Other benefits to ARM vs. Atmega besides lower latency?

[anttipi]

Happy New Year, fellow Megadrumsters!

I have a five year old Atmega 20 MHz based Megadrum and was wondering if there are any other tangible benefits to upgrading besides the ~2 ms lower latency (was thinking perhaps improved triggering performance, ability to use lower MinScan & Latency settings etc.)?

Mind you, latency is currently not a real issue for me even when using 20-30 MinScan and 40 Latency because I have an RME AIO which can be comfortably set at 32 samples/44.1 KHz.

One reason I might jump for the upgrade is to get positional sensing for my Drum-Tec Diabolo mesh pads but I'm not sure if that alone is enough of a reason right now.

[airflamesred]

Latency. Whether you have rme or focusrite 2ms is a long time in drumming terms.
happy new year

[anttipi]

Please correct me if I'm wrong on either account:

1. MinScan is basically dependent upon the pad type, not procesing horsepower (because the pad's first half wave takes a certain time to complete anyway) so upgrading to ARM most probably wouldn't allow me to utilize lower MinScan settings
2. The Latency setting however must be (at least to some degree?) dependent on processing power. I haven't tested if Atmega can handle the lowest Latency setting (10) but probably not as by default it's set as high as 40. I don't have any idea whether ARM could be set at a lower Latency setting vs. the Atmega or not - is there any difference here?

But yeah, the 2 ms gained by the better USB implementation may be a lot in drumming terms but still the upgrade would set me back me roughly 100 euros per just one millisecond gained (prebuilt - I won't be building anything myself). So that's why I want to consider my options carefully. Not at all saying Megadrum isn't fairly priced but it's a bit much just for an upgrade which I may not see that much practical benefit from.

[airflamesred]

Well, all set ups will differ, I have miniscan and latency set to 11. I did notice a considerable improvement on upgrading from Atmiga.

[ignotus]

The biggest improvement I saw actually was with crosstalk. I don't know if the new chips have different or improved code for that but it massively improved. I used to have crosstalk issues and had to set fairly high values for some pads, while now I can leave them at zero. Other than that I find performance to be quite similar, though I reckon the newer boards have potential for significant future improvements, while the atmega boards will likely hit their limit at some point.

[anttipi]

Thanks for the replies! I've personally got no crosstalk issues with the Atmega, probably thanks to most of my pads being attached on sturdy and isolated drum stands.

I will try and experiment to see how low I can go with the Latency/MinScan settings as I've never really tweaked them much. Just a couple more questions regarding that:

1. What's the best way to go on setting Latency and MinScan properly? Start with both on a high setting and lowering gradually while banging a pad repeatedly with consistent force, looking for when weird velocity spikes or drops start to happen?
2. ...or would you rather try lowering either Latency/MinScan first and foremost - or does it matter? As far as I can understand, you could potentially go lower with Latency than MinScan on an ARM processor, because a certain minimum scan time is needed anyway, which depends on the properties of the pad (i.e. mesh vs. rubber, pad size etc.) But Latency I would think depends more on processing speed (and/or bandwidth).

I'm tempted with the upgrade but man, funds are tight right now after the Christmas expenses...

[airflamesred]

I'm on rubber and found a low miniscan better. As for latency, I'd start low and move up if needed. I don't really know what the signs would be.
Looks like you'll have to save for the ARM, at least you can get by with what you have.

[anttipi]

OK, I'll start low on both MinScan and Latency and raise if needed because I'll naturally want them as low as possible. So far for rubber pads I don't think I've seen any difference when going higher than 20 MinScan (haven't tried any lower because the latency has seemed ok).

Like I said, I'm not having any particular problems with Atmega at the moment so I think I'll pass the upgrade for now. Triggering is fine, thresholds can be set as low as needed, no crosstalk issues etc. I wasn't even bothered about latency until I read about the ARM version being 2 ms snappier. And whenever there's performance gains available, however small, I'm naturally tempted!

I think I'll be measuring the roundtrip latency of my setup next, just to appease my obsessive compulsive mind...