Update Eigen to commit:8f2c6f0aa6a59e24df245ef36abfb45ea190766b
Get rid of a lot of cruft in TensorFlow kernels now that Eigen requires c++14 (and thus always supports index lists).
CHANGELOG
=========
7db0ac977 - Remove extraneous ")".
09c0085a5 - Only test pmsub, pnmadd, and pnmsub on signed types.
8f2c6f0aa - Make preciprocal IEEE compliant w.r.t. 1/0 and 1/inf.
d271a7d54 - reduce float warnings (comparisons and implicit conversions)
51311ec65 - Remove inline assembly for FMA (AVX) and add remaining extensions as packet ops: pmsub, pnmadd, and pnmsub.
4e629b3c1 - make casts explicit and fixed the type
ea2c02060 - Add reciprocal packet op and fast specializations for float with SSE, AVX, and AVX512.
4b0926f99 - Prevent heap allocation in diagonal product
a0fc640c1 - Add support for packets of int64 on x86
970640519 - Cleanup
81c928ba5 - Silence some MSVC warnings
c454b8c81 - Improve clang warning suppressions by checking if warning is supported
fb05198bd - Port EIGEN_OPTIMIZATION_BARRIER to soft float arm
937c3d73c - Fix implicit conversion warning in GEBP kernel's packing
49a8a1e07 - Minor correction/clarification to LSCG solver documentation
5fe011572 - Update comment referencing removed macro EIGEN_SIZE_MIN_PREFER_DYNAMIC
ff4dffc34 - fix implicit conversion warning in vectorwise_reverse_inplace
708fd6d13 - Add MMA and performance improvements for VSX in GEMV for PowerPC.
d1bf05639 - fix compilation issue with gcc < 10 and -std=c++2a
PiperOrigin-RevId: 424723393
Change-Id: Idfd5250e558b788bc118dad3b4706d78a9f808ad
85 files changed
