While I think of it, could I just throw in that double precision analysis
is MUCH more interesting to anyone implementing on a modern CPU?
For embedded stuff, single precision and fixed are obviously the big
options (though I can't remember when/why I would have tried implementing a
Actually, I'll go one further: In 2013, single precision is just time
wasting. It's a pathological case for analysis, but it shouldn't
represent
real-world usage.
I'm reminded of a conversation I had with my PhD supervisor 12 years ago,
when showing him some source which caused him to
On 12/11/2013 7:40 PM, Tim Blechmann wrote:
some real-world benchmarks from the csound community imply a performance
difference of roughly 10% [1].
Csound doesn't have a facility for running multiple filters in parallel
though does it? not even 2 in parallel for stereo.
4 biquads in
On Tuesday, November 12, 2013, Vadim Zavalishin wrote:
On 12-Nov-13 09:05, Dave Gamble wrote:
Actually, I'll go one further: In 2013, single precision is just time
wasting. It's a pathological case for analysis, but it shouldn't represent
real-world usage.
I'm reminded of a conversation I
On 12-Nov-13 09:53, Dave Gamble wrote:
PS. Time-varying performance is another word. Nonlinearities is the
third one.
Not criticisms I'm at all familiar with, I'm afraid. Can you expand?
As we are talking about inferiority of DF compared to ZDF, I just
mentioned the other two, which are
Heya,
On Tuesday, November 12, 2013, Vadim Zavalishin wrote:
On 12-Nov-13 09:53, Dave Gamble wrote:
PS. Time-varying performance is another word. Nonlinearities is the
third one.
Not criticisms I'm at all familiar with, I'm afraid. Can you expand?
As we are talking about inferiority of
On 12-Nov-13 10:01, Dave Gamble wrote:
Because switching from double to float will bring extremely small
performance gains in CPU cost, and potentially sizeable problems with
numerical issues.
I'd be very careful with statements like that. There are people with
exactly the opposite
Hi Ross,
The exact answer depends on the exact hardware. It's pretty trivial on the 56k,
of course (the 56-bit accumultor works automatically with the MAC instruction,
quantization happens automatically when saving to 24-bit, just take the
difference and feed it back for the noise shaping). If
On Tuesday, November 12, 2013, Vadim Zavalishin wrote:
On 12-Nov-13 10:01, Dave Gamble wrote:
Because switching from double to float will bring extremely small
performance gains in CPU cost, and potentially sizeable problems with
numerical issues.
I'd be very careful with statements like
On 12-Nov-13 10:10, Dave Gamble wrote:
So let me go out on a limb here: if you take some single precision code and
up it to double, and things get WORSE then there is something very strange
about your original code that merits investigation.
It's very easy. As I mentioned in my other email,
In some cases error feedback methods for floating point would be interesting
if they exist.
The problem is that the error is lost in the floating point hardware—you put in
two floats and get back float of the same size. Something fell into a bit
bucket that you don't have access to.
One word: SIMD
well, when benchmarking my performance code, about 2% of the CPU time is
spent in vector code, while about 60% is spent in scalar filter code.
Hi Tim
I can't believe vector code is running 30 times faster than the scalar
code :-D :-D :-D
scalar/vector parts of the same
On Tue, 12 Nov 2013 10:40:25 +0200, Tim Blechmann t...@klingt.org wrote:
well, when benchmarking my performance code, about 2% of the CPU time is
spent in vector code, while about 60% is spent in scalar filter code. of
course one can run 4 parallel single-precision filters or 2
double-precision
IMHO, a piece of DSP code is generally used in something else, rarely on its
own. Whether it ends up as part of a bigger project or in a plugin, it'd
better not waste CPU.
All I know is that SIMD is made for DSP, whether it's sound or image
processing it's where most of the CPU goes, so if
As an alternative perspective on this, consider the idea that creating a
shipping piece of DSP code requires two (hopefully) distinct phases-
1. Understanding what you're doing, and making it work.
2. Optimising it for a production environment.
I claim that introducing SIMD in 1 is just over
On 12.11.2013, at 10:16, Vadim Zavalishin
vadim.zavalis...@native-instruments.de wrote:
It's very easy. As I mentioned in my other email, switching from float to
double halves the number of available SIMD channels, which means you need to
run your code twice as many times.
Right. But with
Suppose we take a simple 1st order low-pass (or high-pass) filter in the
digital domain, and compare it with the corresponding electronic
implementation, preferably as a network theory example, ignoring
electro-magnetics, Johnson noise, and physical delay of currents through
wires, which are
Hi,
I'm going to have a go at answering, but there will be points where I'm not
sure exactly what you're saying. I'll be as clear as possible and request
clarification.
On Tue, Nov 12, 2013 at 3:56 PM, Theo Verelst theo...@theover.org wrote:
Suppose we take a simple 1st order low-pass (or
On Tue, 12 Nov 2013 16:56:27 +0100, Theo Verelst wrote:
Anyone want to explain to me what their most or least favorite
implementation
of thus simple, single order filter without feedback in the digital domain
will do when we feed it with an impulse, a shifted step (occurring between
samples),
On Tue, Nov 12, 2013 at 4:48 PM, vadim.zavalishin
vadim.zavalis...@native-instruments.de wrote:
PS. Speaking of EE, I believe this simple RC filter *does* have
feedback, from the capacitor's voltage via the resistor's voltage to the
capacitor's charging current. Which is also immediately
Dave Gamble wrote:
On Tue, Nov 12, 2013 at 3:56 PM, Theo Verelsttheo...@theover.org wrote:
Suppose .
So lets take the equivalent of this simple filter in the digital domain
(e.g. as function of a z^-1 network, and see what happens.
Well, there's obviously no direct equivalent. It comes
Dave Gamble wrote:
On Tue, Nov 12, 2013 at 4:48 PM, vadim.zavalishin
vadim.zavalis...@native-instruments.de wrote:
PS. Speaking of EE, I believe this simple RC filter *does* have
feedback, from the capacitor's voltage via the resistor's voltage to the
capacitor's charging current. Which is
Hi,
On Tue, Nov 12, 2013 at 4:52 PM, Theo Verelst theo...@theover.org wrote:
Dave Gamble wrote:
On Tue, Nov 12, 2013 at 3:56 PM, Theo Verelsttheo...@theover.org
wrote:
Suppose .
So lets take the equivalent of this simple filter in the digital
domain
(e.g. as function of a z^-1
Heya,
On Tue, Nov 12, 2013 at 4:54 PM, Theo Verelst theo...@theover.org wrote:
vadim.zavalishin wrote:
On Tue, 12 Nov 2013 16:56:27 +0100, Theo Verelst wrote:
Anyone want to explain to me what their most or least favorite
implementation
of thus simple, single order filter without
On Tue, Nov 12, 2013 at 4:56 PM, Theo Verelst theo...@theover.org wrote:
Dave Gamble wrote:
On Tue, Nov 12, 2013 at 4:48 PM, vadim.zavalishin
vadim.zavalis...@native-instruments.de wrote:
PS. Speaking of EE, I believe this simple RC filter *does* have
feedback, from the capacitor's
On Tue, 12 Nov 2013 17:54:49 +0100, Theo Verelst wrote:
vadim.zavalishin wrote:
I'm not sure this question has very large practical significance within
music DSP context (although I can admit that sometimes it can have one).
Most of the time you are concerned just with the frequency response of
On Tue, 12 Nov 2013 17:54:49 +0100, Theo Verelst wrote:
To me, that is for instance why it all sounds so horribly distorted, so,
in
the phase of the advent of new Electronics with New (better) Converters,
its
spells out the **essence** of my comments, so there you go.
All is one too strong
On Tue, 12 Nov 2013 17:10:15 +, Dave Gamble wrote:
As soon as I see a
pole in the transfer function, I yell FEEDBACK and run around the room
waving my arms. ;)
In doing this, are you trying to compute the residue at the pole by using
the Cauchy integral formula?
Sorry, couldn't resist
On Tue, Nov 12, 2013 at 5:22 PM, vadim.zavalishin
vadim.zavalis...@native-instruments.de wrote:
On Tue, 12 Nov 2013 17:10:15 +, Dave Gamble wrote:
As soon as I see a
pole in the transfer function, I yell FEEDBACK and run around the room
waving my arms. ;)
In doing this, are you
On 6 November 2013 11:45, Andrew Simper a...@cytomic.com wrote:
Here is an updated version of the optimised trapezoidal integrated svf
which bundles up all previous state into equivalent currents for the
capacitors, which is how I solve non-linear circuits (although this
solution is just the
Lubomir I. Ivanov wrote:
On 6 November 2013 11:45, Andrew Simper a...@cytomic.com wrote:
...
actually trapezoidal is all over quantum mechanics because, say in
comparison to midpoint you will get those control and state variable
intermediates that mess calculation quite badly in cases that you
On 10 November 2013 18:12, Dominique Würtz dwue...@gmx.net wrote:
Am Freitag, den 08.11.2013, 11:03 +0100 schrieb Marco Lo Monaco:
I think a crucial point is that besides replicating steady state
response of your analog system, you also want to preserve the
time-varying behavior (modulating
Now for all those people scratching their heads of the whole Zero Delay
Feedback, here is the deal:
Any implicit integration method applied to numerically integrate something
is by its very definition using Zero Delay Feedback, linear or non-linear
this is the case. You can completely ignore that
33 matches
Mail list logo
