[tree-ssa] Merge status 2004-05-03
Diego Novillo
dnovillo@redhat.com
Mon May 3 23:04:00 GMT 2004
The branch is in sync with mainline as of 2004-04-28. The only
remaining items to address should be testsuite failures wrt to mainline.
The following are the merge criteria we had agreed to address and their
status. Feedback on missed items welcome.
* The .dot flowgraph dumps will be renamed to .vcg and the output
will be checked to verify that it can be parsed by VCG.
NOT DONE. I will check this out tomorrow.
* New -f and --param command line options described in
doc/invoke.texi.
DONE.
* New tree codes described in c-tree.texi. Additional internal
documentation files for the new passes, data structures and
functions will also be added.
DONE.
* Verify function and file-level documentation to make sure it's
up-to-date.
DONE.
* Update/finish gfortran.texi.
MOSTLY DONE. Fortran is still in a state of flux, we agreed
that the document is still incomplete and work on it will
continue post-merge.
* Rewrite passes.texi. In particular the FE->optimizer interface.
DONE.
* Regression tests showing no regressions with respect to mainline
before the merge.
On i686 the branch presents these new failures:
==========================================================================
New regressions in gcc
FAIL: gcc.c-torture/compile/20010605-1.c (test for excess errors)
FAIL: gcc.c-torture/compile/20010605-1.c (test for excess errors)
FAIL: gcc.c-torture/execute/20040313-1.c execution
FAIL: gcc.c-torture/execute/20040313-1.c execution
FAIL: gcc.c-torture/execute/20040313-1.c execution
FAIL: gcc.c-torture/execute/20040313-1.c execution
FAIL: gcc.c-torture/execute/20040313-1.c execution
FAIL: gcc.c-torture/execute/20040313-1.c execution
FAIL: gcc.c-torture/execute/921017-1.c compilation
FAIL: gcc.c-torture/execute/921017-1.c compilation
FAIL: gcc.c-torture/execute/921017-1.c compilation
FAIL: gcc.c-torture/execute/921017-1.c compilation
FAIL: gcc.c-torture/execute/921017-1.c compilation
FAIL: gcc.c-torture/execute/921017-1.c compilation
FAIL: gcc.dg/torture/builtin-ctype-1.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-1.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-1.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-1.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-1.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-2.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-2.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-2.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-2.c (test for excess errors)
FAIL: gcc.dg/torture/builtin-ctype-2.c (test for excess errors)
FAIL: 209: expected branch percentages not found: 25
FAIL: gcc.misc-tests/gcov-4b.c gcov: 0 failures in line counts
New regressions in g++
FAIL: 258: expected branch percentages not found: 25
FAIL: g++.dg/gcov/gcov-1.C gcov: 0 failures in line counts
FAIL: 23:is 2:should be 1
FAIL: g++.dg/gcov/gcov-2.C gcov: 1 failures in line counts
FAIL: g++.old-deja/g++.brendan/crash13.C (test for errors
FAIL: g++.old-deja/g++.law/friend5.C (test for errors
FAIL: g++.old-deja/g++.law/friend5.C (test for errors
New regressions in libjava
FAIL: PR4766 -O3 compilation from source
FAIL: Array_3 execution - gij test
FAIL: Array_3 execution - gij test
FAIL: String_overflow -O3 compilation from source
FAIL: err3 output - source compiled test
FAIL: err3 -O3 output - source compiled test
==========================================================================
This is the reason why the branch is frozen now. I think that
these regressions are the only merge blocker now.
* Bootstrapped and tested on:
alphaev67-unknown-linux-gnu
i686-pc-linux-gnu
x86_64-unknown-linux-gnu
powerpc-unknown-linux-gnu
ia64-unknown-linux-gnu
sparc-sun-solaris2.8
i386-unknown-freebsd4.8
DONE. I am not sure what the status is on sparc-sun-solaris2.8
after the last merge. It was working up to that point, I
believe. Gerald had problems with i386-unknown-freebsd4.8 but
he stated that it may not be related to the branch. Gerald, any
news there?
* Built and tested on embedded target mn10300-elf.
DONE. Results for the latest merge point should be available
shortly (tests are still running).
* Applications and Fedora Core packages.
DONE. Pass rate for FC is around 97%. AFAIK, all the other
major applications build (Gerald, Richard G., please check).
* Testcases in the testsuite for all new features.
DONE.
* SPECint and SPECfp performance within 3% of mainline.
DONE. Attached are results i686 and x86-64 as of 2004-05-03.
Base results are for mainline. Peak results are for tree-ssa.
* Bootstrap times within 1-5% of mainline configured with
--disable-libmudflap --without-libbanshee --disable-checking
--enable-languages=c,c++,java,objc
DONE. In absolute terms, tree-ssa bootstrap times are 13.5%
slower. But given that there is ~9% more C code in tree-ssa's
source base, we are within range.
* POOMA, DLV and MICO compile times and memory consumption within
1-5% of mainline.
NOT SURE. I *think* we are within range, but I don't have
recent figures. Could the application owners check that I'm not
babbling nonsense?
I think that one more week should be enough to finish up the remaining
regressions. Some of them may be fixable. Others may need more
extensive work. I would first like to characterize them and decide
whether to fix in branch or post-merge.
Opinions? Thoughts?
Thanks. Diego.
-------------- next part --------------
Estimated Estimated
Base Base Base Peak Peak Peak
Benchmarks Ref Time Run Time Ratio Ref Time Run Time Ratio
------------ -------- -------- -------- -------- -------- --------
164.gzip 1400 211 665* 1400 214 653*
175.vpr 1400 327 428* 1400 329 425*
176.gcc X X
181.mcf 1800 414 435* 1800 425 423*
186.crafty 1000 148 676* 1000 150 666*
197.parser 1800 314 573* 1800 315 572*
252.eon 1300 224 581* 1300 225 577*
253.perlbmk 1800 213 846* 1800 228 790*
254.gap 1100 154 716* 1100 150 733*
255.vortex 1900 224 847* 1900 229 829*
256.bzip2 1500 277 541* 1500 279 538*
300.twolf 3000 565 531* 3000 564 532*
Est. SPECint_base2000 607
Est. SPECint2000 599
Estimated Estimated
Base Base Base Peak Peak Peak
Benchmarks Ref Time Run Time Ratio Ref Time Run Time Ratio
------------ -------- -------- -------- -------- -------- --------
168.wupwise 1600 262 612 * 1600 261 612 *
171.swim 3100 607 511 * 3100 609 509 *
172.mgrid 1800 433 415 * 1800 433 416 *
173.applu 2100 382 550 * 2100 379 554 *
177.mesa 1400 307 456 * 1400 307 456 *
178.galgel X X
179.art 2600 1351 193 * 2600 1379 189 *
183.equake 1300 155 838 * 1300 158 823 *
187.facerec X X
188.ammp 2200 621 354 *
189.lucas X X
191.fma3d X X
200.sixtrack X X
301.apsi X X
Est. SPECfp_base2000 472
Est. SPECfp2000 454
-------------- next part --------------
Estimated Estimated
Base Base Base Peak Peak Peak
Benchmarks Ref Time Run Time Ratio Ref Time Run Time Ratio
------------ -------- -------- -------- -------- -------- --------
164.gzip 1400 280 500 * 1400 279 502 *
175.vpr 1400 319 439 * 1400 321 436 *
176.gcc X X
181.mcf 1800 453 398 * 1800 453 398 *
186.crafty 1000 128 778 * 1000 128 780 *
197.parser 1800 393 458 * 1800 396 455 *
252.eon X X
253.perlbmk 1800 224 803 * 1800 224 805 *
254.gap 1100 131 841 * 1100 130 845 *
255.vortex 1900 224 848 * 1900 224 849 *
256.bzip2 1500 275 545 * 1500 276 544 *
300.twolf 3000 543 552 * 3000 539 556 *
Est. SPECint_base2000 593
Est. SPECint2000 593
Estimated Estimated
Base Base Base Peak Peak Peak
Benchmarks Ref Time Run Time Ratio Ref Time Run Time Ratio
------------ -------- -------- -------- -------- -------- --------
168.wupwise 1600 273 586 * 1600 272 589 *
171.swim 3100 455 682 * 3100 461 672 *
172.mgrid 1800 715 252 * 1800 715 252 *
173.applu 2100 462 454 * 2100 462 455 *
177.mesa 1400 179 782 * 1400 179 782 *
178.galgel X X
179.art 2600 363 717 * 2600 360 723 *
183.equake 1300 159 820 * 1300 159 820 *
187.facerec X X
188.ammp 2200 481 457 * 2200 479 459 *
189.lucas X X
191.fma3d X X
200.sixtrack X X
301.apsi X X
Est. SPECfp_base2000 559
Est. SPECfp2000 560
More information about the Gcc
mailing list