diff --git a/internal/sm2ec/p256_asm_amd64.s b/internal/sm2ec/p256_asm_amd64.s
index 43f1a20..75e126b 100644
--- a/internal/sm2ec/p256_asm_amd64.s
+++ b/internal/sm2ec/p256_asm_amd64.s
@@ -291,7 +291,7 @@ ordSqrLoop:
 	ADCQ DX, acc2
 	ADCQ $0, acc3
 	ADCQ $0, acc0
-	// calculate the positive part: [acc0, acc3, acc2, acc1] - [0, 0x100000000, 1, 0] * t0
+	// calculate the negative part: [acc0, acc3, acc2, acc1] - [0, 0x100000000, 1, 0] * t0
 	MOVQ t0, AX
 	MOVQ t0, DX
 	SHLQ $32, AX
@@ -484,7 +484,7 @@ ordSqrLoopBMI2:
 	ADCQ t1, acc2               // (carry4, acc2) = acc2 + t1 + carry3
 	ADCQ $0, acc3               // (carry5, acc3) = acc3 + carry4
 	ADCQ $0, acc0               //           acc0 = t0 + carry5 
-	// calculate the positive part: [acc0, acc3, acc2, acc1] - [0, 0x100000000, 1, 0] * t0
+	// calculate the negative part: [acc0, acc3, acc2, acc1] - [0, 0x100000000, 1, 0] * t0
 	MOVQ t0, AX
 	//MOVQ t0, DX              // This is not required due to t0=DX already
 	SHLQ $32, AX
@@ -759,66 +759,8 @@ TEXT sm2P256MulInternal(SB),NOSPLIT,$8
 	ADDQ mul0, acc6
 	ADCQ $0, mul1
 	MOVQ mul1, acc7
-	// First reduction step
-	MOVQ acc0, mul0
-	MOVQ acc0, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
+	sm2P256MulReductionInternal()
 	
-	SUBQ mul0, acc1
-	SBBQ mul1, acc2
-	SBBQ mul0, acc3
-	SBBQ mul1, acc0
-	// Second reduction step
-	MOVQ acc1, mul0
-	MOVQ acc1, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc1, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	
-	SUBQ mul0, acc2
-	SBBQ mul1, acc3
-	SBBQ mul0, acc0
-	SBBQ mul1, acc1
-	// Third reduction step
-	MOVQ acc2, mul0
-	MOVQ acc2, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc2, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	
-	SUBQ mul0, acc3
-	SBBQ mul1, acc0
-	SBBQ mul0, acc1
-	SBBQ mul1, acc2
-	// Last reduction step
-	MOVQ acc3, mul0
-	MOVQ acc3, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc3, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	
-	SUBQ mul0, acc0
-	SBBQ mul1, acc1
-	SBBQ mul0, acc2
-	SBBQ mul1, acc3
 	MOVQ $0, BP
 	// Add bits [511:256] of the result
 	ADCQ acc0, acc4
@@ -918,66 +860,7 @@ internalMulBMI2:
 	ADDQ mul0, acc6
 	ADCQ $0, acc7
 
-	// First reduction step
-	MOVQ acc0, mul0
-	MOVQ acc0, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
-	
-	SUBQ mul0, acc1
-	SBBQ mul1, acc2
-	SBBQ mul0, acc3
-	SBBQ mul1, acc0
-	// Second reduction step
-	MOVQ acc1, mul0
-	MOVQ acc1, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc1, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	
-	SUBQ mul0, acc2
-	SBBQ mul1, acc3
-	SBBQ mul0, acc0
-	SBBQ mul1, acc1
-	// Third reduction step
-	MOVQ acc2, mul0
-	MOVQ acc2, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc2, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	
-	SUBQ mul0, acc3
-	SBBQ mul1, acc0
-	SBBQ mul0, acc1
-	SBBQ mul1, acc2
-	// Last reduction step
-	MOVQ acc3, mul0
-	MOVQ acc3, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc3, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	
-	SUBQ mul0, acc0
-	SBBQ mul1, acc1
-	SBBQ mul0, acc2
-	SBBQ mul1, acc3
+	sm2P256MulReductionInternal()
 	MOVQ $0, BP
 	// Add bits [511:256] of the result
 	ADCQ acc0, acc4
diff --git a/internal/sm2ec/p256_asm_arm64.s b/internal/sm2ec/p256_asm_arm64.s
index 6c93e1b..398bde4 100644
--- a/internal/sm2ec/p256_asm_arm64.s
+++ b/internal/sm2ec/p256_asm_arm64.s
@@ -207,54 +207,57 @@ TEXT ·p256FromMont(SB),NOSPLIT,$0
 	LSL $32, acc0, y0
 	LSR	$32, acc0, y1
 
-	ADDS acc0, acc1, acc1
-	ADCS $0, acc2, acc2
-	ADCS $0, acc3, acc3
-	ADC $0, acc0, acc0
-
 	SUBS y0, acc1
 	SBCS y1, acc2
 	SBCS y0, acc3
-	SBC y1, acc0	
+	SBC y1, acc0, y0	
+
+	ADDS acc0, acc1, acc1
+	ADCS $0, acc2, acc2
+	ADCS $0, acc3, acc3
+	ADC $0, y0, acc0
+
 	// Second reduction step
 	LSL $32, acc1, y0
 	LSR	$32, acc1, y1
 
-	ADDS acc1, acc2, acc2
-	ADCS $0, acc3, acc3
-	ADCS $0, acc0, acc0
-	ADC $0, acc1, acc1
-
 	SUBS y0, acc2
 	SBCS y1, acc3
 	SBCS y0, acc0
-	SBC y1, acc1	
+	SBC y1, acc1, y0
+
+	ADDS acc1, acc2, acc2
+	ADCS $0, acc3, acc3
+	ADCS $0, acc0, acc0
+	ADC $0, y0, acc1
+
 	// Third reduction step
 	LSL $32, acc2, y0
 	LSR	$32, acc2, y1
 
-	ADDS acc2, acc3, acc3
-	ADCS $0, acc0, acc0
-	ADCS $0, acc1, acc1
-	ADC $0, acc2, acc2
-
 	SUBS y0, acc3
 	SBCS y1, acc0
 	SBCS y0, acc1
-	SBC y1, acc2
+	SBC y1, acc2, y0
+
+	ADDS acc2, acc3, acc3
+	ADCS $0, acc0, acc0
+	ADCS $0, acc1, acc1
+	ADC $0, y0, acc2
+
 	// Last reduction step
 	LSL $32, acc3, y0
 	LSR	$32, acc3, y1
 
-	ADDS acc3, acc0, acc0
-	ADCS $0, acc1, acc1
-	ADCS $0, acc2, acc2
-	ADC $0, acc3, acc3
-
 	SUBS y0, acc0
 	SBCS y1, acc1
 	SBCS y0, acc2
-	SBC y1, acc3
+	SBC y1, acc3, y0
+
+	ADDS acc3, acc0, acc0
+	ADCS $0, acc1, acc1
+	ADCS $0, acc2, acc2
+	ADC $0, y0, acc3
 
 	SUBS	const0, acc0, t0
 	SBCS	const1, acc1, t1
@@ -967,15 +970,15 @@ TEXT sm2P256MulInternal<>(SB),NOSPLIT,$0
 	LSL $32, acc0, t0
 	LSR	$32, acc0, t1
 
-	ADDS acc0, acc1, acc1
-	ADCS $0, acc2, acc2
-	ADCS $0, acc3, acc3
-	ADC $0, acc0, acc0
-
 	SUBS t0, acc1
 	SBCS t1, acc2
 	SBCS t0, acc3
-	SBC t1, acc0	
+	SBC t1, acc0, t0
+
+	ADDS acc0, acc1, acc1
+	ADCS $0, acc2, acc2
+	ADCS $0, acc3, acc3
+	ADC $0, t0, acc0
 
 	// y[1] * x
 	MUL	y1, x0, t0
@@ -1003,15 +1006,15 @@ TEXT sm2P256MulInternal<>(SB),NOSPLIT,$0
 	LSL $32, acc1, t0
 	LSR	$32, acc1, t1
 
-	ADDS acc1, acc2, acc2
-	ADCS $0, acc3, acc3
-	ADCS $0, acc0, acc0
-	ADC $0, acc1, acc1
-
 	SUBS t0, acc2
 	SBCS t1, acc3
 	SBCS t0, acc0
-	SBC t1, acc1	
+	SBC t1, acc1, t0
+
+	ADDS acc1, acc2, acc2
+	ADCS $0, acc3, acc3
+	ADCS $0, acc0, acc0
+	ADC $0, t0, acc1
 
 	// y[2] * x
 	MUL	y2, x0, t0
@@ -1039,15 +1042,15 @@ TEXT sm2P256MulInternal<>(SB),NOSPLIT,$0
 	LSL $32, acc2, t0
 	LSR	$32, acc2, t1
 
-	ADDS acc2, acc3, acc3
-	ADCS $0, acc0, acc0
-	ADCS $0, acc1, acc1
-	ADC $0, acc2, acc2
-
 	SUBS t0, acc3
 	SBCS t1, acc0
 	SBCS t0, acc1
-	SBC t1, acc2	
+	SBC t1, acc2, t0
+
+	ADDS acc2, acc3, acc3
+	ADCS $0, acc0, acc0
+	ADCS $0, acc1, acc1
+	ADC $0, t0, acc2
 
 	// y[3] * x
 	MUL	y3, x0, t0
@@ -1075,15 +1078,15 @@ TEXT sm2P256MulInternal<>(SB),NOSPLIT,$0
 	LSL $32, acc3, t0
 	LSR	$32, acc3, t1
 
-	ADDS acc3, acc0, acc0
-	ADCS $0, acc1, acc1
-	ADCS $0, acc2, acc2
-	ADC $0, acc3, acc3
-
 	SUBS t0, acc0
 	SBCS t1, acc1
 	SBCS t0, acc2
-	SBC t1, acc3	
+	SBC t1, acc3, t0
+
+	ADDS acc3, acc0, acc0
+	ADCS $0, acc1, acc1
+	ADCS $0, acc2, acc2
+	ADC $0, t0, acc3
 
 	// Add bits [511:256] of the mul result
 	ADDS	acc4, acc0, acc0
diff --git a/internal/sm2ec/p256_common_amd64.s b/internal/sm2ec/p256_common_amd64.s
index 32e874d..614355b 100644
--- a/internal/sm2ec/p256_common_amd64.s
+++ b/internal/sm2ec/p256_common_amd64.s
@@ -207,19 +207,19 @@ TEXT ·p256Mul(SB),NOSPLIT,$0
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc0, acc1
+	SUBQ AX, acc1
+	SBBQ DX, acc2
+	SBBQ AX, acc3
+	MOVQ acc0, AX
+	SBBQ DX, acc0
+
+	ADDQ AX, acc1
 	ADCQ $0, acc2
 	ADCQ $0, acc3
 	ADCQ acc0, acc4
 	ADCQ $0, acc5
 	
-	SUBQ AX, acc1
-	SBBQ DX, acc2
-	SBBQ AX, acc3
-	SBBQ DX, acc4
-	SBBQ $0, acc5
 	XORQ acc0, acc0
-
 	// x * y[1]
 	MOVQ (8*1)(y_ptr), t0
 
@@ -258,19 +258,19 @@ TEXT ·p256Mul(SB),NOSPLIT,$0
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc1, acc2
+	SUBQ AX, acc2
+	SBBQ DX, acc3
+	SBBQ AX, acc4
+	MOVQ acc1, AX
+	SBBQ DX, acc1
+
+	ADDQ AX, acc2
 	ADCQ $0, acc3
 	ADCQ $0, acc4
 	ADCQ acc1, acc5
 	ADCQ $0, acc0
 	
-	SUBQ AX, acc2
-	SBBQ DX, acc3
-	SBBQ AX, acc4
-	SBBQ DX, acc5
-	SBBQ $0, acc0	
 	XORQ acc1, acc1
-
 	// x * y[2]
 	MOVQ (8*2)(y_ptr), t0
 
@@ -309,17 +309,18 @@ TEXT ·p256Mul(SB),NOSPLIT,$0
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc2, acc3
+	SUBQ AX, acc3
+	SBBQ DX, acc4
+	SBBQ AX, acc5
+	MOVQ acc2, AX
+	SBBQ DX, acc2
+
+	ADDQ AX, acc3
 	ADCQ $0, acc4
 	ADCQ $0, acc5
 	ADCQ acc2, acc0
 	ADCQ $0, acc1
 	
-	SUBQ AX, acc3
-	SBBQ DX, acc4
-	SBBQ AX, acc5
-	SBBQ DX, acc0
-	SBBQ $0, acc1	
 	XORQ acc2, acc2
 	// x * y[3]
 	MOVQ (8*3)(y_ptr), t0
@@ -359,17 +360,18 @@ TEXT ·p256Mul(SB),NOSPLIT,$0
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc3, acc4
+	SUBQ AX, acc4
+	SBBQ DX, acc5
+	SBBQ AX, acc0
+	MOVQ acc3, AX
+	SBBQ DX, acc3
+
+	ADDQ AX, acc4
 	ADCQ $0, acc5
 	ADCQ $0, acc0
 	ADCQ acc3, acc1
 	ADCQ $0, acc2
 	
-	SUBQ AX, acc4
-	SBBQ DX, acc5
-	SBBQ AX, acc0
-	SBBQ DX, acc1
-	SBBQ $0, acc2	
 	p256PrimReduce(acc4, acc5, acc0, acc1, acc2, x_ptr, acc3, t0, BX, res_ptr)
 	RET
 
@@ -395,19 +397,19 @@ mulBMI2:
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc0, acc1
+	SUBQ AX, acc1
+	SBBQ DX, acc2
+	SBBQ AX, acc3
+	MOVQ acc0, AX
+	SBBQ DX, acc0
+
+	ADDQ AX, acc1
 	ADCQ $0, acc2
 	ADCQ $0, acc3
 	ADCQ acc0, acc4
 	ADCQ $0, acc5
-	
-	SUBQ AX, acc1
-	SBBQ DX, acc2
-	SBBQ AX, acc3
-	SBBQ DX, acc4
-	SBBQ $0, acc5
-	XORQ acc0, acc0
 
+	XORQ acc0, acc0
 	// x * y[1]
 	MOVQ (8*1)(y_ptr), DX
 	MULXQ (8*0)(x_ptr), AX, BX
@@ -436,19 +438,19 @@ mulBMI2:
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc1, acc2
+	SUBQ AX, acc2
+	SBBQ DX, acc3
+	SBBQ AX, acc4
+	MOVQ acc1, AX
+	SBBQ DX, acc1
+
+	ADDQ AX, acc2
 	ADCQ $0, acc3
 	ADCQ $0, acc4
 	ADCQ acc1, acc5
 	ADCQ $0, acc0
 	
-	SUBQ AX, acc2
-	SBBQ DX, acc3
-	SBBQ AX, acc4
-	SBBQ DX, acc5
-	SBBQ $0, acc0	
 	XORQ acc1, acc1
-
 	// x * y[2]
 	MOVQ (8*2)(y_ptr), DX
 
@@ -477,17 +479,18 @@ mulBMI2:
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc2, acc3
+	SUBQ AX, acc3
+	SBBQ DX, acc4
+	SBBQ AX, acc5
+	MOVQ acc2, AX
+	SBBQ DX, acc2
+
+	ADDQ AX, acc3
 	ADCQ $0, acc4
 	ADCQ $0, acc5
 	ADCQ acc2, acc0
 	ADCQ $0, acc1
 	
-	SUBQ AX, acc3
-	SBBQ DX, acc4
-	SBBQ AX, acc5
-	SBBQ DX, acc0
-	SBBQ $0, acc1	
 	XORQ acc2, acc2
 	// x * y[3]
 	MOVQ (8*3)(y_ptr), DX
@@ -517,17 +520,18 @@ mulBMI2:
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc3, acc4
+	SUBQ AX, acc4
+	SBBQ DX, acc5
+	SBBQ AX, acc0
+	MOVQ acc3, AX
+	SBBQ DX, acc3
+
+	ADDQ AX, acc4
 	ADCQ $0, acc5
 	ADCQ $0, acc0
 	ADCQ acc3, acc1
 	ADCQ $0, acc2
 	
-	SUBQ AX, acc4
-	SBBQ DX, acc5
-	SBBQ AX, acc0
-	SBBQ DX, acc1
-	SBBQ $0, acc2	
 	p256PrimReduce(acc4, acc5, acc0, acc1, acc2, x_ptr, acc3, t0, BX, res_ptr)
 	RET
 
@@ -550,32 +554,35 @@ TEXT ·p256FromMont(SB),NOSPLIT,$0
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc0, acc1
+	SUBQ AX, acc1
+	SBBQ DX, acc2
+	SBBQ AX, acc3
+	MOVQ acc0, AX
+	SBBQ DX, acc0
+
+	ADDQ AX, acc1
 	ADCQ $0, acc2
 	ADCQ $0, acc3
 	ADCQ acc0, acc4
 	
-	SUBQ AX, acc1
-	SBBQ DX, acc2
-	SBBQ AX, acc3
-	SBBQ DX, acc4
 	XORQ acc5, acc5
-
 	// Second stage
 	MOVQ acc1, AX
 	MOVQ acc1, DX
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc1, acc2
+	SUBQ AX, acc2
+	SBBQ DX, acc3
+	SBBQ AX, acc4
+	MOVQ acc1, AX
+	SBBQ DX, acc5
+
+	ADDQ AX, acc2
 	ADCQ $0, acc3
 	ADCQ $0, acc4
 	ADCQ acc1, acc5
 	
-	SUBQ AX, acc2
-	SBBQ DX, acc3
-	SBBQ AX, acc4
-	SBBQ DX, acc5
 	XORQ acc0, acc0
 	// Third stage
 	MOVQ acc2, AX
@@ -583,15 +590,17 @@ TEXT ·p256FromMont(SB),NOSPLIT,$0
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc2, acc3
+	SUBQ AX, acc3
+	SBBQ DX, acc4
+	SBBQ AX, acc5
+	MOVQ acc2, AX
+	SBBQ DX, acc2
+
+	ADDQ AX, acc3
 	ADCQ $0, acc4
 	ADCQ $0, acc5
 	ADCQ acc2, acc0
 	
-	SUBQ AX, acc3
-	SBBQ DX, acc4
-	SBBQ AX, acc5
-	SBBQ DX, acc0
 	XORQ acc1, acc1
 	// Last stage
 	MOVQ acc3, AX
@@ -599,15 +608,16 @@ TEXT ·p256FromMont(SB),NOSPLIT,$0
 	SHLQ $32, AX
 	SHRQ $32, DX
 
-	ADDQ acc3, acc4
-	ADCQ $0, acc5
-	ADCQ $0, acc0
-	ADCQ acc3, acc1
-	
 	SUBQ AX, acc4
 	SBBQ DX, acc5
 	SBBQ AX, acc0
-	SBBQ DX, acc1
+	MOVQ acc3, AX
+	SBBQ DX, acc3
+
+	ADDQ AX, acc4
+	ADCQ $0, acc5
+	ADCQ $0, acc0
+	ADCQ acc3, acc1
 	
 	MOVQ acc4, x_ptr
 	MOVQ acc5, acc3
diff --git a/internal/sm2ec/p256_macros_amd64.s b/internal/sm2ec/p256_macros_amd64.s
index dd5fb3e..976854c 100644
--- a/internal/sm2ec/p256_macros_amd64.s
+++ b/internal/sm2ec/p256_macros_amd64.s
@@ -237,6 +237,72 @@ GLOBL p256one<>(SB), 8, $32
 	CMOVQCS t2, acc6                   \
 	CMOVQCS t3, acc7
 
+#define sm2P256MulReductionInternal() \
+	\// First reduction step
+	MOVQ acc0, mul0                     \
+	MOVQ acc0, mul1                     \
+	SHLQ $32, mul0                     \
+	SHRQ $32, mul1                     \
+	\
+	SUBQ mul0, acc1                     \
+	SBBQ mul1, acc2                     \
+	SBBQ mul0, acc3                     \
+	MOVQ acc0, mul0                     \
+	SBBQ mul1, acc0                     \
+	\
+	ADDQ mul0, acc1                     \
+	ADCQ $0, acc2                     \
+	ADCQ $0, acc3                     \
+	ADCQ $0, acc0                     \
+	\// Second reduction step
+	MOVQ acc1, mul0                     \
+	MOVQ acc1, mul1                     \
+	SHLQ $32, mul0                     \
+	SHRQ $32, mul1                     \
+	\
+	SUBQ mul0, acc2                     \
+	SBBQ mul1, acc3                     \
+	SBBQ mul0, acc0                     \
+	MOVQ acc1, mul0                     \
+	SBBQ mul1, acc1                     \
+	\
+	ADDQ mul0, acc2                     \
+	ADCQ $0, acc3                     \
+	ADCQ $0, acc0                     \
+	ADCQ $0, acc1                     \
+	\// Third reduction step
+	MOVQ acc2, mul0                     \
+	MOVQ acc2, mul1                     \
+	SHLQ $32, mul0                     \
+	SHRQ $32, mul1                     \
+	\
+	SUBQ mul0, acc3                     \
+	SBBQ mul1, acc0                     \
+	SBBQ mul0, acc1                     \
+	MOVQ acc2, mul0                     \
+	SBBQ mul1, acc2                     \
+	\
+	ADDQ mul0, acc3                     \
+	ADCQ $0, acc0                     \
+	ADCQ $0, acc1                     \
+	ADCQ $0, acc2                     \
+	\// Last reduction step
+	MOVQ acc3, mul0                     \
+	MOVQ acc3, mul1                     \
+	SHLQ $32, mul0                     \
+	SHRQ $32, mul1                     \
+	\
+	SUBQ mul0, acc0                     \
+	SBBQ mul1, acc1                     \
+	SBBQ mul0, acc2                     \
+	MOVQ acc3, mul0                     \
+	SBBQ mul1, acc3                     \
+	\
+	ADDQ mul0, acc0                     \
+	ADCQ $0, acc1                     \
+	ADCQ $0, acc2                     \
+	ADCQ $0, acc3
+
 #define p256PointDoubleInit() \
 	MOVOU (16*0)(BX), X0 \
 	MOVOU (16*1)(BX), X1 \
diff --git a/internal/sm2ec/p256_plugin_amd64.s b/internal/sm2ec/p256_plugin_amd64.s
index 8594ca8..3223598 100644
--- a/internal/sm2ec/p256_plugin_amd64.s
+++ b/internal/sm2ec/p256_plugin_amd64.s
@@ -757,67 +757,9 @@ TEXT sm2P256MulInternal(SB),NOSPLIT,$8
 	ADDQ mul0, acc6
 	ADCQ $0, mul1
 	MOVQ mul1, acc7
-	// First reduction step
+
 	PEXTRQ $0, X0, acc0
-	MOVQ acc0, mul0
-	MOVQ acc0, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
-	
-	SUBQ mul0, acc1
-	SBBQ mul1, acc2
-	SBBQ mul0, acc3
-	SBBQ mul1, acc0
-	// Second reduction step
-	MOVQ acc1, mul0
-	MOVQ acc1, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc1, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	
-	SUBQ mul0, acc2
-	SBBQ mul1, acc3
-	SBBQ mul0, acc0
-	SBBQ mul1, acc1
-	// Third reduction step
-	MOVQ acc2, mul0
-	MOVQ acc2, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc2, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	
-	SUBQ mul0, acc3
-	SBBQ mul1, acc0
-	SBBQ mul0, acc1
-	SBBQ mul1, acc2
-	// Last reduction step
-	MOVQ acc3, mul0
-	MOVQ acc3, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc3, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	
-	SUBQ mul0, acc0
-	SBBQ mul1, acc1
-	SBBQ mul0, acc2
-	SBBQ mul1, acc3
+	sm2P256MulReductionInternal()
 	MOVQ $0, mul0
 	// Add bits [511:256] of the result
 	ADCQ acc0, acc4
@@ -918,67 +860,8 @@ internalMulBMI2:
 	ADDQ mul0, acc6
 	ADCQ $0, acc7
 
-	// First reduction step
 	PEXTRQ $0, X0, acc0
-	MOVQ acc0, mul0
-	MOVQ acc0, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
-	
-	SUBQ mul0, acc1
-	SBBQ mul1, acc2
-	SBBQ mul0, acc3
-	SBBQ mul1, acc0
-	// Second reduction step
-	MOVQ acc1, mul0
-	MOVQ acc1, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc1, acc2
-	ADCQ $0, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	
-	SUBQ mul0, acc2
-	SBBQ mul1, acc3
-	SBBQ mul0, acc0
-	SBBQ mul1, acc1
-	// Third reduction step
-	MOVQ acc2, mul0
-	MOVQ acc2, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc2, acc3
-	ADCQ $0, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	
-	SUBQ mul0, acc3
-	SBBQ mul1, acc0
-	SBBQ mul0, acc1
-	SBBQ mul1, acc2
-	// Last reduction step
-	MOVQ acc3, mul0
-	MOVQ acc3, mul1
-	SHLQ $32, mul0
-	SHRQ $32, mul1
-
-	ADDQ acc3, acc0
-	ADCQ $0, acc1
-	ADCQ $0, acc2
-	ADCQ $0, acc3
-	
-	SUBQ mul0, acc0
-	SBBQ mul1, acc1
-	SBBQ mul0, acc2
-	SBBQ mul1, acc3
+	sm2P256MulReductionInternal()
 	MOVQ $0, mul0
 	// Add bits [511:256] of the result
 	ADCQ acc0, acc4