-
Notifications
You must be signed in to change notification settings - Fork 10
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Updated examples directory #71
Conversation
60fefbe
to
1c3b1d9
Compare
Parts of krun need to be compiled. There is a |
You will may need to set a |
1c3b1d9
to
82383f2
Compare
OK, this is now fixed, documented, and I have tested it on my laptop. |
This looks pretty good Sarah. Especially the much needed documentation. I'm going to add some comments. They will be pretty nit-picky -- feel free to accept or reject them as you see fit. |
executed on two VMs (*cPython* and a standard *JVM* such as HotSpot). | ||
Each benchmark is run for 5 iterations on the same VM, then the VM is | ||
restarted and the benchmark is re-run for another 5 iterations. | ||
We say that the benchmarks are *executed* twice, with 5 *iterations*. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe use a different number of iterations and executions, so as to help distinguish them?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
? 5 iterations, 2 executions
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah. Fine.
OK, done. |
There are definite tabs & spaces issues in some of this stuff. We should use 4 spaces unless there's a compelling reason not to. |
Those are likely present in the original benchmarks. I leave it to you guys to decide whether to fix them. Diff distance vs. readability. |
Readability is generally more important (there are lines next to each other with mixed tabs/spaces, which means it looks pretty bad). |
These latest changes need a good squashing, but hopefully they address the issues in the comments. The |
I really think the |
|
||
* Edit /etc/default/grub (e.g. `sudo gedit /etc/default/grub`) | ||
* Add `isolcpus=X` to `GRUB_CMDLINE_LINUX_DEFAULT` (where `X` is an integer > 0) | ||
* Run `sudo update-grub` |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Currently we also require intel pstate to be disabled.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
These are all linux specific details of course.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Well, I've stated "Debian-based systems"; I don't know how anything other than grub works!
Looks good Sarah! Anything to add @ltratt? |
@vext01 you are happy to have this documentation in the example/ directory? |
This configuration can be found in the file `examples/example.krun`. | ||
|
||
## Step 0: prepare the benchmarking machine | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Step 0? Step 1?!
Anyway, I think we should say something like "on Linux, you can optionally set up some kernel arguments to obtain more reliable results". On non-Linuxes, either there are no equivalent options, or there are different ones. We don't want to make it seem it's Linux only though.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK, but the options aren't optional on Linux, krun bails if the environment isn't set up like this :)
@snim2 You could even move it up to the main |
Ready to squash? |
Perhaps my only comment is that we might want to make it a little clearer that some stuff is Linux specific and some isn't. [We'll soon have this running on OpenBSD.] I don't think we have to tell the non-Linux people what to do, as such, but we should make it clear that they're not excluded. |
@ltratt OK. When we have BSD support I'll check that these examples work on the BSD machine, and raise a new PR to make any necessary changes and move the documentation across to the main README. |
… sleeps for one second and exits.
… up. Notes are Linux-specific. Includes current list of supported VMs and platforms.
I'm happy with that approach. |
7c26258
to
cdd16dd
Compare
OK, squashed |
This is an updated version of the old
examples/
directory, which is a cut down version of the warmup benchmarks. The aim is to use this to test the work to resolve Issue #41 andreboot mode
.This isn't quite ready to merge, firstly it needs an
examples/README.md
. Secondly, running this onbencher5
results in:which looks like the
CLASSPATH
has been incorrectly set?