Issues - unladen-swallow - A faster implementation of Python

ListGrid

1 - 61 of 61

ID ▼	Type ▼	Status ▼	Priority ▼	Release ▼	Owner ▼	Summary + Labels ▼
6	Enhancement	Accepted	Medium	----	----	Persist LLVM IR to .pyc files
10	Defect	Accepted	Medium	----	----	Huge expressions are really slow to compile under LLVM Performance
13	Enhancement	Accepted	Medium	----	----	Speed up regular expressions Performance
14	Enhancement	Accepted	Low	----	----	Add a regex benchmark suite StarterProject
15	Enhancement	Started	Medium	Merger	----	Add fuzz-based testing StarterProject
16	Defect	Accepted	Medium	----	jyass...@gmail.com	Simple functions are way bigger than they need to be. Fix that. Performance
19	Defect	Accepted	Medium	----	----	Turn BUILD_TUPLE/BUILD_LIST into memcpy operations. Performance
20	Defect	Accepted	Medium	Merger	----	f_lasti is probably broken Beer
22	Defect	Accepted	Low	----	----	compilation fails with IBM xlc in ceval
29	Defect	Accepted	Low	----	----	Some tests are flaky Beer
33	Defect	Accepted	Medium	----	----	Compiling large functions runs out of memory
34	Enhancement	Accepted	Low	----	----	Make perf.py --track_memory work on other platforms StarterProject PyCon
35	Enhancement	Started	Critical	Merger	----	Need support for debugging LLVM-generated machine code Maintainability
40	Enhancement	Started	Critical	Merger	reid.kle...@gmail.com	Offload JIT compilation to secondary threads Performance
41	Enhancement	Accepted	Low	----	----	Teach the JIT to recompile things Maintainability
48	Enhancement	Accepted	Medium	----	----	Tune hotness function Performance
55	Enhancement	Accepted	Low	----	----	Use Google's MAO tool for extra performance Performance StarterProject
58	Enhancement	Accepted	Low	----	----	Keep track of LLVM's tail call support Performance
59	Enhancement	Accepted	Medium	----	----	Improve the set of LLVM optimization passes Performance
63	Enhancement	Accepted	Medium	Merger	----	Improve oprofile support Instrumentation StarterProject PyCon
64	Defect	Accepted	High	Merger	----	Major regression in startup time Performance
68	Defect	Started	High	Merger	neunon	Memory usage increased significantly
72	Enhancement	Accepted	Low	----	----	FDO: don't compile cold branches Performance
73	Enhancement	Started	Medium	----	----	FDO: specialize operators using type feedback Performance
74	Enhancement	Started	Medium	----	----	FDO: avoid Python function call overhead Performance
75	Enhancement	Accepted	High	----	----	FDO: inline builtins, if profitable Performance
76	Enhancement	Accepted	Low	----	reid.kle...@gmail.com	Modify benchmarks so that we can compare hotness heuristics Instrumentation
80	Enhancement	New	----	----	----	Bitcode loading support
86	Enhancement	Started	Medium	----	e...@4geeks.de	FDO: Python/Python inlining Performance
87	Defect	Accepted	Medium	----	----	FreeBSD 7.1 unladen-2009Q3 build fails
88	Enhancement	Accepted	Medium	----	----	Expand direct calling to more C functions Performance StarterProject PyCon
91	Defect	Accepted	Medium	2009Q4	collinw	Using tracing to jump out of a loop causes the Python object stack to overflow
92	Defect	Accepted	Medium	----	collinw	reproduceable segfault with unladen
97	Enhancement	New	Low	----	----	More control on JIT behaviour
100	Enhancement	Accepted	Medium	----	----	New benchmarking snippet Performance
105	Enhancement	Accepted	Medium	----	----	html5lib no quicker or slower than CPython Performance
109	Defect	New	Medium	2009Q4	----	python -m cherrypy.test.benchmark fails
111	Defect	New	Medium	----	----	Ghost fatal guard failure in simple function
113	Enhancement	Accepted	Medium	----	----	FDO: specialize isinstance() calls Performance
116	Enhancement	Accepted	Medium	----	----	Inline Comparison Operations Performance
118	Defect	Accepted	Critical	Merger	----	Massive increase in binary size
119	Enhancement	Accepted	Medium	----	----	Change optimized binary operations to use llvm overflow intrinsics Performance StarterProject
120	Enhancement	Accepted	Medium	----	----	Make perf.py usable by other Python implementations Performance PyCon
123	Defect	Started	High	Merger	collinw	Ensure -j never has minimal overhead compared to CPython Performance
125	Enhancement	Accepted	Medium	----	alex.gay...@gmail.com	Specialize UNPACK_SEQUENCE based on type feedback Performance PyCon
126	Enhancement	Started	Medium	----	e...@4geeks.de	Improve our alias analysis for LLVM Performance
129	Enhancement	Started	----	----	collinw	Support traditional make?
130	Enhancement	Started	Critical	Merger	jyass...@gmail.com	Improve shared linking support in LLVM
131	Enhancement	Started	Critical	Merger	jyass...@gmail.com	Make LLVM 2.7 enough for Python
132	Defect	Accepted	High	----	----	Improve type-check propagation Performance
133	Enhancement	New	----	----	----	installing u-s in parallel to python-2.6
135	Defect	New	Medium	----	----	Cannot run mod_wsgi
136	Enhancement	Started	----	Merger	collinw	Build LLVM JIT as a module
137	Enhancement	Accepted	Critical	Merger	collinw	Expose ability to turn off JIT via sys module
138	Defect	New	Medium	----	----	Problems with __builtins__ sandboxing technique (RestrictedPython)
139	Enhancement	Accepted	Medium	----	----	Collect feedback only after code objects have been determined to be "warm" Performance
141	Enhancement	Accepted	Medium	----	----	FDO: specialize dict operations PyCon Performance
142	Enhancement	Accepted	High	----	alex.gay...@gmail.com	Specialize FOR_ITER opcode Performance PyCon
143	Enhancement	New	----	----	----	Create functions for Python/getargs.c arg parsing
144	Enhancement	New	----	----	----	Special Build for debugging LLVM IR
145	Defect	New	Medium	----	----	Add some benchmark diagrams

CSV

1 - 61 of 61

Show columns:
♦ ID
♦ Type
♦ Status
♦ Priority
♦ Release
♦ Owner
♦ Summary
Milestone
Attachments
Stars
Opened
Closed
Modified
BlockedOn
Blocking
Blocked
MergedInto
Reporter
Cc
Project
Opsys
Edit column spec...