azillionmonkeys.com · Domains

Re:Because it was written in Seastar or C++ by phantomfive · 2015-09-23 17:19 · Score: 1 · on Cassandra Rewritten In C++, Ten Times Faster

Here you go man, this guy says it much better than I do.

Re:Yes, he was wrong... by phantomfive · 2015-03-11 06:22 · Score: 1 · on Was Linus Torvalds Right About C++ Being So Wrong?

C++ and C can obviously be as fast as each other, but once you start taking advantage of the C++ features (smart pointers, lists instead of arrays, rtti, exceptions, objects), you lose efficiency. C++ ironically was not designed to be efficient. Effective use of cache (i.e., aligning objects correctly in memory), branch prediction, and loop hoisting are the main techniques of efficiency on modern processors, whereas C++ design decisions seem to be aimed at counting instructions, which is drastically misleading (as the section entitled "Performance optimization is achieved primarily by counting the cycles of the assembly code" in the previous link explains).

At that point, you'll find that the STL libraries make C++ a little better.

I wouldn't assert that without timing it (if you did time it, please let me know, I want to hear it). The exact implementation of the library function is what matters (i.e., Microsoft's C++ sort function could be slower than GCC's C function which could be slower than G++'s C function which could be slower than Microsoft's C function. Unless you time it, you don't know).

Someone a couple months ago on Slashdot asserted that they had timed the C++ STL algorithms and they were slower than the Java algorithms. I haven't had time to test it, but it's not surprising. Yeah, it's cool that the STL can automatically generate code to sort an array of shorts, but there are many other considerations that make an an implementation of an algorithm quick or slow, as of course you know.

Re:Let's get C99 right first by Anonymous Coward · 2011-12-24 02:28 · Score: 0 · on ISO Updates C Standard

Unfortunately, many programs have to care about MSVC. Usually, you end up writing compatibility implementations of things they're lacking (yay pstdint.h http://www.azillionmonkeys.com/qed/pstdint.h) and turn them on/off at need.

I really wish Microsoft would bundle clang with Visual Studio - as far as I know there would be no license issues with doing so, and why pay to develop cl.exe when clang already exists - but I rather doubt Microsoft is institutionally capable of the notion of contributing to the clang project instead of developing their own compiler.

Re:Devs can now be more lazy by phantomfive · 2011-08-24 19:32 · Score: 1 · on Java 7: What's In It For Developers

Performance hits for closing files is extremely minimal. By far the slowest part of writing a file is the time it takes to transfer it to disks. It is in fact often on the order of a million times slower than any other part. Worrying about the performance hit of a few extra instructions to close a file is silly in that context. There is a law related to this point, but I can't find it right now. It is described under The General Task of Code Optimization.

Re:As a C programmer by phantomfive · 2009-03-21 06:39 · Score: 1 · on Programming Language Specialization Dilemma

Also, you might want to learn about efficiency, real efficiency, not just counting clock cycles. Check out Paul Hsieh's site.

Basically, your first focus should be good design. After that, you can focus on speeding things up by figuring out where your bottlenecks are. If you randomly try to speed things up, you will probably waste time. If you focus on optimization from the beginning, you will make your program both inefficient and inflexible.

Re:Duh? by phantomfive · 2008-03-25 05:07 · Score: 1 · on More Interest In Parallel Programming Outside the US?

The problem is you, and many other proponents of of multi-threading is that you only consider processing time as a resource. The fact is there are other problems: shared resources, caching, IO, disk access times; check out this discussion on the LKML. Notice how Alan Cox mentions that "as you add threads you generally decrease cache effectiveness."

If you are blocking on IO, it doesn't matter how many threads you are running, you're still blocked. In that case, there is no point in adding more threads, it will only add complexity to your program, without any real benefit. Given that the vast majority of real-world programs are interactive and spend the vast majority of their time waiting on IO (if you don't believe me check your processor usage right now...I'll bet it's below 10%), it is safe to say that the vast majority of real-world programs will experience no boost from being multi-threaded.

Sometimes threads just slow things down. To test this, just last month for work I wrote a server using a threaded model (because it was easier) and then the exact same server using a single threaded model (because my coworker complained that threads are too slow). My purpose was to show that threads are fast enough that if it's easier, you should program that way. I tested them both on a quad-core system, and the single threaded server was indeed faster (but the threaded model handled 10,000 simultaneous connections no problem....go Linux Pthreads!). Also, I am not the only one who has seen this effect.

Use threads when they're useful. Don't use them when they're not. But if you are still under the idea that parallel processing is the magic bullet, you need to learn more about optimization.

Re:tea leaves and biz speak by phantomfive · 2007-09-11 15:04 · Score: 2, Insightful · on Believe the Occupational Outlook Handbook?

It's pretty simple, really. How many jobs have you applied for? Two? Three? If you are serious, you will probably have applied for ten or more. If everyone is applying for ten or more, then every employer will receive a lot of applications, but there will be a lot of duplicates. Three applications is not very many.

As for why you weren't picked, there are a couple possibilities:

You are just not very good. In this case, the solution is simple (if not easy). Join an open source project, start your own, or do whatever it takes to get better. Maybe just reading through some guides like a zillion monkeys or even Joel. Programming is a skill, and can be developed.
Someone else who applied was better. Whatever, it happens no matter how good you are. Move on, apply somewhere else.
You have bad interview skills. It happens. Read What Color is Your Parachute

There are so many companies desperate for programmers out there, you too can find a job. It's just a matter of learning how to do it.

It may in part be related to something I did ... by Ninja+Programmer · 2006-12-01 09:23 · Score: 5, Informative · on Origin of Quake3's Fast InvSqrt()

Here's an old version of one of my webpages:

http://web.archive.org/web/19990210111728/www.geoc ities.com/SiliconValley/9498/sqroot.html

And here's an updated version of the same page:

http://www.azillionmonkeys.com/qed/sqroot.html

It isn't an exact rendering of the code in question, but it explains enough for any skilled hacker to 1) understand what's going on and 2) modify the code to create the resulting code that's in the Quake 3 source. Furthermore this web page has existed since about 1997 (archive.org doesn't go back that far for some reason.)

Now *IF* in fact the code origin comes from someone who took ideas from my site, I should point out that *I* am not the originator of the idea either (though I did write the relevant code). Bruce Holloway (who I credit on the page) was the first person to point out this technique to me at around the 1997 timeframe (prior to this, I created my own method which is similar, but not really as fast). (Vesa Karvonen informed by of the technique (through a code snippet with no explanation) at roughly the same time as well.) It was a technique well known to hard core 3D accelerator and CPU implementors, and follows an intentional design idea from the IEEE-754 specification.

Prof. William Kahan, one of the key people who specified the IEEE-754 standard (the standard for floating point the many CPUs use, starting with Intel's 8087 coprocessor) apparently presented this idea, and is the source for where Bruce Holloway got the idea. The IEEE-754 standard came out around the 1982 time frame. Though, its very likely that these ideas originate from even earlier in computing history.

Re:DirectX: vendor-lock-in and avoid paying to SGI by Jerry+Coffin · 2006-01-31 05:44 · Score: 1 · on Wine vs Windows Benchmarks

Love the 'personal' theories...

I do too -- and yours are better than most.

Microsoft did not even have an alternative to OpenGL in development when Microsoft pulled out of OpenGL.

Microsoft was a member of the ARB until early 2003. Offhand, I don't recall whether DirectX 9 was truly "released" by then, but it had been available at least to developers for quite a while at that point.

Microsoft pressed for OpenGL to enhance low level hardware support with intention of doing more than cad/engineering and supporting 3D rendering environment conducive to gaming and directly access video card hardware for gaming.

What sorts of things that games needed did the ARB (supposedly) tell Microsoft they wouldn't support? I seem to remember going through situations like this Usenet thread discusses.

If you look really carefully, you can find a few points in time at which DirectX has supported something that was slightly more difficult with OpenGL -- just for one example, rendering to a bitmap (instead of the frame buffer). None of these, however, was anything that the ARB told MS (or nVidia, ATI, et al) to pound sand on or anything like it. Quite the contrary -- OpenGL has always had a mechanism for extensions that let people integrate new capabilities into the existing framework. In addition, the ARB has moved reasonably quickly to provide (usually cleaner versions of) those capabilities in the core of newer versions as well.

Microsoft started stringing together a set of technologies that were called WinG, mainly a 2D form of rendering with plans for a new model that was a 3D rendering solution with direct video access on par of what the current DOS based games were used to, but in the Windows environment.

Well, you're at least partly right. WinG was certainly 2D in nature -- in fact, its basic idea was only to allow fast display of device independent bitmaps (as opposed to the device dependent bitmaps they'd primarily supported previously). It was not, however, replaced by DirectX or OpenGL or anything similar. Its real replacement was CreateDIBSection, which is still part of GDI (and takes relatively little advantage of graphics hardware).

If you really want to look into the history of OpenGL vs. DirectX, I'd recommend going to Paul Hsieh's OpenGL Vs. DirectX page. This has blow-by-blow documentation of things that were really happening at the time, including references to other articles, Usenet threads, etc., making it pretty easy to track down things like dates, so you have an idea of when things happened in relation to each other (such as the fact that all of this happened long before MS quit the OpenGL ARB).

When DirectX first existed it was the only game in town for any standardized interface to video for accelerated graphics in gaming. Now it is more than just Video...

This much is true -- OpenGL has stuck to more or less the UNIX philosophy -- do one thing, and do it well. Unsurprisingly, DirectX seems to follow the Microsoft philosophy much more, attempting to integrate everything. Unfortunately, (as usual) it only does the job halfway -- on one hand, DirectX has bits and pieces of almost everything, but on the other, the pieces aren't really very well integrated together. It's branded as a single "product", but it's really separate pieces with little real integration, and even with quite a few dissimilarities between the pieces that are entirely gratuitous.

Just write your own! by Anml4ixoye · 2005-11-23 06:02 · Score: 1 · on Organizing Organic Chemical Reactions?

Surely you know exactly what you want. It shouldn't be too difficult to just throw something together. In fact, some of the languages nowadays are so easy to follow, it should be a breeze. I found a tutorial that should hopefully get you down the right path.

DirectX and id??? by Midnight+Warrior · 2005-11-12 12:15 · Score: 2, Interesting · on Bill Gates' Doom Video From 1995

Since when was id a fan of DirectX, let alone Direct3D? (If this sentence doesn't make any sense, you didn't watch the video) An article from the time clearly show's id's lead engine developer, John Carmack in complete opposition to the Direct3D API. I haven't followed things after that, but let's assume the API hasn't gotten any simpler, especially with all the new shading specifications.

Re:Those are not the laws of optimization I learne by phantomfive · 2005-10-09 21:47 · Score: 1 · on Java Urban Performance Legends

Those rules seem inflexible to me, and sound like rules a professor would tell his students when he is tired of reading their unreadable attempts at optimization. Optimization is really useful if you need your program to go faster. Of course, if you don't know how to optimize, you will have trouble. Here is a very good page by Paul Hsieh about optimization:

http://www.azillionmonkeys.com/qed/optimize.html

Assembly is not slower.... by phantomfive · 2005-07-30 14:08 · Score: 1 · on Choice of Language for Large-Scale Web Apps?

Actually, odds are that hand written assembly will underperform compiled c these days. Hiring or training people that can write better assembler than a modern compiler is very very difficult.

Have you written much assembly? Modern assembly focuses a lot on memory access optimization (caching, etc), and parallel programming, and compilers tend to have trouble with that. On a webpage by Paul Hsieh you may find this quote:

Most assembly programmers are neither knowledgable or good, but those who are knowledgable and good have little trouble beating compilers.

http://www.azillionmonkeys.com/qed/optimize.html

Re:Automate it by Ninja+Programmer · 2005-02-09 21:43 · Score: 1 · on Round 2 of Apple's Lost '1984' Series

Well completely automating it is probable not the right answer. Often people will submit a link to a .HTML page, which itself links to the larger file (like a movie) that causes their server to run out of bandwidth.

The Coral cache will only cache the top-most link -- and its actually left to the HTML author to coral cache the final contents. An example is a page I wrote here:

Thank you

The root html file is not cached at all, but its fairly small, so I'm not worried about it. All those pictures though? There ain't no way I would post such a link in slashdot if I had to pay for the bandwidth to serve up those pictures myself!

So this is really a message to HTML authors and slashdot story submitters.

You mean like this? by Anonymous Coward · 2004-11-22 21:04 · Score: -1 · on Porn Site Sues Google Over Linked Images

gOOgle girl!

Re:the whole thing makes me wonder market shares by swordboy · 2004-11-04 01:26 · Score: 4, Interesting · on Doom 3 Announced for Mac

Please read about Carmack's battle with D3D and Microsoft. The bottom line is that he's the main reason that we've still got OpenGL. This technology enables game developers to deploy 3D gaming engines to multiple platforms quite easily.

Make no bones about it - there is no substantial non-Windows gaming market. No sir. Carmack does this for two reasons - because he can and because it really throws mud in the eye of Microsoft.

As a side note, if you haven't purchased this game, then go out and do so... even if you won't use it. Without id software (and some other keys), gaming would be Windows-only right now.

Re:until by Ninja+Programmer · 2004-11-02 02:48 · Score: 1 · on RC4 Code Achieves 319 MB/s On AMD64 Opteron

GCC is no slouch though [...]

Yes it is

[...] and obviously Intel is performing some tricks that could also be implemented by GCC.

Its not like the Intel compiler group is resting on its laurels. GCC has been getting its ass kicked by Intel's compiler for about 5 years now.

Mod parent down by Anonymous Coward · 2004-09-06 03:26 · Score: 1, Informative · on The End Of DirectX As We Know It

..and me up:

"Microsoft - like any other big company tending a big market - tries to please them, not piss them off!"
Paul Hsieh's OpenGL versus Direct3D

Learn to link. While copypasteing is annoying enough, these anti-page-widening spaces /. inserts make it unbearable.

A tribute to the FCC. by Ninja+Programmer · 2004-03-19 11:39 · Score: 1 · on FCC to Regulate 'Profane' Speech

A tribute to the FCC

Re:In a word, no by Ninja+Programmer · 2003-10-29 23:33 · Score: 1 · on Should Hackers Get Their Own Logo?

If we do have a logo, I think it should be a vector rendered shilouete of a fat, unwashed, unshaven hacker sitting in front of a PC.

Well how about a bitmapped one to start with?

Re:5177 MFLOPS 288 MFLOPS by Attitude+Adjuster · 2003-07-04 11:22 · Score: 0 · on NASA Benchmarks the New G5 Powermac

Someone please explain to me how 5177 MFLOPS and ~300 MFLOPS are even comparable.
As the Mathematica guy said, the competition is no longer high-end PCs, it's now $10,000 UNIX workstations...and the G5 is still faster than any of them.

Argh! Totally the wrong way to interpret the article.

Single CPU scalar benchmark
P4 2.66 Ghz --> 255 MFLOPS
G5 2.00 Ghz --> 254 MFLOPS

Then the guy basically switches on the part of his code (developed on a G4) that use Altivec (Apples SIMD instruction/pipeline), and wammo, reasonably good MFLOPS numbers (2755 MFLOPS) ensue (any modern processor can get over 1000 MFLOPS with suitably written code).

This guys code is parallelized (nice), so with the 2nd processor on he gets more done, hence 5177 MFLOPS.

But there isn't a intel or AMD comparison for the vector tests, cos his code doesn't seem to have been written to use their SIMD instruction sets. The Intel and AMD pundits claim their SIMD architecture is superior to Apple's (e.g. see discussion in this article).

As the author of these benchmarks himself states
Based on an extrapolation of current P4 results, the 2GHz G5 would lag newly announced 3.2GHz P4 systems in Jet3D scalar floating point performance by about 20%, but this kind of comparison is best deferred until G5-aware compiler tools become available (since a 20% performance gain is well within the potential of compiler optimization).

Finally, it is important to note that the current test does not factor machine cost or intended use into the picture, and that can have a large impact, especially in clustering applications.

I.e. a G5 is probably comparable to a top-end desktop 32bit PC for joe blogs applications once the small changes from G4-to-G5 specific optimizations take effect. A comparison of suitably SIMD-aware number crunching applications is still needed to evaluate absolute speed. Processor power per unit dollar cost is another issue that the author hints at, which is probably not in Apple's favor.

Re:Why do Microsoft reviewers always sound... by mav[LAG] · 2003-02-19 08:12 · Score: 1 · on Inside The Development of Windows NT

Rest of the gaming industry? From my viewpoint it was Carmack alone.

He certainly started it but many agreed, and continue to agree with him.

OpenGL certainly wasnt (and still isnt in many cases) faster on consumer level cards.

Speed wasn't the primary concern to him, it was the design and functionality of the two APIs.

An excellent chronology of that period (Christmas 96 to around October 97) can be found here.

By and large 3D gaming was being written for glide, and developers absolutely loved an open api specifically targetted for game development.

At first. But when 3dfx kept pushing our their GL drivers - and actively targeting with lawsuits any developers who dared to write wrappers - the time was ripe for a change, and it came with the TNT.

Re:19" iMac by Anonymous Coward · 2003-01-02 11:13 · Score: 0 · on 17-inch flat-Panel iMac Dead

Spare us of your Apple-propaganda bullshit and go enlighten yourself here.

Hah! Speaking of a "switch" campaign that delivers by Anonymous Coward · 2002-12-22 18:23 · Score: 0 · on All schools In Denmark switching to Linux

While Apple is certainly having a nice product in their hands, they often are too pushy with their adds, and have the tendency to skew the facts...

On the other hand, while Linux OS certainly has its own share of FUD zealotry, this kind of thing certainly brings a smile to my face:

1.1+ Million switchers in one day, try to beat that :-)!

P.S. For Apple zealots out there... you may insert your fav flame below this line:

Re:Spare Cycles @ Work by TPS+Report · 2002-12-10 02:40 · Score: 1 · on Gateway Puts Wasted Cycles to Work

I don't understand why companies don't include such things on new PCs as an option. Just include the .EXE file for Folding@Home (or one of the lesser projects :), a link on the desktop and an explanation of what the user can do with his/her idle CPU time. The number crunching power of millions upon millions of PCs wouldn't go to waste.

It's a good idea, but which ones do you choose? SETI? Folding@Home? Mersenne? There are tons of distributed computing projects - who do you include, and who do you leave out? I would imagine that AOL, Earthlink, and others have paid a set amount to put their icons on the desktop of new PCs. Giving away "desktop real-estate" on their new PC's might reduce the price they can command from other customers. :)

integer square root by Mr+Z · 2002-10-10 08:46 · Score: 2, Interesting · on Math Toolkit for Real-Time Programming

If you still need a decent integer square root algo, check out this page. I used the mborg_isqrt2 variant on that page as a starting point for writing my highly optimized Intellivision version for SDK-1600. My optimized version takes about 600 - 700 cycles for a 16-bit square root, on a machine where most operations take 6 to 8 cycles. (The version I was replacing took 4000 - 10000 cycles.)

This book looks like it might be interesting to me. Here at work, we had our own math expert, but he's retired (or semi-retired). We've contracted with him to do math libraries, and that works for now. But what about 10 years from now? There's a lot of subtlety in some of these algorithms (it's not always just as easy as whipping through a Taylor series expansion), so it's probably time someone in our group started learning. :-)

--Joe

Re:3k or 3 PCs? by Anonymous Coward · 2002-01-28 04:37 · Score: 0 · on Dual 1Ghz G4 PowerMac With Extra Yummy

You mean like this?

-cscx (posting as an AC with no thanks to Fuck-O The Wonder Moderator, who decided to slaughter my karma... thanks for the communist IP ban, guys!)

Re:Numerical FUD by Velex · 2001-11-02 11:02 · Score: 2 · on InfoWorld says WinXP much slower than Win2K

FUD is a fact of life. As a hacker I don't like FUD, either, but it should be clear that you can't fight Microsoft FUD with conventional or even logical arguments. FUD, sadly, is the only way to go.

Microsoft is scared shitless about open source, and what it can do to its unethical monopoly. Calling Microsoft's FUD for what it is hasn't even made them flinch. The only thing that management listens to, after all, is FUD, so what reason should stop Open Source from making FUD arguments against the Evil Empire?

Let's face it, people in management and people who don't know better have only emotions to rely upon. I don't see anything particulary wrong with creating some FUD to cloud those emotions and make them look at the facts that Microsoft is run by a megalomaniac who wants to have legal authority to basically become the first global internet government, supported by taxes like renewable licensing.

Re:For those of us youngins... by Anonymous Coward · 2001-09-29 18:50 · Score: 0 · on Open Watcom Effort Makes First Public Release

Take a look at the Watcom C/C++ FAQ, here. Particularly the answers to questions 1, 2, 6, 10, and 12 should interest you.

Re:but we already have directx by Phleg · 2001-09-24 16:18 · Score: -1, Troll · on 3D Labs Proposes OpenGL 2.0 To Kick DirectX

Direct3D? You mean that horribly inefficient, overly complex, counterintutive API wrapper? Spend a day learning OpenGL, and a day learning Direct3D. By the end of the first day, you'll be creating whirling, swirling explosions of color and light with OpenGL, and you'll still be trying to figure out how to get Direct3D to display in 8-bit color. Not to mention, OpenGL makes much more efficient use of your processor. Go here for more information about this. OpenGL is THE API wrapper of choice if you want ease of use, power, flexibility, and compatibility. Try a game in Direct3D and a game in OpenGL on a GeForce-based card (all the other cards have crap OpenGL drivers, IMHO). OpenGL will not only run faster, but look smoother as well. Direct3D, while having been improved over the years, is still utter trash. Go here for a good source of information and history over the D3D vs. OpenGL debate.

Re:Um....yeah, volum also brought us x86 by Dwonis · 2001-08-19 10:35 · Score: 2 · on What is Happening with OpenGL?

Nobody needs to prove that D3D (that's what we're really talking about, isn't it?) is crap. John Carmack has already done that for us.

Source can be obfuscated. [source is expressive] by Anonymous Coward · 2001-05-02 06:06 · Score: 1 · on Report From The 2600 Appeal Hearing

1. Why and how is a computer program expressive speech? What does it express? 2600's lawyers are entirely familiar with Touretzky's Gallery, so forget about those. Assume you have some C or perl staring at you, any random block of code in any random print-out. What does it express? Why should that code be protected expression?

Source code is obfuscatable. I found a tool called 'shroud' with a quick google search, there are probably other similar tools. According to this FAQ watcom provides it for customers that send in bug reports

If source would be purely functional, then obfuscating it before sending it to an untrusted third party would be meaningless

It's an argument from another point of view, it doesn't require any understanding of software at all

Finally... a story where this is *not* offtopic... by Anonymous Coward · 2000-05-19 20:38 · Score: 5 · on Oxford Yanks Student Page Over Spoof DeCSS

and I'm asleep at the keyboard. Doh!

Oh well... here it is.
Add your own, and spread this far and wide: