diff --git a/.travis.yml b/.travis.yml
index c43927f..86a202e 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -2,8 +2,8 @@ language: go
 
 jobs:
   include:
-#    - arch: ppc64le
-#      go: 1.17.x
+    - arch: ppc64le
+      go: 1.17.x
     - arch: arm64-graviton2
       virt: vm
       os: linux
diff --git a/README.md b/README.md
index 57f7570..f857e00 100644
--- a/README.md
+++ b/README.md
@@ -1,7 +1,10 @@
 
 # GM-Standards SM2/SM3/SM4 for Go
 
-[![Build Status](https://app.travis-ci.com/emmansun/gmsm.svg?branch=main)](https://app.travis-ci.com/emmansun/gmsm) <a href="https://github.com/emmansun/gmsm/actions?query=workflow%3Aci+branch%3Amain"><img src="https://github.com/emmansun/gmsm/workflows/ci/badge.svg" alt="CI" style="max-width: 100%;"></a> [![Documentation](https://godoc.org/github.com/emmansun/gmsm?status.svg)](https://godoc.org/github.com/emmansun/gmsm) [![Release](https://img.shields.io/github/release/emmansun/gmsm/all.svg)](https://github.com/emmansun/gmsm/releases)
+[![Travis CI](https://app.travis-ci.com/emmansun/gmsm.svg?branch=main)](https://app.travis-ci.com/emmansun/gmsm) 
+[![Github CI](https://github.com/emmansun/gmsm/actions/workflows/ci.yml/badge.svg)](https://github.com/emmansun/gmsm/actions/workflows/ci.yml) 
+[![Documentation](https://godoc.org/github.com/emmansun/gmsm?status.svg)](https://godoc.org/github.com/emmansun/gmsm) 
+[![Release](https://img.shields.io/github/release/emmansun/gmsm/all.svg)](https://github.com/emmansun/gmsm/releases)
 
 This is a **SM2 sm2p256v1** implementation whose performance is similar like golang native NIST P256 under **amd64** and **arm64**, for implementation detail, please refer [SM2实现细节](https://github.com/emmansun/gmsm/wiki/SM2%E6%80%A7%E8%83%BD%E4%BC%98%E5%8C%96).
 
diff --git a/sm2/p256.go b/sm2/p256.go
index 5473290..923f46a 100644
--- a/sm2/p256.go
+++ b/sm2/p256.go
@@ -896,20 +896,14 @@ func p256Invert(out, in *[p256Limbs]uint32) {
 	} // 2^128 - 2^64
 	p256Mul(&ftmp, &ftmp, &e64) // 2^128 - 1
 
-	for i := 0; i < 32; i++ {
+	for i := 0; i < 96; i++ {
 		p256Square(&ftmp, &ftmp)
-	} // 2^160 - 2^32
-
-	p256Mul(&ftmp, &ftmp, &e32m2) // 2^160 - 2
-
-	for i := 0; i < 95; i++ {
-		p256Square(&ftmp, &ftmp)
-	} // 2^255 - 2^96
+	} // 2^224 - 2^96
 
 	p256Assign(&ftmp2, &e32m2)
-	for i := 0; i < 223; i++ {
+	for i := 0; i < 224; i++ {
 		p256Square(&ftmp2, &ftmp2)
-	} // 2^255 - 2^224
+	} // 2^256 - 2^225
 
 	p256Mul(&ftmp, &ftmp, &ftmp2) // 2^256 - 2^224 - 2^96
 
diff --git a/sm2/p256_asm.go b/sm2/p256_asm.go
index 5e4b812..7bdbe5e 100644
--- a/sm2/p256_asm.go
+++ b/sm2/p256_asm.go
@@ -389,13 +389,11 @@ func p256Inverse(out, in []uint64) {
 	p256Sqr(out, p32, 32)  //2^64 - 2^32
 	p256Mul(p64, out, p32) // 2^64 - 2^0
 
-	p256Sqr(out, p64, 64)    //2^128 - 2^64
-	p256Mul(out, out, p64)   // 2^128 - 2^0
-	p256Sqr(out, out, 32)    //2^160 - 2^32
-	p256Mul(out, out, p32m2) //2^160 - 2^1
-	p256Sqr(ptmp, out, 95)   //2^255 - 2^96
+	p256Sqr(out, p64, 64)  //2^128 - 2^64
+	p256Mul(out, out, p64) // 2^128 - 2^0
+	p256Sqr(ptmp, out, 96) // 2^224 - 2^96
 
-	p256Sqr(out, p32m2, 223) //2^255 - 2^224
+	p256Sqr(out, p32m2, 224) //2^256 - 2^225
 	p256Mul(ptmp, ptmp, out) //2^256 - 2^224 - 2^96
 
 	p256Sqr(out, p32, 16)  // 2^48 - 2^16