Pro Tools - AVE performance test results

[mik]

Hi Marsian,

Just another information.
I've also made some test for 12.7 and Sierra with the intensity (no genlock).
I had AVE errors with drivers 10.4.3 and over.

All errors disappeared with driver 10.2.3

[MIKEROPHONICS]

Hi. Marsian,

My opinions come because I have done extensive testing with accurate test gear and I have published my results on Pro-Tools-Expert. I tested everything from PT 10 with a mojo, through to the AVE with BM and various AJA peripherals. I took it seriously and bought the SyncOne2 which is accurate to 0.01 of a frame (compared to sync checks 1/16 of a frame - also reviewed on PTE. I gave up on my BM Decklinck Studio because the drivers were hpoor and the AJA IOXT blew it out of the water, so I bought it and I now spend little time faffing these days as the AJA drivers just work with the AVE.
I am not trying to be a dick here, but am offering my experience to be helpful (as you are). I haven't tested all your latest combinations, I have no need as I have a fast and stable system via my thunderbolt IOXT. Many of my friends & peers in the London freelance scene have given up on BM and now use AJA.
If you read my reviews you will find that under some conditions, BMs do indeed perturbate, even genlocked, rather more than you would believe! YMMV, but remember that if genlocked, the AVE can only be adjusted in whole frames, so talk of quarter frames measured on a device that is not very accurate is scientifically gibberish. Hey ho.


Sent from my iPad using Tapatalk

[Postman]

Originally Posted by MIKEROPHONICS

Hi. Marsian,

My opinions come because I have done extensive testing with accurate test gear and I have published my results on Pro-Tools-Expert. I tested everything from PT 10 with a mojo, through to the AVE with BM and various AJA peripherals. I took it seriously and bought the SyncOne2 which is accurate to 0.01 of a frame (compared to sync checks 1/16 of a frame - also reviewed on PTE. I gave up on my BM Decklinck Studio because the drivers were hpoor and the AJA IOXT blew it out of the water, so I bought it and I now spend little time faffing these days as the AJA drivers just work with the AVE.
I am not trying to be a dick here, but am offering my experience to be helpful (as you are). I haven't tested all your latest combinations, I have no need as I have a fast and stable system via my thunderbolt IOXT. Many of my friends & peers in the London freelance scene have given up on BM and now use AJA.
If you read my reviews you will find that under some conditions, BMs do indeed perturbate, even genlocked, rather more than you would believe! YMMV, but remember that if genlocked, the AVE can only be adjusted in whole frames, so talk of quarter frames measured on a device that is not very accurate is scientifically gibberish. Hey ho.


Sent from my iPad using Tapatalk

Hi Mike,

I'm pretty much out of the game now but for the record, Syncheck3, the one that SyncOne2 copied, measures to .01 frame.

[simonchase]

Originally Posted by marsian

Mike,
The question is: simonchase, what was your combination?
Also, when the BM is genlocked, it doesn't drift for shure, right?

I think it was PT 12.4, El Cap and BM10.4.3 but don't quote me on that. It was very obvious when the BM drivers were wrong and when they were right.

I wasn't gen locked but it was very solid (especially compared to desktop video and PT10 either of which is soup) but as Mike says - only tested to 16th of a frame. I was happy with that level of accuracy, it never bit me on the bum.

But I bought a second hand AJA PCIE cheap online and I get close to zero video errors now and on the very rare occasions when I do get them, they resolve themselves much quicker.

YMMV

[marsian]

Hi Mike,

all good. We both just want to share some informations. I think it's tricky these times to find the best performing system, because there are so many options and it's not only the video hardware.

@simonchase: thanks very much.

I think all in all these are good infos for the people that are using Decklinks.

At the moment I'm using BM driver 10.2.3 with OS10.10.5 and PT 12.7 with the videoengine from PT 12.4. It works good, but you never know. It's good to have the AJA option if everything goes wrong with BM.


Cheers
Maik

[Frank Kruse]

Maik,

I've been saying this for years on this board but here goes again:

1) Don't use BM drivers newer than 10.4.3 (there's absolutely nothing to gain from newer versions for us audio people anyway)
2) Genlock your systems

I have a Syncheck 3 and I can compensate my system down to 1ms offset accuracy, zero span and zero drift.

(BMv10.4.3, 300€ BM Decklink SDI card, Sync HD, AJA Gen10)

All the tests I know that show terrible results with BM were either done with non genlocked cards or the well known broken v10.5 and or newer drivers (the ones that drift or play in sync and then suddenly go out of sync etc. by random measures)

There's no way that two independent clocks will maintain sync over time. Without a common clock any two separate devices will drift. It's just a question of time.

Wether or not the drift bothers each individual depends on how long your average play span is. If you need bang on sync for a continuous 2 hours presentations personally I wouldn't work without genlock.

The second issue that could bother you is the start-error. +-3qf for every time you hit start is more than 1 frame of difference in your example. Personally this would drive me crazy when editing ADR, dialogue or hard-fx. But again, YMMV.

And just to clear up some confusion about these figures and names that often lead to misunderstandings in these discussions:

Offset: The overall (constant) delay between video and audio during playback. Pretty much a non-issue and not a sign for general quality because PT can compensate for this or you can do it in your monitor controller (which is more accurate). As in "A card with 4 qf delay is not *better* than a card with 8qf delay for the same format" So generally speaking: Offset is a non-issue unless it gets very extreme.

Drift: Variance in playback speed during playback meaning video and audio run out of sync over time. This is problematic and you can not compensate for it.

Start-error (I call it that way): Hit play 5 times and get a different offset reading every time. Again problematic (probably the worst of them all as you can't trust what you're doing in your timeline).

Span: a figure showing the average or span of the start error after a given amount of measurements. Lets say you measure twice. First time show 4 qf offset, the second time 6qf. that's a span of 2qf.

I've seen high end AJA systems strobe, run out of sync (even when genlocked) all the time in Video satellites in PT12 leading to 10-15 restarts per mix day.

A truly comparable test still hasn't been done to my knowledge. Either they were using the broken BM driver or different OS/PT version when comparing AJA vs. BM or something else was different.

All I'm saying is: I've been using the above setup for years now without any sync issues. I've seen terrible BM setups and terrible AJA setups. The only way to rule out sync issues is to genlock them.

The only issues I do have started with PT AVE going downhill more and more with it's "unable to stop" and "couldn't create thumbnails" and whatever other errors it throws up from time to time.

I've seen all of these on both AJA and BM based systems on edit systems and on truly high end setups in brand new top notch mixing theatres in London and elsewhere. I don't blame these places. They are at the mercy of PT. We didn't have these new issues in PT10 but on the other hand we never had truly accurate sync under PT10 and older either. (At least not with non AVID video hardware)

Frank.

[tom_lowe]

Originally Posted by Frank Kruse

Maik,

I've been saying this for years on this board but here goes again:

1) Don't use BM drivers newer than 10.4.3 (there's absolutely nothing to gain from newer versions for us audio people anyway)
2) Genlock your systems

I have a Syncheck 3 and I can compensate my system down to 1ms offset accuracy, zero span and zero drift.

(BMv10.4.3, 300€ BM Decklink SDI card, Sync HD, AJA Gen10)

All the tests I know that show terrible results with BM were either done with non genlocked cards or the well known broken v10.5 and or newer drivers (the ones that drift or play in sync and then suddenly go out of sync etc. by random measures)

There's no way that two independent clocks will maintain sync over time. Without a common clock any two separate devices will drift. It's just a question of time.

Wether or not the drift bothers each individual depends on how long your average play span is. If you need bang on sync for a continuous 2 hours presentations personally I wouldn't work without genlock.

The second issue that could bother you is the start-error. +-3qf for every time you hit start is more than 1 frame of difference in your example. Personally this would drive me crazy when editing ADR, dialogue or hard-fx. But again, YMMV.

And just to clear up some confusion about these figures and names that often lead to misunderstandings in these discussions:

Offset: The overall (constant) delay between video and audio during playback. Pretty much a non-issue and not a sign for general quality because PT can compensate for this or you can do it in your monitor controller (which is more accurate). As in "A card with 4 qf delay is not *better* than a card with 8qf delay for the same format" So generally speaking: Offset is a non-issue unless it gets very extreme.

Drift: Variance in playback speed during playback meaning video and audio run out of sync over time. This is problematic and you can not compensate for it.

Start-error (I call it that way): Hit play 5 times and get a different offset reading every time. Again problematic (probably the worst of them all as you can't trust what you're doing in your timeline).

Span: a figure showing the average or span of the start error after a given amount of measurements. Lets say you measure twice. First time show 4 qf offset, the second time 6qf. that's a span of 2qf.

I've seen high end AJA systems strobe, run out of sync (even when genlocked) all the time in Video satellites in PT12 leading to 10-15 restarts per mix day.

A truly comparable test still hasn't been done to my knowledge. Either they were using the broken BM driver or different OS/PT version when comparing AJA vs. BM or something else was different.

All I'm saying is: I've been using the above setup for years now without any sync issues. I've seen terrible BM setups and terrible AJA setups. The only way to rule out sync issues is to genlock them.

The only issues I do have started with PT AVE going downhill more and more with it's "unable to stop" and "couldn't create thumbnails" and whatever other errors it throws up from time to time.

I've seen all of these on both AJA and BM based systems on edit systems and on truly high end setups in brand new top notch mixing theatres in London and elsewhere. I don't blame these places. They are at the mercy of PT. We didn't have these new issues in PT10 but on the other hand we never had truly accurate sync under PT10 and older either. (At least not with non AVID video hardware)

Frank.

I know nothing is going to beat a genlocked system, but I really feel that's only important if laying back to tape, or doing very long playbacks.

VideoSlave 3 is the fraction of the cost of a genlocked system and gives incredibly close results, with latest BM drivers on non-genlocked systems.

It seems the problem is with BM drivers and how they interact with the AVE. Given the cost difference for almost the same performance, I think it's a no brainer.

There is a slight delay with video locking to playback, but compared to drifting sync, that's a price I'm willing to pay.

[Frank Kruse]

Originally Posted by tom_lowe

I know nothing is going to beat a genlocked system, but I really feel that's only important if laying back to tape, or doing very long playbacks.

VideoSlave 3 is the fraction of the cost of a genlocked system and gives incredibly close results, with latest BM drivers on non-genlocked systems.

It seems the problem is with BM drivers and how they interact with the AVE. Given the cost difference for almost the same performance, I think it's a no brainer.

There is a slight delay with video locking to playback, but compared to drifting sync, that's a price I'm willing to pay.

Yep, I'm sure it's a good alternative. I have to deal with too many picture revisions sometimes weekly or daily so wrangling that outside the box (or a second box) so I like to keep it all in the session.


hm....

[marsian]

Hey Frank,

I know you posted those things in the past. It's great to have all that information from various people with different setups together. Let's hope that AVID will improve AVE performance in the near future, so that we are able to have solid systems without those well known error messages. There should be a database with known good combinations of soft- and hardware.


Cheers
Maik

[smurfyou]

Originally Posted by marsian

Hey Frank,

I know you posted those things in the past. It's great to have all that information from various people with different setups together. Let's hope that AVID will improve AVE performance in the near future, so that we are able to have solid systems without those well known error messages. There should be a database with known good combinations of soft- and hardware.


Cheers
Maik

I have the exact same specs as you for my home rig. I also find the BM driver v10.2.3 the safest. Blackmagic doesn't necessarily test the legacy cards with each new release. So maybe not surprising to use a driver that was current with the card.

I do prefer Aja because of the driver nonsense but as noted they aren't perfect. Avid really needs a simple playback-only device made specifically for PT. That way they control quality and provide a better guarantee of compatibility. We can always dream. (yes like the Moio but not such a finicky piece)

MacPro 5,1 12-core, 24GB RAM, OSX 10.10.5, Decklink Extreme 3D+