Apollo Sharc Performance Questions

I am trying to understand how the Apollo units handle Sharc processing power when tracking live through Console.

1. When you assign a plugin to an input, my understanding is that a single processor is dedicated to that input. Can you exceed this ceiling and assign a second processor to the input? Some plugins are very demanding on processing resources.
2. If a processor is dedicated to an input and only takes up 10% of the processor power, is the remainder left unavailable to other tasks, thereby rendering it wasted?
3. If the first two interpretations are correct, could live tracking with four inputs utilizing 10% of each Sharc in an Apollo Quad capture the full capacity of the unit, thereby preventing additional inputs from using UAD plugins even though 10% of the processing power being tapped into?
4. When utilizing two Apollo Quad units cascaded via Firewire, are the Sharc processors available as a collective when tracking live through console or are they only available to their respective Apollo unit? If they are captive to the units themselves in a Firewire environment, does this limitation change with a Thunderbolt connection?

I want to be sure that I am fully aware of limitations before I purchase the unit(s). The worst case scenarios above would be incredibly disappointing.

I'm not really sure about every question, but I think the answer to all your questions is: no.

syah said:

I'm not really sure about every question, but I think the answer to all your questions is: no.

Click to expand...

So you cannot use more than processor on a single input but you can spread a single processor out over several inputs? So theoretically with a single Apollo Quad you could use 50% equivalent of a Sharc processor simultaneously on eight inputs?

If the answer to the last question is no then the additional Sharc units on the second Apollo could only be utilized for live tracking on it's internal inputs, correct? If working from the DAW in lieu of the console are the processors of both units available as a collective? In other words, would I have 16 Sharcs available from the DAW if I own an Octo card and two Apollo Quads? In this DAW scenario, would there be any additional latency applied beyond what I currently experience with the Octo card alone? Is the processing power pulled from the cards first with any additional resources pulled secondarily from the Apollos? Again, this is in a Firewire environment.

Poesque said:

So you cannot use more than processor on a single input but you can spread a single processor out over several inputs? So theoretically with a single Apollo Quad you could use 50% equivalent of a Sharc processor simultaneously on eight inputs?

Click to expand...

That's my understanding. I have a Quad Duo and thusfar I record one channel at a time. That channel fully utilizes a single chip and my auxes use other sharcs. So, I'm not 100% sure about your question, but I think I have read that somewhere on this forum.

Poesque said:

If the answer to the last question is no then the additional Sharc units on the second Apollo could only be utilized for live tracking on it's internal inputs, correct?

Click to expand...

Yes. Or you can use it from the DAW in live track mode, but that is thunderbolt only and still has too much latency for most applications.

Poesque said:

If working from the DAW in lieu of the console are the processors of both units available as a collective? In other words, would I have 16 Sharcs available from the DAW if I own an Octo card and two Apollo Quads?

Click to expand...

Yes.

Poesque said:

In this DAW scenario, would there be any additional latency applied beyond what I currently experience with the Octo card alone?

Click to expand...

The latency is compensated by your DAW, so you don't experience anything of it. Maybe you will hear audio a fraction of a second later when you press play, because the latency is larger, but that is not a real problem in my experience

Poesque said:

Is the processing power pulled from the cards first with any additional resources pulled secondarily from the Apollos?

Click to expand...

No. Random order.

Poesque said:

Is the processing power pulled from the cards first with any additional resources pulled secondarily from the Apollos? Again, this is in a Firewire environment.

Click to expand...

it is not, once you open a project in your daw and the plugins get loaded, it seems they load up on the whole availabe dsp sharc processors
You can avoid that they load up on the apollo if you switch it off the uad-2 activity (blue button on the left side next to the icon) once you load the project and later switch it on again.
to me it looks like the dsp power gets filled up without any smart system behind it.. as you can see in my screenshot, there is still total 56% of dsp power but its pretty tight on loading cpu intense plugins like the neve 1073.
it would be a great feature if they could implement a cpu resource sorting feature, that analyse which plugins are running on which cpu and moving them together to shovel space for more cpu-hungry plugins


i´m not sure about your other question, but regarding the console , its only possible to use 1 sharc processor on one channel, not 2.. but the remaining free dsp power from that 1 processor will be used on other plugins on different channels until its full, so the processing power defintly can be utilized to a certain maximum

exoslime said:

to me it looks like the dsp power gets filled up without any smart system behind it.. as you can see in my screenshot, there is still total 56% of dsp power but its pretty tight on loading cpu intense plugins like the neve 1073.
it would be a great feature if they could implement a cpu resource sorting feature, that analyse which plugins are running on which cpu and moving them together to shovel space for more cpu-hungry plugins

Click to expand...

There is actually a coded DSP algorithm that dictates how the DSP's are allocated to each SHARC chip so they load fairly evenly. As new plug-ins get added the allocation will change accordingly. If you are trying to add a DSP heavy plug-in that won't fit as allocated. You can try saving and re-opening a project with one of the UAD devices (blue icon as you noted earlier), not enabled. Once project is loaded, re-enable the empty UAD device to load the new plug-in onto it. UA is known to tweak the DSP loading algo from time to time, as needed.

Best regards,

Billy Buck

if you fill up one channel you can send it to another and have more than 4 plugs in line .

billybk1 said:

There is actually a coded DSP algorithm that dictates how the DSP's are allocated to each SHARC chip so they load fairly evenly. As new plug-ins get added the allocation will change accordingly. If you are trying to add a DSP heavy plug-in that won't fit as allocated. You can try saving and re-opening a project with one of the UAD devices (blue icon as you noted earlier), not enabled. Once project is loaded, re-enable the empty UAD device to load the new plug-in onto it. UA is known to tweak the DSP loading algo from time to time, as needed.

Best regards,

Billy Buck

Click to expand...

thanks billy, thats quite interesting, i noticed on a couple of projects i´m working on, for example i have added a new aux fx bus, lets call it "#8" with Ocean Ways Studio plugin on it, then working on the session, save it, and then close it. The next time i reopen it, i get an error message having not enough DSP resources free (and OWS is disabled), but according my general dsp overview there should be plenty dsp resources left, but the truth is, not on a single sharc processor!
i found a workaround for this kind of situation: i move the #8 FX Bus with Ocean Way Studio at FX Bus #1, so when the project loads, the FX Bus #1 still gets a nice big chunk of free ua-resources and everything loads fine.