On 07/27/2011 12:09 AM, Jon Brisbin wrote:
> I did try that, but it didn't seem to make any difference w.r.t. run
> times.
>
> I've actually decreased the value some as that makes it more scalable.
> I've uploaded a 1GB file so far and that only takes a couple seconds,
> even though memory usage goes through the roof.
>
> Should the readBufferSize match the size I give notifyAvailable()
> (e.g. should I try to keep them in sync, or can they be different)?
I don't think these values should be somehow correlated.
BTW, I see you run perf. tests on localhost, it would also make sense to
try to split client and server to exclude undesirable side-effects and
make test results more reliable.
Thanks.
WBR,
Alexey.
>
> Thanks!
>
> Jon Brisbin
> http//jbrisbin.com
>
>
> ------------------------------------------------------------------------
>
> *From: *"Ryan Lubke" <ryan.lubke_at_oracle.com>
> *To: *users_at_grizzly.java.net
> *Sent: *Tuesday, July 26, 2011 4:20:31 PM
> *Subject: *Re: any way to make this faster/more scalable?
>
> On 7/26/11 11:02 AM, Jon Brisbin wrote:
>
> Can I change the readBufferSize based on Content-Length? This
> is really the question: is there any way (besides eliminating
> the buffer copy) to make this faster or more scalable?
>
> https://github.com/jbrisbin/node-vs-java-shootout/blob/master/src/main/groovy/shootout/UploadHandler.groovy
>
> It does fine now up to >100MB. But trying to upload a 1GB file
> is killing my CPU (GC maybe?). Just wondered if there aren't
> some micro-optimizations I can make to shave a few
> microseconds off here or there.
>
>
> Have you tried decreasing the number of file writes you're
> performing (i.e., increase the value for the number of bytes that
> must be received before invoking the ReadHandler callbacks)?
>
>
>
> Thanks!
>
> Jon Brisbin
> http//jbrisbin.com
>
>
>
>