To me, its all about the Talent that is using the equipment

Great point, but you kind of danced around the question.

Who has the latency issues during the music creation process?
Though this may be a matter of opinion, ALL ELSE BEING EQUAL, Musician A's sound probably kicks more butt too!

Don't handicap Musician A

Wow. This one's still kicking. I've been trying to stay out of it, but ...

I really think that everyone's contributing some valid input. I did think of the MIDI latency, but Rene has backed it up with some actual synth measurements. Hey, we might as well throw in the speed of sound latency that we all share. In a standard triangular near-field setup, we're all adding another 3 ms. minimum, and headphone users aren't completely immune, either.

Edit: Zero latency is impossible, unless in a totally virtual self-contained brain system. [Now would that be software or dedicated hardware?]

The cost differential has been mentioned, and that's a biggie. I can't see that there's a 'secret ingredient' inherent to hardware. Consider the notorious practice of workstation manufacturers to cheeze out on internal memory capacity, and make any upgrade that approaches the current sweet-spot to be impossible or prohibitively expensive. Something's got to give, and that's usually in the form of headroom, sample rate, bit-depth, or compression.

I see an overlap in this thread between playability and actual sound quality; both being highly subjective. We'll never have a definitive answer to either subject, but it's worth noting that these are two entirely different yardsticks. I think that's it's fair to compare this to dedicated gaming machines versus 'gaming' computers. There's a certain efficiency to be gained from tailoring and fine-tuning an O/S along with all of the other components, but you suffer the cost of flexibilty.

If you read any of Rene's contributions, you'll see a software designer that continually struggles against the limitations imposed by the status quo, and yet somehow manages to maintain the highest of sound quality standards. He may run up against speed issues, sample-rate 'ceilings', or the like, but will always find a way to include an optional override in some form for those who desire it. I can't see where something with the quality and feature set like z3ta+ and Dimension could ever be construed as 'inferior', but then again ... the eye of the beholder.

One thing that I haven't seen mentioned [I may have missed it] is instances. Although layering and 'combi' scenarios are possible with hardware, it's not like you're going to be able to line up 19 complete Tritons (for the cost of one) with reasonable (or legal) means. That one alone tips the scales for me.

Sure, preset programming is both great and terrible in boths worlds; sometimes in the included presets of one single synth. But there is a depth odf programmability in software that's difficult to approach with an all-in-one hardware controller, as there needs to circuitry to reflect this, adding to the cost, or a way to easily change and program setups using more than just a pair of increment buttons and a small monochrome display screen. Strip away all of the bells and whistles of the 'marketing' presets, and get down to brass tacks with the sound engines. I think that becomes an eye-opener for anyone from either 'camp'.

I hate to divide this so neatly into two 'camps'; it's obvious that many here use a combination of hardware/software to varying degrees. But this is the mind-set that a 'versus' thread sets you up for: One has got to be better than the other (or does it?). For the record, I still have a lot of hardware that can't be specifically duplcated for one reason or another in software. I keep it around, I'm still using it, but I find that I'm calling on it less & less. I continue that tactic until there's no longer a need to do so.

Who knows? In a few years, when computing power increases, software developers stretch to fill it, and a neKo-type workstation becomes truly affordable, we'll no longer be having this debate: Hardware becomes Software becomes Hardware. In the interim, I'll use anything at my disposal to achieve the sound I need (sometimes "not-so-good" is good enough, thanks to masking.). I love the all-in-ones for what they are, but if anyone were to force me into an either-or situation, this is the one time that I'd be in favor of going totally soft.

ORIGINAL: WhyBe
People shouldn't take the "zero-latency" statement so literal. In a totally software environment near-zero latency isn't possible without some sort of compromise/workaround (all instances of a softsynth active at once, for example). That becomes a musical hinderance. In a hardware environment, it's virtualy a non-issue.

You're assuming softsynths always run on a single computer with a single CPU. Todays workstations are inexpensive enough that you can stack multiple instances of a synth more than one CPU--neigh, more than one workstation and still have negligble latency that will not affect performance. As Tom said, this is prohibitive with strictly hardware (I'd be willing to run another machine to run 30 instances of Dimension between them but I'm not going to buy 30 Fantoms).

People shouldn't take the "zero-latency" statement so literal.

Agreed, Yon. That's one reason that an ancient serial protocol like MIDI is still with us: it's good enough. But we're still dealing with a physiological threshold of about 10 ms. Trained ears can certainly distinguish less, but only as a difference between an original sound source and the 'latent' one. Make it a single source, and we're starting with a 10 ms. base value.

I come close to that on small projects, and have to bump it up some as the projects grow in complexity. I think that it's fair to say that hardware synths do better in this department than soft-synths, but we're only talking a handful of milliseconds. That's barely enough to produce a perceivable comb-filtering/flanging effect without some sort of delay animation, and that even becomes a non-issue when time-based effects are added to the mix in either 'camp'.

I'm not trying to stir things up, Yon; just adding to my opinion here. Hardware surely does perform better than software in this respect, but this difference alone doesn't offset the disadvantages for me. To be on an even playing field, one must compare a single [if layered] instance of a soft-synth to as close to an equal patch that's 'duplicated' in a hardware workstation, if that's even possible.

Thanks for the healthy give-and-take discussion. It's certainly been an informative thread, and somewhat forces the readers into determining where they themselves may stand on the issue. And that can result in nothing but good things.

Musician A - Room full of hardware synths......my numbers hit $50,000 easy

Musician B - Equal money on PC hardware and software?

For that kind of money you could simply contract an engineer to build a custom system with performance stats that just plain anihilate anything in the synth industry. Use the leftover money to buy a few TerraBytes worth of soft synths and run them all at once!

Or something like that.