**- Undocumented Matlab - https://undocumentedmatlab.com -**

Speeding-up builtin Matlab functions – part 3

Posted By Yair Altman On April 6, 2020 | __4 Comments__

A recurring theme in this website is that despite a common misperception, builtin Matlab functions are typically coded for maximal accuracy and correctness, but not necessarily best run-time performance. Despite this, we can often identify and fix the hotspots in these functions and use a modified faster variant in our code. I have shown multiple examples for this in various posts (example1 ^{[1]}, example2 ^{[2]}, many others ^{[3]}).

Today I will show another example, this time speeding up the * mvksdensity* (multi-variate kernel probability density estimate) function, part of the Statistics toolbox since R2016a. You will need Matlab R2016a or newer with the Stats Toolbox to recreate my results, but the general methodology and conclusions hold well for numerous other builtin Matlab functions that may be slowing down your Matlab program. In my specific problem, this function was used to compute the probability density-function (PDF) over a 1024×1024 data mesh.

The builtin * mvksdensity* function took

While we could in theory modify Matlab’s installed m-files if we have administrator privileges, doing this is not a good idea for several reasons. Instead, we should copy and rename the relevant internal files to our work folder, and only modify our local copies.

To see where the builtin files are located, we can use the ** which** function:

```
>> which('mvksdensity')
C:\Program Files\Matlab\R2020a\toolbox\stats\stats\mvksdensity.m
```

In our case, we copy *\toolbox\stats\stats\mvksdensity.m* as *mvksdensity_.m* to our work folder, replace the function name at the top of the file from `mvksdensity`

to `mvksdensity_`

, and modify our code to call * mvksdensity_* rather than

If we run our code, we get an error telling us that Matlab can’t find the * statkscompute* function (in line #107 of our

`statkscompute_`

, and modify our `[fout,xout,u] = statkscompute_(ftype,xi,xispecified,npoints,u,L,U,weight,cutoff,...`

We now repeat the process over and over, until we have all copied all the necessary internal files for the program to run. In our case, it tuns out that in addition to *mvksdensity.m* and *statkscompute.m*, we also need to copy *statkskernelinfo.m*.

Finally, we check that the numeric results using the copied files are exactly the same as from the builtin method, just to be on the safe side that we have not left out some forgotten internal file.

Now that we have copied these 3 files, in practice all our attentions will be focused on the *dokernel* sub-function inside *statkscompute_.m*, since the profiling report (below) indicates that this is where all of the run-time is spent.

Now we run the code through the Matlab Profiler, using the “Run and Time” button in the Matlab Editor, or * profile on/report* in the Matlab console (Command Window). The results show that 99.8% of

Let’s drill into *dokernel* and see where the problems are:

We can immediately see from the profiling results that a single line (#386) in *statkscompute_.m* is responsible for nearly 40% of the total run-time:

`fk = feval(kernel,z);`

In this case, `kernel`

is a function handle to the normal-distribution function in *\stats\private\statkskernelinfo>normal*, which is evaluated 1,488,094 times. Using * feval* incurs an overhead, as can be seen by the difference in run-times: line #386 takes 29.55 secs, whereas the

`kernel`

function directly:`fk = kernel(z);`

Now that we have a local copy of *statkscompute_.m*, we can safely modify the *dokernel* sub-function, specifically line #386 as explained above. It turns out that just bypassing the * feval* call and using the function-handle directly does not improve the run-time (decrease the function-call overhead) significantly, at least on recent Matlab releases (it has a greater effect on old Matlab releases, but that’s a side-issue).

We now recognize that the program only evaluates the normal-distribution kernel, which is the default kernel. So let’s handle this special case by inlining the kernel’s one-line code (from *statkskernelinfo_.m*) directly (note how we move the condition outside of the loop, so that it doesn’t get recomputed 1 million times):

```
...
isKernelNormal = strcmp(char(kernel),'normal'); % line #357
for i = 1:m
Idx = true(n,1);
cdfIdx = true(n,1);
cdfIdx_allBelow = true(n,1);
for j = 1:d
dist = txi(i,j) - ty(:,j);
currentIdx = abs(dist) <= halfwidth(j);
Idx = currentIdx & Idx; % pdf boundary
if iscdf
currentCdfIdx = dist >= -halfwidth(j);
cdfIdx = currentCdfIdx & cdfIdx; %cdf boundary1, equal or below the query point in all dimension
currentCdfIdx_below = dist - halfwidth(j) > 0;
cdfIdx_allBelow = currentCdfIdx_below & cdfIdx_allBelow; %cdf boundary2, below the pdf lower boundary in all dimension
end
end
if ~iscdf
nearby = index(Idx);
else
nearby = index((Idx|cdfIdx)&(~cdfIdx_allBelow));
end
if ~isempty(nearby)
ftemp = ones(length(nearby),1);
for k =1:d
z = (txi(i,k) - ty(nearby,k))./u(k);
if reflectionPDF
zleft = (txi(i,k) + ty(nearby,k)-2*L(k))./u(k);
zright = (txi(i,k) + ty(nearby,k)-2*U(k))./u(k);
fk = kernel(z) + kernel(zleft) + kernel(zright); % old: =feval()+...
elseif isKernelNormal
fk = exp(-0.5 * (z.*z)) ./ sqrt(2*pi);
else
fk = kernel(z); %old: =feval(kernel,z);
end
if needUntransform(k)
fk = untransform_f(fk,L(k),U(k),xi(i,k));
end
ftemp = ftemp.*fk;
end
f(i) = weight(nearby) * ftemp;
end
if iscdf && any(cdfIdx_allBelow)
f(i) = f(i) + sum(weight(cdfIdx_allBelow));
end
end
...
```

This reduced the kernel evaluation run-time from ~30 secs down to 8-9 secs. Not only did we remove the direct function-call overhead, but also the overheads associated with calling a sub-function in a different m-file. The total run-time is now down to **45-55** seconds (expect some fluctuations from run to run). Not a bad start.

Now let’s take a fresh look at the profiling report, and focus separately on the bottom and top parts of the main loop, which you can see above. We start with the bottom part, since we already messed with it in our fix to the kernel evaluation:

The first thing we note is that there’s an inner loop that runs d=2 times (d is set in line #127 of *mvksdensity_.m* – it is the input mesh’s dimensionality, and also the number of columns in the `txi`

data matrix). We can easily vectorize this inner loop, but we take care to do this only for the special case of d==2 and when some other special conditions occur.

In addition, we hoist outside of the main loop anything that we can (such as the constant exponential power, and the weight multiplication when it is constant [which is typical]), so that they are only computed once instead of 1 million times:

```
...
isKernelNormal = strcmp(char(kernel),'normal');
anyNeedTransform = any(needUntransform);
uniqueWeights = unique(weight);
isSingleWeight = ~iscdf && numel(uniqueWeights)==1;
isSpecialCase1 = isKernelNormal && ~reflectionPDF && ~anyNeedTransform && d==2;
expFactor = -0.5 ./ (u.*u)';
TWO_PI = 2*pi;
for i = 1:m
...
if ~isempty(nearby)
if isSpecialCase1
z = txi(i,:) - ty(nearby,:);
ftemp = exp((z.*z) * expFactor);
else
ftemp = 1; % no need for the slow ones()
for k = 1:d
z = (txi(i,k) - ty(nearby,k)) ./ u(k);
if reflectionPDF
zleft = (txi(i,k) + ty(nearby,k)-2*L(k)) ./ u(k);
zright = (txi(i,k) + ty(nearby,k)-2*U(k)) ./ u(k);
fk = kernel(z) + kernel(zleft) + kernel(zright); % old: =feval()+...
elseif isKernelNormal
fk = exp(-0.5 * (z.*z)) ./ sqrt(TWO_PI);
else
fk = kernel(z); % old: =feval(kernel,z)
end
if needUntransform(k)
fk = untransform_f(fk,L(k),U(k),xi(i,k));
end
ftemp = ftemp.*fk;
end
ftemp = ftemp * TWO_PI;
end
if isSingleWeight
f(i) = sum(ftemp);
else
f(i) = weight(nearby) * ftemp;
end
end
if iscdf && any(cdfIdx_allBelow)
f(i) = f(i) + sum(weight(cdfIdx_allBelow));
end
end
if isSingleWeight
f = f * uniqueWeights;
end
if isKernelNormal && ~reflectionPDF
f = f ./ TWO_PI;
end
...
```

This brings the run-time down to **31-32 secs**. Not bad at all, but we can still do much better:

Now let’s take a look at the profiling report’s top part of the main loop:

Again we note is that there’s an inner loop that runs d=2 times, which we can again easily vectorize. In addition, we note the unnecessary repeated initializations of the `true(n,1)`

vector, which can easily be hoisted outside the loop:

```
...
TRUE_N = true(n,1);
isSpecialCase2 = ~iscdf && d==2;
for i = 1:m
if isSpecialCase2
dist = txi(i,:) - ty;
currentIdx = abs(dist) <= halfwidth;
currentIdx = currentIdx(:,1) & currentIdx(:,2);
nearby = index(currentIdx);
else
Idx = TRUE_N;
cdfIdx = TRUE_N;
cdfIdx_allBelow = TRUE_N;
for j = 1:d
dist = txi(i,j) - ty(:,j);
currentIdx = abs(dist) <= halfwidth(j);
Idx = currentIdx & Idx; % pdf boundary
if iscdf
currentCdfIdx = dist >= -halfwidth(j);
cdfIdx = currentCdfIdx & cdfIdx; % cdf boundary1, equal or below the query point in all dimension
currentCdfIdx_below = dist - halfwidth(j) > 0;
cdfIdx_allBelow = currentCdfIdx_below & cdfIdx_allBelow; %cdf boundary2, below the pdf lower boundary in all dimension
end
end
if ~iscdf
nearby = index(Idx);
else
nearby = index((Idx|cdfIdx)&(~cdfIdx_allBelow));
end
end
if ~isempty(nearby)
...
```

This brings the run-time down to **24 seconds**.

We next note that instead of using numeric indexes to compute the `nearby`

vector, we could use faster logical indexes:

```
...
%index = (1:n)'; % this is no longer needed
TRUE_N = true(n,1);
isSpecialCase2 = ~iscdf && d==2;
for i = 1:m
if isSpecialCase2
dist = txi(i,:) - ty;
currentIdx = abs(dist) <= halfwidth;
nearby = currentIdx(:,1) & currentIdx(:,2);
else
Idx = TRUE_N;
cdfIdx = TRUE_N;
cdfIdx_allBelow = TRUE_N;
for j = 1:d
dist = txi(i,j) - ty(:,j);
currentIdx = abs(dist) <= halfwidth(j);
Idx = currentIdx & Idx; % pdf boundary
if iscdf
currentCdfIdx = dist >= -halfwidth(j);
cdfIdx = currentCdfIdx & cdfIdx; % cdf boundary1, equal or below the query point in all dimension
currentCdfIdx_below = dist - halfwidth(j) > 0;
cdfIdx_allBelow = currentCdfIdx_below & cdfIdx_allBelow; %cdf boundary2, below the pdf lower boundary in all dimension
end
end
if ~iscdf
nearby = Idx; % not index(Idx)
else
nearby = (Idx|cdfIdx) & ~cdfIdx_allBelow; % no index()
end
end
if any(nearby)
...
```

This brings the run-time down to **20 seconds**.

We now note that the main loop runs m=1,048,576 (=1024×1024) times over all rows of `txi`

. This is expected, since the loop runs over all the elements of a 1024×1024 mesh grid, which are reshaped as a 1,048,576-element column array at some earlier point in the processing, resulting in a m-by-d matrix (1,048,576-by-2 in our specific case). This information helps us, because we know that there are only 1024 unique values in each of the two columns of `txi`

. Therefore, instead of computing the “closeness” metric (which leads to the `nearby`

vector) for all 1,048,576 x 2 values of `txi`

, we calculate separate vectors for each of the 1024 unique values in each of its 2 columns, and then merge the results inside the loop:

```
...
isSpecialCase2 = ~iscdf && d==2;
if isSpecialCase2
[unique1Vals, ~, unique1Idx] = unique(txi(:,1));
[unique2Vals, ~, unique2Idx] = unique(txi(:,2));
dist1 = unique1Vals' - ty(:,1);
dist2 = unique2Vals' - ty(:,2);
currentIdx1 = abs(dist1) <= halfwidth(1);
currentIdx2 = abs(dist2) <= halfwidth(2);
end
for i = 1:m
if isSpecialCase2
idx1 = unique1Idx(i);
idx2 = unique2Idx(i);
nearby = currentIdx1(:,idx1) & currentIdx2(:,idx2);
else
...
```

This brings the run-time down to **13 seconds**, a total speedup of almost ~6x compared to the original version. Not bad at all.

For reference, here's a profiling summary of the *dokernel* function again, showing the updated performance hotspots:

The 2 vectorized code lines in the bottom part of the main loop now account for 72% of the remaining run-time:

```
...
if ~isempty(nearby)
if isSpecialCase1
z = txi(i,:) - ty(nearby,:);
ftemp = exp((z.*z) * expFactor);
else
...
```

If I had the inclination, speeding up these two code lines would be the next logical step, but I stop at this point. Interested readers could pick up this challenge and post a solution in the comments section below. I haven't tried it myself, so perhaps there's no easy way to improve this. Then again, perhaps the answer is just around the corner - if you don't try, you'll never know...

So far, all the optimization I made have not affected code accuracy, generality or resolution. This is always the best approach if you have some spare coding time on your hands.

In some cases, we might have a deep understanding of our domain problem to be able to sacrifice a bit of accuracy in return for run-time speedup. In our case, we identify the main loop over 1024x1024 elements as the deciding factor in the run-time. If we reduce the grid-size by 50% in each dimension (i.e. 512x512), the run-time decreases by an additional factor of almost 4, down to **~3.5 seconds**, which is what we would have expected since the main loop size has decreased 4 times in size. While this reduces the results resolution/accuracy, we got a 4x speedup in a fraction of the time that it took to make all the coding changes above.

Different situations may require different approaches: in some cases we cannot sacrifice accuracy/resolution, and must spend time to improve the algorithm implementation; in other cases coding time is at a premium and we can sacrifice accuracy/resolution; and in other cases still, we could use a combination of both approaches.

Matlab is composed of thousands of internal functions. Each and every one of these functions was meticulously developed and tested by engineers, who are after all only human. Whereas supreme emphasis is always placed with Matlab functions on their accuracy, run-time performance often takes a back-seat. Make no mistake about this: code accuracy is almost always more important than speed, so I’m not complaining about the current state of affairs.

But when we run into a specific run-time problem in our Matlab program, we should not despair if we see that built-in functions cause slowdown. We can try to avoid calling those functions (for example, by reducing the number of invocations, or decreasing the data resolution, or limiting the target accuracy, etc.), or we could optimize these functions in our own local copy, as I have shown today. There are multiple techniques that we could employ to improve the run time. Just use the profiler and keep an open mind about alternative speed-up mechanisms, and you’d be half-way there. For ideas about the multitude of different speedup techniques that you could use in Matlab, see my book **Accelerating Matlab Performance ^{[4]}**.

Let me know ^{[5]} if you’d like me to assist with your Matlab project, either developing it from scratch or improving your existing code, or just training you in how to improve your Matlab code’s run-time/robustness/usability/appearance.

In the meantime, Happy Easter/Passover everyone, and stay healthy!

Categories: Low risk of breaking in future versions, Stock Matlab function, Toolbox

4 Comments (Open | Close)

Article printed from Undocumented Matlab: **https://undocumentedmatlab.com**

URL to article: **https://undocumentedmatlab.com/articles/speeding-up-builtin-matlab-functions-part-3**

URLs in this post:

[1] example1: **https://undocumentedmatlab.com/articles/speeding-up-builtin-matlab-functions-part-1**

[2] example2: **https://undocumentedmatlab.com/articles/speeding-up-builtin-matlab-functions-part-2**

[3] many others: **https://undocumentedmatlab.com/articles/tag/performance**

[4] Accelerating Matlab Performance: **https://undocumentedmatlab.com/books/matlab-performance**

[5] Let me know: **https://undocumentedmatlab.com/consulting**

[6] Speeding up Matlab-JDBC SQL queries : **https://undocumentedmatlab.com/articles/speeding-up-matlab-jdbc-sql-queries**

[7] Callback functions performance : **https://undocumentedmatlab.com/articles/callback-functions-performance**

[8] Undocumented mouse pointer functions : **https://undocumentedmatlab.com/articles/undocumented-mouse-pointer-functions**

[9] Speeding up compiled apps startup : **https://undocumentedmatlab.com/articles/speeding-up-compiled-apps-startup**

Click here to print.

Copyright © Yair Altman - Undocumented Matlab. All rights reserved.

4 Comments To "Speeding-up builtin Matlab functions – part 3"

#1 CommentByMichelle HirschOn April 7, 2020 @ 22:58Thanks Yair, as always, for the thoughtful post. I’ve forwarded it to the folks from both the Statistics team and the Performance team for their input. Happy Pesach!

#2 CommentByLoren ShureOn April 10, 2020 @ 15:48Hi Yair-

Very nice example showing people how to delve in – plus a nice heads up for us to always go back and look at our shipping code. Probably feval was the only option for evaluation at the time the functionality was first introduced. But I’m only guessing.

Hag Sameach and take care!

–loren

#3 CommentByAdam LangerOn June 17, 2020 @ 00:41Hi Yair,

I had no doubt you would find other ways to keep busy if you were not heading to the States this spring.

It will only make a small difference since in your example this line of code is not called as often, but in “Main Loop – Bottom Part” you can eliminate the repetitive call to

`sqrt()`

in the inner for loop by precalculating it:If I understand your final profile viewer screenshot showing the 13 seconds, the non-vectorized branch under if any(nearby) gets called around 300K times, which looks like 300K calls to sqrt(). This speedup won’t show up in your profiler results (that line is even in the top results now), but it’s another easy speedup some folks may overlook.

In the common case I see that you cleverly avoid using

`sqrt()`

at all by instead dividing by TWO_PI at the end of the snippet.Stay healthy and safe.

#4 CommentByJanOn June 26, 2020 @ 00:55The sad thing is that only old functions can be delved into like that. Many new functions or classes are hidden in p-code encrypted files which is a bummer. Plus they are sealed (in case of classes), so it’s quite complicated to expand or improve their functionality/performance.

This is one of the reasons we’re looking for alternatives like python or Julia.

Don’t get me wrong, it’s not about the price, we’re happy to pay for functionality and documentation.

But not if the trend is to make things more and more unaccessible