And then, what is really important and is known for 10,000 people,

is that they can then do the OC, also with parallel computers and also with Pi.

I have already shown you the Pi.

That means in connection with the implementation of parallel computers,

a multichore CPU, GPU, end-to-end cluster.

So that's the plan. But now everything is closed, the computers are closed,

we have to put them on a

connection post so that they are connected in time, so that they can read again, etc.

And then the whole thing is also C++.

Non-functional properties, as I have already told you,

what do you understand by non-functional? Non-functional, yes.

All properties, running times, energy.

Although we will not start doing energy measurements now.

Okay, and then a big chapter, block 2.

So let's say we do a big one, maybe not, but from the content,

this is an own vlog, the performance optimization.

So optimization with OpenMP and I think I put question marks again,

I have to check again, because I haven't done that myself yet,

but there is the competence on the chair.

For example, you don't have to go to your computer with a stopwatch or an off-tick

to measure how fast it is.

There is a so-called performance counter built into the processors in New German.

I have just translated that here, so performance counter.

That means you press the stopwatch yourself,

and the processor counts the number of bars you need.

And there is a performance counter for many things.

It counts the cache misses and a lot of things like that.

So there is a whole bunch of it, and a certain section of it,

we want to teach you, because that is important for you,

if you later want to do performance optimizations with your program

and want to see how many cache misses I have, how many bar cycles I have used exactly,

and now they are doing parallelization and how much better it gets.

Then you can measure it directly in the program by reading out this performance counter

in hardware.

There are libraries for this and so on.

And there is also the Perf tool, which runs under Linux,

which we have done last year, so you can also do measurements with it.

How fast your program is, where there are gaps, Perf then finds out, for example,

80% or 90% of your program was used in this and that area.

Of course, you know how to loop, right?

But it was used in this and that area, and then you can look in there specifically,

where can I optimize something?

Well, of course, we are back in the computer sector now.

All right.

And then comes a larger complex of topics.

Riemisch 3.

Yes, what applications, how can we actually apply this now?

Okay, in the past, when I did not hold the lecture,

there were usually stencil codes and so on.

You know, possibly, that we at the lecture here at Inf3,