jmeter-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From James Bull <>
Subject Constant throughput timer question
Date Fri, 28 Jul 2006 11:14:50 GMT
Hi all,

I have noticed while using the constant throughput timer that it calculates
its throughput only after the first set of threads has completed.

ie. if you set the throughput to 100 per minute but have 200 threads all
200 threads will run for an iteration and then wait for sufficient time to
achieve the throughput you requested and then run it again.

This is contrary to what I want as I want the load spread out more evenly.
I can achieve this by having a look at the response time and calculating a
more appropriate number of users but then I might as well set the ramp up
time appropriately and ignore the constant throughput timer altogether.

It would be nice if instead of calculating the throughput after the whole
lot of threads had run once it did it after each request.

I tried changing the 'for all active threads in current thread group'
option but this just changes the way the throughput is calculated not
_when_ it is calculated.

I imagine there would probably be a performance penalty associated with
performing the calculation after every request though.

So was it decided to implement this in this fashion for performance
reasons? Would other people find it useful if there was a different option
for when to calculate it?

I don't _need_ it but it does niggle and I think calculating it per request
is more intuitive behavior. I was just wondering what other peoples
thoughts on this were.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message