Is Pro Tools behind the curve?

[Philthy]

Originally Posted by JFreak

RISC usually has less realtime problems


Sent from my iPhone using Tapatalk

Thanks, well that could be good in the long run then!

[Michael Carnes]

Originally Posted by JFreak

RISC usually has less realtime problems


Sent from my iPhone using Tapatalk

Don't think RISC/CISC has any bearing on realtime performance. It's been a long time since there was a pure RISC chip. The simple nature of the architecture often meant that it took a lot of instructions to get something done that could be accomplished by a single instruction build into a wider instruction word. So a lot of the theoretical performance improvements remained only theoretical.

[JFreak]

Was talking about realtime performance. As you said, performance of processing plugin for 1sec may be more efficient with CISC processors which also need less memory, but for realtime - say one sample - this is closer to real DSP as the instructions are shorter and faster.

[deanrichard]

I don't know any specifics about writing code for those chips. But I do know that, in the end, the real time performance almost always depends on the skill and expertise of the programmers writing the drivers and low level code needed.

I don't know where Avid's PT engineering team is located any more (a few years back it was said that the code base moved from Ukraine to Poland) and I don't know how much that engineering team knows about writing low level code for those processors, but let's hope they know a lot or can learn quickly.

Originally Posted by JFreak

Was talking about realtime performance. As you said, performance of processing plugin for 1sec may be more efficient with CISC processors which also need less memory, but for realtime - say one sample - this is closer to real DSP as the instructions are shorter and faster.

[JFreak]

The only low level code Avid needs to write are the drivers. Everything else is complied using the SDK they have zero control over. Here is the big difference between Native and DSP plugins, for example; the DSP plugins are compiled using Avid SDK and they can control how to use the chips.

[Michael Carnes]

Originally Posted by deanrichard

I don't know any specifics about writing code for those chips. But I do know that, in the end, the real time performance almost always depends on the skill and expertise of the programmers writing the drivers and low level code needed.

Certainly true. "Realtime" is a term you have to define for each specific situation. In general it applies to the maximum amount of time a system can take to respond to a stimulus. The stimulus might be a button push, a rising wave in an EKG, etc. When designing a system, you have to determine what that amount of time must be, how long the reaction must take and (ideally) what happens if you fail. So the term may mean one thing in a DAW and another entirely in an automobile. That will guide your code writing and your choice of processor.

[Bob Olhsson]

I'll never forget a Stanford Music Special Interest Group meeting where an Intel contractor slipped up and revealed that the Pentium CPU was actually RISC under the hood. SillyConMan Valley marketing has always been pretty bizarre.

[Darryl Ramm]

Originally Posted by Bob Olhsson

I'll never forget a Stanford Music Special Interest Group meeting where an Intel contractor slipped up and revealed that the Pentium CPU was actually RISC under the hood. SillyConMan Valley marketing has always been pretty bizarre.

Well they really are CISC to mere mortal developers. Intel has had some bizzare approaches to stuff, like at times seemingly trying to deny the processors were microcoded. (if you acknowledge that then then yes the architecture might look more RISC like). I saw a bizarre spat where Intel developer support was telling us the processor were not microcoded. IIRC we sent them a copy of their microcode update included in an OEM firmware fix with a "what's this then" note

[Michael Carnes]

Heck. Even bits of the 68000 were microcoded.

[Darryl Ramm]

Originally Posted by Michael Carnes

Heck. Even bits of the 68000 were microcoded.

Yep it was really bizarre, but they really seemed to not want to talk about microcode for quite a while. In our case our co-founder/architect gave their support folks every clue that we uh might know what we are talking about.. eventually got to a senior person at Intel who was very apologetic for folks wasting our time. We had discovered a subtle CPU bug but noticed it did not happen on all versions of a CPU and then noticed that Intel seemed to have fixed it in some microcode updates that some OEMs had access to.. we'd even unpacked that microcode out of a Dell update. We were just trying to work out how to tell if it was installed/how to tell customer what to do. (... early days of VMware).