FM9 Latency - Around 5ms?

[Aleksi]

Absolutely. Your money is invested well in Fractal. Great products and arguably even better R&D and customer service when it comes to pushing the product forward and correcting issues.

Yep, I should've known to just get an FM9 to replace the Axe-Fx II, would've been cheaper than flipping other stuff . They seem expensive at first, but you get a complete product that keeps evolving.

 

[DLC86]

When I was on the ax8 I found out that also the flanger block introduced a tiny bit of latency on the dry signal, but maybe that has changed on the new gen of hardware.

 

[maxc]

The theoretical latency of the FM3/9 is:
In -> Out - 96 samples = 2ms
In -> Amp -> Out -160 samples = 3.33ms
In -> Amp -> Cab -> Out - 160 samples = 3.33ms (Cab block doesn't add any latency unless IR has leading silence)
in -> Drive -> Amp -> Cab -> Out - 192 samples = 4ms
None of the other blocks add latency.

It appears that perhaps there's a bug in the FM9 and it's adding an extra 64 samples of latency. I will discuss this with the head engineer on that project tomorrow.

By the way, is there a reason Dyna-Cabs appear to have extra ~0.3 ms of leading silence compared to normal factory IRs?




This is consistent with the difference in latency I've measured using RTL Utility.

 

[FractalAudio]

Is 4ms basically the worst case scenario in FM3/9 (not considering eg pitch effects)? I'm wondering whether adding another drive block will add another 0,67ms or if running two amps on the FM9 will add more latency? I would assume two drives in series would add more latency.

Yes, another drive block will add another 32 samples of latency if it's in series.

If you are running two amps chances are they are in parallel so there won't be any added latency.

The only effects that add latency are the amp and drive blocks because they oversample. The amp blocks also add some additional latency beyond the latency due to oversampling because the processing is performed on a dedicated core.

 

[FractalAudio]

By the way, is there a reason Dyna-Cabs appear to have extra ~0.3 ms of leading silence compared to normal factory IRs?

View attachment 131549
View attachment 131550

This is consistent with the difference in latency I've measured using RTL Utility.

Yes, this has been explained before. It's because they are not minimum-phase transformed.

 

[FractalAudio]

When I was on the ax8 I found out that also the flanger block introduced a tiny bit of latency on the dry signal, but maybe that has changed on the new gen of hardware.

It depends on the flanger type. If it's a thru-zero flanger it necessarily adds latency because you have to delay wet and dry for that to work.. Otherwise it does not.

 

[Chromatizm]

This is a tricky subject and there will always be different opinions. He told me at the time that the limit was around 10 ms

Tricky subject indeed. Basically you are trying to say that people who claim to be feeling/affected by <10ms latency are lying. That's quite bold to say the least.

The speed of reflexes is in the range of 25 - 50 ms.

What's this has to do with anything? We are talking different kind of "loop", where you brain has already commanded your fingers to move - way before the actual moment happened - and anticipates a certain sound at a specific point in time, based on his previous experience and training, knowing where the sound source is, and on top of that having additional physical reinforcement. And the values within the brain won't "match". Some don't care, some do.
Brain is not waiting to see how the finger fret the string, then think what to do, react, etc. That's a totally different subject and case.
Yes, you probably won't be able to feel the latency with one chord only, but with fast playing - easy.

And that was ultimately my point. If you're happy with 3.6, you won't hear any difference with 4.2.

I'm not arguing with this specific point. Also, it's 0.6, not 0.4. See - cumulative values start to grow.
I'm saying it is not correct that only "other devices up the chain" add latency which you can finally start to feel. And that 10ms is some "minimum" value for that - it obviously was measured for a very different case, for which it can be absolutely correct.
I'm also saying that 4ms vs 8ms is a big difference people can feel. It's ok if you don't feel it - no need to say other don't.
The latency was the first thing I've noticed after moving from pedals and amps to modelers. Many share this experience.

If you've ever heard Beethoven's Ninth live, for example, you won't ask yourself how the gigantic orchestra, some of whose musicians sit 20 metres apart, can perform work that thrills the audience. These are latencies in the range of 30 - 40 ms. And that's why they need a conductor. And that's the beauty of music.

Come on, too much derailing and jumping between unrelated subjects. At least let a people play fast if they want without telling them it's not musical or sport.

 

[DLC86]

It depends on the flanger type. If it's a thru-zero flanger it necessarily adds latency because you have to delay wet and dry for that to work.. Otherwise it does not.

Pretty sure it was there even on non thru-zero models at the time...

EDIT: yep https://forum.fractalaudio.com/threads/not-a-bug-flanger-block-dry-delay.123368/

EDIT2: that axe-edit 2 screenshot looks so "vintage" now

 

[Ringuz]

Does this sound right? I'm getting around 202 samples / 4.2ms with just an amp and cab block in the grid. This increases to 235 samples / 4.9ms with a drive block. Adding a delay and reverb doesn't seem to affect it. FWIW, just an input and output block shows as 87 samples or 1.81ms. Granted, even with a loaded up lead sound, it only gets to 5.6ms but that's slower than the Kemper's constant latency setting which is around 4.9ms I think. It's often lower than that.

I thought these were supposed to be around 3ms? Is it a firmware thing?

Very strange...I measure it and it's no more than 2,5ms

 

[maxc]

Yes, this has been explained before. It's because they are not minimum-phase transformed.

OK. I haven't seen the explanation, but I guessed it may have something to do with the new alignment technique you've mentioned here:

https://forum.fractalaudio.com/threads/whats-cliff-doing-now.192448/page-7#post-2393957

 

[Random Hero]

Very strange...I measure it and it's no more than 2,5ms

Faster than what Cliff himself says is the latency. Very impressive.

 

[Random Hero]

Yes, another drive block will add another 32 samples of latency if it's in series.

If you are running two amps chances are they are in parallel so there won't be any added latency.

The only effects that add latency are the amp and drive blocks because they oversample. The amp blocks also add some additional latency beyond the latency due to oversampling because the processing is performed on a dedicated core.

Great to know. Thanks!

 

[Randalljax]

Yes, another drive block will add another 32 samples of latency if it's in series.

If you are running two amps chances are they are in parallel so there won't be any added latency.

The only effects that add latency are the amp and drive blocks because they oversample. The amp blocks also add some additional latency beyond the latency due to oversampling because the processing is performed on a dedicated core.

So amp blocks in series Will add to your latency then correct ?

 

[Aleksi]

Yes, another drive block will add another 32 samples of latency if it's in series.

If you are running two amps chances are they are in parallel so there won't be any added latency.

The only effects that add latency are the amp and drive blocks because they oversample. The amp blocks also add some additional latency beyond the latency due to oversampling because the processing is performed on a dedicated core.

What about the boost in the amp model? I'm playing around with my new FM3 and just noticed the feature.

 

[Keybi]

Tricky subject indeed. Basically you are trying to say that people who claim to be feeling/affected by <10ms latency are lying. That's quite bold to say the least.

I did not say that!! For the majority of users <= 10 ms cannot be heard. That's why a lot of people play a modeller with a digital wireless and only get the signal from FOH digital mixer to a monitor with DSP. And they feel very comfortable even though the resulting latency is well over 10 ms. And others still have digital pedals in the FX loop. And that's why I asked a lot of experts about it. And they were usually biologists. You're welcome if you stand out from the crowd. But I was primarily concerned with the 0.x ms. Because latency is unavoidable in digital systems. Nothing more. You misunderstood me or perhaps lost me in translation, as English is not my native language. There is no question that someone is more or less susceptible. But there is a statistical limit somewhere, isn't it ?

And the example of the orchestra also shows that immense latencies do not torpedo the ability of a perfect overall sound. And those violins runs are fast as hell ;-). I also propagate low latencies. Because I also like to play very fast and I like it. I therefore also recommend analogue wireless and monitors without DSP. And I myself always stay around 8 ms. In principle, we have the same point of view and are arguing about tiny milliseconds? I think that's nonsense.
Btw this is put into perspective when I play IN EAR. You have more room to manoeuvre because the cabinet-to-ear distance is eliminated. If, on the other hand, I were to play IN EAR without latency, it would probably be very strange for me, as I probably played around 3000 gigs with tube amps before I switched to modellers.

A few weeks ago I played several gigs with an unknown IN EAR system for a bigger production. That's was weird because the latency introduced with this system was huge. It was pretty annoying. I guess it was overall 20 ms.
Fortunately, the audience didn't care and the fee was very good. But the feeling was less good.

And that's what I mean by my sports comparison. At the end of the day, it's a tool that a lot of players love and get great results, isn't it?

 

[Chromatizm]

I did not say that!! For the majority of users <= 10 ms cannot be heard.

Sorry if I misunderstood you. Didn't mean to be rude. This we can definitely agree on, pretty much none of my friends feel any difference in 4ms vs 8ms when playing through PC. No idea why.

But I was primarily concerned with the 0.x ms

Of course, in itself this value is nothing to be concerned about. But as others mentioned, it will be added to everything else in the chain.
I played FM3->X18 a few times in a 6mx6m room (all cables, not wireless) with backing tracks, VERY loud, and I was going CRAZY about the latency. I basically couldn't play. Lost all patience, asked sound guys to enable "direct monitoring" on my channel. They were laughing at me at first, saying there is no delay in X18. Showed me that all filters and blocks on my channel are bypassed, no processing they said.

Then I plugged directly to the speaker and their smiles disappear. It was obvious for them even just listening to me playing that there is a delay in X18. Maybe FM3 latency affected that, maybe something else, I have no idea. But some of us for whatever reason are really sensitive to that. That's why we ask for low latency stuff. It just doesn't feel right to us, even if we can't "calculate" the delay. I have way less live experience than you, maybe that's the issue - 99% of my guitar time was 1-2 meters from the speaker with analog gear, not walking all over the stage accepting that the sound and latency change also.
Maybe it's just in my bones, that immediate response. I never ever had a digital pedal in my chain before Fractal. All analog, always.

I posted my experience here and many people said I should be wrong cause they don't feel the delay of X18. That I shouldn't been able to feel it because the volume was so high that I didn't hear direct signal from the strings, which many believe is the reason for some to hear the latency.
So we are all different, that's probably the only sure thing about all that.

And the example of the orchestra also shows that immense latencies do not torpedo the ability of a perfect overall sound

I honestly think this example is misplaced because the composer was going for that big-orchestra-natural-chorus-wide-scene sound on purpose. It seems to me that "epic" was the goal.
If he'd want to be precise with timing, he'd go for a much smaller group, I'd say.

 

[Keybi]

Of course, in itself this value is nothing to be concerned about. But as others mentioned, it will be added to everything else in the chain.
I played FM3->X18 a few times in a 6mx6m room (all cables, not wireless) with backing tracks, VERY loud, and I was going CRAZY about the latency. I basically couldn't play. Lost all patience, asked sound guys to enable "direct monitoring" on my channel. They were laughing at me at first, saying there is no delay in X18. Showed me that all filters and blocks on my channel are bypassed, no processing they said.

I don't know the X18 . But I have the X32 and the M32 from Midas here in my studio alongside the RME Fireface.
I've never tested their latency, but maybe I will soon. The X32 probably has a latency of 0.8 ms.
But then you must not loop in any effects, as each effect adds at least 0.7 ms.
I got these values from our sound engineer. Perhaps other users can confirm this?

However, if the console is then used as an interface on a computer and is not routed directly to the OUTS, but via a DAW, then of course there is a latency.

I played with an X32 for a long time and fed the signal (via AUX) to a Yamaha DXR10. That didn't feel so good.
I now use OUT 2 directly on a Redsound. I am sure that there is no additional latency. But I suspect that the DSP of the Yamaha adds latency.
On the one hand because of the AD_DA conversion but mainly because of the FIR filtering. But I've never measured that either.

It's also interesting what you say about your experience of playing live. That's probably the reason why I'm more relaxed about latency.
My first gig was probably around 1985, but I remember it was an acoustic catastrophe. Two Marshall Plexi fullstacks behind me and the drum 5 metres away on a riser. The hall reflected so much that I must have heard the snare drum three times. Once directly and the two reflections of the rear hall wall and the opposite wall. And at some point I asked myself where is the ‘One'. And the situation didn't get much better over the next few years. And it was loud. . So forget it if someone says everything was better in the past

I'm not surprised that my heroes from back then, Lukather, Paul Gilbert, Morse, have serious hearing losses these days. I myself have only suffered mild tinnitus.

Back to the topic. I feel very comfortable playing the FM9 via the Redsound.

 

[FractalAudio]

Pretty sure it was there even on non thru-zero models at the time...

EDIT: yep https://forum.fractalaudio.com/threads/not-a-bug-flanger-block-dry-delay.123368/

EDIT2: that axe-edit 2 screenshot looks so "vintage" now

That doesn't apply to the FM9, which is the product we're discussing here.

 

[unix-guy]

Yes, another drive block will add another 32 samples of latency if it's in series.

If you are running two amps chances are they are in parallel so there won't be any added latency.

The only effects that add latency are the amp and drive blocks because they oversample. The amp blocks also add some additional latency beyond the latency due to oversampling because the processing is performed on a dedicated core.

Interesting on the Drive block since a lot of people like to stack drives.

I typically don't but I do usually have 2 in series for my kitchen sink presets.

 

[mr_fender]

Series routed time based effects running 100% wet will also add latency when enabled (whammy, pitch shifting/virtual capo, vibrato via the chorus or flanger block, etc). Compressor look ahead and through zero flanging will as well.