Yep, many (especially younger) programmers don't get the "feel" for how fast things should run and as a result often "optimize" things horribly by either "scaling out" i.e. running things on clusters way larger than the problem justifies or putting queuing in front and dealing with the wait.
The computers are fast, but you don't know it
101–110 of 819 posts
Re: The computers are fast, but you don't know it
#102I've been lightly banging the drum the last few years that a lot of programmers don't seem to understand how fast computers are, and often ship code that is just miserably slower than it needs to be, like the code in this article, because they simply don't realize that their code ought to be much, much faster. There's still a lot of very early-2000s ideas of how fast computers are floating around. I've wondered how m…
Back in time all you needed for perfect performance is to use C and proper algorithms. It was easy. Nowadays you need vector operations, you need to utilise GPU, you need to utilise various accelerators. For me it is black magic.
To be really fast, yes. Those are optimizations that allow you to go beyond the speed of just C and proper algorithms.
But C and proper algorithms are still fast - Moore's law is going wider, yes, and single-threaded advancements aren't as impressive as they used to be, but solid C code and proper algorithms will still be faster than it was before!
What's not fast is when, instead of using a hashmap when you should have used a B-tree, you instead store half the data in a relational database from one microservice and the other half on the blockchain and query it using a zero-code platform provided by a third vendor.
Re: The computers are fast, but you don't know it
#103Earlier quoted context omitted.
It's not only a matter of 750ms instead of 200ms. I'm astonished every time I open some tool like Visual Studio, SAP Power Designer, or Libre Office that can stay for the most part of a minute on its loading screen. What do those tools even do for that long? They can read enough data from the disk to overflow my computer's main memory a few times during it.
Phone home. I suspect much of the lag is network latency.
Network lag can be worked around with concurrent programming techniques--you don't even have to use a high-performance language to do it. The problem is that concurrent programming is far beyond what the typical Jira jockey can do--bosses would rather hire commodity drones who'll put up with Agile than put up with and pay for the kind of engineers who can write concurrent or parallel programs.
Re: The computers are fast, but you don't know it
#104Earlier quoted context omitted.
> And in the end, the code seems to run "fast enough" and nobody involved really notices that what is running in 750ms really ought to run in something more like 200us. Nobody has created a language that is both thousands of times faster than Python and nearly as straightforward to learn and to use. The closest thing I know of might be Julia, but that has its own performance problems and is tied closely to its AI/ML…
The "iterate from notebook to production" process which is common everywhere but the largest data engineering groups rules out anything with manual memory management from becoming popular with data science work. Some data scientists I know like (or even love) Scala, but that tends to blow up once it's handed over to the data engineers as Scala supports too many paradigms and just a couple DSs will probably manage to…
I love scala and wish it was more popular. I've made piece with java at this point as it slowly adopts my favorite parts of scala but I miss how concise my code was.
Re: The computers are fast, but you don't know it
#105Hmm, interesting that single threaded C++ is 25% of Python exec time. It feels like C++ implementation might have area for improvement. My usual 1-to-1 translations result in C++ being 1-5% of Python exec time, even on combinatorial stuff.
I recently ported some very simple combinatorial code from Python to Rust. I was expecting around 100x speed up. I was surprised when the code ended running only 14 times faster.
Did you use python specific functions like list comprehensions, or "classic" for/while loops? Because I've found the former to be surprisingly fast, while naive for loops are incredibly slow in python.
Re: The computers are fast, but you don't know it
#106> It's crazy how fast pure C++ can be. We have reduced the time for the computation by ~119%!
The pure C++ version is so fast, it finishes before you even start it!
Re: The computers are fast, but you don't know it
#107Earlier quoted context omitted.
Most likely a missued of Pandas. DF are heavy to create, but calculations on them are fast if you stay in the numpy world and stay vectorized.
"It's fast so long as you don't use any of the many parts that aren't fast!" This isn't great.
People create accidentally quadratic code all the time. It's even easier in pandas because the feature set is so huge and finding the right way to do it takes some experience (see stackoverflow for a lot of plain loops over pandas dataframes).
Re: The computers are fast, but you don't know it
#108On a 3GHz CPU, one clock cycle is enough time for light to travel only 10cm. If you hold up a sign with, say, a multiplication, a CPU will produce the result before light reaches a person a few metres away.
I ran across an animation once that showed graphically the time it takes light to travel between the planets and the sun. It's weird, but light doesn't seem that fast anymore.
Re: The computers are fast, but you don't know it
#109Earlier quoted context omitted.
> I agree 100%. I wish every software engineer would spent at least a little time writing some programs in bare C and running them to get a feel for how fast a native executable can start up and run. It is breathtaking if you're used to running scripting languages and VMs. Conversely when 99.9% of the software you use in your daily life is blazing fast C / C++, having to do anything in other stacks is a complete exer…
Conversely when 99.9% of the software you use in your daily life is user friendly Python, having to do anything in C/C++ is a complete exercise in frustration, it feels like going back a few decades in time
I kind of feel both statements.
I like writing things in python. It honestly feels like cheating at times. Being able to reduce things down to a list comprehension feels like wizardry.
I like having things written in C/C++. Because like every deep magic, there's a cost associated with it.
Re: The computers are fast, but you don't know it
#110On a 3GHz CPU, one clock cycle is enough time for light to travel only 10cm. If you hold up a sign with, say, a multiplication, a CPU will produce the result before light reaches a person a few metres away.
Huh, but then I'm pretty sure that there are some paths inside the CPU die that are long enough that speed of light is a consideration at these frequencies. Must require a lot of smart people to design these things, yet it only takes a bunch of junior developers to bog them down.
I say we ban all junior devs.