Digital Igloo (Eric Klein, YGG)

SKU said:
How did Boss manage to get their latency well under 1ms which most other are above 2ms and usually 3+ with some configurations, including Helix?
Click to expand...

I'not DI and I can't give you the complete answer, either, but one part of the answer surely is that at least the GT-1000 is operating at 96kHz internally, whereas most other modelers are using 48kHz. They might as well be able to use smaller buffersizes at the same time and they're probably using somewhat faster converters - but I have no idea which of the latter two would be the case.
 
Gross legacy amps take less resources
Shooting Star GIF
 
SKU said:
Hi, @Digital Igloo! A tough one. How did Boss manage to get their latency well under 1ms which most other are above 2ms and usually 3+ with some configurations, including Helix?
Click to expand...
Just confirmed with Brandon—Helix Floor, Rack, and LT are ~1.83ms from analog in to analog out. HX Stomp and Stomp XL are even shorter (1.17ms) because they don't need to hop between two DSPs. (Hopping between cores is similar, so remember: more cores = more latency.)

BOSS is one of the only MI companies on the planet to make their own silicon, so I'm not entirely sure what's going on. (Yamaha is another.) They also run everything at 96kHz, which lowers the latency of A/D and D/A conversion by a small bit. I doubt latency was the motivating factor here, however, as their marketing attempted to spin that 96kHz sounds better. Spoiler alert: it doesn't. In fact, the algorithms can end up sounding worse, because you're wasting half your horsepower on unnecessarily high sample rates, which means you're sacrificing things elsewhere, like granularity, higher oversampling, number of simultaneous blocks, etc.

Then consider that if you're also recording at a sample rate other than 96kHz, you lose any latency advantage because you need to sample rate convert back to 44.1 or 48.

No data to back this up, but I can only surmise the whole 96kHz thing was either because their silicon was designed to be used for other Roland products like digital mixers or multitrack recording, or... it's a marketing bullet hoping to fool guitarists into thinking 96kHz is better. I can't imagine BOSS building their entire engine around 96kHz to save less than a few hundred microseconds in latency, which even Steve Vai wouldn't notice.
 
Last edited:
JiveTurkey said:
I just wanted bag on Boss
Click to expand...

I know (obviously).

On a serious note, there's really a whole lot of things Boss is doing right. Sometimes *vastly* better than any of the competitors.
- Mobile OS editor. Excellent. No need for any fancy on-device touchscreens. I know, the BTS app itself is quite debatable, but still.
- Lowest latency in its class. Pretty much the only unit allowing you to nest multiple loops serially, maybe even when running some other digital devices, without running into latency trouble all the time. This is a big one for me as I could possibly replace my loop switcher and still integrate my analog pedals the way I like it.
- Global blocks. No need to elaborate, IMO each and every somewhat more complexed modeler should offer such a functionality.
- Sort of gapless switching with delay/verb spillover. Yes, I know about the limitations - but for most situations, they should not limit you all that much. Especially along with global blocks this is a huge one (as you could, say, split a huge kitchen sink preset into one bank without losing gapless switching and global access to some parameters).
- Freely assignable parameter knobs. That's just sooo much what Mr. Franck's doctor ordered, really. Loved it with the GT-10 already (and it's been just 4 knobs with rather limited assignment options).

There's probably more and I'm explicitely not talking about the sounds (whether they'll be working for me is the very thing I plan to find out more thoroughly in 2024, my rather brief encounters yielded mixed results).

Sorry that this is getting so much OT in your part of the forum, Eric, but uhm... couldn't help it. And well, seriously, if the Helix had some/all of the above (ok, at least global blocks) and better visibility, I'd rebuy it tomorrow.
 
Sascha Franck said:
I know (obviously).

On a serious note, there's really a whole lot of things Boss is doing right. Sometimes *vastly* better than any of the competitors.
- Mobile OS editor. Excellent. No need for any fancy on-device touchscreens. I know, the BTS app itself is quite debatable, but still.
- Lowest latency in its class. Pretty much the only unit allowing you to nest multiple loops serially, maybe even when running some other digital devices, without running into latency trouble all the time. This is a big one for me as I could possibly replace my loop switcher and still integrate my analog pedals the way I like it.
- Global blocks. No need to elaborate, IMO each and every somewhat more complexed modeler should offer such a functionality.
- Sort of gapless switching with delay/verb spillover. Yes, I know about the limitations - but for most situations, they should not limit you all that much. Especially along with global blocks this is a huge one (as you could, say, split a huge kitchen sink preset into one bank without losing gapless switching and global access to some parameters).
- Freely assignable parameter knobs. That's just sooo much what Mr. Franck's doctor ordered, really. Loved it with the GT-10 already (and it's been just 4 knobs with rather limited assignment options).

There's probably more and I'm explicitely not talking about the sounds (whether they'll be working for me is the very thing I plan to find out more thoroughly in 2024, my rather brief encounters yielded mixed results).

Sorry that this is getting so much OT in your part of the forum, Eric, but uhm... couldn't help it. And well, seriously, if the Helix had some/all of the above (ok, at least global blocks) and better visibility, I'd rebuy it tomorrow.
Click to expand...
I should probably clarify that I like Roland/BOSS quite a bit. I'd probably still be at Roland US if Japan hadn't completely ignored all my proposals—not even a "Sorry, Eric. We're not interested." They have way more wheelhouses in which to play than YGG does. My MV-7000 and MV-9000 (2007, so yeah, they're mighty ugly today) might've done well, but alas, they were never meant to be:
MV-7k9k.jpg
 
Last edited:
Digital Igloo said:
Then consider that if you're also recording at a sample rate other than 96kHz, you lose any latency advantage because you need to sample rate convert back to 44.1 or 48.
Click to expand...

Nah, that's irrelevant. In-to-Out device latency will stay the same, regardless of the sample rate you record at (assuming that you draw an analog connection to another interface, which is at least what many people are still doing very often).

Digital Igloo said:
I can't imagine BOSS building their entire engine around 96kHz to save less than a few hundred microseconds in latency
Click to expand...

With the Zoom UAD-2 I'm currently using, at 32 samples buffersize I'm getting 4.2ms RTL @ 48kHz vs. 2.7ms @ 96kHz. With some additional I/O routings factored in, that's already within the "could become noticeable for mere mortals as well" realm. And within a dedicated OS, things should possibly scale a bit more evenly, too (as in getting half the latency at double the sample rate).

Digital Igloo said:
Just confirmed with Brandon—Helix Floor, Rack, and LT are ~1.83ms from analog in to analog out. HX Stomp and Stomp XL are even shorter (1.17ms)
Click to expand...

But that's only true for empty paths. From all I know, some blocks are adding their own amounts of latency (very obviously the pitch blocks do, but I think some more commonly used ones do as well). Pretty decent figures, though.
 
Sascha Franck said:
Nah, that's irrelevant. In-to-Out device latency will stay the same, regardless of the sample rate you record at (assuming that you draw an analog connection to another interface, which is at least what many people are still doing very often).
Click to expand...
Analog in > Analog Out won't be affected, correct, but if you're recording via USB (or S/PDIF or AES/EBU if the modeler has it) and monitoring the output of your DAW, SRC negates any latency advantage of 96kHz—unless you're also recording at 96kHz, of course, which almost no one does.
Sascha Franck said:
But that's only true for empty paths. From all I know, some blocks are adding their own amounts of latency (very obviously the pitch blocks do, but I think some more commonly used ones do as well). Pretty decent figures, though.
Click to expand...
Very few blocks in Helix add any appreciable latency. The poly pitch ones do though, yeah.
 
Last edited:
Sascha Franck said:
With the Zoom UAD-2 I'm currently using, at 32 samples buffersize I'm getting 4.2ms RTL @ 48kHz vs. 2.7ms @ 96kHz. With some additional I/O routings factored in, that's already within the "could become noticeable for mere mortals as well" realm. And within a dedicated OS, things should possibly scale a bit more evenly, too (as in getting half the latency at double the sample rate).
Click to expand...
This isn't what DI's talking about, he's talking about the conversion latency which is typically 24 samples in+out. 24/48k - 24/96k = 250 microseconds.
 
Digital Igloo said:
Analog in > Analog Out won't be affected, correct, but if you're recording via USB (or S/PDIF or AES/EBU if the modeler has it) and monitoring the output of your DAW, SRC negates any latency advantage of 96kHz
Click to expand...

While true, rather little folks are doing so. I have been recording at some places using whatever modeler setups I was using at the time, and I never connected things digitally, not even with the Helix, which would've been easier to connect than other things I was using before.
 
Oh dang, totally forgot about the Roland A-61. Hardcore keyboard controller, robust DAW controller (motorized faders of course), audio interface, and monitor controller with talkback and cue feed in one box. Designed for MainStage/Ableton Live music directors. This, a laptop, and monitors/phones is all you'd need.
Sovereign61.jpg

Clearly, my designs have mellowed out somewhat since 2007.
 
Sascha Franck said:
While true, rather little folks are doing so. I have been recording at some places using whatever modeler setups I was using at the time, and I never connected things digitally, not even with the Helix, which would've been easier to connect than other things I was using before.
Click to expand...
My point was that if you're recording and you're concerned with latency, 96kHz impacts latency (and sound quality) way less than unnecessary D/A and A/D conversions, and you should be running digital if possible. If you're not recording—that is, not dealing with additional A/D/As, DAW buffers, interfaces, control panels, and the like—anything under 4-5ms is a nothingburger.

So 96kHz is pointless in a guitar processor either way. The more I think about it, the more I'm convinced their silicon was specced for Roland digital mixers (just like how Helix's architecture was birthed from Stagescape M20d). It's the only thing that makes sense to me from an ROI standpoint.
 
Digital Igloo said:
If you're not recording—that is, not dealing with additional A/D/As, DAW buffers, interfaces, control panels, and the like—anything under 4-5ms is a nothingburger.
Click to expand...

Yes and no. Because in case RTL is 5ms, using a loop (splitting up the digital path of course) would already result in 10ms. And at least for me (plus for quite some other folks I know), this is when things start to feel sort of distracted, at least when using headphones. Using 2 serial loops would be pretty much impossible already. With the GT-1000, even with both loops set up serially, RTL (at least without any blocks adding their own latency - no idea about which those would be in Boss land), we'd still be kicking at 3ms (or even lower, according to Leo Gibson - I rounded his 0.7ms up to 1ms), which is just fabulous.
Now, if you never use digital FX loops, if you never run your signal into another digital device, into an additional digital mixer and/or IEM system, any DSP controlled monitors, all that while using a guitar cable instead of a wireless system - sure, in that case, even 5ms is a decent value.
But personally, I'm constantly running into various scenarios where it's not possible to avoid some of these things, sometimes in combination.
So, even in case you're not Steve Vai (which I certainly am not), it's not exactly too uncommon running beyond the "uh, that's not too much fun anymore" point.
It's the reason why I always bring my own little mixer with me on IEM jobs, so I can at least monitor my own signal without any irregularities (and believe me, I've been there multiple times... FOH folks apparently aren't always the brightest candles in my neck of the woods). And it's also why I rather start with the lowest possible latencies. It's also one of the reasons I kept using Logic, because at least for a long time it's been the only DAW where buffer sizes were pretty much a "set and forget" thing (even if you dialed them in pretty low).
 
Hi,

  1. About helix paths/cores synchronization - if I play with powercab AND real cab I have two paths - one with IR for powercab and second without for real one. Are they in phase? Generally speaking if two paths have different effects sets will they be synchronized at the output?
  2. Please add poweramp simulator
  3. Please keep worldclock in the helix 2.0
  4. Have you considered "advanced" editing mode where you expose more model parameters?
  5. Is it possible to make model editor public? a tool a-la ltspice?
  6. I can't RTFM my way through this but how to make helix auto assign bypass to footswitch as soon as I add new block?
  7. Please give us global blocks
Thank you!
 
chicken said:
About helix paths/cores synchronization - if I play with powercab AND real cab I have two paths - one with IR for powercab and second without for real one. Are they in phase? Generally speaking if two paths have different effects sets will they be synchronized at the output?
Click to expand...

I guess this would be difficult as the Helix can't guess what the IR is doing.

chicken said:
Please add poweramp simulator
Click to expand...

Yes please.

chicken said:
I can't RTFM my way through this but how to make helix auto assign bypass to footswitch as soon as I add new block?
Click to expand...

While I'm all for UI improvements, how would the Helix know which footswitch you'd like to use? Besides, if anything, assigning bypass to a footswitch takes around a second (in case the unit is in reach).

chicken said:
Please give us global blocks
Click to expand...

Yes please!
 
Sascha Franck said:
While I'm all for UI improvements, how would the Helix know which footswitch you'd like to use? Besides, if anything, assigning bypass to a footswitch takes around a second (in case the unit is in reach).
Click to expand...
In headrush it assigns to the first free one. Very handy for quick experiments. And as all UI/UX shortcuts takes only that little time manually yeah. I'm very proficient at hitting Save twice for example. All my presets are New.
 
Sascha Franck said:
I guess this would be difficult as the Helix can't guess what the IR is doing.
Click to expand...
I'm far from DSP field but I guess IR length dictates segment size and n in O( n ). That's why it can limit number of 2048-point IR - knowing length gives estimation of load
 
You must log in or register to reply here.
Back
Top