aboutsummaryrefslogtreecommitdiff
path: root/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp
AgeCommit message (Collapse)Author
2011-06-21Fix some trailing issues from my introduction of MVT::untyped and its use ↵Owen Anderson
for REGISTER_SEQUENCE. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133567 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-18Remove unused but set variables.Benjamin Kramer
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133347 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-15Add a new MVT::untyped. This will be used in future work for modelling ISA ↵Owen Anderson
features like register pairs and lists with "interesting" constraints (such as ARM NEON contiguous register lists or even-odd paired registers). We need to be able to generate these instructions (often from intrinsics), but don't want to have to assign a legal type to them. Instead, we'll use an "untyped" edge to bypass the type-checking and simply ensure that the register classes match. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133106 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-15Added -stress-sched flag in the Asserts build.Andrew Trick
Added a test case for handling physreg aliases during pre-RA-sched. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133063 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-08Remove a temporary test case probe in CheckForLiveRegDef.Andrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132751 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-07Fix a merge bug in preRAsched for handling physreg aliases.Andrew Trick
I've been sitting on this long enough trying to find a test case. I think the fix should go in now, but I'll keep working on the test case. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@132701 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-26Be careful about scheduling nodes above previous calls. It increase usages ofEvan Cheng
more callee-saved registers and introduce copies. Only allows it if scheduling a node above calls would end up lessen register pressure. Call operands also has added ABI restrictions for register allocation, so be extra careful with hoisting them above calls. rdar://9329627 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130245 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-26Fix typoEvan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@130190 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-14In the pre-RA scheduler, maintain cmp+br proximity.Andrew Trick
This is done by pushing physical register definitions close to their use, which happens to handle flag definitions if they're not glued to the branch. This seems to be generally a good thing though, so I didn't need to add a target hook yet. The primary motivation is to generate code closer to what people expect and rule out missed opportunity from enabling macro-op fusion. As a side benefit, we get several 2-5% gains on x86 benchmarks. There is one regression: SingleSource/Benchmarks/Shootout/lists slows down be -10%. But this is an independent scheduler bug that will be tracked separately. See rdar://problem/9283108. Incidentally, pre-RA scheduling is only half the solution. Fixing the later passes is tracked by: <rdar://problem/8932804> [pre-RA-sched] on x86, attempt to schedule CMP/TEST adjacent with condition jump Fixes: <rdar://problem/9262453> Scheduler unnecessary break of cmp/jump fusion git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@129508 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-13Recommit r129383. PreRA scheduler heuristic fixes: VRegCycle, TokenFactor ↵Andrew Trick
latency. Additional fixes: Do something reasonable for subtargets with generic itineraries by handle node latency the same as for an empty itinerary. Now nodes default to unit latency unless an itinerary explicitly specifies a zero cycle stage or it is a TokenFactor chain. Original fixes: UnitsSharePred was a source of randomness in the scheduler: node priority depended on the queue data structure. I rewrote the recent VRegCycle heuristics to completely replace the old heuristic without any randomness. To make the ndoe latency adjustments work, I also needed to do something a little more reasonable with TokenFactor. I gave it zero latency to its consumers and always schedule it as low as possible. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@129421 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-12Revert 129383. It causes some targets to hit a scheduler assert.Andrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@129385 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-12PreRA scheduler heuristic fixes: VRegCycle, TokenFactor latency.Andrew Trick
UnitsSharePred was a source of randomness in the scheduler: node priority depended on the queue data structure. I rewrote the recent VRegCycle heuristics to completely replace the old heuristic without any randomness. To make these heuristic adjustments to node latency work, I also needed to do something a little more reasonable with TokenFactor. I gave it zero latency to its consumers and always schedule it as low as possible. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@129383 91177308-0d34-0410-b5e6-96231b3b80d8
2011-04-07Added a check in the preRA scheduler for potential interference on aAndrew Trick
induction variable. The preRA scheduler is unaware of induction vars, so we look for potential "virtual register cycles" instead. Fixes <rdar://problem/8946719> Bad scheduling prevents coalescing git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@129100 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-25Fix for -pre-RA-sched=source.Andrew Trick
Yet another case of unchecked NULL node (for physreg copy). May fix PR9509. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@128266 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-23Ensure that def-side physreg copies are scheduled above any other usesAndrew Trick
so the scheduler can't create new interferences on the copies themselves. Prior to this fix the scheduler could get stuck in a loop creating copies. Fixes PR9509. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@128164 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-23whitespaceAndrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@128163 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-21Grammar-o.Eric Christopher
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@128004 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-10Re-commit 127368 and 127371. They are exonerated.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127380 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-09Revert 127368 and 127371 for now.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127376 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-09Change the definition of TargetRegisterInfo::getCrossCopyRegClass to be moreEvan Cheng
flexible. If it returns a register class that's different from the input, then that's the register class used for cross-register class copies. If it returns a register class that's the same as the input, then no cross- register class copies are needed (normal copies would do). If it returns null, then it's not at all possible to copy registers of the specified register class. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127368 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-09Fix typo, make helper static.Benjamin Kramer
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127335 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-08Fix some latent bugs if the nodes are unschedulable. We'd gotten awayEric Christopher
with this before since none of the register tracking or nightly tests had unschedulable nodes. This should probably be refixed with a special default Node that just returns some "don't touch me" values. Fixes PR9427 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127263 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-08Further improvements to pre-RA-sched=list-ilp.Andrew Trick
This change uses the MaxReorderWindow for both height and depth, which tends to limit the negative effects of high register pressure. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127203 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-07Move getRegPressureLimit() from TargetLoweringInfo to TargetRegisterInfo.Cameron Zwarich
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127175 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-06Typo.Eric Christopher
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127131 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-06Disable a couple of experimental heuristics to get the best results from the ↵Andrew Trick
current implementation of -pre-RA-sched=list-ilp. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127113 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-05Be explicit with abs(). Visual Studio workaround.Andrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127075 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-05Missing comment.Andrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127068 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-05Increased the register pressure limit on x86_64 from 8 to 12Andrew Trick
regs. This is the only change in this checkin that may affects the default scheduler. With better register tracking and heuristics, it doesn't make sense to artificially lower the register limit so much. Added -sched-high-latency-cycles and X86InstrInfo::isHighLatencyDef to give the scheduler a way to account for div and sqrt on targets that don't have an itinerary. It is currently defaults to 10 (the actual number doesn't matter much), but only takes effect on non-default schedulers: list-hybrid and list-ilp. Added several heuristics that can be individually disabled for the non-default sched=list-ilp mode. This helps us determine how much better we can do on a given benchmark than the default scheduler. Certain compute intensive loops run much faster in this mode with the right set of heuristics, and it doesn't seem to have much negative impact elsewhere. Not all of the heuristics are needed, but we still need to experiment to decide which should be disabled by default for sched=list-ilp. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@127067 91177308-0d34-0410-b5e6-96231b3b80d8
2011-03-04Minor pre-RA-sched fixes and cleanup.Andrew Trick
Fix the PendingQueue, then disable it because it's not required for the current schedulers' heuristics. Fix the logic for the unused list-ilp scheduler. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@126981 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-04Introducing a new method of tracking register pressure. We can'tAndrew Trick
precisely track pressure on a selection DAG, but we can at least keep it balanced. This design accounts for various interesting aspects of selection DAGS: register and subregister copies, glued nodes, dead nodes, unused registers, etc. Added SUnit::NumRegDefsLeft and ScheduleDAGSDNodes::RegDefIter. Note: I disabled PrescheduleNodesWithMultipleUses when register pressure is enabled, based on no evidence other than I don't think it makes sense to have both enabled. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@124853 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-27Remove a temporary workaround for a lencod miscompile. Depends on the fix in ↵Andrew Trick
r124442. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@124443 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-24Temporarily workaround JM/lencod miscompile (SIGSEGV).Andrew Trick
rdar://problem/8893967 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@124137 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-21Enable support for precise scheduling of the instruction selectionAndrew Trick
DAG. Disable using "-disable-sched-cycles". For ARM, this enables a framework for modeling the cpu pipeline and counting stalls. It also activates several heuristics to drive scheduling based on the model. Scheduling is inherently imprecise at this stage, and until spilling is improved it may defeat attempts to schedule. However, this framework provides greater control over tuning codegen. Although the flag is not target-specific, it should have very little affect on the default scheduler used by x86. The only two changes that affect x86 are: - scheduling a high-latency operation bumps the current cycle so independent operations can have their latency covered. i.e. two independent 4 cycle operations can produce results in 4 cycles, not 8 cycles. - Two operations with equal register pressure impact and no latency-based stalls on their uses will be prioritized by depth before height (height is irrelevant if no stalls occur in the schedule below this point). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123971 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-21Convert -enable-sched-cycles and -enable-sched-hazard to -disableAndrew Trick
flags. They are still not enable in this revision. Added TargetInstrInfo::isZeroCost() to fix a fundamental problem with the scheduler's model of operand latency in the selection DAG. Generalized unit tests to work with sched-cycles. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123969 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-20Selection DAG scheduler register pressure heuristic fixes.Andrew Trick
Added a check for already live regs before claiming HighRegPressure. Fixed a few cases of checking the wrong number of successors. Added some tracing until these heuristics are better understood. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123892 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-14Support for precise scheduling of the instruction selection DAG,Andrew Trick
disabled in this checkin. Sorry for the large diffs due to refactoring. New functionality is all guarded by EnableSchedCycles. Scheduling the isel DAG is inherently imprecise, but we give it a best effort: - Added MayReduceRegPressure to allow stalled nodes in the queue only if there is a regpressure need. - Added BUHasStall to allow checking for either dependence stalls due to latency or resource stalls due to pipeline hazards. - Added BUCompareLatency to encapsulate and standardize the heuristics for minimizing stall cycles (vs. reducing register pressure). - Modified the bottom-up heuristic (now in BUCompareLatency) to prioritize nodes by their depth rather than height. As long as it doesn't stall, height is irrelevant. Depth represents the critical path to the DAG root. - Added hybrid_ls_rr_sort::isReady to filter stalled nodes before adding them to the available queue. Related Cleanup: most of the register reduction routines do not need to be templates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123468 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-24Minor cleanup related to my latest scheduler changes.Andrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122545 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-24Fix a few cases where the scheduler is not checking for phys reg copies. The ↵Andrew Trick
scheduling node may have a NULL DAG node, yuck. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122544 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-24Various bits of framework needed for precise machine-level selectionAndrew Trick
DAG scheduling during isel. Most new functionality is currently guarded by -enable-sched-cycles and -enable-sched-hazard. Added InstrItineraryData::IssueWidth field, currently derived from ARM itineraries, but could be initialized differently on other targets. Added ScheduleHazardRecognizer::MaxLookAhead to indicate whether it is active, and if so how many cycles of state it holds. Added SchedulingPriorityQueue::HasReadyFilter to allowing gating entry into the scheduler's available queue. ScoreboardHazardRecognizer now accesses the ScheduleDAG in order to get information about it's SUnits, provides RecedeCycle for bottom-up scheduling, correctly computes scoreboard depth, tracks IssueCount, and considers potential stall cycles when checking for hazards. ScheduleDAGRRList now models machine cycles and hazards (under flags). It tracks MinAvailableCycle, drives the hazard recognizer and priority queue's ready filter, manages a new PendingQueue, properly accounts for stall cycles, etc. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122541 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-23flags -> glue for selectiondagChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122509 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-23Reorganize ListScheduleBottomUp in preparation for modeling machine cycles ↵Andrew Trick
and instruction issue. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122491 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-23Converted LiveRegCycles to LiveRegGens. It's easier to work with and allows ↵Andrew Trick
multiple nodes per cycle. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122474 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-23In CheckForLiveRegDef use TRI->getOverlaps.Andrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122473 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-23Fixes PR8823: add-with-overflow-128.llAndrew Trick
In the bottom-up selection DAG scheduling, handle two-address instructions that read/write unspillable registers. Treat the entire chain of two-address nodes as a single live range. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122472 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-21In DelayForLiveRegsBottomUp, handle instructions that read and writeAndrew Trick
the same physical register. Simplifies the fix from the previous checkin r122211. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122370 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-21whitespaceAndrew Trick
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122368 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-21rename MVT::Flag to MVT::Glue. "Flag" is a terrible name forChris Lattner
something that just glues two nodes together, even if it is sometimes used for flags. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122310 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-20Fix a bug in the scheduler's handling of "unspillable" vregs.Chris Lattner
Imagine we see: EFLAGS = inst1 EFLAGS = inst2 FLAGS gpr = inst3 EFLAGS Previously, we would refuse to schedule inst2 because it clobbers the EFLAGS of the predecessor. However, it also uses the EFLAGS of the predecessor, so it is safe to emit. SDep edges ensure that the right order happens already anyway. This fixes 2 testsuite crashes with the X86 patch I'm going to commit next. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122211 91177308-0d34-0410-b5e6-96231b3b80d8
2010-12-20the result of CheckForLiveRegDef is dead, remove it.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@122209 91177308-0d34-0410-b5e6-96231b3b80d8
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-09-05 20:43:25 +0000