bkochuna / ners570f24-SpMV

3 stars 0 forks source link

Storage tests for COO #51

Open KyleVaughn opened 1 month ago

KyleVaughn commented 1 month ago

Description:

Storage test evaluates if the correct values are read from the input format and stored in the COO format:

  1. rowIdx: (array of int) the row ID of the non-zero element
  2. colIdx: (array of int) the column ID of the non-zero element
  3. value: (array of double/float): the value of the non-zero element

Tasks:

(Each task is currently blocked by the previous task)

Definition of done:

The values can be read from the input format and stored in the COO format properly, and all tasks are checked off.

xianyu2023umich commented 2 weeks ago

I'm reading it. Very comprehensive definition of done.

xianyu2023umich commented 1 week ago

@athenafhl Hey I'm reviewing you test it fails the compile at line 34: SpMV::SparseMatrix* ptr_A = new SpMV::SparseMatrix_COO(n, n); it's probably cuz you declared fp_type in the template instead of T.

athenafhl commented 1 week ago

@athenafhl Hey I'm reviewing you test it fails the compile at line 34: SpMV::SparseMatrix* ptr_A = new SpMV::SparseMatrix_COO(n, n); it's probably cuz you declared fp_type in the template instead of T.

Hi! I fix all the bugs! I think the rest of the errors come from the matvec test, not the storage test. I can do nothing until the other people update their branches.

xianyu2023umich commented 1 week ago

@athenafhl Hey I'm reviewing you test it fails the compile at line 34: SpMV::SparseMatrix* ptr_A = new SpMV::SparseMatrix_COO(n, n); it's probably cuz you declared fp_type in the template instead of T.

Hi! I fix all the bugs! I think the rest of the errors come from the matvec test, not the storage test. I can do nothing until the other people update their branches.

Tried your test again passed the compiler but the test failed:

[ctest] Site: gl-login3.arc-ts.umich.edu [ctest] Build name: Linux-g++ [ctest] Test project /home/xianyu/hw3/ners570f24-SpMV/build [ctest] Start 1: test_COO_storage_test [ctest] 1/1 Test #1: test_COO_storage_test ............Subprocess aborted***Exception: 5.43 sec [ctest] test_COO_storage_test: /home/xianyu/hw3/ners570f24-SpMV/tests/COO_storage_test.cpp:58: void random_nxn() [with long unsigned int n = 100; fp_type = double]: Assertion `I[i] == rowIdx[i]' failed. [ctest] [ctest] [ctest] 0% tests passed, 1 tests failed out of 1 [ctest] [ctest] Total Test time (real) = 5.45 sec [ctest] [ctest] The following tests FAILED: [ctest] 1 - test_COO_storage_test (Subprocess aborted) [ctest] Errors while running CTest

Is this due to the matvec?

xianyu2023umich commented 1 week ago

@athenafhl Hey I'm reviewing you test it fails the compile at line 34: SpMV::SparseMatrix* ptr_A = new SpMV::SparseMatrix_COO(n, n); it's probably cuz you declared fp_type in the template instead of T.

Hi! I fix all the bugs! I think the rest of the errors come from the matvec test, not the storage test. I can do nothing until the other people update their branches.

Tried your test again passed the compiler but the test failed:

[ctest] Site: gl-login3.arc-ts.umich.edu [ctest] Build name: Linux-g++ [ctest] Test project /home/xianyu/hw3/ners570f24-SpMV/build [ctest] Start 1: test_COO_storage_test [ctest] 1/1 Test #1: test_COO_storage_test ............Subprocess aborted***Exception: 5.43 sec [ctest] test_COO_storage_test: /home/xianyu/hw3/ners570f24-SpMV/tests/COO_storage_test.cpp:58: void random_nxn() [with long unsigned int n = 100; fp_type = double]: Assertion `I[i] == rowIdx[i]' failed. [ctest] [ctest] [ctest] 0% tests passed, 1 tests failed out of 1 [ctest] [ctest] Total Test time (real) = 5.45 sec [ctest] [ctest] The following tests FAILED: [ctest] 1 - test_COO_storage_test (Subprocess aborted) [ctest] Errors while running CTest

Is this due to the matvec?

Oh no I think it's due to the storage has nothing to do with matvec.

athenafhl commented 1 week ago

@athenafhl Hey I'm reviewing you test it fails the compile at line 34: SpMV::SparseMatrix* ptr_A = new SpMV::SparseMatrix_COO(n, n); it's probably cuz you declared fp_type in the template instead of T.

Hi! I fix all the bugs! I think the rest of the errors come from the matvec test, not the storage test. I can do nothing until the other people update their branches.

Tried your test again passed the compiler but the test failed: [ctest] Site: gl-login3.arc-ts.umich.edu [ctest] Build name: Linux-g++ [ctest] Test project /home/xianyu/hw3/ners570f24-SpMV/build [ctest] Start 1: test_COO_storage_test [ctest] 1/1 Test #1: test_COO_storage_test ............Subprocess aborted***Exception: 5.43 sec [ctest] test_COO_storage_test: /home/xianyu/hw3/ners570f24-SpMV/tests/COO_storage_test.cpp:58: void random_nxn() [with long unsigned int n = 100; fp_type = double]: Assertion `I[i] == rowIdx[i]' failed. [ctest] [ctest] [ctest] 0% tests passed, 1 tests failed out of 1 [ctest] [ctest] Total Test time (real) = 5.45 sec [ctest] [ctest] The following tests FAILED: [ctest] 1 - test_COO_storage_test (Subprocess aborted) [ctest] Errors while running CTest Is this due to the matvec?

Oh no I think it's due to the storage has nothing to do with matvec.

Yeah you are right. I delete the matvec part and the test did not pass. Did you see the assemble method for COO? Would it be the problem of creating the storage? It looks pretty weird. I feel that it should be index++ instead of index += index

xianyu2023umich commented 1 week ago

@athenafhl Hey I'm reviewing you test it fails the compile at line 34: SpMV::SparseMatrix* ptr_A = new SpMV::SparseMatrix_COO(n, n); it's probably cuz you declared fp_type in the template instead of T.

Hi! I fix all the bugs! I think the rest of the errors come from the matvec test, not the storage test. I can do nothing until the other people update their branches.

Tried your test again passed the compiler but the test failed: [ctest] Site: gl-login3.arc-ts.umich.edu [ctest] Build name: Linux-g++ [ctest] Test project /home/xianyu/hw3/ners570f24-SpMV/build [ctest] Start 1: test_COO_storage_test [ctest] 1/1 Test #1: test_COO_storage_test ............Subprocess aborted***Exception: 5.43 sec [ctest] test_COO_storage_test: /home/xianyu/hw3/ners570f24-SpMV/tests/COO_storage_test.cpp:58: void random_nxn() [with long unsigned int n = 100; fp_type = double]: Assertion `I[i] == rowIdx[i]' failed. [ctest] [ctest] [ctest] 0% tests passed, 1 tests failed out of 1 [ctest] [ctest] Total Test time (real) = 5.45 sec [ctest] [ctest] The following tests FAILED: [ctest] 1 - test_COO_storage_test (Subprocess aborted) [ctest] Errors while running CTest Is this due to the matvec?

Oh no I think it's due to the storage has nothing to do with matvec.

Yeah you are right. I delete the matvec part and the test did not pass. Did you see the assemble method for COO? Would it be the problem of creating the storage? It looks pretty weird.

yeah I don't really understand that loop (especially the head) but definitely it shouldn't be index += index should be +=1

athenafhl commented 1 week ago

@athenafhl Hey I'm reviewing you test it fails the compile at line 34: SpMV::SparseMatrix* ptr_A = new SpMV::SparseMatrix_COO(n, n); it's probably cuz you declared fp_type in the template instead of T.

Hi! I fix all the bugs! I think the rest of the errors come from the matvec test, not the storage test. I can do nothing until the other people update their branches.

Tried your test again passed the compiler but the test failed: [ctest] Site: gl-login3.arc-ts.umich.edu [ctest] Build name: Linux-g++ [ctest] Test project /home/xianyu/hw3/ners570f24-SpMV/build [ctest] Start 1: test_COO_storage_test [ctest] 1/1 Test #1: test_COO_storage_test ............Subprocess aborted***Exception: 5.43 sec [ctest] test_COO_storage_test: /home/xianyu/hw3/ners570f24-SpMV/tests/COO_storage_test.cpp:58: void random_nxn() [with long unsigned int n = 100; fp_type = double]: Assertion `I[i] == rowIdx[i]' failed. [ctest] [ctest] [ctest] 0% tests passed, 1 tests failed out of 1 [ctest] [ctest] Total Test time (real) = 5.45 sec [ctest] [ctest] The following tests FAILED: [ctest] 1 - test_COO_storage_test (Subprocess aborted) [ctest] Errors while running CTest Is this due to the matvec?

Oh no I think it's due to the storage has nothing to do with matvec.

Yeah you are right. I delete the matvec part and the test did not pass. Did you see the assemble method for COO? Would it be the problem of creating the storage? It looks pretty weird.

yeah I don't really understand that loop (especially the head) but definitely it shouldn't be index += index should be +=1

Ok I'll tell him to debug that. Thank you for your help!