Re: fatal error: config.h: No such file or directory
On 12/21/2014 02:38 AM, Bruce Korb wrote: > Shouldn't the configure step have made config.h? It's probably because you are building in srcdir. That is not supported. Andrew.
Sebastian Huber appointed RTEMS Co-Maintainer
Hi I am pleased to announce that the steering committee has appointed Sebastian Huber as co-maintainer of the RTEMS target in GCC. This is a reflection of the work has done and community involvement Sebastian has already had. We are looking forward to more of this in the future. Please adjust the MAINTAINERS file accordingly, and Happy Hacking! -- Joel Sherrill, Ph.D. Director of Research & Development joel.sherr...@oarcorp.com On-Line Applications Research Ask me about RTEMS: a free RTOS Huntsville AL 35805 Support Available (256) 722-9985
Register Allocation with Instruction Scheduling.
Hello All: I was going through the following article " Register Allocation with instruction scheduling: a new approach" by Pinter etal. The phase ordering of register allocation and Instruction scheduling is important topic. The scheduling before register allocator increases the register pressure whereas the instruction scheduling after register allocation will be seeing the false dependencies created by Register allocator. The false dependencies between the instructions will not be able to schedule the two instruction simultaneously reducing the Instruction Level Parallelism. To achieve better ILP for the instruction scheduling after register allocation the false dependencies created by the register allocator should be reduced. Let Gs(V,E) is the scheduling Graph and G'(V,E) is the Scheduling Graph after the register allocation. The Graph G'(V,E) has the false dependencies created by Register Allocator Another graph is created with Gf(V,E) where the edges represents the false dependencies. To make sure the false dependencies Graph which is the complement of Gs(V,E) and the edges of Gf are the intersection of the edge in the G'(V,E) and the edges in the complement of Gs scheduling graph. With the availability of the false dependencies graph and the interference graph, The edges in the interference graph doesn't intersects with the edges in the false dependencies graph, this edge will be added to the interference graph and since this edge interfere and the assigning phase will assign different register. Thus the false dependencies can be removed and the scheduling graph after the register allocator will not have false dependencies and the better ILP and scheduling is achieved. This overcomes the disadvantages of instruction scheduling after register allocator. This looks feasible to implement in GCC and I would like to implement the above if not done in the GCC. Please let me know what do you think. Thanks & Regards Ajit
gcc-5-20141221 is now available
Snapshot gcc-5-20141221 is now available on ftp://gcc.gnu.org/pub/gcc/snapshots/5-20141221/ and on various mirrors, see http://gcc.gnu.org/mirrors.html for details. This snapshot has been generated from the GCC 5 SVN branch with the following options: svn://gcc.gnu.org/svn/gcc/trunk revision 218997 You'll find: gcc-5-20141221.tar.bz2 Complete GCC MD5=db8bfce2c4dcf109e88dc7a4719362c9 SHA1=14eb2557a2e76af97a64e25609d68779f7ebf8b3 Diffs from 5-20141214 are available in the diffs/ subdirectory. When a particular snapshot is ready for public consumption the LATEST-5 link is updated and a message is sent to the gcc list. Please do not use a snapshot before it has been announced that way.
simplifying the gcc configure script to allow building of compilers that supports offloading to multiple accelerator in one go
As in the offloading documentation at gcc.gnu.org, currently one must run ./configure and make for each accelerator and the host itself. This is very inconvenient and prone to error. Would it be possible to change the configure script so that all the user needs to do is to: * specify the target host architecture (if different from the build machine architecture) * list all accelerator architectures (such as KNL and PTX) * specify the prefix Then the compiler should build for the host, and all accelerator architectures, and install binaries for both host and offload supoorts into the prefix, taking care of relative positions between the host and accelerator binaries. Would this be feasible?
Re: Sebastian Huber appointed RTEMS Co-Maintainer
Hello, thanks for your trust. On 21/12/14 17:46, Joel Sherrill wrote: Hi I am pleased to announce that the steering committee has appointed Sebastian Huber as co-maintainer of the RTEMS target in GCC. This is a reflection of the work has done and community involvement Sebastian has already had. We are looking forward to more of this in the future. Please adjust the MAINTAINERS file accordingly, and Happy Hacking! Done in Revision 219003. -- Joel Sherrill, Ph.D. Director of Research & Development joel.sherr...@oarcorp.com On-Line Applications Research Ask me about RTEMS: a free RTOS Huntsville AL 35805 Support Available (256) 722-9985 -- Sebastian Huber, embedded brains GmbH Address : Dornierstr. 4, D-82178 Puchheim, Germany Phone : +49 89 189 47 41-16 Fax : +49 89 189 47 41-09 E-Mail : sebastian.hu...@embedded-brains.de PGP : Public key available on request. Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
