search for: insnclass

...ok to >>> commit or does it not apply to trunk properly ? >>> I can fix it if that's the problem. >>> >>> Ivan >>> >>> On 20/06/2012 19:33, Anshuman Dasgupta wrote: >>>>> Thanks for reviewing this. I added a top comment for AddInsnClass >>>> and I fixed the violation of column numbers. >>>> >>>> Great. Looks good to me. >>>> >>>>> IMO, it wil be nice to keep it alive for performance comparisons. >>>> Given the overall performance >>>>> is rat...

...s thread ;-). Is this patch ok to >> commit or does it not apply to trunk properly ? >> I can fix it if that's the problem. >> >> Ivan >> >> On 20/06/2012 19:33, Anshuman Dasgupta wrote: >>>> Thanks for reviewing this. I added a top comment for AddInsnClass >>> and I fixed the violation of column numbers. >>> >>> Great. Looks good to me. >>> >>>> IMO, it wil be nice to keep it alive for performance comparisons. >>> Given the overall performance >>>> is rather determined by transition...

...t or does it not apply to trunk properly ? >>>> I can fix it if that's the problem. >>>> >>>> Ivan >>>> >>>> On 20/06/2012 19:33, Anshuman Dasgupta wrote: >>>>>> Thanks for reviewing this. I added a top comment for AddInsnClass >>>>> and I fixed the violation of column numbers. >>>>> >>>>> Great. Looks good to me. >>>>> >>>>>> IMO, it wil be nice to keep it alive for performance comparisons. >>>>> Given the overall performance &gt...

...ce 13! states. Fortunately, many of those states are reused :-) but it still takes up to 11min to build the entire table. This patch reduces the build time to 5min, giving a speed-up factor greater than 2. It contains small changes: - Transitions are stored in a set for quicker searches - canAddInsnClass() API is split in two API's: - canAddInsnClass() which perform a quick verification about the possibility of having new states for a given InsnClass - AddInsnClass() performs the actual computation of possible states. I've regenerated the DFA table of Hexagon and all seems to be ok....

Hi Anshu, Thanks for reviewing this. I added a top comment for AddInsnClass and I fixed the violation of column numbers. On 15/06/2012 21:31, Anshuman Dasgupta wrote: > Hi Ivan, > > The patch looks good to me. I have a couple of minor comments: > > +void State::AddInsnClass(unsigned InsnClass, > Add a top level comment describing the function > >...

Hi Ivan, The patch looks good to me. I have a couple of minor comments: +void State::AddInsnClass(unsigned InsnClass, Add a top level comment describing the function + std::map<State*, std::set<Transition*, ltTransition>, ltState> stateTransitions; You should be able to use SmallSet here. Also, this line exceeds 80 columns. On a related note, is the CachedTable mechanism in DFA...

Hi Anshu, Just in case you have forgotten this thread ;-). Is this patch ok to commit or does it not apply to trunk properly ? I can fix it if that's the problem. Ivan On 20/06/2012 19:33, Anshuman Dasgupta wrote: > > > Thanks for reviewing this. I added a top comment for AddInsnClass > and I fixed the violation of column numbers. > > Great. Looks good to me. > > > IMO, it wil be nice to keep it alive for performance comparisons. > Given the overall performance > > is rather determined by transition searches on the current state, > for small DFA...

> Thanks for reviewing this. I added a top comment for AddInsnClass and I fixed the violation of column numbers. Great. Looks good to me. > IMO, it wil be nice to keep it alive for performance comparisons. Given the overall performance > is rather determined by transition searches on the current state, for small DFA tables may not be a win > and it...

...have forgotten this thread ;-). Is this patch ok to > commit or does it not apply to trunk properly ? > I can fix it if that's the problem. > > Ivan > > On 20/06/2012 19:33, Anshuman Dasgupta wrote: >> >> > Thanks for reviewing this. I added a top comment for AddInsnClass >> and I fixed the violation of column numbers. >> >> Great. Looks good to me. >> >> > IMO, it wil be nice to keep it alive for performance comparisons. >> Given the overall performance >> > is rather determined by transition searches on the current s...

...Packetizer fails at assert(CachedTable.count(StateTrans) != 0); in the following function: // reserveResources - Reserve the resources occupied by a MCInstrDesc and // change the current state to reflect that change. void DFAPacketizer::reserveResources(const llvm::MCInstrDesc *MID) { unsigned InsnClass = MID->getSchedClass(); const llvm::InstrStage *IS = InstrItins->beginStage(InsnClass); unsigned FuncUnits = IS->getUnits(); UnsignPair StateTrans = UnsignPair(CurrentState, FuncUnits); ReadTable(CurrentState); assert(CachedTable.count(StateTrans) != 0); CurrentState = CachedTa...

...f those states > are reused :-) but it still takes up to 11min to build the entire > table. This patch reduces the build time to 5min, giving a speed-up > factor greater than 2. > > It contains small changes: > - Transitions are stored in a set for quicker searches > - canAddInsnClass() API is split in two API's: > - canAddInsnClass() which perform a quick verification about the > possibility of having new states for a given InsnClass > - AddInsnClass() performs the actual computation of possible states. > > I've regenerated the DFA table of Hexagon a...