View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0029038 | Open CASCADE | OCCT:Modeling Algorithms | public | 2017-08-22 19:06 | 2019-01-24 12:23 |
| Reporter | Assigned To | bugmaster | |||
| Priority | normal | Severity | minor | ||
| Status | closed | Resolution | fixed | ||
| Target Version | 7.3.0 | Fixed in Version | 7.3.0 | ||
| Summary | 0029038: Modeling - IntTools_EdgeEdge::FindParameters() hangs on bad curve | ||||
| Description | The issue is diagnosed on a case initially reported in #26528: during creation of offset on surface with degenerated corner, the procedure hangs in function IntTools_EdgeEdge::FindParameters(). The cause is that input curve (generated as boundary of offset surface at degeneration point) is poorly defined; it has very high estimated maximum derivative, and hence very low resolution. Method IntTools_EdgeEdge::FindParameters() iterates along the curve by steps defined by resolution, so it takes incredibly long for it to go along the curve (my estimate was 10 min, but it did not finish in ~ 30 min in Release mode). In practice, this means that procedure hangs. | ||||
| Steps To Reproduce | restore [locate_data_file bug29038_bfuse_hang.brep] a explode a bfuse r a_1 a_2 | ||||
| Tags | No tags attached. | ||||
| Test case number | bugs modalg_7 bug29038 | ||||
|
2017-08-22 19:10 manager |
bug29038_bfuse_hang.brep (57,413 bytes) |
|
|
Branch CR29038 has been created by abv. SHA-1: 0a67b41da501174fdce490c21253036679b17f28 Detailed log of new commits: Author: abv Date: Wed Aug 23 12:49:38 2017 +0300 0029038: Modeling - IntTools_EdgeEdge::FindParameters() hangs on bad curve Avoid using global resolution of curve when computing step for iterating along it; length of derivative at the current point is used instead. |
|
|
Branch CR29038 has been updated by abv. SHA-1: 25afa22379e0dc1638284c2cbfecf81202f9cdc8 Detailed log of new commits: Author: abv Date: Thu Aug 24 08:24:33 2017 +0300 Step is reduced ten times to avoid regressions |
|
|
The first version of the fix caused regressions on 58 cases. To avoid this, I have reduced step by 10 times. Most of regressions have gone, however some remained: bugs modalg_5 bug24981 (hanged -- killed by CPU limit) bugs modalg_6 bug26938_1 bug26938_2 bug26938_3 bug27021 bugs modalg_7 bug28892_1 Apparently we need to have more sophisticated approach to choice of the iteration step. It can be reasonable to have some standard and reusable tools for that. Such tools can be naturally put into package CPntts which already implements similar (but different) algorithms. |
|
|
Branch CR29038_1 has been created by msv. SHA-1: a9f18cf1bd284dac8d9f7e224553634142ab8fde Detailed log of new commits: Author: msv Date: Fri Sep 1 09:10:36 2017 +0300 0029038: Modeling - IntTools_EdgeEdge::FindParameters() hangs on bad curve Make possibility to dramatically increase the step when iteration on the curve if the global resolution of the curve is too small. Correct the functions FindParameters, FindBestSolution, FindDistPC. |
|
|
The branch CR29038_1 contains alternative and more safe solution. |
|
|
Branch CR29038_1 has been updated forcibly by msv. SHA-1: ef012c8d68f4c004488e21901cfe14b657d4e164 |
|
|
Branch CR29038_1 has been updated forcibly by msv. SHA-1: 76f46c4dbfc8f42de2f55f8055fe843d1fa49976 |
|
|
Branch CR29038_1 has been updated forcibly by msv. SHA-1: 396671dc4365b06096b99c66bdadfe07c431caa4 |
|
|
Dear Eugeny, please review the branch CR29038_1. Jenkins job CR29038-master-MSV showed no regressions. |
|
|
Branch CR29038_1 has been updated forcibly by msv. SHA-1: 5281bd40e1dcb34c2317a1543e99474a0cc5c4c3 |
|
|
Reviewed. |
|
|
Combination - OCCT branch : CR29038_1 SHA-1: 5281bd40e1dcb34c2317a1543e99474a0cc5c4c3 Products branch : master was compiled on Linux, MacOS and Windows platforms and tested on optimize mode. http://jenkins-test-10.nnov.opencascade.com/view/CR29038-master-MSV/ Number of compiler warnings: OCCT : Linux: 0 (0 on master) Windows: 0 (0 on master) MacOS : 0 (0 on master) Products : Linux: 4 (4 on master) Windows: 0 (0 on master) MacOS : 0 (0 on master) Regressions/Differences/Improvements: No regressions/differences CPU differences: Windows: OCCT CPU bugs moddata_3 bug24886: 0.3744024 / 4.5552292 [-91.78%] .. CPU bugs modalg_7 bug27160: 1.6536106 / 2624.2176218 [-99.94%] ... CPU bugs modalg_7: 221.67742100000004 / 2840.7470098000026 [-92.20%] ... Total CPU difference: 18289.838041798714 / 20591.476795798557 [-11.18%] Producst Total CPU difference: 7688.758886599963 / 7720.302288799964 [-0.41%] Linux: OCCT ... CPU bugs moddata_3 bug24886: 0.42 / 3.34 [-87.43%] ... CPU bugs modalg_7 bug27160: 1.79 / 1883.2 [-99.90%] ... CPU bugs modalg_7: 231.38000000000008 / 2112.4499999999985 [-89.05%] .... Total CPU difference: 20424.210000000145 / 22080.990000000125 [-7.50%] Products Total CPU difference: 7703.40000000009 / 7725.120000000072 [-0.28%] Please confirm performance improving !!!! Image differences : A lot of difference in images in OCCT on Linux and Windows platforms, especially on location of vertex during intersection of 2 curves http://occt-tests/CR29038-master-MSV-OCCT/Debian70-64/diff-Debian70-64-image.html http://occt-tests/CR29038-master-MSV-OCCT/Windows-64-VC10/diff-Windows-64-VC10-image.html Please confirm !!! Memory differences : No differences that require special attention |
|
|
I confirm performance improvement. It is the main purpose of the fix. I have wrote about it in commit message. A side effect (positive one) of the fix is more precise calculation of the zone of intersection of two edges and the exact intersection point. So, it is expected that some vertices can have slightly changed positions. But this must not impact the quality of the result shapes. If there are suspicious cases they must be thoroughly checked. Please tell me the names of test cases where you think can be incorrect change of the picture. |
|
|
Differences in performance and images have been confirmed by developer |
|
|
Branch CR29038 has been deleted by kgv. SHA-1: 25afa22379e0dc1638284c2cbfecf81202f9cdc8 |
|
|
Branch CR29038_1 has been deleted by kgv. SHA-1: 5281bd40e1dcb34c2317a1543e99474a0cc5c4c3 |
|
occt: master d89c7d60 2017-09-01 06:10:36
Committer: bugmaster Details Diff |
0029038: Modeling - IntTools_EdgeEdge::FindParameters() hangs on bad curve Make possibility to dramatically increase the step during iteration on the curve when the global resolution of the curve is too small. Correct the functions FindParameters, FindBestSolution, FindDistPC. The change caused improvement of the test case "boolean bfuse_complex E4", so the bug #27697 has been fixed. The change caused dramatic improvement of the performance of the test "bugs modalg_7 bug27160", and the bug 0029080 has been fixed. |
Affected Issues 0029038 |
|
| mod - src/BOPTools/BOPTools_AlgoTools.cxx | Diff File | ||
| mod - src/IntTools/IntTools_EdgeEdge.cxx | Diff File | ||
| mod - src/IntTools/IntTools_EdgeEdge.hxx | Diff File | ||
| mod - tests/boolean/bfuse_complex/E4 | Diff File | ||
| add - tests/bugs/modalg_7/bug29038 | Diff File | ||
|
occt: master 28c08a7d 2017-09-01 06:10:36
Committer: bugmaster Details Diff |
0029038: Modeling - IntTools_EdgeEdge::FindParameters() hangs on bad curve Make possibility to dramatically increase the step during iteration on the curve when the global resolution of the curve is too small. Correct the functions FindParameters, FindBestSolution, FindDistPC. The change caused improvement of the test case "boolean bfuse_complex E4", so the bug #27697 has been fixed. The change caused dramatic improvement of the performance of the test "bugs modalg_7 bug27160", and the bug 0029080 has been fixed. |
Affected Issues 0029038 |
|
| mod - src/BOPTools/BOPTools_AlgoTools.cxx | Diff File | ||
| mod - src/IntTools/IntTools_EdgeEdge.cxx | Diff File | ||
| mod - src/IntTools/IntTools_EdgeEdge.hxx | Diff File | ||
| mod - tests/boolean/bfuse_complex/E4 | Diff File | ||
| add - tests/bugs/modalg_7/bug29038 | Diff File | ||
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2017-08-22 19:06 |
|
New Issue | |
| 2017-08-22 19:06 |
|
Assigned To | => msv |
| 2017-08-22 19:10 |
|
File Added: bug29038_bfuse_hang.brep | |
| 2017-08-22 19:11 |
|
Steps to Reproduce Updated | |
| 2017-08-22 19:11 |
|
Description Updated | |
| 2017-08-23 12:50 | git | Note Added: 0069777 | |
| 2017-08-24 08:24 | git | Note Added: 0069807 | |
| 2017-08-24 08:30 |
|
Note Added: 0069809 | |
| 2017-09-04 14:37 | git | Note Added: 0070164 | |
| 2017-09-04 14:37 |
|
Status | new => assigned |
| 2017-09-04 14:38 |
|
Note Added: 0070165 | |
| 2017-09-04 16:42 | git | Note Added: 0070169 | |
| 2017-09-04 17:23 |
|
Relationship added | related to 0029080 |
| 2017-09-04 18:30 | git | Note Added: 0070218 | |
| 2017-09-05 11:59 | git | Note Added: 0070232 | |
| 2017-09-05 15:06 |
|
Note Added: 0070241 | |
| 2017-09-05 15:06 |
|
Assigned To | msv => emv |
| 2017-09-05 15:06 |
|
Status | assigned => resolved |
| 2017-09-05 15:29 | git | Note Added: 0070244 | |
| 2017-09-05 15:48 |
|
Note Added: 0070245 | |
| 2017-09-05 15:48 |
|
Assigned To | emv => bugmaster |
| 2017-09-05 15:48 |
|
Status | resolved => reviewed |
| 2017-09-05 19:48 | bugmaster | Note Added: 0070248 | |
| 2017-09-05 19:48 | bugmaster | Assigned To | bugmaster => msv |
| 2017-09-05 19:48 | bugmaster | Status | reviewed => assigned |
| 2017-09-06 09:33 |
|
Note Added: 0070256 | |
| 2017-09-06 09:33 |
|
Assigned To | msv => bugmaster |
| 2017-09-06 09:33 |
|
Status | assigned => feedback |
| 2017-09-06 10:35 | bugmaster | Note Added: 0070258 | |
| 2017-09-06 10:35 | bugmaster | Status | feedback => tested |
| 2017-09-06 10:36 | bugmaster | Test case number | => bugs modalg_7 bug29038 |
| 2017-09-18 13:43 | bugmaster | Relationship added | related to 0025081 |
| 2017-09-21 19:33 | bugmaster | Changeset attached | => occt master d89c7d60 |
| 2017-09-21 19:33 | bugmaster | Status | tested => verified |
| 2017-09-21 19:33 | bugmaster | Resolution | open => fixed |
| 2017-09-21 19:48 | apn | Target Version | 7.4.0 => 7.2.0 |
| 2017-09-22 12:05 | bugmaster | Target Version | 7.2.0 => 7.4.0 |
| 2017-09-29 13:26 | bugmaster | Changeset attached | => occt master 28c08a7d |
| 2017-09-29 17:11 | git | Note Added: 0071013 | |
| 2017-09-29 17:11 | git | Note Added: 0071014 | |
| 2018-02-20 12:59 |
|
Target Version | 7.4.0 => 7.3.0 |
| 2018-06-29 21:15 |
|
Fixed in Version | => 7.3.0 |
| 2018-06-29 21:19 |
|
Status | verified => closed |
