| [eigen-commits] commit/eigen: 5 new changesets |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen-commits Archives
]
- To: eigen-commits@xxxxxxxxxxxxxxxxxxx
- Subject: [eigen-commits] commit/eigen: 5 new changesets
- From: Bitbucket <commits-noreply@xxxxxxxxxxxxx>
- Date: Wed, 12 Oct 2016 16:30:56 -0000
- Dkim-signature: v=1; a=rsa-sha1; c=relaxed; d=bitbucket.org; h=mime-version:content-type:content-transfer-encoding:subject:from:to; s=s1; bh=whFQC8s/VxUeVzaMpKekEqb3jTA=; b=2WI2wvm1NRv3wxp4DRs9Fd3 9tJOHsyNzHuV0hxeobDCtkGMdeWV7O34HInnNAkuD0YzoqPZMIUBxlR7Vj1BytQo zCCAAUEMOvpKnISn1KDMbSRT71lnNwP+VcaadA0ekhBmnzEFGQC0UxNrnTTow5KR PsID3W1uTd0shtNoZ9P0=
5 new commits in eigen:
https://bitbucket.org/eigen/eigen/commits/ea3d21db5a50/
Changeset: ea3d21db5a50
User: rmlarsen
Date: 2016-10-04 21:22:56+00:00
Summary: Fix a bug in the implementation of Carmack's fast sqrt algorithm in Eigen (enabled by EIGEN_FAST_MATH), which causes the vectorized parts of the computation to return -0.0 instead of NaN for negative arguments.
Benchmark speed in Giga-sqrts/s
Intel(R) Xeon(R) CPU E5-1650 v3 @ 3.50GHz
-----------------------------------------
SSE AVX
Fast=1 2.529G 4.380G
Fast=0 1.944G 1.898G
Fast=1 fixed 2.214G 3.739G
This table illustrates the worst case in terms speed impact: It was measured by repeatedly computing the sqrt of an n=4096 float vector that fits in L1 cache. For large vectors the operation becomes memory bound and the differences between the different versions almost negligible.
Affected #: 3 files
https://bitbucket.org/eigen/eigen/commits/2b11d472d9f7/
Changeset: 2b11d472d9f7
User: rmlarsen
Date: 2016-10-04 22:08:41+00:00
Summary: Update comment for fast sqrt.
Affected #: 1 file
https://bitbucket.org/eigen/eigen/commits/24587e3541cf/
Changeset: 24587e3541cf
User: rmlarsen
Date: 2016-10-04 22:09:11+00:00
Summary: Update comment for fast sqrt.
Affected #: 1 file
https://bitbucket.org/eigen/eigen/commits/c8a3ef8107c9/
Changeset: c8a3ef8107c9
User: rmlarsen
Date: 2016-10-12 15:34:39+00:00
Summary: Fix copy-paste error: Must use _mm256_cmp_ps for AVX.
Affected #: 1 file
https://bitbucket.org/eigen/eigen/commits/fd9765d4abad/
Changeset: fd9765d4abad
User: ggael
Date: 2016-10-12 16:30:51+00:00
Summary: Merged in rmlarsen/eigen (pull request #230)
Fix a bug in psqrt for SSE and AVX when EIGEN_FAST_MATH=1
Affected #: 3 files
Repository URL: https://bitbucket.org/eigen/eigen/
--
This is a commit notification from bitbucket.org. You are receiving
this because you have the service enabled, addressing the recipient of
this email.