>> Which measurement results would you accept for further discussion?
>
> If you don't know how to show your results, then this whole discussion
> is pointless.
I could show test results. But it could happen that I would choose insufficient
variations for execution environments.
I am unsure
On Sun, Jan 03, 2016 at 09:10:45PM +0100, SF Markus Elfring wrote:
> >> Can such fine-tuning result in positive effects for the run-time behaviour?
> >
> > If you can not benchmark and show the proof, don't even start to claim
> > such a thing.
>
> Which measurement results would you accept for
>> Can such fine-tuning result in positive effects for the run-time behaviour?
>
> If you can not benchmark and show the proof, don't even start to claim
> such a thing.
Which measurement results would you accept for further discussion?
>> My suggestions can result in measurable differences.
>
On Sun, Jan 03, 2016 at 07:50:18PM +0100, SF Markus Elfring wrote:
> >> I am a bit surprised that you do not like such source code fine-tuning.
> >
> > It's moving stuff around for no real reason, why would I like it?
>
> Can such fine-tuning result in positive effects for the run-time
>> I am a bit surprised that you do not like such source code fine-tuning.
>
> It's moving stuff around for no real reason, why would I like it?
Can such fine-tuning result in positive effects for the run-time behaviour?
> Reading and reviewing and applying this type of stuff takes away from
>
On Sun, Jan 03, 2016 at 07:16:49PM +0100, SF Markus Elfring wrote:
> >> I prefer that assignments for variables like "card" and "slic_regs"
> >> will only be performed immediately before the corresponding content will be
> >> read again (after a few condition checks were executed).
> >>
> >>
>> I prefer that assignments for variables like "card" and "slic_regs"
>> will only be performed immediately before the corresponding content will be
>> read again (after a few condition checks were executed).
>>
>> Another description could be this view:
>> I suggest to move the variable
On Sun, Jan 03, 2016 at 06:48:17PM +0100, SF Markus Elfring wrote:
> >> Replace explicit initialisation for two local variables at the beginning
> >> by assignments.
> >
> > Why?
>
> I prefer that assignments for variables like "card" and "slic_regs"
> will only be performed immediately before
>> Replace explicit initialisation for two local variables at the beginning
>> by assignments.
>
> Why?
I prefer that assignments for variables like "card" and "slic_regs"
will only be performed immediately before the corresponding content will be
read again (after a few condition checks were
On Sun, 3 Jan 2016, SF Markus Elfring wrote:
> From: Markus Elfring
> Date: Sun, 3 Jan 2016 17:25:59 +0100
>
> Replace explicit initialisation for two local variables at the beginning
> by assignments.
Why?
julia
> Signed-off-by: Markus Elfring
> ---
> drivers/staging/slicoss/slicoss.c | 6
From: Markus Elfring
Date: Sun, 3 Jan 2016 17:25:59 +0100
Replace explicit initialisation for two local variables at the beginning
by assignments.
Signed-off-by: Markus Elfring
---
drivers/staging/slicoss/slicoss.c | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git
>> Which measurement results would you accept for further discussion?
>
> If you don't know how to show your results, then this whole discussion
> is pointless.
I could show test results. But it could happen that I would choose insufficient
variations for execution environments.
I am unsure
From: Markus Elfring
Date: Sun, 3 Jan 2016 17:25:59 +0100
Replace explicit initialisation for two local variables at the beginning
by assignments.
Signed-off-by: Markus Elfring
---
drivers/staging/slicoss/slicoss.c | 6 --
1
>> I prefer that assignments for variables like "card" and "slic_regs"
>> will only be performed immediately before the corresponding content will be
>> read again (after a few condition checks were executed).
>>
>> Another description could be this view:
>> I suggest to move the variable
On Sun, Jan 03, 2016 at 07:16:49PM +0100, SF Markus Elfring wrote:
> >> I prefer that assignments for variables like "card" and "slic_regs"
> >> will only be performed immediately before the corresponding content will be
> >> read again (after a few condition checks were executed).
> >>
> >>
>> I am a bit surprised that you do not like such source code fine-tuning.
>
> It's moving stuff around for no real reason, why would I like it?
Can such fine-tuning result in positive effects for the run-time behaviour?
> Reading and reviewing and applying this type of stuff takes away from
>
>> Replace explicit initialisation for two local variables at the beginning
>> by assignments.
>
> Why?
I prefer that assignments for variables like "card" and "slic_regs"
will only be performed immediately before the corresponding content will be
read again (after a few condition checks were
On Sun, Jan 03, 2016 at 06:48:17PM +0100, SF Markus Elfring wrote:
> >> Replace explicit initialisation for two local variables at the beginning
> >> by assignments.
> >
> > Why?
>
> I prefer that assignments for variables like "card" and "slic_regs"
> will only be performed immediately before
On Sun, 3 Jan 2016, SF Markus Elfring wrote:
> From: Markus Elfring
> Date: Sun, 3 Jan 2016 17:25:59 +0100
>
> Replace explicit initialisation for two local variables at the beginning
> by assignments.
Why?
julia
> Signed-off-by: Markus Elfring
On Sun, Jan 03, 2016 at 07:50:18PM +0100, SF Markus Elfring wrote:
> >> I am a bit surprised that you do not like such source code fine-tuning.
> >
> > It's moving stuff around for no real reason, why would I like it?
>
> Can such fine-tuning result in positive effects for the run-time
>> Can such fine-tuning result in positive effects for the run-time behaviour?
>
> If you can not benchmark and show the proof, don't even start to claim
> such a thing.
Which measurement results would you accept for further discussion?
>> My suggestions can result in measurable differences.
>
On Sun, Jan 03, 2016 at 09:10:45PM +0100, SF Markus Elfring wrote:
> >> Can such fine-tuning result in positive effects for the run-time behaviour?
> >
> > If you can not benchmark and show the proof, don't even start to claim
> > such a thing.
>
> Which measurement results would you accept for
22 matches
Mail list logo