Actually fix the patch

Author: Flamefire

easybuild/easyconfigs/p/PyTorch/PyTorch-1.9.0-foss-2020b.eb

@@ -46,7 +46,7 @@ checksums = [
     '7a6e512274f0b8673f4f207a5bc53387d88be7e79833f42d20365668b2118071',
     # PyTorch-1.9.0_avoid-failures-in-test_unary_ufuncs.patch
     'f600e6831f8a03af007845687d1e0f65b2394ca89a9dab5178e2cdc9bd384d43',
-    'd1f85b2f0f03b500a61b2456d6ec63fccb62f7edc350a820553e2891ec23bf13',  # PyTorch-1.9.0_fix-vsx-vector-functions.patch
+    'a4733b6b16a0db4ee5f85f2b103abc29bd711cfc5253f8dd8494d2b0c1509516',  # PyTorch-1.9.0_fix-vsx-vector-functions.patch
     # PyTorch-1.9.0_skip-lstm-serialization-test.patch
     '0fc14e29bd7530bcc09f4212df3c846072b1313216da86b827e102b85d695f49',
 ]

easybuild/easyconfigs/p/PyTorch/PyTorch-1.9.0-fosscuda-2020b-imkl.eb

@@ -64,7 +64,7 @@ checksums = [
     '8e8b417782e2f3004462c32338e12685e7296d15207f3e3087dcb8015e648f98',  # PyTorch-1.9.0_fix-testnn-on-A100.patch
     # PyTorch-1.9.0_fix-use-after-destruct-in-cudaipctypes.patch
     '67960bf9140baf004b07e29f7c2b338e7bc4e4e4f2c931768be44f58526e605f',
-    'd1f85b2f0f03b500a61b2456d6ec63fccb62f7edc350a820553e2891ec23bf13',  # PyTorch-1.9.0_fix-vsx-vector-functions.patch
+    'a4733b6b16a0db4ee5f85f2b103abc29bd711cfc5253f8dd8494d2b0c1509516',  # PyTorch-1.9.0_fix-vsx-vector-functions.patch
     # PyTorch-1.9.0_increase-test-cuda-tolerance.patch
     '73de855ab1ed38043c7fb2a983927786b83d7547aefed926f19e554e2214838a',
     # PyTorch-1.9.0_increase-tolerance-for-distributed-tests.patch

easybuild/easyconfigs/p/PyTorch/PyTorch-1.9.0-fosscuda-2020b.eb

@@ -63,7 +63,7 @@ checksums = [
     '8e8b417782e2f3004462c32338e12685e7296d15207f3e3087dcb8015e648f98',  # PyTorch-1.9.0_fix-testnn-on-A100.patch
     # PyTorch-1.9.0_fix-use-after-destruct-in-cudaipctypes.patch
     '67960bf9140baf004b07e29f7c2b338e7bc4e4e4f2c931768be44f58526e605f',
-    'd1f85b2f0f03b500a61b2456d6ec63fccb62f7edc350a820553e2891ec23bf13',  # PyTorch-1.9.0_fix-vsx-vector-functions.patch
+    'a4733b6b16a0db4ee5f85f2b103abc29bd711cfc5253f8dd8494d2b0c1509516',  # PyTorch-1.9.0_fix-vsx-vector-functions.patch
     # PyTorch-1.9.0_increase-test-cuda-tolerance.patch
     '73de855ab1ed38043c7fb2a983927786b83d7547aefed926f19e554e2214838a',
     # PyTorch-1.9.0_increase-tolerance-for-distributed-tests.patch

easybuild/easyconfigs/p/PyTorch/PyTorch-1.9.0_fix-vsx-vector-functions.patch

@@ -160,6 +160,35 @@ index 2a1a87aa72..5bcf818232 100644
    }
    Vec256<float> C10_ALWAYS_INLINE ceil() const {
      return {vec_ceil(_vec0), vec_ceil(_vec1)};
+@@ -593,27 +593,7 @@ class Vec256<float> {
+   }
+ 
+   Vec256<float> C10_ALWAYS_INLINE pow(const Vec256<float>& exp) const {
+-    auto x = *this;
+-    auto sign_bit = (*this) & sign_mask;
+-    // |b|
+-    auto exp_abs = exp.abs();
+-    auto exp_trunc = exp.trunc();
+-    Vec256<float> odd_mask;
+-    odd_mask._vecb0 = (vec_signed(exp._vec0) & vi_1) != vi_0;
+-    odd_mask._vecb1 = (vec_signed(exp._vec1) & vi_1) != vi_0;
+-    // using ln fuction
+-    auto temp = (abs().log() * exp).exp();
+-
+-    // is odd or even check from Sleef
+-    auto is_int = (exp == exp_trunc) | (exp_abs >= vcheck);
+-    auto is_odd = odd_mask & is_int & (exp_abs < vcheck);
+-    // if even then then pow result should be absolute
+-    auto temp_sign = temp | sign_bit; // copy_sign
+-    auto out = blendv(temp, temp_sign, is_odd);
+-    // x<0 and y != N, then NAN
+-    auto out1 = blendv(out, v_nan, ((exp.floor() != exp) & (x < zero)));
+-    // y = 0 then 1
+-    return blendv(out1, one, (exp_abs == zero));
++    return {Sleef_powf4_u10vsx(_vec0, exp._vec0), Sleef_powf4_u10vsx(_vec1, exp._vec1)};
+   }
+ 
+   Vec256<float> fmod(const Vec256<float>& b) const {
 @@ -653,8 +616,8 @@ class Vec256<float> {
    DEFINE_MEMBER_OP(operator-, float, vec_sub)
    DEFINE_MEMBER_OP(operator*, float, vec_mul)