Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[ZBridges]

There's overhead associated with the loop itself.  Some people prefer to just write the instructions repeatedly in order to avoid this overhead.

[bloop_bleep]

The thing is, NG, although the number of times each function is called is the same in each case, in the rolled loop version you have a counter you're incrementing, comparing, and branching based on every time through the loop, which is where the inefficiency can come in theoretically. But also, the unrolled version makes your code bigger, so the processor can keep less other memory such as code in its cache, which is a memory area that it can access quickly, meaning more memory accesses take longer.

[Arx]

Getting developers to comprehend that not everyone trades out their computer every 2 years like they do (because waiting on the compiler is an arduous task, that their time is better spent doing things elsewhere), and thus the real-world environment their code will be running in is not the bleeding cutting edge they are always using (and where their code runs fabulously!), further gets them in a pissy mood when you bring it up.

The maxims I always hear are "CPU is cheap. Ram is Cheap."  Or, in other words, "You should have a 24 core threadripper, and 50gb of RAM, like I do."

That's a disingenuous argument. You can't complain about the approach devs take to space/time optimisation and then claim devs don't optimise at all.

The fact of the matter is that RAM is cheaper than CPU power these days, thanks to thermal dissipation limits and whatnot. Moore's Law died in 2007 (or maybe it was buried then? Don't remember), exchanging CPU usage for more RAM usage is entirely justified. My desktop's CPU is a little over twice as fast as my 7 year old laptop's; I have 8 times as much RAM. CPU is expensive. RAM is cheap.

Mass market tools are developed for the mass market. If you want things that run hyper-optimised and close to the wire, low-RAM and low-CPU, they exist; but don't expect companies to freeze feature development to make things lighter in a capitalist market.

As an alternative, I suppose you could try to persuade Mozilla (for instance) that they should always just release from source so that you can use an optimising compiler tuned to your PC... but I'm not sure they'll bite.

Final comment:

The first half of that quote really annoys me, because it's meaningless dick-waving to make yourself feel better than developers. I assure you, as someone who works on a product targeting feature phones among others and having taught software engineering, only the dumbest developers don't consider customer machines. "It works on your machine? So you'll be sending your machine to the client, then?" or snark to that effect is something I've heard many times. It's like complaining that your local supermarket has shelves higher than you can reach because they're a bunch of tall people, and they get pissy every time you bring it up because they're resistant to change: no, the reason they're getting annoyed is that they have to have the shelves that high to stock products and you're wasting their time.

[bloop_bleep]

But it's not just amount of RAM that matters, it's cache size that's also important, which is a CPU metric. A memory fetch can be something like 200 times slower than an L1 fetch. So if you use too much memory for the sake of improving performance (such as gratuitous lookup tables) you can gum up your cache and run slower.

[Arx]

But it's not just amount of RAM that matters, it's cache size that's also important, which is a CPU metric. A memory fetch can be something like 200 times slower than an L1 fetch. So if you use too much memory for the sake of improving performance (such as gratuitous lookup tables) you can gum up your cache and run slower.

Absolutely. I'm just very skeptical of wierd's one-man crusade against loop unrolling as the single main cause of RAM bloat, and a little ticked off by the fact that he ascribes this to developer incompetence or passive malice.

Surely it couldn't be that programs use more RAM these days because they ever higher resolution assets, more features, and more stuff in general. Surely it couldn't be that a cache miss is a blip on the radar compared to a spinning disk read, so it makes sense to preload as much as possible (note: the latter doesn't imply the former of course) now that most customers have the RAM to make it possible. Nope, Firefox uses three times more RAM than it used to because of loop unrolling.

[wierd]

Oh, it's not the single source, no.  It is *A* source.

Others include gratuitous use of libraries that are not needed, just to avoid having to implement a single common function, and a host of others.

--AAAANNNDD-- Gratitous preloading like you own the whole computer, is how many things get shipped into virtual memory, Because that is what the OS's memory manager is for. (Unless you blatantly and brazenly allocate exclusively from the nonpagedd pool, and then fuck the system in OTHER ways.) 


(and you can be butthurt about it all you want.  It is not disingenuous when they straight up tell you that your time as a user is not important to them.)

[bloop_bleep]

Well tbf library functions that are not used would probably just sit around in main memory and not bother anyone. They may even be swapped to disk. Having a lot of total memory is not much of a problem, having a lot of frequently-accessed memory can be a problem.

[LordBaal]

Wow.... all this discussion from a little post.

What I meant is that generally now developers slack on memory usage because the norm is that there is plenty. From a technical point of view is wasteful. In the sense of not just because you have plently of gasoline youll leave your car engine runing whenever you park it at night (not the best analogy but I guess it gets the point).

However from an economic point of view having your software developers optimize code to reduce ram usage when is not one of the objectives of current development is wasteful on time and effort. If an computer aplication consumes say 40mb of ram, even if say you optimize it to only use 30mb, it wont matter much because is a negiable ammount in current computers. So if you have to pay hundreds of dollars for that, would you?

This of course is not only ram usage but library development, cache usage, even bugs and quality of code.....

And not to talk about game developers that push out prepaid games with chunks of blocked code/content and barely functional features with the hopes/promises of patching everything afterwards.

Of course no complex software will ever be perfect from the start. But now sometimes is ridiculous.

[wierd]

zero-day patching is governed by a whole other set of nasty things.  Usually involving management, and bullshit deadlines that were intentionally short.  That is not really the programmer's fault.

[Arx]

(and you can be butthurt about it all you want.  It is not disingenuous when they straight up tell you that your time as a user is not important to them.)

Whatever dev told you that is a jerk, and you're being silly by generalizing it to all developers. You should absolutely know better than to do that, and I shouldn't need to explain why.

[wierd]

It hasn't been "just one."

[LordBaal]

zero-day patching is governed by a whole other set of nasty things.  Usually involving management, and bullshit deadlines that were intentionally short.  That is not really the programmer's fault.

Yeah, I know most coders would rather take their time, but execs tend to have no patience and want money NOW.

Is sad the current state of the gaming industry because this is now endemic. Shallow games with quick developement times and yearly release of the same game with different titles, with more and more simple and steamroled mechanics geared toward 5 minutes multiplayer rounds........ bug ridden using the players that are gullible enough to prebuy the thing as bugtesters, or sometimes the whole player base as bug testers. Pushing games outside despite knowing it barely or even just outrigth doesnt work.... bahh

Many of those are the things that pushed me to Dwarf Fortress.

[Reelya]

zero-day patching is governed by a whole other set of nasty things.  Usually involving management, and bullshit deadlines that were intentionally short.  That is not really the programmer's fault.

People also talk about how in the old days they didn't need patches and fixes, things worked out of the box. But that's really good old days thinking. When a game shipped on a couple of floppies or a Nintendo ROM cartridge, they were tiny, and simple enough that one person could know every line of code in the thing. It's a lot different to say you shipped an executable that runs in 64K of memory with all assets drawn by the programmer, and runs on a set piece of hardware, and there were no bugs on launch day, vs shipping a program that's 50GB worth of code and data and runs on any one of possibly billions of system configurations, and there were no bugs on launch day. Sure they could maintain the no-bugs-on-launch day thing, but the problem with that is that with the ever-increasing complexity, nothing would ever get shipped and the companies would just collapse. At some point, you need to ship just to stay in business. For example, consider if Notch didn't release Minecraft until it was "finished".

So yeah, deadlines are shorter than programmers would like, but the programmers aren't necessarily seeing the bigger picture, all the balls that are in the air. So those over-short deadlines aren't necessarily bullshit, even though they were too short for some purposes. If the trade off is between the project having a high risk of total failure, vs the product being perfect at launch, you skimp on being perfect at launch. For example, maybe the managers demanded that this particular product launch by Christmas, so there were bugs in it and a zero-day patch to try and compensate for that. Well, what actually happens if you don't launch by Christmas? It's not just pride, this is about project survival. You end up being Duke Nukem Forever.

Or, consider Daikatana. If Romero had merely launched the game quickly, warts and all rather than a drawn-out project with delays and growing expectations, it would have blown over much more quickly. Those two projects are a result of what happens when the programmers get their way and aren't constrained by money or the pencil pushers. Or it would be one of those games that's never released but people talk about with almost religious reverence, how "it would have been great". Categorically, whatever is it almost certainly would not have been great, it would have the substance of a Peter Molyneux game pitch. But, because nobody was pushing on the "just get it done!" button, the project failed so people can pretend how amazing it would have been. The exact same forces that cause the game to to be launched with some bugs also shape the game as a whole, including the parts people liked, and people don't consider that. It's sort of a mythology that developers are in an ivory tower carefully sculpting their master work and then the evil producers come in and say "launch today" but the brave creators say "but it's not finished, think of the public!" like a scene in a movie. However it doesn't really work like that. What's actually in the game is a drawn-out negotiation between different parties, to design a product that many people will like - so the features that many people like in a game are the same features those "just get it shipped" people were also demanding. They get it wrong, sometimes, but they get it right more often. By definition, or they wouldn't stay in business. Developers left alone without input might perfectly easily make a perfect game that nobody actually wants, or at a time nobody wants it.

[methylatedspirit]

Guy who called me handsome decided that throwing me into a large group of girls (alright, there was 1 boy, but still) would lead to anything productive. I don't know what he was expecting. Was it socialization? Testing my sexuality? Trying to get me to flirt with them?

Was he expecting that I would socialize with them? That clearly didn't happen; I felt they were too engrossed in their own stuff for me to reasonably try and barge in to the conversation. Plus, with a seemingly nonexistent common ground, what could I talk about with them? Barging in only works if you can identify common ground. In the end, I was like "fuck this" and just left. It's the least productive interaction I've had in a while, almost like it was the identity function of interactions. I didn't gain much from it, and I highly doubt they got anything from me.

Was he trying to test my sexuality? Newsflash, buddy: it's really fucking weak for people I don't know. I could be mistaken for asexual (or gay, as is more likely) given how weak it is. Or more precisely, it's heavily suppressed for strangers, and it only weakens under high trust or direct permission. I think it's a good way to not be creepy; just hide your horniness! Works every time. Still wouldn't suggest doing it yourself. I blame my conservative upbringing and a literal reading of the "rules"*.

Even with that out of the way, it's not like I was attracted to any of these identical-looking faces (I'm not being anything-ist here; I just couldn't give less of a shit). I'm more strongly attracted to personality than anything else, and I need a large dataset to play with to have a chance to get attracted. What fucking chance was there that I'd even try asking any of them out? Hell, I don't know how to flirt because I consider it a bullshit move. Why not just be direct? Not one of them threw themselves at me, so that is pretty good evidence that I'm not as attractive as the guy thinks I am. Not one. Or they're suppressing their sexuality, but I don't probe strangers. It's beyond creepy in my book.

All I can say is, "What were you expecting?". Can anyone here decipher what his intentions were? I certainly can't.

*(It doesn't actually say that you have to (pretend to) be asexual and aromantic, but I figured that's the easiest way to not get in trouble. It helps that my sexuality doesn't have much power (except when it does), and it's really picky. It's not like I'd gain hugely from a romantic (and/or sexual, if it comes to that) relationship that I or the other party don't mutually like, so I guess I'll wait or something. I have better things to do. Maybe I do need to re-evaluate my sexuality.)

[Mathel]

I will assume he did not throw you literaly.

My guess is your first guess. That he was trying to get you to socialise with them.