On 23/07/2025 17:29, Alexander Monakov wrote:
On Wed, 23 Jul 2025, Jan Hubicka via Gcc wrote:
Thank you for the careful explanation. It seems like all the "innermost" blocks
would need to be instrumented, and in most cases the rest can be assumed. Unfortunately,
that probably means instrument
On 23/07/2025 17:05, Jan Hubicka wrote:
Am 23.07.2025 um 16:42 schrieb Andrew Stubbs :
On 23/07/2025 15:22, Michael Matz wrote:
Hello,
On Tue, 22 Jul 2025, Andrew Stubbs wrote:
Don’t you need to instrument more (or at least different) edges when
only having visited bits ? With counters yo
On Wed, 23 Jul 2025, Jan Hubicka via Gcc wrote:
> > > Thank you for the careful explanation. It seems like all the "innermost"
> > > blocks would need to be instrumented, and in most cases the rest can be
> > > assumed. Unfortunately, that probably means instrumenting the hottest
> > > blocks,
>
>
> > Am 23.07.2025 um 16:42 schrieb Andrew Stubbs :
> >
> > On 23/07/2025 15:22, Michael Matz wrote:
> >> Hello,
> >> On Tue, 22 Jul 2025, Andrew Stubbs wrote:
> Don’t you need to instrument more (or at least different) edges when
> only having visited bits ? With counters you can
> Am 23.07.2025 um 16:42 schrieb Andrew Stubbs :
>
> On 23/07/2025 15:22, Michael Matz wrote:
>> Hello,
>> On Tue, 22 Jul 2025, Andrew Stubbs wrote:
Don’t you need to instrument more (or at least different) edges when
only having visited bits ? With counters you can derive coverage
On 23/07/2025 15:22, Michael Matz wrote:
Hello,
On Tue, 22 Jul 2025, Andrew Stubbs wrote:
Don’t you need to instrument more (or at least different) edges when
only having visited bits ? With counters you can derive coverage of
related edges by diffing counters (consider a simple diamond).
I
Hello,
On Tue, 22 Jul 2025, Andrew Stubbs wrote:
> > Don’t you need to instrument more (or at least different) edges when
> > only having visited bits ? With counters you can derive coverage of
> > related edges by diffing counters (consider a simple diamond).
>
> I have not yet got this deep
On 22/07/2025 16:45, Richard Biener wrote:
Am 22.07.2025 um 16:56 schrieb Andrew Stubbs :
Hi all,
Question: Would it be acceptable to introduce a new "counter" variety, together with
options and UI support, that simply records a "visited" state?
Have there been any previous efforts in thi
> Am 22.07.2025 um 16:56 schrieb Andrew Stubbs :
>
> Hi all,
>
> Question: Would it be acceptable to introduce a new "counter" variety,
> together with options and UI support, that simply records a "visited" state?
>
> Have there been any previous efforts in this space that I've missed?
>
Hi all,
Question: Would it be acceptable to introduce a new "counter" variety,
together with options and UI support, that simply records a "visited" state?
Have there been any previous efforts in this space that I've missed?
I'm currently looking at the feasibility of reducing the memory
fo
10 matches
Mail list logo
