Freesound Forums

Not bad the trick to do it on the client side,
but the problem first is that not every one that have an high-speed connection
that's make a problem problem only with big files,
after the computation must be done with and
client language programming Java/Flash,
for short samples that's doesn"t matter,
but with more longer samples the DFT/FFT taking more time,
and again for peoples having CPU <= 800Mhz that be very slow.
Again loading the files from the server increase network packets transfers
so that can slow the server, how much, i don't know? :?

An concept we can do with that is to do that
when the calculation preview picture is done,
The Client Transfer The Preview Pictures(Small and Big) To The Server,
so when you would the Spectral Previews of the sample
and you are the first you load it and do the calculation for
the benefit of everyones and after you return check the preview and
members don"t need to re-compute the Spectral Previews again and again :twisted:

hello_flowers
Why not cut off the new server for a few days to configure all of the images on there, then when the servers synch all the new samples [from the "old server"] are automatically put back into moderation, that way, when they pass moderation they are automaticaly assigned amplitude and spectral images via some script on the new server to say all samples must have both images to represent them.

dobroide
Some form of distributed computing (Seti-like) could do the trick. Anyway, I doubt the coolness of having spectral views justifies the effort to implement this...

Am I right in thinking that the overhead (for Bram) of developing a C application using FFTW to do this would be relatively low given the current spectral-centroid image generating codebase, which is presumably doing most of the same calculations already? It seems to me that while there might be benefits, implementing something client side in java / flash might be substantially more work and possibly annoying for users not to be able to use straight ftp... having said that, I do believe there is a place for more rich stuff on the client side, generally.

I can see that using the same server that is hosting the site in order to do all the processing of the existing database would be a problem, but I don't see the problem if a grid is dedicated to it for a while and the server is used to process new samples... not that I have much real experience of such things.

Problem #1:
Do the DFT or the FFT Spectral take some part of CPU/MPU power.

Problem #2:
How process the 32000 Sounds already here without steeling
all the CPU power...

Problem #3:
This is not everyone that use Spectral Graphic or that know that exist...

But when I propose this, FreeSound Has only one server
that run with an CPU power average of 75%.

But now FreeSound have two server with
individual CPU power average of 30%...

For now Bram seem to have not the time to implement this
and have another problems to solve like make running the new server http://www.clicksmilies.com/s1106/sauer/angry-smiley-053.gif