Booniers drum machine got me thinking about drums again so tonight I made this.

It's 3 techno drums that are quite efficient. The kick and snare contain no oscillators but use a wrapped line segment. Total DSP count - 4 multiplies and a cos. This trick is extended in the snare with a feedback around a second function which results in a chaotic band of noise around the center frequency. Becuase we use [vline~] and always start on the same phase they sound pretty tight.

best,

andy

http://www.pdpatchrepo.info/hurleur/efficient-drum.zip

Boy, do you ever teach obiwan. I will learn much from your pwness of BigO drum synth.

By eck they sound Guuurgeous. They sound like CR78 or Rhythm Ace

Hey hey is that what one calls a Bridged-T Network? I was wondering how to do this is Pd.

ps. I've added a few funky features to the drum machine like optional sample or synthesis for each channel. Its quite nice to drop in loops and envelope them to chopp them up. Also theres a fill for each step to make funky little flam/glitch effects...When its ready i'll post it up and you can rip it apart and build it far more efficiently

Hi folks
two things, obi how do i trigger the drum sounds ? and Boonier have you posted your new funky featured drum machine if so where is it ?
Many thanks Alan

no quite yet - needs a few more hours
sorry

@boonier said:

no quite yet - needs a few more hours
sorry

Hey boonier, i guess those hours are already over Any results you'd like to shared with the world?

If we wanted to take it even further, we could cache the drum sounds in a table, update the table when the params change, and simply trigger the sample in the table. But I must hand it to you obi, those are some pretty tight drums.

bz, that's an awesome idea.

@domien said:

@boonier said:

no quite yet - needs a few more hours
sorry

Hey boonier, i guess those hours are already over Any results you'd like to shared with the world?

Do you know, I had forgotten about this! I've been working on other stuff

I suppose I could upload the current version, but theres no form of presetting for either the drum voicing or the patterns, and I need to look at the table writing mechanism, but that would go hand in hand with the presetting as I see it.

But does work nicely....would you like to take a look?

Dear Obi, share this patch wasn't a good idea. Actually I'm wondering if is really necessary go to sleep 'cause I've a lot of ideas for my next performances. Good stuff guy! Thanks a lot!

@boonier said:

But does work nicely....would you like to take a look?

Mmm... What did you guess That is, if you don't mind of course...

here yugo -

improve, add bits, post it back for everyone - thats all i ask

have fun, S

http://www.pdpatchrepo.info/hurleur/seq16_drum.zip

are all the bits there? Couldn't find seq16_drum :/

oops! sorry

start with drumtest.pd

http://www.pdpatchrepo.info/hurleur/seq16_drum.zip

@hardoff said:

bz, that's an awesome idea.

Or be super elite and use nothing more than an [expr] and counter to fill the table... although that might require more of a refresh time (delay?) than simply sampling it would.

By the way! WHY does Obi's patch here require 20% cpu load on my 2.4ghz dual core macbook pro, but only 8% on my 1.5ghz G4???? Anyone else able to see this?

That's interesting Bz - the machine it was made on is a 500MHz box and it runs at about 30% on that. Sounds like the code is being interpreted into some ugly instructions on the macbook. Can you isolate the part that is being hungry - it's an interesting exercise because the patch contains only [+~], [*~], [cos~] and [phasor~] IIRC - (apart from the hihat which uses [osc~] which is a memory lookup operation - maybe that's it). Do other Pd patches give similar performance ratios on the two machines? Or just this one?

@obiwannabe said:

That's interesting Bz - the machine it was made on is a 500MHz box and it runs at about 30% on that. Sounds like the code is being interpreted into some ugly instructions on the macbook. Can you isolate the part that is being hungry - it's an interesting exercise because the patch contains only [+~], [*~], [cos~] and [phasor~] IIRC - (apart from the hihat which uses [osc~] which is a memory lookup operation - maybe that's it). Do other Pd patches give similar performance ratios on the two machines? Or just this one?

I see this behavior with every Pd file. I'll do a more in depth look at this file when I get home from work tonight. From what I can see right now, Pd-extended is eating up 17% cpu load just being open with compute audio turned off. This could imply that your patch only takes 3% usage, which sounds more fair. I'll post a new topic about it and link it here. Check out the attachment for a screenshot of the activity monitor during while Pd-extended is open w/o any patches. Hans, where you at? You should know of this, I got pd-extended from your autobuilds, but you don't give any contact info to report bugs to

http://www.pdpatchrepo.info/hurleur/PdHungry.png

EDIT: Sorry I just opened the attached png! :/

What does your Activity monitor say?

Yesterday I discovered that an item called PrintJobMgr was chomping 90% of the cpu! I just killed it and its like a new machine. Its the reason i was getting such cruddy performance for midi activity over IAC bus a couple of weeks ago.

This is quite common judging by the dicussions on the interweb. Its like a print job that never makes it to the printer, but is so determined to print that it will use most of the computers power to do so. I havent even connected to print for over a year, so you can imagine, I just thought my computer was really slow (well it is compared to todays standards)

**OSX users** for best performance in Pd, download Jack OS X, run Pd in Jack mode and you'll be able to get buffer sizes down to 64/128. I run a 1ghz g4 currently and its not even popping or stuttering.

boonier I think you might be on to something w/ the whole Jack thing. I've also noticed that when I use my Firewire MOTU 828MKII, my cpu usage is even higher. I'll try Jack, and report back.

(reporting back)
boonier you're my hero. I just switched on Jack, and Pd-extended is sitting at 0-1percent cpu w/ no patches, and 2percent while running Obi's patch

Hey obiwannabe, I'm resurrecting this old thread because I just found it, and this patch is teaching me a lot. But I have several questions about the snare:

a) was it from trial and error, or is there some theory available that explains the principles involved? Because I know a bit about ring modulation, a bit about cosine waveshapers, and a bit about feedback loops, but I would never have come up with the RM-cos-feedback structure you have here.
What's the [del 5] for, and why is it connected to a [90( and [70( that both go into the same inlet?