ATLAS - build failure

[ferdinand.gruebler@…]

Hi,

I would like to install py25-matplotlib and py25-numpy . But they depend on ATLAS.

And ATLAS won"t compile on my machine saying:

--->  Building atlas
Error: Target org.macports.build returned: shell command failed (see log for details)
Log for atlas is at: /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_release_ports_math_atlas/main.log
Error: Status 1 encountered during processing.
To report a bug, see <http://guide.macports.org/#project.tickets>

I am working on 10.6.6 (snow leopard), xcode installed - macbook pro i5. MacPorts version 1.9.2.

Here is a part of the logfile:

:info:build IN STAGE 1 INSTALL:  SYSTEM PROBE/AUX COMPILE
:info:build 
:info:build 
:info:build    Level 1 cache size calculated as 32KB
:info:build    dFPU: Separate multiply and add instructions with 5 cycle pipeline.
:info:build          Apparent number of registers : 17
:info:build          Register-register performance=5784.86MFLOPS
:info:build    sFPU: Separate multiply and add instructions with 4 cycle pipeline.
:info:build          Apparent number of registers : 17
:info:build          Register-register performance=5650.25MFLOPS
:info:build 
:info:build 
:info:build IN STAGE 2 INSTALL:  TYPE-DEPENDENT TUNING
:info:build 
:info:build 
:info:build STAGE 2-1: TUNING PREC='d' (precision 1 of 4)
:info:build 
:info:build 
:info:build    STAGE 2-1-1 : BUILDING BLOCK MATMUL TUNE
:info:build make -f Makefile INSTALL_LOG/dMMRES pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dMMSEARCH.LOG
:info:build       The best matmul kernel was ATL_dmm2x2x128_sse2.c, NB=56, written by R. Clint Whaley
:info:build       Performance: 9213.85MFLOPS (346.39 percent of of detected clock rate)
:info:build         (Gen case got 4586.80MFLOPS)
:info:build make -f Makefile INSTALL_LOG/dNCNB pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dMMSEARCH.LOGmake -f Makefile INSTALL_LOG/dbestNN_56x56x56 pre=d nb=56 2>&1 | ./xatlas_tee INSTALL_LOG/dMMSEARCH.LOG      NCgemmNN : muladd=0, lat=4, pf=1, nb=56, mu=14, nu=1 ku=4,
:info:build                  ForceFetch=1, ifetch=15 nfetch=1
:info:build                  Performance = 3683.41 (39.98 of copy matmul, 138.47 of clock)
:info:build make -f Makefile INSTALL_LOG/dbestNT_56x56x56 pre=d nb=56 2>&1 | ./xatlas_tee INSTALL_LOG/dMMSEARCH.LOG      NCgemmNT : muladd=0, lat=8, pf=1, nb=56, mu=14, nu=1 ku=56,
:info:build                  ForceFetch=1, ifetch=15 nfetch=1
:info:build                  Performance = 3770.41 (40.92 of copy matmul, 141.74 of clock)
:info:build make -f Makefile INSTALL_LOG/dbestTN_56x56x56 pre=d nb=56 2>&1 | ./xatlas_tee INSTALL_LOG/dMMSEARCH.LOG      NCgemmTN : muladd=0, lat=1, pf=1, nb=56, mu=14, nu=1 ku=56,
:info:build                  ForceFetch=1, ifetch=15 nfetch=1
:info:build                  Performance = 3741.57 (40.61 of copy matmul, 140.66 of clock)
:info:build make -f Makefile INSTALL_LOG/dbestTT_56x56x56 pre=d nb=56 2>&1 | ./xatlas_tee INSTALL_LOG/dMMSEARCH.LOG      NCgemmTT : muladd=0, lat=1, pf=1, nb=56, mu=14, nu=1 ku=56,
:info:build                  ForceFetch=1, ifetch=15 nfetch=1
:info:build                  Performance = 3045.51 (33.05 of copy matmul, 114.49 of clock)
:info:build make -f Makefile MMinstall pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dMMSEARCH.LOG
:info:build 
:info:build 
:info:build    STAGE 2-1-2: CacheEdge DETECTION
:info:build make -f Makefile INSTALL_LOG/atlas_cacheedge.h pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dMMCACHEEDGE.LOG
:info:build 
:info:build 
:info:build    STAGE 2-1-3: LARGE/SMALL CASE CROSSOVER DETECTION
:info:build 
:info:build 
:info:build    STAGE 2-1-3: COPY/NO-COPY CROSSOVER DETECTION
:info:build make -f Makefile INSTALL_LOG/dXover.h pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dMMCROSSOVER.LOG
:info:build       done.
:info:build 
:info:build 
:info:build    STAGE 2-1-4: LEVEL 3 BLAS TUNE
:info:build make -f Makefile INSTALL_LOG/atlas_dtrsmXover.h pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dL3TUNE.LOG
:info:build make -f Makefile dcblaslib 2>&1 | ./xatlas_tee INSTALL_LOG/dL3TUNE.LOG
:info:build       done.
:info:build 
:info:build 
:info:build    STAGE 2-1-5: GEMV TUNE
:info:build make -f Makefile INSTALL_LOG/dMVRES pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dMVTUNE.LOG
:info:build       gemvN : chose routine 3:ATL_gemvN_1x1_1a.c written by R. Clint Whaley
:info:build               Yunroll=32, Xunroll=1, using 96 percent of L1
:info:build               Performance = 1565.78 (16.99 of copy matmul, 58.86 of clock)
:info:build       gemvT : chose routine 105:ATL_gemvT_2x16_1.c written by R. Clint Whaley
:info:build               Yunroll=2, Xunroll=16, using 96 percent of L1
:info:build               Performance = 1457.26 (15.82 of copy matmul, 54.78 of clock)
:info:build 
:info:build 
:info:build    STAGE 2-1-6: GER TUNE
:info:build make -f Makefile INSTALL_LOG/dR1RES pre=d 2>&1 | ./xatlas_tee INSTALL_LOG/dR1TUNE.LOG
:info:build       ger : chose routine 1:ATL_ger1_axpy.c written by R. Clint Whaley
:info:build             mu=16, nu=1, using  0.87 percent of L1 Cache
:info:build               Performance = 1135.36 (12.32 of copy matmul, 42.68 of clock)
:info:build 
:info:build 
:info:build STAGE 2-2: TUNING PREC='s' (precision 2 of 4)
:info:build 
:info:build 
:info:build    STAGE 2-2-1 : BUILDING BLOCK MATMUL TUNE
:info:build make -f Makefile INSTALL_LOG/sMMRES pre=s 2>&1 | ./xatlas_tee INSTALL_LOG/sMMSEARCH.LOG
:info:build       The best matmul kernel was ATL_smm14x1x84_sse.c, NB=84, written by R. Clint Whaley
:info:build       Performance: 13507.18MFLOPS (507.79 percent of of detected clock rate)
:info:build         (Gen case got 4217.26MFLOPS)
:info:build make -f Makefile INSTALL_LOG/sNCNB pre=s 2>&1 | ./xatlas_tee INSTALL_LOG/sMMSEARCH.LOGmake -f Makefile INSTALL_LOG/sbestNN_48x48x48 pre=s nb=48 2>&1 | ./xatlas_tee INSTALL_LOG/sMMSEARCH.LOG      NCgemmNN : muladd=1, lat=4, pf=513, nb=48, mu=7, nu=1 ku=4,
:info:build                  ForceFetch=0, ifetch=7 nfetch=1
:info:build                  Performance = 3195.87 (23.66 of copy matmul, 120.15 of clock)
:info:build make -f Makefile INSTALL_LOG/sbestNT_48x48x48 pre=s nb=48 2>&1 | ./xatlas_tee INSTALL_LOG/sMMSEARCH.LOG      NCgemmNT : muladd=1, lat=6, pf=513, nb=48, mu=7, nu=1 ku=48,
:info:build                  ForceFetch=0, ifetch=7 nfetch=1
:info:build                  Performance = 3041.97 (22.52 of copy matmul, 114.36 of clock)
:info:build make -f Makefile INSTALL_LOG/sbestTN_48x48x48 pre=s nb=48 2>&1 | ./xatlas_tee INSTALL_LOG/sMMSEARCH.LOG      NCgemmTN : muladd=1, lat=3, pf=513, nb=48, mu=7, nu=1 ku=48,
:info:build                  ForceFetch=0, ifetch=7 nfetch=1
:info:build                  Performance = 3699.20 (27.39 of copy matmul, 139.07 of clock)
:info:build make -f Makefile INSTALL_LOG/sbestTT_48x48x48 pre=s nb=48 2>&1 | ./xatlas_tee INSTALL_LOG/sMMSEARCH.LOG      NCgemmTT : muladd=1, lat=4, pf=513, nb=48, mu=7, nu=1 ku=48,
:info:build                  ForceFetch=0, ifetch=7 nfetch=1
:info:build                  Performance = 3226.16 (23.88 of copy matmul, 121.28 of clock)
:info:build make -f Makefile MMinstall pre=s 2>&1 | ./xatlas_tee INSTALL_LOG/sMMSEARCH.LOG
:info:build 
:info:build 
:info:build    STAGE 2-2-2: CacheEdge DETECTION
:info:build 
:info:build 
:info:build    STAGE 2-2-3: LARGE/SMALL CASE CROSSOVER DETECTION
:info:build 
:info:build 
:info:build    STAGE 2-2-3: COPY/NO-COPY CROSSOVER DETECTION
:info:build make -f Makefile INSTALL_LOG/sXover.h pre=s 2>&1 | ./xatlas_tee INSTALL_LOG/sMMCROSSOVER.LOG
:info:build       done.
:info:build 
:info:build 
:info:build    STAGE 2-2-4: LEVEL 3 BLAS TUNE
:info:build make -f Makefile INSTALL_LOG/atlas_strsmXover.h pre=s 2>&1 | ./xatlas_tee INSTALL_LOG/sL3TUNE.LOG
:info:build make -f Makefile scblaslib 2>&1 | ./xatlas_tee INSTALL_LOG/sL3TUNE.LOG
:info:build       done.
:info:build 
:info:build 
:info:build    STAGE 2-2-5: GEMV TUNE
:info:build make -f Makefile INSTALL_LOG/sMVRES pre=s 2>&1 | ./xatlas_tee INSTALL_LOG/sMVTUNE.LOG
:info:build       gemvN : chose routine 3:ATL_gemvN_1x1_1a.c written by R. Clint Whaley
:info:build               Yunroll=32, Xunroll=1, using 100 percent of L1
:info:build               Performance = 2874.33 (21.28 of copy matmul, 108.06 of clock)
:info:build       gemvT : chose routine 105:ATL_gemvT_2x16_1.c written by R. Clint Whaley
:info:build               Yunroll=2, Xunroll=16, using 100 percent of L1
:info:build               Performance = 1978.01 (14.64 of copy matmul, 74.36 of clock)
:info:build 
:info:build 
:info:build    STAGE 2-2-6: GER TUNE
:info:build make -f Makefile INSTALL_LOG/sR1RES pre=s 2>&1 | ./xatlas_tee INSTALL_LOG/sR1TUNE.LOG
:info:build       ger : chose routine 1:ATL_ger1_axpy.c written by R. Clint Whaley
:info:build             mu=16, nu=1, using  0.56 percent of L1 Cache
:info:build               Performance = 4207.98 (31.15 of copy matmul, 158.19 of clock)
:info:build 
:info:build 
:info:build STAGE 2-3: TUNING PREC='z' (precision 3 of 4)
:info:build 
:info:build 
:info:build    STAGE 2-3-1 : BUILDING BLOCK MATMUL TUNE
:info:build make -f Makefile INSTALL_LOG/zMMRES pre=z 2>&1 | ./xatlas_tee INSTALL_LOG/zMMSEARCH.LOG
:info:build       The best matmul kernel was ATL_dmm4x2x128_sse2.c, NB=56, written by Whaley & Voronenko
:info:build       Performance: 7511.28MFLOPS (282.38 percent of of detected clock rate)
:info:build         (Gen case got 4424.50MFLOPS)
:info:build make -f Makefile INSTALL_LOG/zNCNB pre=z 2>&1 | ./xatlas_tee INSTALL_LOG/zMMSEARCH.LOGmake -f Makefile INSTALL_LOG/zbestNN_36x36x36 pre=z nb=36 2>&1 | ./xatlas_tee INSTALL_LOG/zMMSEARCH.LOG      NCgemmNN : muladd=0, lat=2, pf=1, nb=36, mu=13, nu=1 ku=4,
:info:build                  ForceFetch=1, ifetch=13 nfetch=1
:info:build                  Performance = 3323.76 (44.25 of copy matmul, 124.95 of clock)
:info:build make -f Makefile INSTALL_LOG/zbestNT_36x36x36 pre=z nb=36 2>&1 | ./xatlas_tee INSTALL_LOG/zMMSEARCH.LOG      NCgemmNT : muladd=0, lat=6, pf=1, nb=36, mu=13, nu=1 ku=36,
:info:build                  ForceFetch=1, ifetch=13 nfetch=1
:info:build                  Performance = 3182.73 (42.37 of copy matmul, 119.65 of clock)
:info:build make -f Makefile INSTALL_LOG/zbestTN_36x36x36 pre=z nb=36 2>&1 | ./xatlas_tee INSTALL_LOG/zMMSEARCH.LOG      NCgemmTN : muladd=0, lat=1, pf=1, nb=36, mu=13, nu=1 ku=36,
:info:build                  ForceFetch=1, ifetch=13 nfetch=1
:info:build                  Performance = 3733.61 (49.71 of copy matmul, 140.36 of clock)
:info:build make -f Makefile INSTALL_LOG/zbestTT_36x36x36 pre=z nb=36 2>&1 | ./xatlas_tee INSTALL_LOG/zMMSEARCH.LOG      NCgemmTT : muladd=0, lat=6, pf=1, nb=36, mu=13, nu=1 ku=36,
:info:build                  ForceFetch=1, ifetch=13 nfetch=1
:info:build                  Performance = 3271.50 (43.55 of copy matmul, 122.99 of clock)
:info:build make -f Makefile MMinstall pre=z 2>&1 | ./xatlas_tee INSTALL_LOG/zMMSEARCH.LOG
:info:build 
:info:build 
:info:build    STAGE 2-3-2: CacheEdge DETECTION
:info:build make -f Makefile INSTALL_LOG/atlas_zdNKB.h pre=z 2>&1 | ./xatlas_tee INSTALL_LOG/zMMCACHEEDGE.LOG
:info:build make[1]: *** [build] Error 255
:info:build make: *** [build] Error 2
:info:build shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_math_atlas/work/atlas-3.8.3/build" && /usr/bin/make build " returned error 2
:error:build Target org.macports.build returned: shell command failed (see log for details)
:debug:build Backtrace: shell command failed (see log for details)
    while executing
"command_exec build"
    (procedure "portbuild::build_main" line 8)
    invoked from within
"$procedure $targetname"
:info:build Warning: the following items did not execute (for atlas): org.macports.activate org.macports.build org.macports.destroot org.macports.install
:notice:build Log for atlas is at: /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_release_ports_math_atlas/main.log

[jmroot (Joshua Root)]

Please remember to fill in the Port field and cc the maintainer.

[ferdinand.gruebler@…]

Cc Me!

[don.jackson@…]

I have been having problems building this port for the last week or so. I have seen problems as described above, and I have experienced the build hang for many hours (like overnight) and not complete.

[don.jackson@…]

I withdraw my previous comment. I did another clean, and install, and left my machine alone for 6+ hours . On my return, atlas had built. So, given enough time/hours, it did build for me. I am guessing I was too impatient after only 1-2 hours previously. I did at one point have the same (or similar) build issue as the initial reporter, but the port clean atlas seemed to fix it.

$ sudo port clean atlas
Password:
--->  Cleaning atlas
$ sudo port install atlas
--->  Computing dependencies for atlas
--->  Fetching atlas
--->  Verifying checksum(s) for atlas
--->  Extracting atlas
--->  Applying patches to atlas
--->  Configuring atlas
--->  Building atlas
--->  Staging atlas into destroot
--->  Installing atlas @3.8.3_4+gcc44
--->  Activating atlas @3.8.3_4+gcc44
--->  Cleaning atlas

[ryandesign (Ryan Carsten Schmidt)]

Is this still a problem with atlas 3.9.33?

[afisynte@…]

I think I'm experiencing this as well on Mac OS X 10.4.11 PPC. I'll attach my /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_release_ports_math_atlas/main.log which can also be found here: ​http://pastebin.com/BXX28gVH

[afisynte@…]

Cc Me!

[afisynte@…]

These bugs appear to be similar:
#27897
#28189
#28413

Hopefully vince@... uploading 3.9.35 will fix this:
ticket:28441:2

[Veence (Vincent)]

This should be fixed now.