Kurzweil CC-1 Continuous Control Pedal 89.00. No interest if paid in full within 12 months. TipTop Audio Valhalla Shimmer Reverb ZDSP Cartridge 75.00.Compare.Click to expand.They TOTALLY program DSP in assembly language!05.09.20. 6-month Special Financing Available. Z-DSP Effects Cartridge with 8 Onboard Algorithms Including 6 Shimmer Reverb Types and 2 Special Pitch Effects. Tiptop Audio Valhalla DSP Shimmer Reverb Cartridge for Z-DSP. No interest if paid in full within 12 months. TipTop Audio PERC Card 26.00.
![]() Valhalla Shimmer Cartridge Full Within 12If I told you to pull over on the side of the interstate and get out of your car. The efficiency of the code and the way the processor can read and execute it is generally the biggest detriment to speed.One professor put it this way. It explains to the processor where to drive. The car itself can go 200, but that's in a straight line, on a perfect road, without traffic, and all systems operating at an optimum level.Coding is the road. Limitations, to be sure, but limitations can inspire creativity.Think of processor or clock speed as a sports car that can go 200mph. Once you spend a few months to a year working with the language, it is fairly easy to work with. Star shower movie magicThere can be elegant simplicity in code and most of the time that's a good thing. Honestly, I've seen coders that have an artfulness to their approach. How that code is written and how quickly the processor can actually read and respond to the code is the hard part. It would pull over and get out of the car going 70mph.All of that information in between those two responses is coding. Look to make sure another car wasn't coming, and then finally get out of your car.If I told a computer to pull over and get out of the car. On a fixed point processor, this can result in internal distortion, and this can cause the output to sound weird (and deviate away from actually being allpass). This is a HUGE deal, and comes down to how the memory system is implemented.- Calculates two ramp and two sine/cosine LFOs- Calculates the delay interpolation coefficients for any delays that are being interpolated with these LFOsIn addition, there are all sorts of operations that take 1 or 2 cycles, that would take more cycles on general purpose processors.1 cycle: Delay read w/gain coefficient (a delay MAC, if you will), delay write, reading pot valuesThe drawback to having all this stuff built in, is that calculations that aren't built into the hardware can be FAR more expensive.For example, Direct Form II allpass delays have very high internal gains. It's always "double nickels on the dime." However, the processor itself is pretty smart, so if you ask it to pull over and get out, it knows to pull over while looking for other cars, slow down, stop, all that.In actual reverb terms: The FV-1 (and a lot of older reverb hardware, and possibly some newer reverb hardware) does a lot of things automatically on a per sample basis:- Increments all delay pointers. I'm gonna read up on DSP and assembly.Click to expand.To add to your post: Something like the FV-1 can't ever go 200 MPH. It also allows you to throw more resources at a solution to get it to sound better, instead of stepping back and thinking about if a different approach would sound better (at least this is my experience working with open ended floating point processing).To add to your post: Something like the FV-1 can't ever go 200 MPH. Which is cool, but sometimes a blank piece of paper is intimidating. It can do pretty much anything. Instead of starting with an algorithm, and trying to shoehorn it into the processor, it is best to start with a good understanding of what the FV-1 is good at and what it is bad at, and think about what would work best there.A 333 MHz SHARC, by comparison, is a blank piece of paper. Scattering matrices for FDNs are doable, but they aren't particularly cheap, especially compared to the cost of allpass delays.The above strengths and weaknesses, in my experience, end up shaping what is actually done on the FV-1. So, if you have an algorithm that requires allpass coefficients that are fairly close to +/-1.0, you need to figure out a good solution.Other examples of "expensive" operations on the FV-1 are good random noise generation, and truly random delay access (for granular). You won't hear bad resonance, metallic artifacts as chorusing or randomization were generally created to avoid them. If you hear a whole mix thru it (rather than a single instrument) you'll hear the big difference the 70/80/81 Concert Hall is more an instrument/single track kind of reverb.A whole mix looses focus, strength and the modulation isn't nice for all the tracked sources in the mix.Random Hall makes the mix come together and the delays randomization doesn't appear to sound like a chorus. So chorusing and randomization are taken to a stronger level there.Random Hall is a different thing. The PCM80/81 is true stereo I/O and also has Spin. ![]() It only uses the first three programs of Echo+Reverb, Echo/Repeat+Reverb, and Chorus+Reverb.
0 Comments
Leave a Reply. |
AuthorRenate ArchivesCategories |