search for: inlto

...rt). In this situation, PGO data is useless for all `static` >>> functions (and as a byproduct results in a huge volume of warnings). >>> >> >> >> Rong, I was just looking at implementing a fix for this, but noticed >> something. Can we get rid of the "InLTO" argument to getPGOFuncName if we >> unconditionally apply the funcname metadata to all functions? >> > > Are you proposing always emitting the meta data for internal functions? > Yes; actually for all functions. If I understand correctly, the InLTO argument fixes a speci...

...>> thing to support). In this situation, PGO data is useless for all `static` >> functions (and as a byproduct results in a huge volume of warnings). >> > > > Rong, I was just looking at implementing a fix for this, but noticed > something. Can we get rid of the "InLTO" argument to getPGOFuncName if we > unconditionally apply the funcname metadata to all functions? > Are you proposing always emitting the meta data for internal functions? David > > -- Sean Silva > > >> >> 2. In different TU's, pre-instr inlining might ma...

...one or two dashes. For compatibility, lld has to support both. For newer and lld specific options, we don't have such compatibility problem. I'd suggest we reject one-dash long options to avoid collision with short options. For example, * -lto-emit-obj can be read as -l to-emit-obj * -thinlto-cache-dir= means -t -h inlto-cache-dir= in GNU ld I created a patch https://reviews.llvm.org/D79371 to disallow some one-dash long options. These options are carefully chosen: I can't find anything depending on their one-dash forms. lld has a spell corrector for option names so a misspelled -...

...g build (which is a reasonable > thing to support). In this situation, PGO data is useless for all `static` > functions (and as a byproduct results in a huge volume of warnings). > Rong, I was just looking at implementing a fix for this, but noticed something. Can we get rid of the "InLTO" argument to getPGOFuncName if we unconditionally apply the funcname metadata to all functions? -- Sean Silva > > 2. In different TU's, pre-instr inlining might make different inlining > decisions (for example, different functions may be available for inlining), > causing ha...

Jake and I have been integrating IRPGO on PS4, and we've identified 3 remaining work items. - Driver changes We'd like to make IRPGO the default on PS4. We also think that it would be beneficial to make IRPGO the default PGO on all platforms (coverage would continue to use FE instr as it does currently, of course). In previous conversations (e.g. http://reviews.llvm.org/D15829) it has