feat: add vector addition

.github/workflows/hw1.yaml

@@ -4,7 +4,7 @@
 
 # TODO
 #jobs:
-#  test:
+#  vector_addition:
 #    runs-on: ${{ matrix.os }}
 #    strategy:
 #      fail-fast: false
@@ -21,6 +21,6 @@
 #      - name: dependencies
 #        working-directory: ./homeworks/hw1
 #        run: go mod tidy
-#      - name: test
+#      - name: vector_addition
 #        working-directory: ./homeworks/hw1
-#        run: go test -v ./...
+#        run: go vector_addition -v ./...

homeworks/hw1/cmd/watchdog/tests/watchdog_integration_test_server.go

@@ -36,7 +36,7 @@ func main() {
 	}
 
 	if testEnv != "TEST_ENV_DATA" {
-		panic("test inh mismatch")
+		panic("vector_addition inh mismatch")
 	}
 
 	reader := bufio.NewReader(os.Stdin)

open_lessons/asm/simd/vector_addition/vector_addition_arm64.s

@@ -0,0 +1,32 @@
+#include "textflag.h"
+
+TEXT ·vectorAddition(SB), NOSPLIT, $0
+    LDP first_base+0(FP), (R0, R1)
+    LDP second_base+24(FP), (R2, R3)
+    LDP dst_base+48(FP), (R4, R5)
+
+    MOVD $0, R7
+    MOVD R4, R11
+    MOVD R0, R12
+    MOVD R2, R13
+
+loop:
+    CMP R5, R7
+    BGE done
+
+    VLD1 (R12), [V1.S4]
+    VLD1 (R13), [V2.S4]
+    VADD V1.S4, V2.S4, V3.S4
+
+    VST1 [V3.S4], (R11)
+
+    ADD $4, R7
+
+    ADD $16, R11
+    ADD $16, R12
+    ADD $16, R13
+
+    B loop
+
+done:
+    RET

open_lessons/asm/simd/vector_addition/vector_addition_test.go

@@ -0,0 +1,55 @@
+package main
+
+import (
+	"math/rand/v2"
+	"testing"
+)
+
+func vectorAddition(first, second, dst []uint32)
+
+func vectorAdditionV0(first, second, dst []uint32) {
+	for i := 0; i < len(first); i++ {
+		dst[i] = first[i] + second[i]
+	}
+}
+
+func BenchmarkAdd(b *testing.B) {
+	b.Run("SIMD vector addition", func(b *testing.B) {
+		b.StopTimer()
+		f, s, dst := getData()
+		b.StartTimer()
+
+		for i := 0; i < b.N; i++ {
+			vectorAddition(f, s, dst)
+		}
+	})
+
+	b.Run("simple vector addition", func(b *testing.B) {
+		b.StopTimer()
+		f, s, dst := getData()
+		b.StartTimer()
+
+		for i := 0; i < b.N; i++ {
+			vectorAdditionV0(f, s, dst)
+		}
+	})
+}
+
+// assume alignment
+func getData() ([]uint32, []uint32, []uint32) {
+	first := make([]uint32, 1_000_000)
+
+	for i := 0; i < len(first); i++ {
+		first[i] = rand.N[uint32](5)
+	}
+
+	second := make([]uint32, 1_000_000)
+
+	for i := 0; i < len(second); i++ {
+		second[i] = rand.N[uint32](5)
+	}
+
+	dst := make([]uint32, 1_000_000)
+
+	return first, second, dst
+}