify_f16(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%h) -> %v != %v", i, #procedure, d.v, r, d.e))
+	}
+
+	/* Check all subnormals (exponent 0, 10-bit significand non-zero) */
+	for i :u16 = 1; i < 0x400; i += 1 {
+		v :f16 = transmute(f16)i
+		r = classify_f16(v)
+		e :Float_Class: Subnormal
+		tc.expect(t, r == e, fmt.tprintf("i:%d %s(%h) -> %v != %v", i, #procedure, v, r, e))
+	}
+}
+
+@test
+test_classify_f32 :: proc(t: ^testing.T) {
+
+	using math
+	using Float_Class
+
+	r: Float_Class
+
+	Datum :: struct {
+		i: int,
+		v: f32,
+		e: math.Float_Class,
+	}
+	@static data := []Datum{
+		{ 0, 1.2, Normal },
+		{ 1, 0h0000_0001, Subnormal },
+		{ 2, 0.0, Zero },
+		{ 3, -0.0, Neg_Zero },
+		{ 4, SNAN_F32, NaN },
+		{ 5, QNAN_F32, NaN },
+		{ 6, INF_F32, Inf },
+		{ 7, NEG_INF_F32, Neg_Inf },
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = classify_f32(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%h) -> %v != %v", i, #procedure, d.v, r, d.e))
+	}
+}
+
+@test
+test_classify_f64 :: proc(t: ^testing.T) {
+
+	using math
+	using Float_Class
+
+	r: Float_Class
+
+	Datum :: struct {
+		i: int,
+		v: f64,
+		e: math.Float_Class,
+	}
+	@static data := []Datum{
+		{ 0, 1.2, Normal },
+		{ 1, 0h0000_0000_0000_0001, Subnormal },
+		{ 2, 0.0, Zero },
+		{ 3, -0.0, Neg_Zero },
+		{ 4, SNAN_F64, NaN },
+		{ 5, QNAN_F64, NaN },
+		{ 6, INF_F64, Inf },
+		{ 7, NEG_INF_F64, Neg_Inf },
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = classify_f64(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%h) -> %v != %v", i, #procedure, d.v, r, d.e))
+	}
+}
+
+@test
+test_trunc_f16 :: proc(t: ^testing.T) {
+
+	using math
+
+	r, v: f16
+
+	Datum :: struct {
+		i: int,
+		v: f16,
+		e: f16,
+	}
+	@static data := []Datum{
+		{ 0, 10.5, 10 }, // Issue #1574 fract in linalg/glm is broken
+		{ 1, -10.5, -10 },
+
+		{ 2, F16_MAX, F16_MAX },
+		{ 3, -F16_MAX, -F16_MAX },
+		{ 4, F16_MIN, 0.0 },
+		{ 5, -F16_MIN, -0.0 },
+		{ 6, 0.0, 0.0 },
+		{ 7, -0.0, -0.0 },
+		{ 8, 1, 1 },
+		{ 9, -1, -1 },
+		{ 10, INF_F16, INF_F16 },
+		{ 11, NEG_INF_F16, NEG_INF_F16 },
+
+		/* From https://en.wikipedia.org/wiki/Half-precision_floating-point_format */
+		{ 12, 0h3C01, 1 }, // 0x1.004p+0 (smallest > 1)
+		{ 13, -0h3C01, -1 },
+		{ 14, 0h3BFF, 0.0 }, // 0x1.ffcp-1 (largest < 1)
+		{ 15, -0h3BFF, -0.0 },
+		{ 16, 0h0001, 0.0 }, // 0x0.004p-14 (smallest subnormal)
+		{ 17, -0h0001, -0.0 },
+		{ 18, 0h03FF, 0.0 }, // 0x0.ffcp-14 (largest subnormal)
+		{ 19, -0h03FF, -0.0 },
+
+		{ 20, 0hC809, -8 }, // -0x1.024p+3
+		{ 21, 0h4458, 4 }, // 0x1.16p+2
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = trunc_f16(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%h) -> %h != %h", i, #procedure, d.v, r, d.e))
+	}
+
+	v = SNAN_F16
+	r = trunc_f16(v)
+	tc.expect(t, is_nan_f16(r), fmt.tprintf("%s(%f) -> %f != NaN", #procedure, v, r))
+
+	v = QNAN_F16
+	r = trunc_f16(v)
+	tc.expect(t, is_nan_f16(r), fmt.tprintf("%s(%f) -> %f != NaN", #procedure, v, r))
+}
+
+@test
+test_trunc_f32 :: proc(t: ^testing.T) {
+
+	using math
+
+	r, v: f32
+
+	Datum :: struct {
+		i: int,
+		v: f32,
+		e: f32,
+	}
+	@static data := []Datum{
+		{ 0, 10.5, 10 }, // Issue #1574 fract in linalg/glm is broken
+		{ 1, -10.5, -10 },
+
+		{ 2, F32_MAX, F32_MAX },
+		{ 3, -F32_MAX, -F32_MAX },
+		{ 4, F32_MIN, 0.0 },
+		{ 5, -F32_MIN, -0.0 },
+		{ 6, 0.0, 0.0 },
+		{ 7, -0.0, -0.0 },
+		{ 8, 1, 1 },
+		{ 9, -1, -1 },
+		{ 10, INF_F32, INF_F32 },
+		{ 11, NEG_INF_F32, NEG_INF_F32 },
+
+		/* From https://en.wikipedia.org/wiki/Single-precision_floating-point_format */
+		{ 12, 0h3F80_0001, 1 }, // 0x1.000002p+0 (smallest > 1)
+		{ 13, -0h3F80_0001, -1 },
+		{ 14, 0h3F7F_FFFF, 0.0 }, // 0x1.fffffep-1 (largest < 1)
+		{ 15, -0h3F7F_FFFF, -0.0 },
+		{ 16, 0h0000_0001, 0.0 }, // 0x0.000002p-126 (smallest subnormal)
+		{ 17, -0h0000_0001, -0.0 },
+		{ 18, 0h007F_FFFF, 0.0 }, // 0x0.fffffep-126 (largest subnormal)
+		{ 19, -0h007F_FFFF, -0.0 },
+
+		/* From libc-test src/math/sanity/truncf.h */
+		{ 20, 0hC101_11D0, -8 }, // -0x1.0223ap+3
+		{ 21, 0h408B_0C34, 4 }, // 0x1.161868p+2
+		{ 22, 0hC106_1A5A, -8 }, // -0x1.0c34b4p+3
+		{ 23, 0hC0D1_0378, -6 }, // -0x1.a206fp+2
+		{ 24, 0h4114_45DE, 9 }, // 0x1.288bbcp+3
+		{ 25, 0h3F29_77E8, 0.0 }, // 0x1.52efdp-1
+		{ 26, 0hBED0_2E64, -0.0 }, // -0x1.a05cc8p-2
+		{ 27, 0h3F0F_CF7D, 0.0 }, // 0x1.1f9efap-1
+		{ 28, 0h3F46_2ED8, 0.0 }, // 0x1.8c5dbp-1
+		{ 29, 0hBF2D_C375, -0.0 }, // -0x1.5b86eap-1
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = trunc_f32(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%h) -> %h != %h", i, #procedure, d.v, r, d.e))
+	}
+
+	v = SNAN_F32
+	r = trunc_f32(v)
+	tc.expect(t, is_nan_f32(r), fmt.tprintf("%s(%f) -> %f != NaN", #procedure, v, r))
+
+	v = QNAN_F32
+	r = trunc_f32(v)
+	tc.expect(t, is_nan_f32(r), fmt.tprintf("%s(%f) -> %f != NaN", #procedure, v, r))
+}
+
+@test
+test_trunc_f64 :: proc(t: ^testing.T) {
+
+	using math
+
+	r, v: f64
+
+	Datum :: struct {
+		i: int,
+		v: f64,
+		e: f64,
+	}
+	data := []Datum{
+		{ 0, 10.5, 10 }, // Issue #1574 fract in linalg/glm is broken
+		{ 1, -10.5, -10 },
+
+		{ 2, F64_MAX, F64_MAX },
+		{ 3, -F64_MAX, -F64_MAX },
+		{ 4, F64_MIN, 0.0 },
+		{ 5, -F64_MIN, -0.0 },
+		{ 6, 0.0, 0.0 },
+		{ 7, -0.0, -0.0 },
+		{ 8, 1, 1 },
+		{ 9, -1, -1 },
+		{ 10, INF_F64, INF_F64 },
+		{ 11, NEG_INF_F64, NEG_INF_F64 },
+
+		/* From https://en.wikipedia.org/wiki/Double-precision_floating-point_format */
+		{ 12, 0h3FF0_0000_0000_0001, 1 }, // 0x1.0000000000001p+0 (smallest > 1)
+		{ 13, -0h3FF0_0000_0000_0001, -1 },
+		{ 14, 0h3FEF_FFFF_FFFF_FFFF, 0.0 }, // 0x1.fffffffffffffp-1 (largest < 1)
+		{ 15, -0h3FEF_FFFF_FFFF_FFFF, -0.0 },
+		{ 16, 0h0000_0000_0000_0001, 0.0 }, // 0x0.0000000000001p-1022 (smallest subnormal)
+		{ 17, -0h0000_0000_0000_0001, -0.0 },
+		{ 18, 0h000F_FFFF_FFFF_FFFF, 0.0 }, // 0x0.fffffffffffffp-1022 (largest subnormal)
+		{ 19, -0h000F_FFFF_FFFF_FFFF, -0.0 },
+
+		/* From libc-test src/math/sanity/trunc.h */
+		{ 20, 0hC020_2239_F3C6_A8F1, -8 }, // -0x1.02239f3c6a8f1p+3
+		{ 21, 0h4011_6186_8E18_BC67, 4 }, // 0x1.161868e18bc67p+2
+		{ 22, 0hC020_C34B_3E01_E6E7, -8 }, // -0x1.0c34b3e01e6e7p+3
+		{ 23, 0hC01A_206F_0A19_DCC4, -6 }, // -0x1.a206f0a19dcc4p+2
+		{ 24, 0h4022_88BB_B0D6_A1E6, 9 }, // 0x1.288bbb0d6a1e6p+3
+		{ 25, 0h3FE5_2EFD_0CD8_0497, 0.0 }, // 0x1.52efd0cd80497p-1
+		{ 26, 0hBFDA_05CC_7544_81D1, -0.0 }, // -0x1.a05cc754481d1p-2
+		{ 27, 0h3FE1_F9EF_9347_45CB, 0.0 }, // 0x1.1f9ef934745cbp-1
+		{ 28, 0h3FE8_C5DB_097F_7442, 0.0 }, // 0x1.8c5db097f7442p-1
+		{ 29, 0hBFE5_B86E_A811_8A0E, -0.0 }, // -0x1.5b86ea8118a0ep-1
+	}
+
+	for d, i in data {
+		assert(i == d.i)
+		r = trunc_f64(d.v)
+		tc.expect(t, r == d.e, fmt.tprintf("i:%d %s(%h) -> %h != %h", i, #procedure, d.v, r, d.e))
+	}
+
+	v = SNAN_F64
+	r = trunc_f64(v)
+	tc.expect(t, is_nan_f64(r), fmt.tprintf("%s(%f) -> %f != NaN", #procedure, v, r))
+
+	v = QNAN_F64
+	r = trunc_f64(v)
+	tc.expect(t, is_nan_f64(r), fmt.tprintf("%s(%f) -> %f != NaN", #procedure, v, r))
+}