summaryrefslogtreecommitdiffstats
path: root/TODO
blob: 2c8017d14ec0ef0eaca4eabde7dfda9cfb799f5a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
#
# This is just a random list of things that I think need to be done
# to rteval
#

1. Figure out how to add more loads. Seems that we could do it two ways:

   - create a wrapper python module just like is done for kcompile and hackbench
     then add the new module to a config file so rteval loads it on startup
     (i.e. a plugin architecture)

   - figure out some generic load wrapper so that arbitrary commands may be
     used as loads. 

2. Should we add a hwlatdetect option?

3. Potential new loads/measurements:
   - dbench
   - AMQP latencytest/perftest (over loopback)
   - bonnie/bonnie++
   - iozone
   - KVM running guest OS

4. Add an option to run 'perf record' while  a run is in progress. This is
   mainly a diagnostic option to debug *why* something is performing 
   poorly.