From 0cbcc7200ba57096efb8bf0b5a66b09ad27f49f7 Mon Sep 17 00:00:00 2001
From: Peilin Wang <peilin.wang@huawei.com>
Date: Fri, 22 May 2020 17:03:19 -0400
Subject: [PATCH] made shuffle determinisitc for subsequent epochs

---
 .../dataset/engine/datasetops/shuffle_op.cc   |  12 ++++++------
 .../golden/test_2ops_shuffle_repeat.npz       | Bin 3831 -> 3831 bytes
 tests/ut/python/dataset/test_2ops.py          |   4 ++--
 3 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/mindspore/ccsrc/dataset/engine/datasetops/shuffle_op.cc b/mindspore/ccsrc/dataset/engine/datasetops/shuffle_op.cc
index d0ec89671f..69f7e09986 100644
--- a/mindspore/ccsrc/dataset/engine/datasetops/shuffle_op.cc
+++ b/mindspore/ccsrc/dataset/engine/datasetops/shuffle_op.cc
@@ -83,14 +83,14 @@ ShuffleOp::ShuffleOp(int32_t shuffle_size, uint32_t shuffle_seed, int32_t op_con
 // itself rather than waiting for the reset driven from operators above it in the pipeline.
 Status ShuffleOp::SelfReset() {
   MS_LOG(DEBUG) << "Shuffle operator performing a self-reset.";
-  // If ReshuffleEachEpoch is false, then we always use the same seed for every
+  // If reshuffle_each_epoch is false, then we always use the same seed for every
   // epoch.
-  // If ReshuffleEachEpoch is true, then the first epoch uses the given seed,
-  // and all subsequent epochs will then reset the seed based on random device.
-  if (reshuffle_each_epoch_) {
-    shuffle_seed_ = GetNewSeed();
+  // If reshuffle_each_epoch is true, then the first epoch uses the given seed,
+  // and all subsequent epochs will then keep on using the rng_ without resetting it
+  if (!reshuffle_each_epoch_) {
+    rng_ = std::mt19937_64(shuffle_seed_);
   }
-  rng_ = std::mt19937_64(shuffle_seed_);
+
   shuffle_buffer_ = std::make_unique<TensorTable>();
   buffer_counter_ = 0;
   shuffle_last_row_idx_ = 0;
diff --git a/tests/ut/data/dataset/golden/test_2ops_shuffle_repeat.npz b/tests/ut/data/dataset/golden/test_2ops_shuffle_repeat.npz
index 44af6db750edbb783296317290ffc3986218f6f0..5c39c0e64b87b77844a10c529ff628368029b4bb 100644
GIT binary patch
delta 1442
zcmZ9MS!|S57=~wsX{Rk(T9wk$z5SqU?(D^IQRqtn*$ezE(s2XC83iwhnI?)t@5q0}
zjUn;Ia6v-2AtZVy;Krij!e&vyh22EC((!%I>6C;?{^a?d?=0^*^UvI?yH{7a1o1>w
z#D5wi@g3<Ihbxao^9dv~L3XSSq=tI^h*H!VN|7ei87eS}FwT%d38CIF0wW0xhBA~B
z8V!%&Q9_fU5~B&thALDOS_}{2VM41RhB%?kf^2)y+7ZO6qj|sZLcb@+^l;3g2O}-^
zE3mzU?dOX-EcKgHFQZ;=s6`!%6^3!B&yZG9uPSc1+9#y$rC#GtP#<L1`eV|sqYn&^
zVGN7)h8m0w|A{x1wV;)v&(MbUuq83A9RBLIx$8KRP3(&0{ZT(1-s3peI#CRIawsTQ
z`HcQ^K3_a|#xIA@_!ZlyY-o?u{>pWlGxTTqd~Gm{r})M&0Uc_6o2Nf#n21TNzB5e5
zl#CkR)BnKdN5eEsr#NqzfthMupuT9Bh1o1F8RnqVO}R|}6Q4mt0~*=8VrW9Mru<C(
zi^1Q=ukJqFl&iKO$X;Xfx8m*n?ru+W{-FM|_|;q=`fBKJ(BI^9%P<v>Q`|N@fhRTP
z4)tBbQ+S%iUxsJ!Z1`MiUfX;up!nC&4QSkd99J+bL{BEjMWT_2VG$OG2U8zS^)GC$
z(n-C}95Gr&kgIaey}r3>$9&7ffs&p1&6|Q;GYvKoS_Cf=S_Ll?+5{OwyP%I?f|m&6
z1zQOd1g{c01X~Cb1+Ne$30@#fCUniM?mE=K-yQ2Iu1<D^`O-su$zeTH`o!%*VbBko
z<%ZSl-+#c@vt8Zv!O@Mrp5yBK%XYeY)b-sBzCYL1^{m5v>Dom5oeu}Y!|9G>=il5i
zyprCPk1@<Lj_Qu|7YgL%WF!eI1S1J61qs3`L7cE!FoMua=qcn$YotmDYXzl*b%GQj
z5JU;<1zKo>KpSt2h8M~<y9`_Cx*58kdaL@%uuY&0uL(4DyFeLU7bruP&|N6F6gf$y
zctfBRZwi!Rhd?QI3bfQ)0;LE)F5liOc!%cxA0x%P8loV(1q$+>K(pT$D99dxg6t&>
fq?kk>-N}GDCE?o@V`7Mg2P$e3`Wp^EshIvhw!M`R

delta 1366
zcmY+ENlesd7{+IY85vd;K?Fe-5nctg+FDUmOgx}G;8GXd#+9)u+RCKq#lw(P+s4*I
z^FMhpCSE+5s2AhGoBO`CRa?c@wYGIh)RT_S^UY#Q@+H6b_kG{HeEi`}<(tZ)`N#?u
z`;tF3zU<duwHzus=0_Mg(Ma!bLl7ZCg<$})36+L?6c9!j0>~m%8459wP;Dr}5JHV%
zC`t*nhFs)%Re{p{UEx%XKKPGJ4b=PNgD{v!I+B-2ofhexYpKHX+|4NUeCh@1qMs#e
zQ7@!kWEh4r28#{l7_QSxsJqflFHJU0y^MOfp#qf*Rv1R0s#M2U(nkz+7{y?fp&kw1
zyR6#$Vw6yHyE(k3U~|cT+vm<!H~f*_;Pybg8Z|U047C`koF|#{R66HrmoqAVhW;$q
zIYV;v^O?~vP+v48t6oaJo~~S`zruAjJ?OQ}px3EyXrj1A`q_=-%KXBOU%74?KEed<
z-7-wX$4Ymb`ZvQTn8e_B!>5?+uH2!&%hjKL{d<|$|AYF^^y}YGy*_V0pnu5q$hy#s
zDKw7_Q_-R_Pw4+LOv7{rPYpBB>dO30{}0!{hS`|Iy=R7YKv$kqzc6&5lfg^FT+H+4
zglE+oMpL{sj6tK$z2murp$TK7k(kf#^BKlrymvT!p!xs5p;(dI)ECN=Lli_}#m>18
zI#%MC-?X<Y=j(WMZ6r352I~oRf^Nbn!RLf}!8$^NU?af<n+c-@TL@zW8wibp9zv5~
z6Je}iEnys?ZDvW^!D{|7FdpyXP@5OeJ=pz0x1@CX*fbXpe7ZArZMuuAGVu(>t*<g^
zTit%`-lJ=hWoEj#;raf4@29+qVELXs`@BPWjiFP|5{Z8AUf%Y2kj*UQDFg_M1X+Z|
zf&qjj0zaWkkWE-B$RR8vbSC1Y<x(NS3PCPmr65d*5ZZlxi3E9-r0T5}sNNdCcRPQ5
zhg?tEZ4s!4tpfG%g+M)gDNqmF1nQxe(4I)RE@F})Z+AgisGZaIc?Sx1Ojhb|w4ri4
z1WLVApw!<Al=?e?QhzVV@p=l&>NL?03N_J>0!_4=@ZpQ|_7#=~^{>-AUO44FP4|6x

diff --git a/tests/ut/python/dataset/test_2ops.py b/tests/ut/python/dataset/test_2ops.py
index cdf59ecf22..ef60a42e27 100644
--- a/tests/ut/python/dataset/test_2ops.py
+++ b/tests/ut/python/dataset/test_2ops.py
@@ -47,7 +47,7 @@ def test_2ops_repeat_shuffle():
     save_and_check(data1, parameters, filename, generate_golden=GENERATE_GOLDEN)
 
 
-def skip_test_2ops_shuffle_repeat():
+def test_2ops_shuffle_repeat():
     """
     Test Shuffle then Repeat
     """
@@ -159,7 +159,7 @@ def test_2ops_shuffle_batch():
 
 if __name__ == '__main__':
     test_2ops_repeat_shuffle()
-    # test_2ops_shuffle_repeat()
+    test_2ops_shuffle_repeat()
     test_2ops_repeat_batch()
     test_2ops_batch_repeat()
     test_2ops_batch_shuffle()