GNU MPFR 3.1.3 Release Candidate
The release of GNU MPFR 3.1.3 ("canard à l'orange" patch level 3)
is imminent. Please help to make this release as good as possible
by downloading and testing this release candidate:
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.tar.xz
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.tar.bz2
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.tar.gz
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.zip
The MD5's:
f8a0180c2b31c0f9ce32dc44e1035fae mpfr-3.1.3-rc1.tar.bz2
b1a32a83549164b0aa7ffd6678449ac2 mpfr-3.1.3-rc1.tar.gz
e7f64614e5b080bcfccdaf8c9c3d972d mpfr-3.1.3-rc1.tar.xz
f57a88396cba13bef4cb2f80c70390c5 mpfr-3.1.3-rc1.zip
The SHA1's:
515131a35a93e699c4f7de272f9fb509703e0179 mpfr-3.1.3-rc1.tar.bz2
005021d9b167b5f94f23a7983a5f24719a714166 mpfr-3.1.3-rc1.tar.gz
625010d3617c653b2f0caf87e7d9391baedad5a1 mpfr-3.1.3-rc1.tar.xz
0ac3edc7446fe3ab31e105a83d6aa93744516049 mpfr-3.1.3-rc1.zip
The signatures:
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.tar.xz.asc
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.tar.bz2.asc
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.tar.gz.asc
http://www.mpfr.org/mpfr-3.1.3/mpfr-3.1.3-rc1.zip.asc
Each tarball is signed by Vincent Lefèvre. This can be verified
using the DSA key ID 98C3739D; this key can be retrieved with:
gpg --recv-keys 98C3739D
or by downloading it from .
The key fingerprint is:
07F3 DBBE CC1A 3960 5078 094D 980C 1976 98C3 739D
The signatures can be verified with: gpg --verify
You should check that the key fingerprint matches.
Changes from version 3.1.2 to version 3.1.3:
- Better support for Automake 1.13+ (now used to generate the tarball).
- Improved MPFR manual.
- Bug fixes (see and ChangeLog file).
Please send success and failure reports with "./config.guess" output
to .
If no problems are found, GNU MPFR 3.1.3 should be released
around 2015-06-20.
Regards,
--
Vincent Lefèvre - Web:
Proposal of new Unrolling degree before/after the allocated Register Allocation is done in GCC.
All: Given a Data Dependency Graph(DDG) the unrolling degree proposed by Monica Lam et.al calculates the unrolling degree as follows. Unrolling degree = Length of Longest Live range/ Number of cycles in the kernel ( Initiation Interval). The unrolling degree based on the Above leads to more register Spills and Fetch inside the Loops. Given the Data Dependency Graph(DDG) and the unrolling degree calculated as above we build the reuse graph. Lot has been presented On the Reuse graph. Each node nodes labelled the DDG nodes that writes into registers and the arc(u,v) represents that the destination Share same registers if there is no dependency between the between the iterations of DDG nodes. As there is no dependency as represented by the Reuse graph the life time will not overlap and assign the same registers. Each arc will Have the weight that describes the allocated registers and the Unrolling as proposed by Monica Lam given above The reuse graph will be Partitioned based on the heuristics and the Sum of weights of each arc in each partitioned is calculated. Then based on sum of weights of Each partitioned reuse graphs the least common factor of all the weights will be the new Unrolling degree. This new unrolling degree will lead to better register allocation and reduction of register pressure and because the destination registers is Shared between different iterations because of no dependency as described by the Data dependency distance given in DDG, the allocated registers will lead to less spill and Fetch. The calculation of getting the Unrolling degree(new) as above will lead to register allocation for loops ( that has the candidate of lots of Parallelism ) on assigning the Same destination registers between the DDG nodes having no dependency. Some of the Logic for calculation of new unrolling degree is taken from the proposed Unrolling degree based on Periodic register allocation Albert Cohen et.al. This will efficiently use before and after the register allocation based in gcc so that weights of each reuse graph that describes the allocated Registers Will be known after the register allocation and the weights that describes unrolling degree is done before register allocation and the new Unrolling degree efficiently calculated that reduces the register pressure. Suggestions? Thanks & Regards Ajit
Re: Documentation complex logarithm https://gcc.gnu.org/onlinedocs/gfortran/LOG.html
On Tue, 2 Jun 2015, C.Friedrich wrote: > The LOG function returns the principal value of the complex logarithm > whose imaginary part omega must be in the range -pi < omega <= pi. > > GFORTRAN does it correctly but the online documentation > https://gcc.gnu.org/onlinedocs/gfortran/LOG.html gives -pi <= omega <= > pi. Perhaps this should be corrected. Let me redirect this to the fort...@gcc.gnu.org list. Gerald
Re: Errors Summary
On Mon, 11 May 2015, 18d53b+c826wv11dt...@guerrillamail.com wrote: > Is there an option to print out a summary of line numbers with > errors/warnings? Example output: > > > main.cpp : 2, 3, 17 > lex.cpp : 4, 6, 8 > > I often don't need an error message to fix the error, so this would save > me much time. I am not aware of such an option, and it would be too specific to add to GCC itself. You can, however, post-process the output of GCC using a scripting language of your choice to provide what you are looking for. Gerald
