Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

24-core CPU and I can’t move my mouse

randomascii.wordpress.com

Post reply on HN
Print view View original

271–280 of 509 posts

Re: 24-core CPU and I can’t move my mouse

#271

Post by gkya »

fredley wrote: post #20
weinzierl wrote: post #16

I grew up on the Commodore 64 (1 Core, 1 hyper-thread :-), almost 1 MHz clock freq, almost 64 K usable RAM). The machine was usually pretty responsive, but when I typed too quickly in my word processor it sometimes got stuck and ate a few characters. I used to think: "If computers were only fast enough so I could type without interruption...". If you'd asked me back then for a top ten list what I wished computers cou…

It's all just trade offs. However fast or powerful your machine is, software will use as much of that resource as possible, up to the point where it occasionally interferes with input (but not too much or you'll switch to something else).

That's is better rewritten as "however fast or powerful your machine is, software will waste as much of that resource as possible".

Re: 24-core CPU and I can’t move my mouse

#272

Post by pjmlp »

ableal wrote: post #242
pjmlp wrote: post #132

Earlier quoted context omitted.

Bjarne's made the mistake of having C++ rely on C's linker model, so it meant no modules. Now we can see it as a big mistake, but on those days probably it was one of the reasons why C++'s adoption took off. Also while C lacked modules, most Algol and PL/I derived languages supported them since the late 60's. Swift's case has the issue of mixing type inference with subtyping, so lots of time is spent there. All in al…

> Bjarne's made the mistake of having C++ rely on C's linker model, so it meant no modules. > Now we can see it as a big mistake, but on those days probably it was one of the reasons why C++'s adoption took off. No mistake, just no choice - the original (1986 or so) C++ cfront compiled C++ to C which it fed to the C compiler and linker chain.

> "but on those days probably it was one of the reasons why C++'s adoption took off."

It is a mistake with 2017 eyes, because build times are now insupportable.

Of course it was the right decision in 1986 when trying to get adoption inside AT&T.

Also I have read "Design and Evolution of C++" back when it was published, and know C++ since Turbo C++ 1.0 for MS-DOS, so I grew with the language.

Which is also a reason why I still select it as member for my Java, .NET and C++ toolbox trio.

However the first ANSI C++ was approved in 1998, and many of us were expecting to get some kind of module support in C++0x.

Re: 24-core CPU and I can’t move my mouse

#273

Post by mtanski »

mpweiher wrote: post #105
titzer wrote: post #45

Full disclosure: I work for Google on Chrome. A Chrome build is truly a computational load to be reckoned with. Without the distributed build, a from-scratch build of Chrome will take at least 30 minutes on a Macbook Pro--maybe an hour(!). TBH I don't remember toughing out a full build without resorting to goma. Even on a hefty workstation, a full build is a go-for-lunch kind of interruption. It will absolutely own a…

> Damn, are we doing this wrong? Yes, and I mean "we" as in "this industry". I just recently talked to someone whose Swift framework(s) were compiling at roughly 16 lines per second. Spurred on by Jonathan Blow's musings on compile times for Jai[1], I started tinkering with tcc a little. It compiled a generated 200KLOC file (lots of small functions) in around 200 ms. Then there are Smalltalk and Lisp systems that are…

This whole sub-thread tangent-ed into C++ is bad because it's compilation causes this problem. Problems with C++ builds are well understood...

This specific issue prob impact other batch workloads with lots of small tasks (processes). There's no reason this should be happening on a 24 core machine.

Re: 24-core CPU and I can’t move my mouse

#274

Post by innocenat »

nneonneo wrote: post #112
titzer wrote: post #45

Full disclosure: I work for Google on Chrome. A Chrome build is truly a computational load to be reckoned with. Without the distributed build, a from-scratch build of Chrome will take at least 30 minutes on a Macbook Pro--maybe an hour(!). TBH I don't remember toughing out a full build without resorting to goma. Even on a hefty workstation, a full build is a go-for-lunch kind of interruption. It will absolutely own a…

These numbers don't seem abnormal; I recall building Safari many years ago, and having multi-GB of intermediate products shrink down to a 30MB executable (plus a few hundred MB of debug symbol). So, I have a thought: if we're spending all this time to compile functions (particularly template functions) that are just thrown away later, why are we performing all our optimization passes up-front? Surely, optimization pa…

I think that, especially in templated code, you only know which part of code can be thrown away because of the optimization.

Re: 24-core CPU and I can’t move my mouse

#275

Post by sixstringtheory »

coldcode wrote: post #256

Try running Carthage update for an iOS app with 80 frameworks. Ever seen a Mac hit 55GB of ram? MacOS kills it. It takes many of these runs, and all this does is fetch prebuilt binaries. Yes, everything I said here is beyond stupid.

Is it necessary to update _all_ of them at once? I think that's the same problem as OP... rebuilding the whole world is slow, so why not do it incrementally?

Also, 80 frameworks ⊙⊙ I'm assuming some of these are internal, and written in Swift, meaning they can't be compiled into static libs (easily).

Re: 24-core CPU and I can’t move my mouse

#276

Post by sp332 »

ash wrote: post #213
andrepd wrote: post #193

Earlier quoted context omitted.

Are you serious? You want to make a product: a Web browser. What technology are you going to choose? The one that makes your browser fast but gives you more work, or the one that makes your browser slower but makes compilation less of a nuisance to you ? It's mind boggling that some would openly say that, hey, who cares about performance that much, these compilation times bother me, the developer. On a browser of all…

Why not fast browser and fast compilation at the same time?

This might be theoretically possible, but I don't think it's been done yet. All the fast browsers so far take a while to build.

Re: 24-core CPU and I can’t move my mouse

#277

Post by bluGill »

andrepd wrote: post #193
mpweiher wrote: post #105

Earlier quoted context omitted.

> Damn, are we doing this wrong? Yes, and I mean "we" as in "this industry". I just recently talked to someone whose Swift framework(s) were compiling at roughly 16 lines per second. Spurred on by Jonathan Blow's musings on compile times for Jai[1], I started tinkering with tcc a little. It compiled a generated 200KLOC file (lots of small functions) in around 200 ms. Then there are Smalltalk and Lisp systems that are…

Are you serious? You want to make a product: a Web browser. What technology are you going to choose? The one that makes your browser fast but gives you more work, or the one that makes your browser slower but makes compilation less of a nuisance to you ? It's mind boggling that some would openly say that, hey, who cares about performance that much, these compilation times bother me, the developer. On a browser of all…

> Would you be okay with your browser being 2x or 3x slower?

No, I will switch to firefox or something. I'm a user, I don't care how hard it is for developers I can about my workflow which is using a browser on various machines, some of which are very slow.

Re: 24-core CPU and I can’t move my mouse

#278

Post by _arvin »

Eridrus wrote: post #268
_arvin wrote: post #260

Earlier quoted context omitted.

My Linux desktop never locks up. I'd suggest messing around with a different kernel, double checking graphic drivers, etc.

I was probably a bit imprecise; I do sometimes get full lock ups where I have to reboot, mostly when running TensorFlow locally, but usually I just get lock ups that last It's at the level of "annoying, but doesn't impact my work", so I just live with it.

Ah I see. Which scheduler are using?

$ cat /sys/block/sda/queue/scheduler

noop deadline cfq [bfq]

I've been very pleased with the responsiveness of BFQ when multitasking.

And actually, I shouldn't say I never get lock-ups. There's been a couple times where the DE just freezes, and I'll have to hit Ctrl+Alt+F2 to switch to another tty and restart the display manager. But I attribute that to running bleeding-edge version of things and enabling experimental features, so that's fair.

Lastly, my MacBook Pro (2010 6,2) would also experience random freezes on Ubuntu (would have to power off/on) and upgrading the kernel from the Ubuntu default to latest mainline solved that problem completely.

Re: 24-core CPU and I can’t move my mouse

#279

Post by striking »

gizmo wrote: post #209
andrepd wrote: post #193

Earlier quoted context omitted.

Are you serious? You want to make a product: a Web browser. What technology are you going to choose? The one that makes your browser fast but gives you more work, or the one that makes your browser slower but makes compilation less of a nuisance to you ? It's mind boggling that some would openly say that, hey, who cares about performance that much, these compilation times bother me, the developer. On a browser of all…

False dilemma. You can have both. It's okay if a fully optimized release mode binary takes a bit longer to compile, but compiling a few million lines of code for a debug build shouldn't take more than a second or two. Also consider the leverage factor. Improvements to the compiler benefit all users of the programming language, so it's worthwhile to invest in high quality compilers.

In what language can we have both?

Yes, we can use caching compilers (https://wiki.archlinux.org/index.php/ccache) to speed up builds with few changes. We can lower optimization levels (although that barely gives you an increase in compile speed compared to the runtime speed you lose and the fact that it makes your program do slightly different things).

There's no slider from "pessimum" to "optimum". You need to do wildly different things to optimize past this point for compile speed. Erlang hot-reload and at-runtime-code-gen from other langs come to mind. But that will almost definitely slow down your program because of the new infrastructure your code has to deal with.

I have observed that there can be a nice balance with Java and the auto-reloading tools that are available for it. But I am unaware of their limitations and how a web browser might trigger those limitations.

Re: 24-core CPU and I can’t move my mouse

#280

Post by cat199 »

Beltiras wrote: post #122
elorant wrote: post #49

We've come to the point where building the browser from scratch takes more than building the OS itself.

Interesting metric actually. Building Windows takes 12 hours [1]. It's a bit harder to find metrics on Linux. You can build the kernel in 60 seconds apparently [2] but that is not a complete operating system. [1] https://stackoverflow.com/questions/226377/operating-system-... [2] http://www.phoronix.com/scan.php?page=news_item&px=MTAyNjU

Can build base+X11 BSD systems in a handful of minutes on modern hardware in one command. I don't have modern hardware, and build over NFS, so I can't give a precise figure.

But yes, windows includes lots of API's and frameworks and GUI apps and such. Probably something comparable complexity wise would be Base+X11+(1 of KDE or Gnome)..

Post reply on HN