Is XITE applicable for me ?

[dante]

I just saw that recently a member of the forum (fra) has maxed out the DSPs of XITE for the first time by developing modified instruments dedicated for each numbered DSP in XITE. This is why I regard XITE's firmware buggy and unsupported.

Sorry my best effort remains to wait for a scope 6 improvement.

I doubt maxing out DSP's has much to do with firmware. Correct me if I'm wrong anyone but I was under the impression that the DSP's are loaded via the host Scope 'software'.

If Scope 6 features for you are the tipping point in a buying decision then that's ok, in fact it would be good thing if Scope 6 generated more sales for S|C so I'm all for that concept. Even more bonus for existing users !

[jksuperstar]

That's because those instruments have massive amounts of internal routing (many LFOs, envelopes, sequencers, routed to oscillators, each other, multiple fitlers, etc). Routing that many control signals between DSPs, given the amount of power of each DSP (and the number of voices attainable), is simply beyond the hardware architecture.

So the "work around", was not a specialized device for each DSP, but rather a limited number of voices set to each DSP, with common controls going to all DSPs. This kept all the massive routing *inside* a given DSP, not between them, and allowed for more voices to be attained.

[fra77x]

Dante than man doesn't knows what he is talking about.

What has "firmware" has to do with dsp allocation? In any multi processor system there is a procedure that allows communication between the processors. That needs some resources. Until some years ago most native systems had only one processor. Nowadays we have multiprocessor systems as cores and programmers assign different threads so the os can assign these threads to different processors. Windows is not a realtime os. Scope is. Scope has to accomplish anything in 1 ms in 96kHz lower ulli setting. Scope can load a full XITE (18 dsps) and do it well with manual assignment. Scope was a multiprocessor system way back from 1995. What else is different? Scope is a multiprocessor system that communicates at the same time with the host computer. That needs communication channels. Beneath that, it is not a fixed system but lets you use whatever you need anytime you 'd like. It would be very easy to fullfil the dsps with a certain configuration and say goodbye to "dsp" allocation problems. But you 'll had a fixed behavior and function after that. There is no magic anywhere as we know, except in minds without knowledge. Express card, or the PCI have/had a certain amount of bandwidth and that also depends on other aspects of the hardware/software configuration. The dsp allocation is controlled by an algorithm. That should decide during project changes, where each module should be loaded and make a balance between dsp communication channels and host->scope communication channels. That is almost impossible because hardware specifications are different and there is no way for the algorithm to know exactly what is the available communication resources. If anyone think with some understanding that matter will conclude that such an algorithm has a very difficult job to take after or it would take too long and that will lead again to some "user-unfriendly" case. So the automatic dsp allocation algo leaves unused some of the dsps or "complains" often. That is a trade off for user friendlyness and to the "i connect things as i like behaviour" which is no professional because it doesn't knows what it does. (it also introduces phase problems). At the same time the manual allocation lets you use all the dsp power if you follow the demands of the hardware and the communication with the host computer.
Everything else is magic. An audio signal at a specific sample rate is a reality. Communication with memory is a reality. Virtual cables that carry signals around are carrying real stuff. Connecting things and making channels in a native software is happening inside specific processors and happen with no or few cost. The same thing happens inside a specific scope dsp. Travelling signals around have a processing cost. That is all what users experience. They are just using an external dsp hardware which has it's own needs and is different from what they are used to from native processing. All external dsp farms from all companies have analogous behaviours. Nord modular complained very often about memory/dsp amount. Every professional producer, mixer, audio engineer, musician knows how to adapt to their tools and understands that their is no reason to complain about things that everybody would like to be better but at certain situations trade offs are anavoidable.

[Bud Weiser]

Dante than man doesn't knows what he is talking about.

Yep,- this is the assumptions thread.

B.t.w., your post below is excellent value and very informative.

Express card, or the PCI have/had a certain amount of bandwidth and that also depends on other aspects of the hardware/software configuration. The dsp allocation is controlled by an algorithm. That should decide during project changes, where each module should be loaded and make a balance between dsp communication channels and host->scope communication channels. That is almost impossible because hardware specifications are different and there is no way for the algorithm to know exactly what is the available communication resources.

Now I wonder,- when using p.ex. a socket 2011 mobo, Sandy Bridge E / Ivy Bridge E processor and fast RAM,- can I expect better behaviour/ more communication channels because of more lanes/data throughput on the computer mobo and much more processor L3 cache,- or is it the XITE-1 PCIe card hardware setting the terms (or being a potential bottleneck) because of it´s PCIe x1 design ?

... which is no professional because it doesn't knows what it does. (it also introduces phase problems). At the same time the manual allocation lets you use all the dsp power if you follow the demands of the hardware and the communication with the host computer.

Communication with memory is a reality.
Virtual cables that carry signals around are carrying real stuff.
Connecting things and making channels in a native software is happening inside specific processors and happen with no or few cost.

The same thing happens inside a specific scope dsp.
Travelling signals around have a processing cost.

I understand all very well, but I believe we´re talking about limited ressources of the host machine as well as the limited ressources of XITE-1 itself.

When getting error messages,- most are complaining about the communication channels (SAT connections) not the available DSP power (DSP %).

That´s why I asked for speeding up the computer hardware above.

Inside XITE-1, usage of internal RAM might help in addition.

Otherwise,- I don´t know how much headroom in regards of data throughput a PCIe x1 card offers compared to the demand of communication channels needed w/ a fully loaded XITE-1 and some count of ASIO channels and some sequencer MIDI drivers running.
Could that be better w/ a PCIe x4 or x8 card in use or is it overkill,- I don´t know but would like to hear about ?

best

Bud

[fra77x]

It is Pcie x1 internally.

I have checked my xite with different laptops and i have found differences at the SAT performance. Some laptops are utilizing < 250Mb express speed. Perhaps (i have not thouroghly tested) some laptops can offer the max speed (= 250 Mb/s) and i 'm not really sure what that translates to actual communication channels.

The current laptop i use is a fujitsu i-5 which has inferior SAT performance compared to a lenovo i have used some time ago. I work at 96kHz which means i have to be very careful on connections because i easily fall on SAT errors. But for my needs it's ok. Of course i always use the Xite optimized versions of Masterverb and for delay fx (i will upload that fx soon). And very small asio if any. When i want asio performance i create a different project. I use VDAT. (in my case it is possible to avoid asio because i have a scope PCI system -21 dsps- which i use as a sampler etc). I use reaper as a midi sequencer, or Logic 5.5 as a pure midi sequencer through XP VirtualBox with cooperlan.
I know that is an idiosyncratic setup but that's what i feel comfortable with. I have no other instruments. My music - production is full Scope powered. With Xite i can finally do my whole music production without bouncing and have everything on midi or on sampler. I do a whole bunch of different styles, but my main style is psy trance. I'm really satisfied with the result.

[dante]

Is any of your psy-trance on Soundcloud ?

[jksuperstar]

It is Pcie x1 internally.
I have checked my xite with different laptops and i have found differences at the SAT performance. Some laptops are utilizing < 250Mb express speed. Perhaps (i have not thouroghly tested) some laptops can offer the max speed (= 250 Mb/s) and i 'm not really sure what that translates to actual communication channels.

My Lenovo W520 has a PCIe 2.0 expresscard slot (or maybe 2.1). Through masterverb tests compared to another laptop of mine that is PCIe 1.0, the bandwidth available was increased. (PCIe Expresscard slot is always 1x lane, but PCIe 2.0 is 500MB/s, while PCIe 1.0 is 250MB/s).

Here is a link to a previous conversation:
http://forums.planetz.com/viewtopic.php ... 1x#p277951