improve documentation of the 'name' directive and the 'workload' mechanism
Commit Message
Hi,
here is a patch improving the 'name' directive and the 'workload' mechanism
(for "make bench"). Feedback is welcome. I will submit separately later on
some new workload traces for sin, exp, pow, sinf128, expf128, powf128.
Paul Zimmermann
From b8ff91bf0bfb26114d1b4e5d30f210f41a4ff58d Mon Sep 17 00:00:00 2001
From: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Date: Tue, 4 Aug 2020 13:27:39 +0200
Subject: [PATCH] improve documentation of the 'name' directive and the
'workload' mechanism
---
benchtests/README | 20 ++++++++++++++------
1 file changed, 14 insertions(+), 6 deletions(-)
Comments
On 8/4/20 7:31 AM, Paul Zimmermann wrote:
> Hi,
>
> here is a patch improving the 'name' directive and the 'workload' mechanism
> (for "make bench"). Feedback is welcome. I will submit separately later on
> some new workload traces for sin, exp, pow, sinf128, expf128, powf128.
>
> Paul Zimmermann
Paul,
This looks good and I've pushed this for glibc 2.32. Thanks.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
> From b8ff91bf0bfb26114d1b4e5d30f210f41a4ff58d Mon Sep 17 00:00:00 2001
> From: Paul Zimmermann <Paul.Zimmermann@inria.fr>
> Date: Tue, 4 Aug 2020 13:27:39 +0200
> Subject: [PATCH] improve documentation of the 'name' directive and the
> 'workload' mechanism
>
> ---
> benchtests/README | 20 ++++++++++++++------
> 1 file changed, 14 insertions(+), 6 deletions(-)
>
> diff --git a/benchtests/README b/benchtests/README
> index f440f3295a..44736d7e63 100644
> --- a/benchtests/README
> +++ b/benchtests/README
> @@ -125,17 +125,25 @@ math functions perform computations at different levels of precision (64-bit vs
> performance of these functions. One could separate inputs for these domains in
> the same file by using the `name' directive that looks something like this:
>
> - ##name: 240bit
> + ##name: 240bits
>
> -See the pow-inputs file for an example of what such a partitioned input file
> -would look like.
> +All inputs after the ##name: 240bits directive and until the next `name'
> +directive (or the end of file) are part of the "240bits" benchmark and
> +will be output separately in benchtests/bench.out. See the pow-inputs file
> +for an example of what such a partitioned input file would look like.
>
> -It is also possible to measure throughput of a (partial) trace extracted from
> -a real workload. In this case the whole trace is iterated over multiple times
> -rather than repeating every input multiple times. This can be done via:
> +It is also possible to measure latency and reciprocal throughput of a
> +(partial) trace extracted from a real workload. In this case the whole trace
> +is iterated over multiple times rather than repeating every input multiple
> +times. This can be done via:
>
> ##name: workload-<name>
>
> +where <name> is simply used to distinguish between different traces in the
> +same file. To create such a trace, you can simply extract using printf()
> +values uses for a specific application, or generate random values in some
> +interval. See the expf-inputs file for an example of this workload mechanism.
> +
> Benchmark Sets:
> ==============
>
>
@@ -125,17 +125,25 @@ math functions perform computations at different levels of precision (64-bit vs
performance of these functions. One could separate inputs for these domains in
the same file by using the `name' directive that looks something like this:
- ##name: 240bit
+ ##name: 240bits
-See the pow-inputs file for an example of what such a partitioned input file
-would look like.
+All inputs after the ##name: 240bits directive and until the next `name'
+directive (or the end of file) are part of the "240bits" benchmark and
+will be output separately in benchtests/bench.out. See the pow-inputs file
+for an example of what such a partitioned input file would look like.
-It is also possible to measure throughput of a (partial) trace extracted from
-a real workload. In this case the whole trace is iterated over multiple times
-rather than repeating every input multiple times. This can be done via:
+It is also possible to measure latency and reciprocal throughput of a
+(partial) trace extracted from a real workload. In this case the whole trace
+is iterated over multiple times rather than repeating every input multiple
+times. This can be done via:
##name: workload-<name>
+where <name> is simply used to distinguish between different traces in the
+same file. To create such a trace, you can simply extract using printf()
+values uses for a specific application, or generate random values in some
+interval. See the expf-inputs file for an example of this workload mechanism.
+
Benchmark Sets:
==============