On 2/28/20 12:34 AM, Gerald Pfeifer wrote:
Okay? Or does this qualify as obvious?
The patch seems to me obvious. Please install it.
Martin
On Thu, 27 Feb 2020, Gerald Pfeifer wrote:
>> This (or rather its predecessor?) breaks bootstrap on 32-bit
>> i386-unknown-freebsd11.3.
>>
>> /scratch/tmp/gerald/gcc10-devel-work/gcc-10-20200223/gcc/value-prof.c: In
>> function 'void dump_histogram_value(FILE*, histogram_value)':
>> /scratch/tmp
On Mon, 24 Feb 2020, Gerald Pfeifer wrote:
> This (or rather its predecessor?) breaks bootstrap on 32-bit
> i386-unknown-freebsd11.3.
>
> /scratch/tmp/gerald/gcc10-devel-work/gcc-10-20200223/gcc/value-prof.c: In
> function 'void dump_histogram_value(FILE*, histogram_value)':
> /scratch/tmp/geral
On Mon, 17 Feb 2020, Martin Liška wrote:
> Patch can bootstrap on x86_64-linux-gnu and survives regression tests.
This (or rather its predecessor?) breaks bootstrap on 32-bit
i386-unknown-freebsd11.3.
/scratch/tmp/gerald/gcc10-devel-work/gcc-10-20200223/gcc/value-prof.c: In
function 'void dump_
On 1/30/20 5:13 PM, Jan Hubicka wrote:
@@ -330,7 +334,7 @@
stream_out_histogram_value (struct output_block *ob, histogram_value hist)
/* When user uses an unsigned type with a big value, constant converted
to gcov_type (a signed type) can be negative. */
gcov_type value = hist->hvalue.counter
On 2/13/20 2:35 PM, Jan Hubicka wrote:
Otherwise the patch is OK and thanks for working on this!
Honza
Hello.
There's updated version of the patch that comes with the
new option name.
If there's no comment I'm going to install it tomorrow.
Patch can bootstrap on x86_64-linux-gnu and survives
> On 1/30/20 5:13 PM, Jan Hubicka wrote:
> > Martin, I welcome your opinion on the patch
>
> Ok, recently we've made quite some experiments with Honza about
> TOP N counters. Based on the numbers, it shows that tracking a negative
> value per-counter value does not help much. So that I suggest to
On 1/30/20 5:13 PM, Jan Hubicka wrote:
Martin, I welcome your opinion on the patch
Ok, recently we've made quite some experiments with Honza about
TOP N counters. Based on the numbers, it shows that tracking a negative
value per-counter value does not help much. So that I suggest to use
only on
Hi,
this patch implements more careful merging of topn values (tracking regression
we got by the reproducible profiling work). Instead of giving up on the
counter on the overflow we use sign bits to track
1) if there are some vlaues lost
2) if given target was not having probability at least 1/T
