Timestamp support?

[Peterjk]

Just upgraded to Power Mac 2x1.2 Ghz, OS X and PT 6.0 (PT Mix3).

Need a Midi interface. Should I go for Digidesign's? or should it be Emagic AMT8.

Does both support midi timestamping?

Peter

[shaggy]

...along these same lines.

1) Motu released OS_X drivers for their boxes.
Is this really an Apple compatibility issue? Since the core MIDI engine is within the kernal? Question then is will these also support timestamping?

2) Then the 001 midi interface should also support timestamping?

I mean, if the timestamping occurs in the kernal, sans delay, and placed accurately within the Pro Tools session, does this really require any new hardware? i.e. timestamping must pass the other way as well, out of PTs to the kernal---for the kernal to then send it out in sync with the Session audio?

Anybody follow me here? So it may not effect the interfaces at all. Just no MIDI getting qued up in a cache somewhere and being spit out late, as it seems to be now?

The timestamping can be stripped away in the kernal on it's way out...to the playback device. So it's just plain old MIDI when it hits the sampler?

Is that what's happening? What about Roland's GS MIDI?

[accession]

This one still has me a little stumped.

Doesn't the 'timestamping' buffer have to be in the MIDI interface itself? That is, to eliminate the realtime management issues of the USB, or the like?

That being the case, did Digi, Emagic (which is now Apple, so I may have a fighting chance), and Steinberg, implement the same compatible buffering electronics that will allow native OSX MIDI time stamping?

Seems too good to be true.

I have the Emagic AMT8 (purchased before the Digi version, and providing me 'time stamping' should I have 'jumped' to Logic - thank goodness for you guys, Digi, you released the 32 track beta in the nick of time).

Still, I would love to continue using my AMT8 with Pro Tools into OSX with the full advantages of time stamping.

Digi, is there any input you can offer on the subject?

Phil

PS. It struck me the other day my MIDI timing was sounding good. Bumped the Pro Tools CPU usage up to 85% and the poor timing returned.

I might do a few more tests in the next couple of days to see if assigning more CPU resources to the OS rather than Pro Tools provides signifcant timing improvemnet.

That makes sense though, if we're led to believe it's the OS that doesn't prioritise MIDI too well when under strain.

It might also explain why some users experience no issues whatsoever on identical MHz Macs.

However, before I get too carried away, let me test this theory.

Phil

[shaggy]

What would the clock reference to? Itself? I guess each midi track would be relative, but still, you'd have to nudge it forwards are backwards depending on the delay of the traffic to PTs.

You're probably right and the stamp gets set at the interface. Kind've like a router. And the kernal supports this larger packet and routes it to the application. Then the interface strips it off on the way out, and it's plain old midi again.

Quicktime 6 supports Roland's MIDI GS system. Does PTs 6?

Since there's timestamping I guess it would be to much to ask if MTC sync is tightened up.

rant: Sure be nice if 001 interface timestamped with a firmware update. What is that extra serial interface for? A new MIDI interface? Please say yes. I'm stuck with these Opcode jobs, that have worked great so far. Even the USB version.

Wishful thinking...

Back to subject

AMT8 or Digi interface for 6? Can the Digi HD interface even be used with Mix (or Le)?

[accession]

Ted,

Thank you immensely for your insight into MIDI and OSX.

I gather, therefore, with the OSX drivers for my AMT8 interface, 'time stamping' should be active with Pro Tools? Well, maybe.

I now wonder whether Emagic, in the conception of the AMT8, knew what OSX's CoreMIDI would need to know in order to send MIDI ahead of time? While the USB drivers may be OSX updated, the firmware within the unit can't.

Is CoreMIDI designed to recognise the various manufacturer's own 'time stamping' specs? Aren't they all different (based on the thinking each was sequencer-dependent in OS9)?

Hmmm...

I guess the answer lies in whether the AMT8's 'time stamping' isn't outdated by OSX, and whether, in fact, it relies on CoreMIDI, or Apple/Emagics own special system to keep the two compatible.

Well at least we know Digidesign's MIDI I/O works. Slick looking box too. [img]images/icons/cool.gif[/img]

Sorry for the rambled thoughts.

Phil

[Digi Engineering]

Hi guys--

On OSX, each midi device tells CoreMIDI whether it wants the midi data "ahead of time" or "on time." In the case of devices that implement hardware timestamping (like our MIDI I/O), we ask for data ahead of time. On the MIDI I/O there are buffers and a very accurate clock. We buffer up the early data on the device, then send it out the ports according to it's timestamp. Since this is all done on the device itself, we are able to obtain much lower jitter.

Another thing to note is that Pro Tools always sends midi data ahead of time to CoreMIDI. Then CoreMIDI either 1) funnels this data directly to the device for the device to schedule (as in our MIDI I/O), or 2) schedules the midi data itself and sends it "on time" to the device.

All of this is done according to the CoreMIDI spec, so our MIDI I/O will work with any app that sends out timestamped data, and Pro Tools will work with other devices that implement hardware timestamping.

-Ted

[Peterjk]

Thanks Ted. For your insight on timestamping.

I am still left with my question though.

Digi Midi og Emagic ATM8? I asume that the later implement timestaping.

Question: The Digi Midi WILL WORK with my old 888/24 and old USD right. It's a standalone USB midi interface RIGHT??

(ATM og DIGI, I think they are about the same price)??

Any thoughts guys.....(Digi?)


Peter

[shaggy]

Ted,

Is there any advantage using the DIGI interface? I.E. does it 'know' when the session starts so that the interface clock is referenced to ticks accurately. Or is it a freewheel thing, on the incoming MIDI like before, you have to nudge a bit? But once its in place...it will output accurately---I understand that part.

Any thoughts on something like the AV timepeice or AMT-8---would they behave similarly with the new OS-X drivers?

Also, is the timestamp stripped away by PTs, or will it be effected by pasting to a new location? I.E. MTC based stamping or tick based? Would then PTs correct the stamp according to the new 'appended' location? And could these stamps be used to place events back into a session?

One more quicky---Does all this mean we'll have much tighter MTC? (for syncing) Or will there still be the 1/4 frame slop?

[shaggy]

...even 'though I'm an ol dawg to MIDI (having rewound enough tape so the dang DX drum machine counted the FSK pulses correctly....)

I would still like to know. How does a Pro Tool's 6 Session communicate to the DIGI interface, if indeed it does? And will 1/4 frame slop be a thing of the past?

[Digi Engineering]

Timestamps are generated "on the fly" by PT. In other words, midi events don't have a permanent timestamp attached to them.

In terms of MTC, all MTC events have timestamps, and are sent ahead-of-time, so MTC will have the same sub-millisecond accuracy as other midi data.

-Ted