Skip to content
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

Performance measurements #1

Closed
mohanr opened this issue May 22, 2015 · 3 comments
Closed

Performance measurements #1

mohanr opened this issue May 22, 2015 · 3 comments

Comments

@mohanr
Copy link

mohanr commented May 22, 2015

I have a few doubts.

Why haven't you quantified the data based on USL models and other such statistical and queueing models ?
Don't Error bars denote the error in measurements ?
Why do you show 101% of some measurements ?

@mogill
Copy link
Owner

mogill commented May 23, 2015

The diagrams indicate the minimum and maximum time measured for a single operation, and the arithmetic mean time over all operations. The 101% referred to in the header of the example is to offset used to the top of the block illustrating the mean time 1% higher from the bottom of the block. If the block is 0 height it is not displayed, the offset guarantees a non-zero block height.

@mohanr
Copy link
Author

mohanr commented May 23, 2015

I didn't understand the part related to 101%. The graph image link is
broken. Couldn't you use nanoseconds to get a better idea ?
I meant that standard error is used to draw error bars. Right ?

Which set of measurements are for cluster ? Which are for asynch ?

Thanks,
Mohan

On Sat, May 23, 2015 at 9:06 AM, Jace Mogill notifications@github.com
wrote:

The diagrams indicate the minimum and maximum time measured for a single
operation, and the arithmetic mean time over all operations. The 101%
referred to in the header of the example is to offset used to the top of
the block illustrating the mean time 1% higher from the bottom of the
block. If the block is 0 height it is not displayed, the offset guarantees
a non-zero block height.


Reply to this email directly or view it on GitHub
#1 (comment)
.

@mogill
Copy link
Owner

mogill commented May 24, 2015

I fixed the image link so maybe things will become more clear. I used a Candlestick Chart which requires two ranges, however, my experiments have a min-max range and the other value is a scalar value. In order for the chart to display the scalar value I must turn it into a range, thus the block ranges from 100-101% of average, solely to make it drawable.

If you take that CSV file and load it into a spreadsheet like Excel or Google Sheets you can change the 101% value to be the same as the 100% value and you'll see the solid block (indicating the average time) disappears from the chart. The left average column is 100% of average, that is, it is the computed average. The column to the right of it is 101% of the actual average.

@mogill mogill closed this as completed Jun 13, 2015
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants