similar to: [LLVMdev] Implementing devirtualization

On Thu, Dec 8, 2011 at 9:56 AM, Vitor Luis Menezes <vitor at utexas.edu> wrote: > Hello all, > > Our compilers class has been using LLVM, and a partner and I decided to > implement devirtualization of virtual C++ calls in LLVM as a class project. > We quickly realized that existing debug metadata generated by Clang didn't > give us enough info to (precisely) implement

We've got the following test case: class A { public: int x; A(int x) : x(x) {} int hoo() {return 4;} virtual int foo() {return x;} virtual int goo() {return foo()+10;} virtual int operator+(A &a) { return x + a.x; } }; class B : public A { public: B(int x) : A(x) {} int hoo() {return 2;} virtual int foo() {return A::foo()*2;} }; int main() { A* a = new A(1);

On Thu, Dec 8, 2011 at 2:11 PM, Vitor Luis Menezes <vitor at utexas.edu> wrote: > We've got the following test case: > > > class A { > public: >   int x; >   A(int x) : x(x) {} >   int hoo() {return 4;} >   virtual int foo() {return x;} >   virtual int goo() {return foo()+10;} >   virtual int operator+(A &a) { >     return x + a.x; >   } > };

----- Original Message ----- > From: "Hal Finkel" <hfinkel at anl.gov> > To: "Chandler Carruth" <chandlerc at google.com> > Cc: llvmdev at cs.uiuc.edu > Sent: Thursday, July 16, 2015 1:58:02 AM > Subject: Re: [LLVMdev] Improving loop vectorizer support for loops > with a volatile iteration variable > ----- Original Message ----- > >

Vitor Luis Menezes wrote: > Hello all, > > Our compilers class has been using LLVM, and a partner and I decided to > implement devirtualization of virtual C++ calls in LLVM as a class > project. We quickly realized that existing debug metadata generated by > Clang didn't give us enough info to (precisely) implement this, and as > such have already begun modifying Clang to

On Thu, Dec 8, 2011 at 4:26 PM, Eli Friedman <eli.friedman at gmail.com> wrote: > On Thu, Dec 8, 2011 at 2:11 PM, Vitor Luis Menezes <vitor at utexas.edu> wrote: >> We've got the following test case: >> >> >> class A { >> public: >>   int x; >>   A(int x) : x(x) {} >>   int hoo() {return 4;} >>   virtual int foo() {return x;}

On Dec 8, 2011, at 10:03 PM, Nick Lewycky wrote: > Noalias returns, nocapture, SCC refinement, linkonce_odr and > available_externally were added with the goal of making devirtualization > in LLVM happen, but as orthogonal independent optimizations. I think > LLVM should continue with this design. If you want to implement a single > substantial optimization pass, I would suggest

Hi Gerolf, I think we have several (perhaps separable) issues here: 1. Do we have a canonical form for loops, preserved through the optimizer, that allows naturally-constructed loop nests to remain separable? 2. Do we forbid non-lowering transformations that turn vectorizable loops into non-vectorizable loops? 3. How do we detect cases where transformations cause a negative answer to either

Chandler et al, I have been playing with the new PM pipeline, being particularly interested in how it can handle devirtualization. Now, I discovered what I believe is a "regression" vs old PM on a rather simple one-translation-unit testcase. clang is able to devirtualize it with -O3 and fails to do so with -fexperimental-new-pass-manager added. It looks like a pipeline issue,

Hi, I've recently run into a problem of missing !dbg metadata after the code has been optimized (even opt -O1). The original code was generated using llvmlite python package and I've verified that all instructions have their !dbg metadata present. After optimizing the module (even opt -O1) I see that some instructions (~25%) don't have any dbg metadata. These are mostly getelementptr

The frontend code is a pretty simple for loop, that counts from i = 0; i != 10; i += 1 It gets optimized into and endless loop. export fn entry() void { var array: [10]Bar = undefined; var x = for (array) |elem, i| { if (i == 1) break elem; } else bar2(); } Here's the generated IR: ; ModuleID = 'test' source_filename = "test" target datalayout =

I'm having an issue where SROA will throw away no-alias information on some loads after inlining, because the loads are derived from a store to an alloca which can be removed after inlining. The pointers that were originally stored into the alloca do *not *have any aliasing information - the only context that allowed me to assert aliasing was that the inlined-function guaranteed it to be so.

Yes, this looks broken in the new PM. The DevirtSCCRepeatedPass::run method first scans the functions in the SCC to collect value handles for indirect calls, runs the CGSCC pass pipeline and then checks if any of the call value handles now point to a direct call, in which case it runs the pipeline again (which should inline the devirtualized call) . The problem is scanning the initial SCC for

Using LLVM ToT and Hal's helpful slide deck [1], I've been trying to use `llvm.assume` to communicate pointer alignment guarantees to vector load and store instructions. For example, in [2] %5 and %9 are guaranteed to be 32-byte aligned. However, if I run this IR through `opt -O3 -datalayout -S`, the vectorized loads and stores are still 1-byte aligned [3]. What's going wrong? Do I

On Mon, Jun 19, 2017 at 12:06 PM, Mehdi AMINI <joker.eph at gmail.com> wrote: > Hi, > > 2017-06-19 8:45 GMT-07:00 Andrew Kelley via llvm-dev < > llvm-dev at lists.llvm.org>: > >> Greetings, >> >> I have a Zig implementation of ceil which is emitted into LLVM IR like >> this: >> >> ; Function Attrs: nobuiltin nounwind >> define

John McCall wrote: > On Dec 8, 2011, at 10:03 PM, Nick Lewycky wrote: >> Noalias returns, nocapture, SCC refinement, linkonce_odr and >> available_externally were added with the goal of making devirtualization >> in LLVM happen, but as orthogonal independent optimizations. I think >> LLVM should continue with this design. If you want to implement a single >>

Hey list, I've recently done the first test run of bumping our Burst compiler from LLVM 10 -> 11 now that the branch has been cut, and have noticed an apparent loop vectorization codegen regression for X86 with AVX or AVX2 enabled. The following IR example is vectorized to 4 wide with LLVM 11 and trunk whereas in LLVM 10 it (correctly as per what we want) vectorized it 8 wide matching the

Tried a bunch of them there (x86-64, haswell, znver2) and they all defaulted to 4-wide - haswell additionally caused some extra loop unrolling but still with 8-wide pows. Cheers, -Neil. On Thu, Jul 16, 2020 at 2:39 PM Roman Lebedev <lebedev.ri at gmail.com> wrote: > Did you specify the target CPU the code should be optimized for? > For clang that is -march=native/znver2/... /

I'm going to show the same function, first normally, and then as a coroutine, and show how gdb can see the variable when it's a normal function, but not when it's a coroutine. I'd like to understand if this can be improved. I'm trying to debug a real world problem, but the lack of debug info on variables in coroutines is making it difficult. Should I file a bug? Is this a

I'm trying to track down the problem with the assertion failure in SelectionDAGBuilder.cpp. This is the code: *case* *Intrinsic*::gcroot: *if* (GFI) { *const* Value *Alloca = I.getArgOperand(0); *const* Constant *TypeMap = cast<Constant>(I.getArgOperand(1)); * FrameIndexSDNode *FI = cast<FrameIndexSDNode>(getValue(Alloca).getNode());*