diff --git a/sm4/aesni_macros_ppc64x.s b/sm4/aesni_macros_ppc64x.s
index ac0809d..c2a37f8 100644
--- a/sm4/aesni_macros_ppc64x.s
+++ b/sm4/aesni_macros_ppc64x.s
@@ -1,22 +1,14 @@
 #define LOAD_CONSTS(baseAddrReg, offsetReg) \
-	LXVD2X (baseAddrReg)(R0), M0; \
+	LXVD2X (baseAddrReg)(R0), REVERSE_WORDS; \
 	MOVD $0x10, offsetReg; \
-	LXVD2X (baseAddrReg)(offsetReg), M1; \
-	MOVD $0x20, offsetReg; \
-	LXVD2X (baseAddrReg)(offsetReg), M2; \
-	MOVD $0x30, offsetReg; \
-	LXVD2X (baseAddrReg)(offsetReg), M3; \
-	MOVD $0x40, offsetReg; \
-	LXVD2X (baseAddrReg)(offsetReg), REVERSE_WORDS; \
-	MOVD $0x50, offsetReg; \
 	LXVD2X (baseAddrReg)(offsetReg), NIBBLE_MASK; \
-	MOVD $0x70, offsetReg; \
+	MOVD $0x20, offsetReg; \
 	LXVD2X (baseAddrReg)(offsetReg), M1L; \
-	MOVD $0x80, offsetReg; \
+	MOVD $0x30, offsetReg; \
 	LXVD2X (baseAddrReg)(offsetReg), M1H; \
-	MOVD $0x90, offsetReg; \
+	MOVD $0x40, offsetReg; \
 	LXVD2X (baseAddrReg)(offsetReg), M2L; \
-	MOVD $0xa0, offsetReg; \
+	MOVD $0x50, offsetReg; \
 	LXVD2X (baseAddrReg)(offsetReg), M2H
 
 #ifdef GOARCH_ppc64le
diff --git a/sm4/asm_ppc64x.s b/sm4/asm_ppc64x.s
index 4b97bf7..d639258 100644
--- a/sm4/asm_ppc64x.s
+++ b/sm4/asm_ppc64x.s
@@ -8,29 +8,19 @@
 
 DATA ·rcon+0x00(SB)/8, $0x0b0a09080f0e0d0c // byte swap per word
 DATA ·rcon+0x08(SB)/8, $0x0302010007060504
-DATA ·rcon+0x10(SB)/8, $0x0001020310111213 // Permute for transpose matrix
-DATA ·rcon+0x18(SB)/8, $0x0405060714151617
-DATA ·rcon+0x20(SB)/8, $0x08090a0b18191a1b
-DATA ·rcon+0x28(SB)/8, $0x0c0d0e0f1c1d1e1f
-DATA ·rcon+0x30(SB)/8, $0x0001020304050607
-DATA ·rcon+0x38(SB)/8, $0x1011121314151617
-DATA ·rcon+0x40(SB)/8, $0x08090a0b0c0d0e0f
-DATA ·rcon+0x48(SB)/8, $0x18191a1b1c1d1e1f
-DATA ·rcon+0x50(SB)/8, $0x0c0d0e0f08090a0b // reverse words
-DATA ·rcon+0x58(SB)/8, $0x0405060700010203
-DATA ·rcon+0x60(SB)/8, $0x0F0F0F0F0F0F0F0F // nibble mask
-DATA ·rcon+0x68(SB)/8, $0x0F0F0F0F0F0F0F0F
-DATA ·rcon+0x70(SB)/8, $0x000D0A0704010E0B // inverse shift rows
-DATA ·rcon+0x78(SB)/8, $0x0805020F0C090603
-DATA ·rcon+0x80(SB)/8, $0x691CA0D5B6C37F0A // affine transform matrix m1 low
-DATA ·rcon+0x88(SB)/8, $0x53269AEF8CF94530
-DATA ·rcon+0x90(SB)/8, $0x009837AF6CF45BC3 // affine transform matrix m1 high
-DATA ·rcon+0x98(SB)/8, $0xAB339C04C75FF068
-DATA ·rcon+0xa0(SB)/8, $0x616EF1FE050A959A // affine transform matrix m2 low
-DATA ·rcon+0xa8(SB)/8, $0xF5FA656A919E010E
-DATA ·rcon+0xb0(SB)/8, $0x00A4E044CD692D89 // affine transform matrix m2 high
-DATA ·rcon+0xb8(SB)/8, $0xA50145E168CC882C
-GLOBL ·rcon(SB), RODATA, $192
+DATA ·rcon+0x10(SB)/8, $0x0c0d0e0f08090a0b // reverse words
+DATA ·rcon+0x18(SB)/8, $0x0405060700010203
+DATA ·rcon+0x20(SB)/8, $0x0F0F0F0F0F0F0F0F // nibble mask
+DATA ·rcon+0x28(SB)/8, $0x0F0F0F0F0F0F0F0F
+DATA ·rcon+0x30(SB)/8, $0x691CA0D5B6C37F0A // affine transform matrix m1 low
+DATA ·rcon+0x38(SB)/8, $0x53269AEF8CF94530
+DATA ·rcon+0x40(SB)/8, $0x009837AF6CF45BC3 // affine transform matrix m1 high
+DATA ·rcon+0x48(SB)/8, $0xAB339C04C75FF068
+DATA ·rcon+0x50(SB)/8, $0x616EF1FE050A959A // affine transform matrix m2 low
+DATA ·rcon+0x58(SB)/8, $0xF5FA656A919E010E
+DATA ·rcon+0x60(SB)/8, $0x00A4E044CD692D89 // affine transform matrix m2 high
+DATA ·rcon+0x68(SB)/8, $0xA50145E168CC882C
+GLOBL ·rcon(SB), RODATA, $112
 
 #define REVERSE_WORDS V19
 #define M1L V20
@@ -38,12 +28,7 @@ GLOBL ·rcon(SB), RODATA, $192
 #define M2L V22
 #define M2H V23
 #define V_FOUR V24
-#define M0 V25
-#define M1 V26
-#define M2 V27
-#define M3 V28
 #define NIBBLE_MASK V29
-#define INVERSE_SHIFT_ROWS V30
 // For instruction emulation
 #define ESPERMW  V31 // Endian swapping permute into BE
 
diff --git a/sm4/cbc_ppc64x.s b/sm4/cbc_ppc64x.s
index 4766eef..8371224 100644
--- a/sm4/cbc_ppc64x.s
+++ b/sm4/cbc_ppc64x.s
@@ -12,12 +12,7 @@
 #define M2L V22
 #define M2H V23
 #define V_FOUR V24
-#define M0 V25
-#define M1 V26
-#define M2 V27
-#define M3 V28
 #define NIBBLE_MASK V29
-#define INVERSE_SHIFT_ROWS V30
 // For instruction emulation
 #define ESPERMW  V31 // Endian swapping permute into BE
 
diff --git a/sm4/ecb_ppc64x.s b/sm4/ecb_ppc64x.s
index afac15a..14b3316 100644
--- a/sm4/ecb_ppc64x.s
+++ b/sm4/ecb_ppc64x.s
@@ -12,12 +12,7 @@
 #define M2L V22
 #define M2H V23
 #define V_FOUR V24
-#define M0 V25
-#define M1 V26
-#define M2 V27
-#define M3 V28
 #define NIBBLE_MASK V29
-#define INVERSE_SHIFT_ROWS V30
 // For instruction emulation
 #define ESPERMW  V31 // Endian swapping permute into BE