Print of bytecode dump when using sparse.COO.dot
mayalinetsky-kryon opened this issue · comments
Describe the bug
While using sparse.COO.dot
function between two sparse.COO
instances a really long red text is printed to the console, which includes "bytecode dump", "stack", "dispatch", "pending", "changing phismap" and more.
To Reproduce
import sparse
matrix1 = sparse.random((1,3,14),nnz=9,format='coo')
matrix2 = sparse.random((1,2,14), nnz=8, format='coo')
matrix1.dot(matrix2.transpose((0,2,1)))
Small snippet of the text
Loading extension: init = sparse._numba_extension:_init_extension
bytecode dump:
> 0 NOP(arg=None, lineno=984)
2 LOAD_FAST(arg=0, lineno=984)
4 UNPACK_SEQUENCE(arg=2, lineno=984)
6 STORE_FAST(arg=7, lineno=984)
8 STORE_FAST(arg=8, lineno=984)
10 LOAD_GLOBAL(arg=0, lineno=986)
12 LOAD_FAST(arg=0, lineno=987)
14 LOAD_FAST(arg=1, lineno=987)
16 LOAD_CONST(arg=1, lineno=987)
18 BINARY_SUBSCR(arg=None, lineno=987)
20 LOAD_FAST(arg=2, lineno=987)
22 LOAD_CONST(arg=1, lineno=987)
24 BINARY_SUBSCR(arg=None, lineno=987)
26 LOAD_FAST(arg=5, lineno=987)
28 LOAD_FAST(arg=6, lineno=987)
30 CALL_FUNCTION(arg=5, lineno=986)
32 STORE_FAST(arg=9, lineno=986)
34 LOAD_GLOBAL(arg=1, lineno=989)
36 LOAD_ATTR(arg=2, lineno=989)
38 LOAD_CONST(arg=2, lineno=989)
40 LOAD_FAST(arg=9, lineno=989)
42 BUILD_TUPLE(arg=2, lineno=989)
44 LOAD_GLOBAL(arg=1, lineno=989)
46 LOAD_ATTR(arg=3, lineno=989)
48 LOAD_CONST(arg=3, lineno=989)
50 CALL_FUNCTION_KW(arg=2, lineno=989)
52 STORE_FAST(arg=10, lineno=989)
54 LOAD_GLOBAL(arg=1, lineno=990)
56 LOAD_ATTR(arg=2, lineno=990)
58 LOAD_FAST(arg=9, lineno=990)
60 LOAD_DEREF(arg=0, lineno=990)
62 LOAD_CONST(arg=3, lineno=990)
64 CALL_FUNCTION_KW(arg=2, lineno=990)
66 STORE_FAST(arg=11, lineno=990)
68 LOAD_GLOBAL(arg=1, lineno=991)
70 LOAD_METHOD(arg=4, lineno=991)
72 LOAD_FAST(arg=8, lineno=991)
74 LOAD_CONST(arg=4, lineno=991)
76 CALL_METHOD(arg=2, lineno=991)
78 STORE_FAST(arg=12, lineno=991)
80 LOAD_GLOBAL(arg=1, lineno=992)
82 LOAD_ATTR(arg=5, lineno=992)
84 LOAD_FAST(arg=8, lineno=992)
System
- OS and version: Windows 10 Pro Version 21H1
sparse
version '0.13.0'- NumPy version '1.21.5'
- Numba version '0.55.1'
Added NUMBA_DISABLE_JIT=1 to the environment variables and it does not print anything anymore,
For more information, go here.
JIT is an important part of speeding up the code. Let's keep this issue open.
I cannot reproduce this issue, I assume it was an issue in your environment, please comment if necessary.