--unimportant part--
Soundcard buffers data.
CPU grabs data and sucks it along the buses in the system, to memory.
CPU sequentially reads from memory to it's data registers and processes it with the algorithms of each plugin and writes it back to memory.
In an ideal situation, a single CPU core could handle each track and then you could have as many tracks as CPU cores, but you can't. So you have to add more tracks to each CPU. This adds processing time. (And you can't always have multi-core processing work that well)
Then the CPU ships the processed data along the bus from memory to the soundcard.

--important part--
The "bottleneck" is the first part of the system unable to keep up with the load placed on it.
This varies per system although CPU, buses, and memory, are more likely to be a bottleneck because more demand is placed on them (but they're also more powerful, so...)

Codemonkey:Have you thought of running for President of RO? The reason I'm asking is that was one of the best replies that did not really answer the question. It might be any of the three. LOL Besides I'd vote for you. I am a member of the Monkey Party.

BTW Mathius I believe the TASCAM US-122L is an interface and not a soundcard. All the processing for the effects takes place in the computer and the interface simply changes the finished product from digital to analog. When you say "live" do mean monitoring during additional tracking? Or on stage performance? If it is monitoring the interfaces usually have zero latency meaning that you hear all incoming and processed tracks at the same time. There are other ways to reduce plugin draw such as recording the tracks over inside the DAW so that it is already processed (you end up with a raw and a finished track and use the finished one for mixing while retaining the raw track if you wish to make changes), some programs can freeze tracks which essentially stores the track in a processed state but does not require nearly as much processor or RAM draw.

He asked what limits the number of effects... I'd say that answered it though indirectly.

I thought I made clear that the effects are processed on the CPU and that if you can supply the data fast enough, the CPU will be the bottleneck. The soundcard simply outputs what is in it's buffer at the specified sample rate, it's up to the rest of the system to fill it.

Also the impression I got was that the OP wants to process things while recording.
If it's during playback only then just take the 4096 buffer size hit, you won't be ruined by the slight processing delay.

Does that mean you aren't running for Pres? I don't believe RO has a born in the USA clause :-)

Yeah maybe that is what he meant, processing during recording when he said live.
I am disappointed you aren't running for President, however.
Sorry I did read your post but it wasn't clear to me that the bottleneck would be the CPU, but I'm computer illiterate in at least seven languages.

Hey guys thanks for the replies..

To clarify -

When I record a guitar track, I obviously want to hear it in the final form; in this case the final form involves a software called "Guitar Rig" which emulates amps and guitar effects. If I have this and say, a gate and an EQ, (a total of three effects on the one track) then with an older soundcard it skips when the ASIO latency is set to 2ms. I had to bump it out to 4 or even 6, which involves a perceptible delay between hitting the strings, and the final processed sound recorded into the sequencer, or coming out of the speakers.

So obviously my old system was found wanting. I think it reported buffer underruns from memory..

So what I need is a soundcard that can handle several inputs, with several effects on each channel, with a <2ms latency. I am not sure if the processor is also to blame? and I am really unsure what spec to look for in a soundcard..

Cheers
Matt

My memory just kicked in somewhat - am I to understand that the buffer size is inversely proportional to the latency? And that if I specify a larger buffer, then that will increase the latency - which on a weaker card will increase enough to become noticeable?

The smaller the buffer samples the less amount of latency.

Most folks do not record with plugins. They record a dry track which they then process within the DAW either digitally or re-amping the original track or a combination of both. For live FX there are pedals and external FX boxes and such that process the sound before it hits the console and DAW.

Or maybe folks do it differently these days.

I guess I should add that the smaller buffer samples often generate digital artifacts while recording that are undesirable. A sampling of these might include pops or clicks or just outright freezing of the track. A larger buffer will allow more plugins to be utilized but more latency at the headphones unless your interface is also doing the processing for the FX.

TheJackAttack wrote: The smaller the buffer samples the less amount of latency.

Most folks do not record with plugins. They record a dry track which they then process within the DAW either digitally or re-amping the original track or a combination of both. For live FX there are pedals and external FX boxes and such that process the sound before it hits the console and DAW.

Or maybe folks do it differently these days.

Hmm.. thanks - that would reduce the load on my card, but it would also deprive me of the wealth of different sounds I can get from a guitar via the plugin.. I'll consider that as an option though thanks.

TheJackAttack wrote: A larger buffer will allow more plugins to be utilized but more latency at the headphones unless your interface is also doing the processing for the FX.

Hmm.. so some soundcards do the processing, while others rely on the processor?

I am sorry but I am at a loss here. Where does a soundcard enter into this? Are you monitoring through the computer? Or are you monitoring through the Tascam?

I believe he is monitoring at the Tascam but he is monitoring the processed tracks from the DAW rather than the direct input from the Tascam itself.

jg49 wrote: I am sorry but I am at a loss here. Where does a soundcard enter into this? Are you monitoring through the computer? Or are you monitoring through the Tascam?

I'll try to clarify.. I don't own anything yet, but I imagine that soundcard buffer size is vital in applying real-time effects.. zero latency is something you pay a lot for..

I haven't actually bought a Tascam or anything yet, I was wondering what would do what I ask - and here is what I hope to achieve:

-I plug my guitar straight into the card
-I then apply guitar effects which are processed and or recorded with an imperceptible delay - near zero latency

My Audigy2 zs used to do this, but only if I applied a minimum of effects over the guitar track. Any more resulted in a higher latency. But I don't know what this bottleneck is yet... so selecting a card/interface is a mystery.

So someone correct me if I am wrong I am going to parade my lack of understanding here perhaps. I may just learn something. I thought that all latency in real time occured in the computer itself. In how fast the CPU and RAM can process the signal and return it. That the interface had nothing to do with it unless the effect that was being utilized was being processed in the interface itself. The Tascam unit Mathius listed does not do this.
Melodyne
Though there are interfaces that do like TC Konnekt6 which has a reverb unit built into the interface. While reverb is often a resource hog I doubt that shifting the reverb duty from Guitar Rig to something like this will have much impact on that plugin's CPU usage.
I am a guitarist also and I can understand the need to achieve one's "sound" in order to play well but is the gate which frankly I have never used in getting my own sound or the EQ absolutely necessary to be heard in real time. Guitar Rig has bass, mid, and treble controls IIRC. These other two effects could be utilized in post production, I think, with out greatly restricting your sound.
I guess one way to solve this is by purchasing a computer with processing speed sufficient to your task, another way and probably less expensive is to use a processor for the guitar prior to the computer. Line6 pod units come to mind. They even make interfaces like the pod studio UX2. This has a feature (quote from Line6) "Fix it in the mix. Record a dry signal while monitoring a fully processed POD® tone. Don’t commit to a tone until you’re ready." Which means that even if you can only dial this modeler in to something pretty close to your sound so that you are comfortable playing you can use GR3 in post production to tailor your sound exactly without latency.

jg49 wrote: In how fast the CPU and RAM can process the signal and return it. That the interface had nothing to do with it unless the effect that was being utilized was being processed in the interface itself. The Tascam unit Mathius listed does not do this.

I have never used in getting my own sound or the EQ absolutely necessary to be heard in real time. Guitar Rig has bass, mid, and treble controls IIRC. These other two effects could be utilized in post production, I think, with out greatly restricting your sound.

As I understood it, the soundcard has a dramatic effect on the latency you experience, in the same way that a graphics card handles the processing - if the CPU did it all then all soundcards would be equal given the same CPU? And I was also under the impression that the Tascam was an ASIO device.. it must be a soundcard of sorts?

And well.. I just play for fun, and playing with a huge array of sounds and effects inspires me to experiment with styles as well. Just personal preference I guess...

C'mon everyone. Code Monkey would have to be at least 35 to run for Pres.

Just to clarify, for the purposes of a clear discussion let us isolate terms. For the purpose here let's use the term soundcard as a PCI device that is an integral part of the computer and interface as an external device plugged into the computer by either firewire or USB. The interface on the output side is mostly just a digital/analog converter. While I am quite certain that there is some variance in individual units a quality interface should not add more than 2-3ms (milliseconds or 0.002- 0.003 seconds) which is for the most part undetectable by most humans. Most references say that 12ms is where latency begins to become discernible, so the amount of latency occuring due to the interface is at best minimal.
This is an excerpt from Sound on Sound Jan 2005
"The audio signals also have to pass through the A-D and D-A converters on their way into and out of the computer, which adds a small further delay.

Converters usually add a delay of about 1ms each, but there may be other hidden delays as the signals pass through the soundcard, caused by the addition of features such as sample-rate conversion for rates of 32kHz or lower, which are sometimes not directly supported by modern converters. So 'zero'-latency monitoring actually means about 2ms latency, which is still almost undetectable in most situations. It's incredibly useful during recording, because the performers can listen to their live performance without any audible delay, which makes it far easier to monitor via headphones, for instance. However, it does have one unfortunate side-effect: you can't hear the input signal with plug-in effects, which prevents you from giving vocalists some reverb 'in the cans' to help with pitching, or guitarists some distortion, for example. One way around this is to buy an interface with on-board DSP effects and use them instead. Another is to use an external hardware unit to add effects. In both cases it's generally possible to connect them up such that you can hear the performance with effects, but still record it 'dry', for maximum flexibility during later mixdowns."
The use of DSP effects is not what you are seeking for guitar, but the use of the Line6 type system is essentially adding external hardware to process your signal. So IMO opinion your quest for low latency has little to do with the soundcard or interface and is almost exclusively with your computer's ability to process.

Without discussing specific equipment there isn't a complete definitive answer. The Tascam does not do any DSP at all. It digitizes and routes. If the OP wants to then listen to the FX in realtime then those DSP FX are being created in the computer itself.

Herein is the delay. The computer requires resources to create reverb or whatever which can then cause the bottlenecks that in turn cause latency.

If the DSP were happening in a dedicated box like the Konnekt 24 or whatever then the latency delay is less because 1)the signal can be monitored directly at the box instead of sent to and from the computer and 2) the DSP box is NOT running Windows or OS* or a monitor or antivirus or anything else.

There are live shows (Broadway, pop bands) that run real time FX but those are from dedicated DSP boxes and the latency is minimal enough to not be a concern.

That's interesting..

For the sake of the discussion, I'll just add that I swear that with my old PCI Audigy 2 ZS I could apply real time reverb and distortion etc via Amplitube or Guitar Rig (guitar modelling), with it set to 2ms. But say I wanted to also add a better eq, or a gate (to allow a clean palm-mute with heavy distortion) then I would start to get what was called 'buffer under-runs", which neccesitated me to set the buffer to a larger size. A larger buffer allowed more effects, but also had the undesirable side-effect of increasing the latency. I was told that a better soundcard would allow a greater buffer at a <2ms latency.

Of course you are dead right that hardware effects would solve this, but I would also lose a lot of flexibility.. I know there are external soundcards out there that will do the trick but I also know most will also cost a lot. What I am completely unsure of, is what spec of a soundcard demonstrates this ability...?

Sorry if my line of questioning annoys you :) but I would love to know, before I buy my external soundcard.. BTW I have also just learned that firewire and Expresscard/PCMCIA have a much better bandwidth than USB or PCI, so I will be looking for a PCMCIA card I think..

http://en.wikipedia.org/wiki/Software_effect_processor

This is the principle I am talking about... sorry if I have confused everyone, I probably should learn industry terminology first :)

Better hardware has more effective drivers. I suspect the soundcard applies the algorithm mentioned in the wiki link as well..

PCMCIA is for laptops. The current "version" (and much faster) of this technology is Express Card.

A PCI card fits in a slot on a destop computer. The newer and much much faster version of this is PCIe (e for express).

Firewire has always been better for audio recording and indeed anything bandwidth intensive. I must type this statement 50 times a month on these boards. USB is what is termed a "dumb" interface. The actual throughput does not match the published specs for a variety of reasons not worth getting into.

Now, later this year there are two new protocols due to be utilized. USB 3.0 and a 'new' Firewire protocol. I am fairly confident the USB protocol will appear. The other? eh. I am not sure if the drawbacks of USB 1 and 2 will be rectified. At any rate, if we do move to the next level then bottleneck will not be data transfer at all but squarely reside at the computer itself.

TheJackAttack wrote: PCMCIA is for laptops. The current "version" (and much faster) of this technology is Express Card.

A PCI card fits in a slot on a destop computer. The newer and much much faster version of this is PCIe (e for express).

Yes - my old desktop is dead, my old PCI card is useless to me - so now I have a laptop with better spec than that, so I suspect if I select the right external soundcard I'll be much better off..

I was going to go USB, but I see that Expresscards are out there with better bandwidth so I think I'll ignore USB. I think I remember from memory that the older PCMCIA runs off the USB bus, so I think I'll avoid that as well..

Cheers for the clarification

Techically PCMCIA ran off of a PCI bus. Often there was a USB port connected to the same bus as well which just added to data transfer bottlenecks.

Usually on a laptop I recommend that you use the USB port (for external hard drive) farthest physically from the Card Bus or Express Card slot in hopes of eliminating conflicts.

On your laptop you will definitely want/need an external hard drive as a destination for your audio. It could be a firewire drive chained onto your interface or a USB drive on a different IRQ.

I know JackAttack has already said this but I thought I said it twice before, too...

In case we're still unclear,

==I might be tired. I only just got up.
==I don't like politics.
==I'm only 20.
==The soundcard (properly known as interface) does NOTHING except receive a buffer of audio from the CPU, then output that buffer through the DAC.
==The CPU handles all the processing of effects, unless your interface happens to have an effect processor, but these are accessed through the interface software, and AFAIK are not controllable by DAW software.
==ASIO as a driver architecture, has greater timing accuracy than MME or DirectSound. The driver is just a way of getting the CPU to talk to the interface, and does nothing to the audio except give it directions to the interface's memory space.

Thanks for the information and the discussion guys - I have a lot of food for thought, and I really appreciate it.

I must admit though, I am a bit baffled as to the differences between soundcards/interfaces with regards to latency using a "software effect processor".

Perhaps it is simply bandwidth between the card and the memory and CPU? Although I did find this quote on HowStuffWorks:

"Digital Signal Processor (DSP): Like a graphics processing unit (GPU), a DSP is a specialized microprocessor. It takes some of the workload off of the computer's CPU by performing calculations for analog and digital conversion. DSPs can process multiple sounds, or channels, simultaneously. Sound cards that do not have their own DSP use the CPU for processing."

Your software effect processor is more commonly known as a plugin. It is taking the place of a hardware FX box. Where this processing takes place is the crux of your problem. On a cheap sound card or interface ALL audio processing is handled by the computer CPU.

I think you are right, John, the way to achieve truly low latency in real-time sound processing must be to bypass the USB/PCI etc buses and find a card that will process it on the card itself..

Thanks for all your time guys

Is that what you said??

I just meant that he is right that where the processing takes place is the crux of the issue, the rest is my extrapolation..