An error occurred fetching the project authors.
- 04 Feb, 2016 2 commits
-
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
- 03 Feb, 2016 4 commits
-
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
First issue is that the assembly logging previously worked on the assembler buffer, which is reasonable but has issues with the bjit which will update the assembly post-relocation (ie by initializing the patchpoints). Also add some extra flags to separate commenting for rewriter-level comments ("this is the beginning of a setattr") and the new bjit ones ("this is a set local").
-
- 14 Dec, 2015 1 commit
-
-
Kevin Modzelewski authored
-
- 08 Dec, 2015 2 commits
-
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
- 24 Nov, 2015 4 commits
-
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
- 21 Nov, 2015 1 commit
-
-
Kevin Modzelewski authored
Current issue is that this is triggering the "mark IC as something being inside", which trips up the bjit's "overwrite the last few bytes of the previous block, assuming that they are just a jump".
-
- 20 Nov, 2015 3 commits
-
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
-
- 18 Nov, 2015 1 commit
-
-
Kevin Modzelewski authored
-
- 13 Nov, 2015 1 commit
-
-
Kevin Modzelewski authored
-
- 02 Nov, 2015 2 commits
-
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
and a couple of the other related functions: - addRTFunction is now an overload of FunctionMetadata::addVersion - boxFunctionMetadata is now createFunctionFromMetadata (it doesn't return a box) - createRTFunction was just a wrapper around new FunctionMetadata(), so get rid of it - boxRTFunction is now FunctionMetadata::create "CLFunction" never really made sense as a name, and some of the other related functions didn't make sense either (especially after a couple refactorings happened). Hopefully this makes it less confusing.
-
- 20 Oct, 2015 1 commit
-
-
Marius Wachtler authored
LLVM tier: adjust num of IC slots depending on the num of times the IC got rewritten in the bjit tier
-
- 12 Oct, 2015 1 commit
-
-
Marius Wachtler authored
-
- 08 Oct, 2015 1 commit
-
-
Marius Wachtler authored
-
- 07 Oct, 2015 2 commits
-
-
Marius Wachtler authored
-
Marius Wachtler authored
-
- 01 Oct, 2015 1 commit
-
-
Dong-hee Na authored
-
- 01 Sep, 2015 1 commit
-
-
Kevin Modzelewski authored
(motivated by namedtuple) This involves two main changes: - changing the calling convention to pass `globals` as an argument if needed (this only applies going into compiled code, it's already passed into the interpreter) - changing the llvm irgenerator to use the new globals object
-
- 30 Aug, 2015 1 commit
-
-
Rudi Chen authored
-
- 28 Aug, 2015 2 commits
-
-
Kevin Modzelewski authored
-
Kevin Modzelewski authored
I don't think it's that important overall but it seems to be a decent percentage of the time to enter the interpreter/bjit; this commit saves about 10% in a silly ubenchmark.
-
- 24 Aug, 2015 2 commits
-
-
Kevin Modzelewski authored
By moving the deopt logic into an inner frame. I think we trigger the isDeopt() block twice since there's a destructor that has to get run.
-
Marius Wachtler authored
-
- 18 Aug, 2015 2 commits
-
-
Kevin Modzelewski authored
It had gotten to the point that a large part of the cost of throwing an eception was allocating the std::string's to represent the function and filename. Most of the other strings have already been converted to being represented as BoxedString's (ie python strings), so do that conversion here and a few more places.
-
Kevin Modzelewski authored
Instead of relying on stack unwinding to find it.
-
- 14 Aug, 2015 1 commit
-
-
Kevin Modzelewski authored
We're keeping them around anyway
-
- 12 Aug, 2015 2 commits
-
-
Kevin Modzelewski authored
This lets us avoid checking equality if the hash values didn't match. Checking equality can involve calling into user code, so it can be both a perf drain and a noticeable behavioral difference.
-
Vinzenz Feenstra authored
-
- 11 Aug, 2015 1 commit
-
-
Kevin Modzelewski authored
-
- 10 Aug, 2015 1 commit
-
-
Kevin Modzelewski authored
if we OSR out of a function, make sure we use the same exception style for the OSR'd version. We could also have the OSR exit code deal with switching exception styles, but for now it doesn't feel worth it.
-