#
d9b646e1 |
|
19-Aug-2019 |
Reto Achermann <reto.achermann@inf.ethz.ch> |
x86_64: make all targets compile Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>
|
#
9bd8d1d5 |
|
03-Apr-2019 |
Reto Achermann <reto.achermann@inf.ethz.ch> |
replacing umlaute and fixing address in headers Haldeneggsteig -> Universitaetsstrasse Signed-off-by: Reto Achermann <reto.achermann@inf.ethz.ch>
|
#
b3813360 |
|
27-Nov-2014 |
Reto Achermann <reto.achermann@inf.ethz.ch> |
moved spawn flags into spawndomain.h
|
#
6d834aae |
|
29-Apr-2012 |
pravin shindep@inf.ethz.ch <pravin shindep@inf.ethz.ch> |
Added local loopback device
|
#
28572628 |
|
25-Apr-2012 |
Antoine Kaufmann <antoinek@student.ethz.ch> |
net_latency now uses core info from SKB Relying on core information from SKB is the way to go. Until now we just used cores 0 through 15, but that is only correct for sbrinz1. --- usr/bench/net_latency/Hakefile | 4 +++- usr/bench/net_latency/main.c | 17 ++++++++++++++--- 2 files changed, 17 insertions(+), 4 deletions(-)
|
#
975aa2c1 |
|
25-Apr-2012 |
Antoine Kaufmann <antoinek@student.ethz.ch> |
net_latency benchmark now relies on spawn_wait_core() This allows us to wait for the benchmark runs to terminate, and removes the hack of using sleep. --- usr/bench/net_latency/main.c | 12 +++++------- 1 files changed, 5 insertions(+), 7 deletions(-)
|
#
342c0182 |
|
25-Apr-2012 |
Antoine Kaufmann <antoinek@student.ethz.ch> |
Added support for memory affinity and dry runs in ethernet latency benchmark The dry runs are just normal runs whose result is thrown out. --- lib/net_queue_manager/ethernet_latency_benchmark.c | 28 +++++++++- usr/bench/net_latency/Hakefile | 2 +- usr/bench/net_latency/main.c | 55 +++++++++++++++----- 3 files changed, 68 insertions(+), 17 deletions(-)
|
#
0b084633 |
|
16-Apr-2012 |
Antoine Kaufmann <antoinek@student.ethz.ch> |
net_latency benchmark extended to run more benchmarks --- usr/bench/net_latency/main.c | 51 ++++++++++++++++++++++++++++++++---------- 1 files changed, 39 insertions(+), 12 deletions(-)
|
#
a71eceb5 |
|
16-Apr-2012 |
Antoine Kaufmann <antoinek@student.ethz.ch> |
Extended latency benchmarks to actually do something useful The ethernet latency benchmark now does a statically configured number of runs and then calculates some values on the result. Also net_latency now runs the benchmark on each core for a start. But since spawn_exit seems to be broken somehow on other cores than 0, we just sleep a bit and then spawn the next one. :-/ --- lib/net_queue_manager/Hakefile | 3 +- lib/net_queue_manager/ethernet_latency_benchmark.c | 54 +++++++++++++--- usr/bench/net_latency/Hakefile | 2 +- usr/bench/net_latency/main.c | 28 +++++--- usr/bench/net_latency/sleep.c | 68 ++++++++++++++++++++ usr/bench/net_latency/sleep.h | 16 +++++ usr/drivers/e10k/Hakefile | 2 +- 7 files changed, 150 insertions(+), 23 deletions(-) create mode 100644 usr/bench/net_latency/sleep.c create mode 100644 usr/bench/net_latency/sleep.h
|
#
da457383 |
|
16-Apr-2012 |
Antoine Kaufmann <antoinek@student.ethz.ch> |
net_latency benchmark suite started At the moment its pretty much just proof of concept, to see how well multiple runs of the ethernet latency benchmark work. Looks good for now. --- lib/net_queue_manager/ethernet_latency_benchmark.c | 2 + usr/bench/net_latency/Hakefile | 16 +++++++ usr/bench/net_latency/main.c | 44 ++++++++++++++++++++ 3 files changed, 62 insertions(+), 0 deletions(-) create mode 100644 usr/bench/net_latency/Hakefile create mode 100644 usr/bench/net_latency/main.c
|