Ticket #5311 (closed defect: fixed)
[with spkg, positive review] Update to ATLAS 3.8.3 (latest upstream release)
| Reported by: | mabshoff | Owned by: | mabshoff |
|---|---|---|---|
| Priority: | critical | Milestone: | sage-3.3 |
| Component: | packages: standard | Keywords: | |
| Cc: | Work issues: | ||
| Report Upstream: | Reviewers: | ||
| Authors: | Merged in: | ||
| Dependencies: | Stopgaps: |
Description
Clint writes:
I have released 3.8.3. This is, of course, mainly a bugfix release. However, I have also backported some assembly kernels from the 3.9 series for the Core2 and AMD64K10h architectures. We have architectural defaults for these systems, as well as the new Corei7 (64 bit only). For Core2-based systems, the speedup is substantial (K10h speedup is modest). You should also be aware of one error, *which has not been fixed in 3.8.3* http://math-atlas.sourceforge.net/errata.html#scal0 Cheers, Clint
Change History
comment:2 Changed 4 years ago by mabshoff
- Status changed from new to assigned
- Summary changed from Update to ATLAS 3.8.3 (latest upstream release) to [with spkg, needs review] Update to ATLAS 3.8.3 (latest upstream release)
The spkg at
http://sage.math.washington.edu/home/mabshoff/release-cycles-3.3/rc3/atlas-3.8.3.spkg
updates to ATLAS 3.8.3. The following patches were rebased/changed:
- archinfo_linux.c - only the Itanium hunk is needed
- archinfo_x86.c - only case 6 PIV hunk is needed
- Make.top - identical
- probe_comp.c - needs to be rebased to account for SiCortex? fix
Other than that the Core i7 support works, it identifies the Itanium2 boxen on SkyNet? correctly (vanilla ATLAS doesn't) and this spkg also fixes #1641.
Cheers,
Michael
comment:3 Changed 4 years ago by ghtdak
- Summary changed from [with spkg, needs review] Update to ATLAS 3.8.3 (latest upstream release) to [with spkg, positive review] Update to ATLAS 3.8.3 (latest upstream release)
Happily installs and runs on 3.3.alpha4 (what I had laying around on the i7 box).
Other comments recommended by mabshoff:
(12:51:46 PM) mabs: Well, it is a very clean spkg with 3 beautiful commits.
(12:52:06 PM) mabs: This is the best spkg ever. Michael needs to get a gold star :)
(12:52:41 PM) mabs: I tested it widely, so it will work :)
seems about right
