search for: 73bfa4a

All, If we can speculatively execute a load instruction, why isn’t it safe to hoist it out by promoting it to a scalar in LICM pass? There is a comment in LICM pass that if a load/store is conditional then it is not safe because it would break the LLVM concurrency model (See commit 73bfa4a). It has an IR test for checking this in test/Transforms/LICM/scalar-promote-memmodel.ll However, I have a sample code where GCC is able to promote the memory to scalar and hoist/sink load/store out of loop but LLVM cannot. Is GCC being aggressive here or LLVM missing out an opportunity? Here is...

...an speculatively execute a load instruction, why isn’t it safe to hoist it out by promoting it to a scalar in LICM pass? >> >> >> There is a comment in LICM pass that if a load/store is conditional then it is not safe because it would break the LLVM concurrency model (See commit 73bfa4a). >> It has an IR test for checking this in test/Transforms/LICM/scalar-promote-memmodel.ll >> >> However, I have a sample code where GCC is able to promote the memory to scalar and hoist/sink load/store out of loop but LLVM cannot. >> Is GCC being aggressive here or LLVM m...

...rg> wrote: All, If we can speculatively execute a load instruction, why isn’t it safe to hoist it out by promoting it to a scalar in LICM pass? There is a comment in LICM pass that if a load/store is conditional then it is not safe because it would break the LLVM concurrency model (See commit 73bfa4a). It has an IR test for checking this in test/Transforms/LICM/scalar-promote-memmodel.ll However, I have a sample code where GCC is able to promote the memory to scalar and hoist/sink load/store out of loop but LLVM cannot. Is GCC being aggressive here or LLVM missing out an opportunity? Here is...