Nyquist in Audacity is much slower on Linux than on Windows XP

[Stevethefiddle]

When running Audacity on Windows XP as a guest OS in Virtualbox,
Nyquist plug-ins run very much quicker than they do on the host Linux
(Debian Squeeze) machine. I've not measured the difference but I'd
estimate that Nyquist is running about 500% faster on the Windows
guest - or to put it another way, dog slow on Linux.

Currently I'm running Audacity 2.0.2 on both OS's, but this has been
the case with much older versions too.

Does anyone know why this is or if anything can be done to make
Nyquist run quicker on Linux?

Steve

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Audacity-nyquist mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/audacity-nyquist

[Gale]

| From Steve the Fiddle <[hidden email]>
| Tue, 25 Sep 2012 06:06:41 +0100
| Subject: [Audacity-nyquist] Nyquist in Audacity is much slower on Linux than on Windows XP

> When running Audacity on Windows XP as a guest OS in Virtualbox,
> Nyquist plug-ins run very much quicker than they do on the host Linux
> (Debian Squeeze) machine. I've not measured the difference but I'd
> estimate that Nyquist is running about 500% faster on the Windows
> guest - or to put it another way, dog slow on Linux.
>
> Currently I'm running Audacity 2.0.2 on both OS's, but this has been
> the case with much older versions too.
>
> Does anyone know why this is or if anything can be done to make
> Nyquist run quicker on Linux?

I thought this list was closing?

Is it possible to compare the Nyquist IDE (rather than Audacity-Nyquist)
on the two platforms? Is the result the same?



Gale  

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Audacity-nyquist mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/audacity-nyquist

[rbd]

On 9/26/12 10:46 AM, Gale Andrews wrote:

> | From Steve the Fiddle <[hidden email]>
> | Tue, 25 Sep 2012 06:06:41 +0100
> | Subject: [Audacity-nyquist] Nyquist in Audacity is much slower on Linux than on Windows XP
>> When running Audacity on Windows XP as a guest OS in Virtualbox,
>> Nyquist plug-ins run very much quicker than they do on the host Linux
>> (Debian Squeeze) machine. I've not measured the difference but I'd
>> estimate that Nyquist is running about 500% faster on the Windows
>> guest - or to put it another way, dog slow on Linux.
>>
>> Currently I'm running Audacity 2.0.2 on both OS's, but this has been
>> the case with much older versions too.
>>
>> Does anyone know why this is or if anything can be done to make
>> Nyquist run quicker on Linux?
> I thought this list was closing?

I thought so too, but here we are.

The speed of Nyquist should be very similar on both. One big performance
issue I've run into is that any signal that decays (exponential
envelopes, feedback delay, reverb) eventually becomes denormalized, and
depending on how things were compiled, the cpu, etc., this can invoke a
software handler at which point floating point becomes VERY slow. I
should probably make a simple test to see if this is happening. It would
also make sense to make some simple tests to see if the speed difference
is related to allocation, floating point, the XLISP interpreter, etc.

-Roger

>
> Is it possible to compare the Nyquist IDE (rather than Audacity-Nyquist)
> on the two platforms? Is the result the same?
>
>
>
> Gale
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Audacity-nyquist mailing list
> [hidden email]
> https://lists.sourceforge.net/lists/listinfo/audacity-nyquist

------------------------------------------------------------------------------
How fast is your code?
3 out of 4 devs don\\\'t know how their code performs in production.
Find out how slow your code is with AppDynamics Lite.
http://ad.doubleclick.net/clk;262219672;13503038;z?
http://info.appdynamics.com/FreeJavaPerformanceDownload.html
_______________________________________________
Audacity-nyquist mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/audacity-nyquist

[Stevethefiddle]

>> I thought this list was closing?
> I thought so too, but here we are.
since it's not closed yet ....


> The speed of Nyquist should be very similar on both.
but it's not.
Running the following in the Nyquist Prompt effect:

(dotimes (i 10000000)
(setq x i))
"done"

Windows XP Guest in Virtualbox: about 2.5 seconds.
Debian host: about 12 seconds.

Steve

[rbd]

On 9/26/12 9:18 PM, Stevethefiddle wrote:

>>> I thought this list was closing?
>> I thought so too, but here we are.
> since it's not closed yet ....
>
>
>> The speed of Nyquist should be very similar on both.
> but it's not.
> Running the following in the Nyquist Prompt effect:
>
> (dotimes (i 10000000)
> (setq x i))
> "done"
>
> Windows XP Guest in Virtualbox: about 2.5 seconds.
> Debian host: about 12 seconds.
>
> Steve

This is a great test because there's no signal processing. What happens
here is:
     memory allocation and garbage collection (all but small integers go
on the heap)
     occasional calls to oscheck() to check for interrupts
It's possible, depending on compilers and alignment, that XLISP cells
are bigger on Debian.
It would be really helpful to profile the code. Also, it would be
interesting to run the same in nyquist on both systems (not within
audacity).

-Roger



------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://ad.doubleclick.net/clk;258768047;13503038;j?
http://info.appdynamics.com/FreeJavaPerformanceDownload.html
_______________________________________________
Audacity-nyquist mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/audacity-nyquist

[Stevethefiddle]

Stevethefiddle wrote

Running the following in the Nyquist Prompt effect:

(dotimes (i 10000000)
(setq x i))
"done"

Windows XP Guest in Virtualbox: about 2.5 seconds.
Debian host: about 12 seconds.

Standalone Nyquist in Debian:

(defun test ()
(dotimes (i 10000000)
(setq x i))
(print "done"))

{note: this counts to 10,000,000}
about 6 seconds - counting to 1,000,000 was too fast to time, about half a second.

Steve

[rbd]

On 9/26/12 9:43 PM, Stevethefiddle wrote:

> Stevethefiddle wrote
>> Running the following in the Nyquist Prompt effect:
>>
>> (dotimes (i 10000000)
>> (setq x i))
>> "done"
>>
>> Windows XP Guest in Virtualbox: about 2.5 seconds.
>> Debian host: about 12 seconds.
> Standalone Nyquist in Debian:
>
> (defun test ()
> (dotimes (i 10000000)
> (setq x i))
> (print "done"))
>
> {note: this counts to 10,000,000}
> about 6 seconds - counting to 1,000,000 was too fast to time, about half a
> second.
>
> Steve
>

Of course, the standalone runt ime should match the Audacity runtime, at
least in this case, where there's no substantial interaction with
anything outside of XLISP. So now we have 2 disparities to resolve. I
think the next step is profiling under Debian to see where the time is
going. I assume everything here is compiled for 32 bit architectures.
There might be a big difference between running 64-bit vs 32-bit code
(not to mention that Nyquist doesn't support 32 bit code -- but actually
the latest version in SVN is running under 64-bit OS X and I'm in the
process of testing others.)

-Roger


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://ad.doubleclick.net/clk;258768047;13503038;j?
http://info.appdynamics.com/FreeJavaPerformanceDownload.html
_______________________________________________
Audacity-nyquist mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/audacity-nyquist

[Stevethefiddle]

Yes this is 32-bit Debian Squeeze.

Profiling to see where the time is going sounds like a good approach but sadly I'm not a programmer and I've no idea how to do that.

Steve

[rbd]

On 9/26/12 11:33 PM, Stevethefiddle wrote:
> Yes this is 32-bit Debian Squeeze.
>
> Profiling to see where the time is going sounds like a good approach but
> sadly I'm not a programmer and I've no idea how to do that.
Thanks for your help -- I'll look at this. -Roger
>
> Steve
>

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://ad.doubleclick.net/clk;258768047;13503038;j?
http://info.appdynamics.com/FreeJavaPerformanceDownload.html
_______________________________________________
Audacity-nyquist mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/audacity-nyquist