GNOME Bugzilla – Bug 780226
hash mismatches in docs makes build fail
Last modified: 2017-05-11 17:58:18 UTC
When building the docs, GEGL says OK on a lot of the lines but for some of the lines it says FAIL. followed by a summary where it says: with opencl acceleration Maybe see bug https://bugzilla.gnome.org/show_bug.cgi?id=777XXX 3 operations not producing the expected result: gegl:unsharp-mask ecedf51ab754c690b8eecf815105a059 != 9868efefcaf96dddb6ea7e9f8388a717 gegl:spachropyzer 72b74a0b7b7fe745f31ae389ec418bef != cc9c401d7c8bcea7147499ecbeda9986 gegl:value-invert 463c60bce2542a48a63aef9e91aa127a != 489d46de415cb7308b59014449ff76af This means that for some operations the rendered result does not match MD5 hashes of rendered reference images. This can be due to differences code paths or cpy/system capabilities. There is mechanisms for registering multiple reference hashes, this might end up being the way to deal with a couple of different well known expected variations that can be well documented. The current hashes, which the continous build server is agreeing with are produced with, and without OpenCL support. processor : 1 vendor_id : GenuineIntel cpu family : 6 model : 76 model name : Intel(R) Celeron(R) CPU N3050 @ 1.60GHz fpu : yes fpu_exception : yes cpuid level : 11 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 movbe popcnt tsc_deadline_timer aes rdrand lahf_lm 3dnowprefetch epb tpr_shadow vnmi flexpriority ept vpid tsc_adjust smep erms dtherm ida arat The format of the hashes is made to be easily copy-pasted into this bug report, if making attachments to this bug report, please clearly indicate how the system is special.
With beignet-dev:amd64 1.3.0-2 installed on the same development system, I get the following mismatches: with opencl acceleration Maybe see bug https://bugzilla.gnome.org/show_bug.cgi?id=780226 10 operations producing unexpected hashes: gegl:unsharp-mask ecedf51ab754c690b8eecf815105a059 != 9868efefcaf96dddb6ea7e9f8388a717 gegl:spachropyzer 72b74a0b7b7fe745f31ae389ec418bef != cc9c401d7c8bcea7147499ecbeda9986 gegl:value-invert 463c60bce2542a48a63aef9e91aa127a != 489d46de415cb7308b59014449ff76af gegl:bilateral-filter bec401d05eb63683d87f15f7b0d0674d != 6504361bfb5d99e9a36cfe8cc9ca52b1 gegl:edge-laplace 555444872f9d129a177f2c7c1b49f912 != b965f78e9ebd7afb421464b211c70f31 gegl:noise-reduction c322c1174f985766bc8178d562c5f88e != f6549ca26f1f7b9706efbf9a6a83f052 gegl:oilify 947805eaab49a2573633a9f987afe408 != 96545d17b7e3bfe05cb47015c51de738 gegl:snn-mean de12841d20667472c288cc184923860f != 116f36a0a62eed60441a35b5e8a10df9 gegl:simplex-noise ffb3b528cc9bb8b16bdca390c07962ec != d6c535d254ebf7cb3213fdb26527f16b gegl:cell-noise f261dddd5b9e5765b1f665135f2465f9 != 24a97f60de37791a65ef8931c3ca3342
vendor_id : GenuineIntel model name : Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz Linux Fedora 25 64 bit, Nvidia Quadro graphic card with proprietary drivers with opencl acceleration Maybe see bug https://bugzilla.gnome.org/show_bug.cgi?id=780226 18 operations producing unexpected hashes: gegl:simplex-noise ffb3b528cc9bb8b16bdca390c07962ec != d6c535d254ebf7cb3213fdb26527f16b gegl:cell-noise f261dddd5b9e5765b1f665135f2465f9 != 24a97f60de37791a65ef8931c3ca3342 gegl:color-to-alpha ac91f678c9f6109e4b9255dcbed6f175 != f110613097308e0fe96ac29f54ca4c2e gegl:gray 6f1ee8b1802e1f5bf4225884800b55a2 != 43ddd80572ab34095298ac7c36368b0c gegl:value-invert d722b464a2e97c0ab929704bd07c9e7a != 489d46de415cb7308b59014449ff76af gegl:antialias ad2e4a275fce1c12155366f71c1313c8 != 51a4e204e61decfe6803cac37222c0ea gegl:bilateral-filter 8d5f84811623385dd2fad699bde50e8e != 6504361bfb5d99e9a36cfe8cc9ca52b1 gegl:edge-laplace 1b1b1edc6948603908f55e33fa2fb541 != b965f78e9ebd7afb421464b211c70f31 gegl:edge-sobel 4e364d9f6455b66b53c6461d95aec7ee != 50c7c77a4a130343facd199362a0d4fb gegl:mosaic 83dc5617930615b598e218f98505eec0 != 952d01e1ea3388e5ddee4fd33d52db3a gegl:motion-blur-zoom 68e1302199ff5c7bf97be1d3d7f842d6 != 9708f693b8f9628f4e54bc2ebeae9926 gegl:noise-reduction 36e98949386e6edf35233733017eccab != f6549ca26f1f7b9706efbf9a6a83f052 gegl:oilify 581d77894a9765e7f8c1eea8cae1f680 != 96545d17b7e3bfe05cb47015c51de738 gegl:snn-mean de12841d20667472c288cc184923860f != 116f36a0a62eed60441a35b5e8a10df9 gegl:value-propagate 0a50915d5fdb24304641af5f6591852d != 0bce414c416e2d705d0b9ecd8333d93b gegl:mantiuk06 9739955a609e4e6cad24afbc414b4607 != cd4aecc6eeb94085424c7adb6d43e92a gegl:reinhard05 f4e972f857d40616569d52eaba179425 != 82d46893705823deabfe2511f0561b68 gegl:unsharp-mask ecedf51ab754c690b8eecf815105a059 != 9868efefcaf96dddb6ea7e9f8388a717
I see this on an HP Elitebook 725G2 with Fedora 25 and this graphic card: 00:01.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Kaveri [Radeon R6/R7 Graphics] And CPU: AMD A10 PRO-7350B R6, 10 Compute Cores 4C+6G gegl:unsharp-mask OK without opencl acceleration Maybe see bug https://bugzilla.gnome.org/show_bug.cgi?id=780226 1 operations producing unexpected hashes: gegl:fattal02 d324c996af0fe584f1f2972d5312a4ea != e2260e134cdaa92773f200986bb3c09b Makefile:777: recipe for target 'examples' failed make[2]: *** [examples] Error 255 make[2]: Leaving directory '/home/kmaraas/.cache/jhbuild/build/gegl/docs' Makefile:637: recipe for target 'all-recursive' failed make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory '/home/kmaraas/.cache/jhbuild/build/gegl' Makefile:544: recipe for target 'all' failed make: *** [all] Error 2 *** Error during phase build of gegl: ########## Error running make -j 5 *** [154/231]
Interesting to observe that many OpenCL implementations disagree with the cpu implementation in the same way. The divergence in results is also likely related to the double vs float precision issue in babl - discussed by Thomas Manni in bug 779893.
I have now made it so such mismatches do not trigger build failure - the report and reference to this bug report are however left in.