Re: [PATCH] mid-end: Use integral time intervals in timevar.cc

2023-08-04 Thread Richard Biener via Gcc-patches
On Fri, 4 Aug 2023, Matthew Malcomson wrote: > Hopefully last update ... > > > Specifically, please try compiling with > >-ftime-report -fdiagnostics-format=sarif-file > > and have a look at the generated .sarif file, e.g. via > >python -m json.tool foo.c.sarif > > which will pretty-print

[PATCH] mid-end: Use integral time intervals in timevar.cc

2023-08-04 Thread Matthew Malcomson via Gcc-patches
Hopefully last update ... > Specifically, please try compiling with >-ftime-report -fdiagnostics-format=sarif-file > and have a look at the generated .sarif file, e.g. via >python -m json.tool foo.c.sarif > which will pretty-print the JSON to stdout. Rebasing onto the JSON output was quit

Re: [PATCH] mid-end: Use integral time intervals in timevar.cc

2023-08-03 Thread Richard Biener via Gcc-patches
On Thu, 3 Aug 2023, Matthew Malcomson wrote: > > > > I think this is undesriable. With fused you mean we use FMA? > > I think you could use -ffp-contract=off for the TU instead. > > > > Note you can't use __attribute__((noinline)) literally since the > > host compiler might not support this. >

Re: [PATCH] mid-end: Use integral time intervals in timevar.cc

2023-08-03 Thread David Malcolm via Gcc-patches
On Thu, 2023-08-03 at 15:54 +0100, Matthew Malcomson wrote: > On 8/3/23 15:09, David Malcolm wrote: > > > > Hi Matthew.  I recently touched the timevar code (in r14-2881- > > g75d623946d4b6e) to add support for serializing the timevar data in > > JSON form as part of the SARIF output (PR analyzer/

Re: [PATCH] mid-end: Use integral time intervals in timevar.cc

2023-08-03 Thread Matthew Malcomson via Gcc-patches
On 8/3/23 15:09, David Malcolm wrote: Hi Matthew. I recently touched the timevar code (in r14-2881- g75d623946d4b6e) to add support for serializing the timevar data in JSON form as part of the SARIF output (PR analyzer/109361). Looking at your patch, it looks like the baseline for the patch se

Re: [PATCH] mid-end: Use integral time intervals in timevar.cc

2023-08-03 Thread David Malcolm via Gcc-patches
On Thu, 2023-08-03 at 14:38 +0100, Matthew Malcomson via Gcc-patches wrote: > > > > I think this is undesriable.  With fused you mean we use FMA? > > I think you could use -ffp-contract=off for the TU instead. > > > > Note you can't use __attribute__((noinline)) literally since the > > host compi

[PATCH] mid-end: Use integral time intervals in timevar.cc

2023-08-03 Thread Matthew Malcomson via Gcc-patches
> > I think this is undesriable. With fused you mean we use FMA? > I think you could use -ffp-contract=off for the TU instead. > > Note you can't use __attribute__((noinline)) literally since the > host compiler might not support this. > > Richard. > Trying to make the timevar store integral