As long as I've got an SB16 or similarly open-hardware card in my machine, there's no reason I can't basically write a driver to sit there and read what the card is getting, and save it to a raw file (which can later be mp3'd).
My (vague) understanding of unfuck.exe is that it actually intercepts the audio somewhere in the windoze pipeline (therefore architecture independent) -- this is also pretty easy under Linux.
On the (admittedly short) consideration I've given this, I just don't see a way around that problem for the secure music bastards of the world.
It's just 16bit. It works fine. The 286 can take up to (IIRC) 16megs of RAM (tho few MB's will take it).
IBM originally came out with OS/2 for the 286.
Interestingly, Intel never provided a way to switch out of 286 protected mode -- they figured you'd boot in real mode, switch to protected as the OS loads, and never look back (they forgot they had M$ to contend with:). To get your 286 out of protected mode you triple-fault it, which causes the processor to reset.
the original IBM 101-key keyboard. I have a pile of them, ranging in years from the mid-80s thru the early 90s, and all my friends use them. They have an amazing tactile-click, and they're completely indestructible. I've taken them apart -- the grounding strap in there could take a direct lightning strike. You can also pop off all the key caps (which fit over the actually keys) and remap them very easily.
I've seen them selling (refurbished) for around $80.
BTW, when Fire and Darkness was still a DOS game, you used Alt-SysRq to switch to the text mode console. Ever since we switched to Windows, it's been Scroll Lock. So we like all those extra keys.
I have to say, I do like Sun keyboards with the Ctrl where the Capslock is on all other keyboards (except the original 88-key that came with my IBM PC...now that was old school...)
Granted, the only look I took at SO5 was a hacked english version back when only the german version was out -- i found a partial english ver on one of the german technical university servers (mighta been clausthal) and fixed its missing parts with the german package. My impression on the k6-200 i was using then was, kinda slow, crashy, and way too MS-like.
I very much prefer Corel's much-superior (stability and speedwise IMHE) wordperfect suite, which already exists for Solaris (anyone else want Corel to release a free-for-personal-use ver for Sun? Would top off a 3GS nicely;). I worry that the reason Sun would want StarDivision is that SO5 had (in theory) a Java version.... yuck... wouldn't want to try that.
Corel also has a pretty darn good record of supporting Free stuff, for a commercial-software company.
For that matter, I intend to get LyX and Gnumeric and AbiWord on my two SPARCs and live a happy life. I've had it with commercial word processing. Bleah. @#$% MS-Word files. Fight the man with your word processor!
That is, if you can solve one NP-complete problem, you can use it to do the others in polynomial type. A Hamiltonian oracle can solve 3-coloring, circuit-sat, traveling salesmen, and the others, fast
...
I don't really know if quantum computers can solve NP problems, or even factoring for that matter, since I don't know how to phrase algos in quantum computable terms....although, like parallel algorithms, it'll be a booming area of research, you can bet.
the technical details described in the article, but I do have (imho;) a decent understanding of the computational principle behind quantum computing. I've seen some pretty wrong (or just confused) things in the comments...and/. discussions of encryption/complexity related issues frequently get mired in BS;)
Quantum computers do not really translate to higher FPS in Quake3 or more realistic flightsims. You don't get quantum computers that have amazing SPECfp numbers.
Quantum computers do operations that normally take one order of magnitude in a much lower order time. Example: searching a list of items for a value on von Neumann machine is O(n) (length of the list), since you have to look at each element till you find the one you want. You can do this in O(1) on a parallel machine. This is a significant improvement. IIRC there was a quantum computer that could search a 4-element list in O(1).
Obviously, that's a pretty trivial problem. Even for very long lists, O(n) isn't bad, and you can do it in parallel.
What about fundamentally much more complex problems, such as factoring, or better, something NP-complete such as circuit-sat or Hamilton paths (that's what the article was talking about, I think. Those guys think their quantum computer is a Hamilton oracle, if I'm right. So you have to pose questions in the form of graphs...not a serious problem since all NP is orthogonal). A parallel machine doesn't help there. But a quantum computer can take NP problems from exponential to polynomial time.
All of a sudden factoring (not NP, just decently hard) goes from taking 100 trillion years to taking hours.
Yes, all encryption that uses factoring as a trapdoor (one-way function) would be toast.
No, it wouldn't necessarily be any better at running first person shooters.
Still, an NP oracle machine IMHO would be much more significant than some bloody nanotech robots.
My browser couldn't find the internet due to momentary enormous stupidity, and decided to pull up its last cached copy of/., from late last year IIRC when I was at home and still used this machine.
Jon Katz was ruminating unpopularly in the first of his articles playing with a linux machine and how weird and confusing it was.
GNOME vs. KDE flamewars were going on fast and furious in the unmoderated comments.
No one had heard of, or cared about, Linux the Holy OS or Open Sourceness or User Friendly or anything.
Man, we've come a long way since then. I've been reading/. since nearly the beginning, but even I couldn't have imagined the community that grew up around the little Nerd News sight.
I've posted a few stories myself. I've wasted a lot of time writing comments, and searching out comments by resident cool, funny guys Bruce Perens and Alan Cox.
Now, it ain't news unless it's on/. You can wear the hat and the tshirt. I start my day off in the office with/., then UF. Everything I need in life is a/box. When we won IGF, the first thing I did after seeing the news in a more-or-less reputable story I could link to was skip class (251, no less) and post it on/.
I'd be interested to see if anyone has a copy, HTML or PS or hard copy or whatever, of the earliest/. page we can find.
This is user # 3273... I'll be back tomorrow morning, as always... man, I really am a nerd:)
Alright. Fine. Linux is not good at striping across multiple NICs (or at least across 4). IMHO that's not a normal thing to do anyhow...
There are really two parts two heavy-duty serving: files and web.
a) Web content You don't need 4 100Mb NICs to serve real webcontent (no, corporations using "intranets" to be buzzword compliant doesn't count. when i see a serious intranet implementation i may believe it. till then it's just buzzword compliance;).
No one serves more than cdrom.com, and if they're the only ones straining 100Mb (on 100% static content, at that) 100Mb is good enough fer anyone. Next issue: dynamic content. I want a fair benchmark of IIS and (mod_perl) Apache on PERL and ASP (using mod_asp).
How good is mod_asp ennyhoo? Does IIS do PERL? Does anyone have serious site implementations (on the order of/. or Excite or similar mad dynamic pagegen) that could be forced to work on ASP or PERL under both IIS & Apache? Would be fascinating
b) File serving I'm still not convinced in the real world you need more than a 100Mb card -- if your fileserver can sustain 7 megs/sec (realistic on ethernet, peak is 10 megs/sec) you're doing pretty well with that RAID array, since no HD i've seen can do that alone, especially not when doing simultaneous serves (and therefore not reading contiguous blocks). I'm bothered by the SMB test setup in that sense -- if it was tiny reads out of the cache, big deal. I wanna know who can move real volumes of data, and fast.
Of course, I know the answer. Sun, and after them SGI's Origin2k boxes (the fastest Windows networking fileservers in the world;)....
No. Just no. And for good reason.
on
The Factoid
·
· Score: 2
Why? 'Cos mine will run linux. And will do what I tell it to do. Like the rest of my computing hardware and technology. I use it to make my life better, but I'm (reasonably) certain I run it and not the other way around.
My Factoid:
-- Collects bizcards of everyone I ever met. This means no more flipping thru the huge company stack'o'bizcard fun for the one I need. And yes, Mark, this means your frikkin' bizcard database (lovingly coded & entered by hand) is obsolete:) -- Remembers what I did today So I can check stuff off my list. -- Can pick up important announcements but since I filter it will only keep the ones i think are important. So my friends' factoids could tell mine what's being planned, but I can ignore the specials at WalMart. -- Runs Linux And crashes M$ Factoids, especially those that want marketing data from me:)
Those guys at WRL are just cool.
Remember. Your hardware can only screw you over if you let The Man tell it what to do. Which is why we tell it what to do ourselves.
Tho that would be an impressively minimal linux system... I guess not reallly any different than a PC-on-a-SIMM...
-------
"Now look in that big clustered computer and find the 1GHz Alpha 21264 that says `Badass Mutherfucker' on the heatsink. That's my badass mutherfucker." --- Seen on a Valley bumper sticker
I apologize if this has been said before... @#$% MSIE 3 can't handle rendering the discussion quite right, and its poorly implemented smooth-scrolling is bloody slow on my p133...yuck....
No one ever guaranteed that Linux was proof against commercial applications, Anything Free, and we can port it. Doesn't matter if the original developer only has eyes for one distro, if it's free the work can be done.
But MetroWerks CodeWarriorisn't Free. It's commercial, and if all they give out is an RPM that makes all kinds of assumptions about how your system is set up... well, you might hack it into place with Slack or Debian, but good luck to you.
On the other hand, CodeWarrior has nothing in the compiler end (everyone knows that). Just that a lot of people like their IDE.
Well, I have a new concept for Linux:
We don't need to be scared of any luser inferface
It's really an idea that's been mentioned before: UI's a fundamentally not a cutting edge thing. They require careful planning and execution, but no massive technology breakthroughs. Hence Gtk+ and Qt, massively superior in flexibility and power to M$'s MFC. In fact the last really good UI I saw on Windoze was DevStudio, which is all custom widgets anyhow.
The point is, we don't need their frikkin' IDE. We can write our own better ones. And our own better spreadsheets, word processors, desktops, and everything else. Miguel and the crew wrote all of GNOME in a year. These things can be developed fast, and with spread out
The only applications that IMHO can deliver real value to Linux are big-iron server-side things that require heavy-duty technology and years of experience (and millions of $$ of screw-ups and mistakes) to get right. Oracle. Lotus Notes.
We can compete in some of these areas, but I'm often less than impressed with Apache or the ftpd's in terms of efficiency and robustness. Take a positive example, CODA, the long awaited high-tech replacement for NFS. It's good. Very good. But it's taken a lot of development, significant resources of CMU, and it's still not quite done. Even our beloved/. depends on a Beer- (not Speech-) licensed SQL database. It'll take some work before the Free community can deliver truly excellent server side applications.
But even so, these are not the kinds of applications that will Balkanize the distros. They depend on more fundamental, standard services of the OS. They've run on other Unices and the effort to making them cross-Linux is miniscule next to the effort of bringing them in from Solaris or HP-UX or AIX (or worse:).
In conclusion, CodeWarrior sucks. Go support you favorite GPL'd IDE today!
I've read a lot of game programming books over the years, and have yet to see one I would call "good", let alone "excellent". They all share the same small set of problems: out of date technologically, clumsy (or just plain wrong) code/solutions, poorly written. Not to sound cynical, but if the people who write game programming books were good enough to write real games, I suspect they would.
There are good books on networking technology (Tannenbaum), project structuring (Design Patterns, many others), 3D graphics (ask Rosenthal for recommendations here:), but you have to figure out how to make a game out of them yourself.
Game programming under linux: - use X, it's not that bad - use Glide or GLX (or Mesa if you don't care about speed so much) for 3D.
1. Large files does xfs support _big_ files? Is xfs what SGI uses on their BigIron origin servers (the fastest WinNT servers on the face of the planet, thanx to Samba;)
3. NFS Cats and kids -- NFS sucks. As in, "will never be worth a damn." Use something good. Coda, anyone? Actually, the Coda guy at CMU just wrote a distributed file system (in PERL!!) that runs on top of whatever FS the OS is already using -- would be very nice on top of xfs. It's a bit easier to maintain/add functionality, since it's ~8k lines to Coda's 500k. It's called InterMezzo.
4. DEs GNOME & KDE blow 4dwm, Win95, Mac, and everything else away. 'Nuff said.
5. HA clustering so how good is TurboLinux's? They must be convinced to OS it. Or we'll have to rewrite it:)
a) RAID controllers IIRC there are some RAID controllers which work beautifully in Linux and others which are in alpha drivers (such as the one MindCraft used...). Does anyone know which ones are which, and where the one ZD used fits in?
b) Througput for Linux peaked at exactly 200Mbps. Anyone else find that suspicious? As if they only had 2 NICs going in Linux? Why on earth should the kernel choose such a nice round number at which to pan out?
c) Static Pages this has been mentioned before, but it's very pertinent. The only thing that counts is dynamic content. Anyone know how the Apache mod_asp performs?
d) Multiprocessing i386 I'm sorry. When you're spending $20k on a computer, you buy a Sun Ultra-60, run Solaris on it, and end the question there. Intel machines suck at high-end multiprocessing. And Linux will kick anyone's ass on a dual box:)
Um, "markup language", "text formatting language", what's the diff?
I recall a debate about a year ago on/. on whether HTML was Turing complete. While the tiny amount I know isn't:), with garbage like Java or VBScript it most certainly is.
I extract data from TeX documents all the time. And my interpreter never has any problems. Sure, a TeX document can be invalid, but so can HTML (and w/Javascript it can be downright hostile). PostScript is (I'm pretty sure) Turing complete too, and we never have any problems using that for all our documents at some point in their existence...
I don't necessarily say that the file format transferred should even be LaTeX, I just said it should be TeX-based, so we could write pages the same civilized way we write everything else.
It's possible a somewhat more general TeX like language could be useful, since TeX is really for document processing rather than general publishing, but there's no reason to choose a standard that relies on humans writing such unreadable gobbledygook as HTML.
I'm kinda peeved at all y'all who keep referring to writing HTML "code".
WHAT !?!?
C/C++ is code. Smalltalk is code. LISP is code. ML is code.
PERL is code. Tck/Tk is code. Python is code.
hell, even Java is (probably:) code.
....
HTML is a @#$%^& text formatting language, for God's sake!. I'll even concede that writing CGI/Perl web stuff is code. But I'm pretty sure you do that in emacs or vi, not Dreamweaver or whatever. No, Javascript doesn't count.
I had to write some HTML once. It sucked. It's a pain. It's terrible. What lunatic decided that HTML was an appropriate language with which to invent the Web? (rhetorical question, i know the history behind the http).
Hell, i'm using annoying HTML formatting in this post.
Referring to HTML as code puts you in the same catagory IMHO as Al Gore's "Open Source" website.
Speaking as a coder, I use LaTeX for all my text formatting needs. Wouldn't the web be much better if it was all LaTeX based?
Who's with me? Who wants to bring the glorious coders' revolution? We have nothing to lose but our chains!
by the level of animosity here. It's not professional, and it's clear most of you are running your mouths without having a clue what you are talking about.
1. Glide doesn't suck It's not a perfect API, but at least it's fast, easy and flexible. In fact, as far as we can tell (using VTune to benchmark) it's actually physically impossible to make a D3D card faster than a V3-2000 running Glide -- M$'s code is slower than Glide, so even if the card was infinitely fast, it the program would run only as fast as Glide on a V3 (it turns out D3D runs code in two chunks, the M$ part and the card driver part).
2. Yes releasing full specs would be better Matrox has done it. Which means that 3Dfx can be convinced to do it. We do need to fight the culture of hardware secrecy that permeates many parts of the electronics industry
3. But binary only is fine for now At least they're making an effort at OS support. It's their hardware, and Glide is their code. They should have a choice about how they want it used. Why should we turn down binary support for an important API?
4. Writing our own drivers, even from the specs, it hard I know they're doing it for Matrox with GLX. It remains to be seen whether it will ever be a usably fast GL tho. 3DCards are about the most complex peripheral you could put in your computer. Writing drivers for them isn't trivial, and the drivers need to be really good to be usable -- right now the G200 drivers aren't even near there yet. I would like to see the Open Driver movement prove itself with truly excellent G200 drivers before we go demanding the right to write drivers for other hardware.
The way I see it, the ball's in our court, not theirs. I'm sure they'll find the right person with all the resumes they'll get after a/. posting. Meanwhile, let's demonstrate that Open Drivers is a viable option. That's the best way to convince companies like 3Dfx and nVidia to open their specs.
will suck 5 years down the line, it's just that there will still be games produced for it (like N64 now).
I would just consider $100-$150 (no need to spend more than that on 3D hardware) a part of the price of a new computer every 18-24 mos. Granted you may end up spending more, but that's because computer games continually take advantage of the latest technology that's available. A good thing rather than a bad thing IMHO.
It's not clear to me that pushing more GPixels is meaningful.
the highest end Voodoo3 can fill 1600x1200 at 60fps. That's _plenty_ of fill rate to go around. And the next generation will be faster.
You have to take these tech spec numbers carefully. they get big very fast, but they don't always mean much.
Don't get me wrong, I don't work for 3Dfx. But I'm a game developer. The PC 3D card companies work hard to support game developers, while devkits for consoles run like $50k, so obviously there's no love lost.
it's really only FP performance now, there's plenty of bandwidth left on the AGP bus.
Napalm will have mad geometry acceleration, 3Dfx isn't stupid. Their line on the whole issue is "well, we guess you can use the CPU for physics or something unimportant like that":)
a) Performance: the numbers I've seen on these things suggest 20million polygons per second. But this has to be taken with a grain of salt. The current leader is almost definitely the Voodoo3, which claims to be capable of 6-8million polys/sec. But our benchmarks of a V3-2000 indicate it gets about 500k best case in the real world. But fine, PsxII will be extremely impressive if it can actually do 5-10million polys/sec.
b) Obsolescence: this thing will be out in Q1'00 in Japan, and not in the States till probably Q4'00. And that's the introductory price -- they will have to keep selling this thing for years (as Nintendo still sells the N64, which came out in like '95?'96? only now is Nintendo making money on the hardware...). But that's an absurdly long time to keep a computer -- we expect them to be obsolete at the best every 2 years.
c) Competition: this thing goes head to head with the current leaders in game performance, nVidia and 3Dfx. But 3Dfx will have their new Napalm boards for this Xmas -- 1st completely new architecture board yet. All kinds of spiffy features (they talked about it at their developers conference at GDC this year). Cycle time in 3D hardware is 9 months, forget Moore's law. N64 & Psx are pretty pathetic now, and they didn't have any 3D hardware competition till partway thru their product lifespan. PsxII will enter a madly competitive, established field.
d) Price: Sure, new computers appear more expensive than Sony's announced price targets for PsxII. But remember, computers right now have a lot more functionality. The wholesale (or pricewatch.com) price of a computer without HD or monitor starts to push Sony's targets, and when you consider that OEM's always have to be making a profit on their machines, since they can't sell the same configuration for very long, a computer is decently price competitive.
This is not to discount what Sony is doing. It's interesting, and I'd be fascinated to know more about the hardware in there (yeah right! Japanese technology...we may never know. Be easier to get an account on the NSA's supercomputer:). But they've certainly carved out some hard work for themselves.
that the biggest corporation in the industry can't do any better than to let me search their website haphazardly. I search for "performance tuning" on www.microsoft.com. Yes, I found some articles. Some were good. Some were idiotic. They weren't particularly well organized.
Does Linux have this problem? To some degree, for those uninitiates who don't know whom/how to ask. When I had a question about video hardware on Suns, I mailed Jim Mintha of UltraLinux. When I had a question about SMBfs support in the kernel, I mailed the maintainer. The documentation that is there is easily organized in howto's on common subjects. And there's a whole community of USENET junkies ready to answer questions as soon as you ask them.
Repositories with papers on tuning specific packages, organized to show you exactly what's available, will help a lot.
Slashdot Mirror
as an Orientation Counsellor. They're a pretty sharp bunch.
The class of 2001 has 14 women.
The class of '02 has 28 (my year). We were impressed by that.
Way to go geek chicks!
Seriously, I'm impressed that CMU has come this far without resorting to quotas, only recruiting heavily in the areas they feel need more diversity.
As long as I've got an SB16 or similarly open-hardware card in my machine, there's no reason I can't basically write a driver to sit there and read what the card is getting, and save it to a raw file (which can later be mp3'd).
My (vague) understanding of unfuck.exe is that it actually intercepts the audio somewhere in the windoze pipeline (therefore architecture independent) -- this is also pretty easy under Linux.
On the (admittedly short) consideration I've given this, I just don't see a way around that problem for the secure music bastards of the world.
Questions/comments/snide remarks?
The 286 has a full multitasking protected mode.
:). To get your 286 out of protected mode you triple-fault it, which causes the processor to reset.
It's just 16bit. It works fine. The 286 can take up to (IIRC) 16megs of RAM (tho few MB's will take it).
IBM originally came out with OS/2 for the 286.
Interestingly, Intel never provided a way to switch out of 286 protected mode -- they figured you'd boot in real mode, switch to protected as the OS loads, and never look back (they forgot they had M$ to contend with
the original IBM 101-key keyboard. I have a pile of them, ranging in years from the mid-80s thru the early 90s, and all my friends use them. They have an amazing tactile-click, and they're completely indestructible. I've taken them apart -- the grounding strap in there could take a direct lightning strike. You can also pop off all the key caps (which fit over the actually keys) and remap them very easily.
I've seen them selling (refurbished) for around $80.
BTW, when Fire and Darkness was still a DOS game, you used Alt-SysRq to switch to the text mode console. Ever since we switched to Windows, it's been Scroll Lock. So we like all those extra keys.
I have to say, I do like Sun keyboards with the Ctrl where the Capslock is on all other keyboards (except the original 88-key that came with my IBM PC...now that was old school...)
Granted, the only look I took at SO5 was a hacked english version back when only the german version was out -- i found a partial english ver on one of the german technical university servers (mighta been clausthal) and fixed its missing parts with the german package. My impression on the k6-200 i was using then was, kinda slow, crashy, and way too MS-like.
;). I worry that the reason Sun would want StarDivision is that SO5 had (in theory) a Java version .... yuck ... wouldn't want to try that.
I very much prefer Corel's much-superior (stability and speedwise IMHE) wordperfect suite, which already exists for Solaris (anyone else want Corel to release a free-for-personal-use ver for Sun? Would top off a 3GS nicely
Corel also has a pretty darn good record of supporting Free stuff, for a commercial-software company.
For that matter, I intend to get LyX and Gnumeric and AbiWord on my two SPARCs and live a happy life. I've had it with commercial word processing. Bleah. @#$% MS-Word files. Fight the man with your word processor!
No.
No, No, No.
NP-complete problems are hard.
All of them.
Equally hard.
That is, if you can solve one NP-complete problem, you can use it to do the others in polynomial type. A Hamiltonian oracle can solve 3-coloring, circuit-sat, traveling salesmen, and the others, fast
...
I don't really know if quantum computers can solve NP problems, or even factoring for that matter, since I don't know how to phrase algos in quantum computable terms....although, like parallel algorithms, it'll be a booming area of research, you can bet.
the technical details described in the article, but I do have (imho ;) a decent understanding of the computational principle behind quantum computing. I've seen some pretty wrong (or just confused) things in the comments...and /. discussions of encryption/complexity related issues frequently get mired in BS ;)
Quantum computers do not really translate to higher FPS in Quake3 or more realistic flightsims. You don't get quantum computers that have amazing SPECfp numbers.
Quantum computers do operations that normally take one order of magnitude in a much lower order time. Example: searching a list of items for a value on von Neumann machine is O(n) (length of the list), since you have to look at each element till you find the one you want. You can do this in O(1) on a parallel machine. This is a significant improvement. IIRC there was a quantum computer that could search a 4-element list in O(1).
Obviously, that's a pretty trivial problem. Even for very long lists, O(n) isn't bad, and you can do it in parallel.
What about fundamentally much more complex problems, such as factoring, or better, something NP-complete such as circuit-sat or Hamilton paths (that's what the article was talking about, I think. Those guys think their quantum computer is a Hamilton oracle, if I'm right. So you have to pose questions in the form of graphs...not a serious problem since all NP is orthogonal). A parallel machine doesn't help there. But a quantum computer can take NP problems from exponential to polynomial time.
All of a sudden factoring (not NP, just decently hard) goes from taking 100 trillion years to taking hours.
Yes, all encryption that uses factoring as a trapdoor (one-way function) would be toast.
No, it wouldn't necessarily be any better at running first person shooters.
Still, an NP oracle machine IMHO would be much more significant than some bloody nanotech robots.
Not a linux community, a nerd community. I agree.
:)
/. does an amazing job of representing exactly the interests of people like the (rising) sophomore CS class CMU.
Still, the Holy OS is pretty important to this community. We spend a lot of time on our computers...and waste a lot of time configuring them...
My browser couldn't find the internet due to momentary enormous stupidity, and decided to pull up its last cached copy of /., from late last year IIRC when I was at home and still used this machine.
/. since nearly the beginning, but even I couldn't have imagined the community that grew up around the little Nerd News sight.
/. You can wear the hat and the tshirt. I start my day off in the office with /., then UF. Everything I need in life is a /box. When we won IGF, the first thing I did after seeing the news in a more-or-less reputable story I could link to was skip class (251, no less) and post it on /.
/. page we can find.
... I'll be back tomorrow morning, as always... man, I really am a nerd :)
Jon Katz was ruminating unpopularly in the first of his articles playing with a linux machine and how weird and confusing it was.
GNOME vs. KDE flamewars were going on fast and furious in the unmoderated comments.
No one had heard of, or cared about, Linux the Holy OS or Open Sourceness or User Friendly or anything.
Man, we've come a long way since then. I've been reading
I've posted a few stories myself. I've wasted a lot of time writing comments, and searching out comments by resident cool, funny guys Bruce Perens and Alan Cox.
Now, it ain't news unless it's on
I'd be interested to see if anyone has a copy, HTML or PS or hard copy or whatever, of the earliest
This is user # 3273
Alright. Fine. Linux is not good at striping across multiple NICs (or at least across 4). IMHO that's not a normal thing to do anyhow...
;).
/. or Excite or similar mad dynamic pagegen) that could be forced to work on ASP or PERL under both IIS & Apache? Would be fascinating
;)....
There are really two parts two heavy-duty serving: files and web.
a) Web content
You don't need 4 100Mb NICs to serve real webcontent (no, corporations using "intranets" to be buzzword compliant doesn't count. when i see a serious intranet implementation i may believe it. till then it's just buzzword compliance
No one serves more than cdrom.com, and if they're the only ones straining 100Mb (on 100% static content, at that) 100Mb is good enough fer anyone. Next issue: dynamic content. I want a fair benchmark of IIS and (mod_perl) Apache on PERL and ASP (using mod_asp).
How good is mod_asp ennyhoo? Does IIS do PERL? Does anyone have serious site implementations (on the order of
b) File serving
I'm still not convinced in the real world you need more than a 100Mb card -- if your fileserver can sustain 7 megs/sec (realistic on ethernet, peak is 10 megs/sec) you're doing pretty well with that RAID array, since no HD i've seen can do that alone, especially not when doing simultaneous serves (and therefore not reading contiguous blocks). I'm bothered by the SMB test setup in that sense -- if it was tiny reads out of the cache, big deal. I wanna know who can move real volumes of data, and fast.
Of course, I know the answer. Sun, and after them SGI's Origin2k boxes (the fastest Windows networking fileservers in the world
Why? 'Cos mine will run linux. And will do what I tell it to do. Like the rest of my computing hardware and technology. I use it to make my life better, but I'm (reasonably) certain I run it and not the other way around.
:) :)
... I guess not reallly any different than a PC-on-a-SIMM...
My Factoid:
-- Collects bizcards of everyone I ever met.
This means no more flipping thru the huge company stack'o'bizcard fun for the one I need. And yes, Mark, this means your frikkin' bizcard database (lovingly coded & entered by hand) is obsolete
-- Remembers what I did today
So I can check stuff off my list.
-- Can pick up important announcements
but since I filter it will only keep the ones i think are important. So my friends' factoids could tell mine what's being planned, but I can ignore the specials at WalMart.
-- Runs Linux
And crashes M$ Factoids, especially those that want marketing data from me
Those guys at WRL are just cool.
Remember. Your hardware can only screw you over if you let The Man tell it what to do. Which is why we tell it what to do ourselves.
Tho that would be an impressively minimal linux system
-------
"Now look in that big clustered computer and find the 1GHz Alpha 21264 that says `Badass Mutherfucker' on the heatsink. That's my badass mutherfucker."
--- Seen on a Valley bumper sticker
I apologize if this has been said before ... @#$% MSIE 3 can't handle rendering the discussion quite right, and its poorly implemented smooth-scrolling is bloody slow on my p133...yuck....
... well, you might hack it into place with Slack or Debian, but good luck to you.
/. depends on a Beer- (not Speech-) licensed SQL database. It'll take some work before the Free community can deliver truly excellent server side applications.
:).
No one ever guaranteed that Linux was proof against commercial applications, Anything Free, and we can port it. Doesn't matter if the original developer only has eyes for one distro, if it's free the work can be done.
But MetroWerks CodeWarriorisn't Free. It's commercial, and if all they give out is an RPM that makes all kinds of assumptions about how your system is set up
On the other hand, CodeWarrior has nothing in the compiler end (everyone knows that). Just that a lot of people like their IDE.
Well, I have a new concept for Linux:
We don't need to be scared of any luser inferface
It's really an idea that's been mentioned before: UI's a fundamentally not a cutting edge thing. They require careful planning and execution, but no massive technology breakthroughs. Hence Gtk+ and Qt, massively superior in flexibility and power to M$'s MFC. In fact the last really good UI I saw on Windoze was DevStudio, which is all custom widgets anyhow.
The point is, we don't need their frikkin' IDE. We can write our own better ones. And our own better spreadsheets, word processors, desktops, and everything else. Miguel and the crew wrote all of GNOME in a year. These things can be developed fast, and with spread out
The only applications that IMHO can deliver real value to Linux are big-iron server-side things that require heavy-duty technology and years of experience (and millions of $$ of screw-ups and mistakes) to get right. Oracle. Lotus Notes.
We can compete in some of these areas, but I'm often less than impressed with Apache or the ftpd's in terms of efficiency and robustness. Take a positive example, CODA, the long awaited high-tech replacement for NFS. It's good. Very good. But it's taken a lot of development, significant resources of CMU, and it's still not quite done. Even our beloved
But even so, these are not the kinds of applications that will Balkanize the distros. They depend on more fundamental, standard services of the OS. They've run on other Unices and the effort to making them cross-Linux is miniscule next to the effort of bringing them in from Solaris or HP-UX or AIX (or worse
In conclusion, CodeWarrior sucks. Go support you favorite GPL'd IDE today!
I've read a lot of game programming books over the years, and have yet to see one I would call "good", let alone "excellent". They all share the same small set of problems: out of date technologically, clumsy (or just plain wrong) code/solutions, poorly written. Not to sound cynical, but if the people who write game programming books were good enough to write real games, I suspect they would.
:), but you have to figure out how to make a game out of them yourself.
There are good books on networking technology (Tannenbaum), project structuring (Design Patterns, many others), 3D graphics (ask Rosenthal for recommendations here
Game programming under linux:
- use X, it's not that bad
- use Glide or GLX (or Mesa if you don't care about speed so much) for 3D.
Things are improving...
;)
:)
1. Large files does xfs support _big_ files? Is xfs what SGI uses on their BigIron origin servers (the fastest WinNT servers on the face of the planet, thanx to Samba
3. NFS Cats and kids -- NFS sucks. As in, "will never be worth a damn." Use something good. Coda, anyone? Actually, the Coda guy at CMU just wrote a distributed file system (in PERL!!) that runs on top of whatever FS the OS is already using -- would be very nice on top of xfs. It's a bit easier to maintain/add functionality, since it's ~8k lines to Coda's 500k. It's called InterMezzo.
4. DEs GNOME & KDE blow 4dwm, Win95, Mac, and everything else away. 'Nuff said.
5. HA clustering so how good is TurboLinux's? They must be convinced to OS it. Or we'll have to rewrite it
_The Great Lafcadio_, definitely my favorty. Also features Uncle Shelby.
about this review.
:)
a) RAID controllers IIRC there are some RAID controllers which work beautifully in Linux and others which are in alpha drivers (such as the one MindCraft used...). Does anyone know which ones are which, and where the one ZD used fits in?
b) Througput for Linux peaked at exactly 200Mbps. Anyone else find that suspicious? As if they only had 2 NICs going in Linux? Why on earth should the kernel choose such a nice round number at which to pan out?
c) Static Pages this has been mentioned before, but it's very pertinent. The only thing that counts is dynamic content. Anyone know how the Apache mod_asp performs?
d) Multiprocessing i386 I'm sorry. When you're spending $20k on a computer, you buy a Sun Ultra-60, run Solaris on it, and end the question there. Intel machines suck at high-end multiprocessing. And Linux will kick anyone's ass on a dual box
Um, "markup language", "text formatting language", what's the diff?
/. on whether HTML was Turing complete. While the tiny amount I know isn't :), with garbage like Java or VBScript it most certainly is.
...
I recall a debate about a year ago on
I extract data from TeX documents all the time. And my interpreter never has any problems. Sure, a TeX document can be invalid, but so can HTML (and w/Javascript it can be downright hostile). PostScript is (I'm pretty sure) Turing complete too, and we never have any problems using that for all our documents at some point in their existence
I don't necessarily say that the file format transferred should even be LaTeX, I just said it should be TeX-based, so we could write pages the same civilized way we write everything else.
It's possible a somewhat more general TeX like language could be useful, since TeX is really for document processing rather than general publishing, but there's no reason to choose a standard that relies on humans writing such unreadable gobbledygook as HTML.
for expressing this. Oh well.
:) code.
I'm kinda peeved at all y'all who keep referring to writing HTML "code".
WHAT !?!?
C/C++ is code. Smalltalk is code. LISP is code. ML is code.
PERL is code. Tck/Tk is code. Python is code.
hell, even Java is (probably
....
HTML is a @#$%^& text formatting language, for God's sake!. I'll even concede that writing CGI/Perl web stuff is code. But I'm pretty sure you do that in emacs or vi, not Dreamweaver or whatever. No, Javascript doesn't count.
I had to write some HTML once. It sucked. It's a pain. It's terrible. What lunatic decided that HTML was an appropriate language with which to invent the Web? (rhetorical question, i know the history behind the http).
Hell, i'm using annoying HTML formatting in this post.
Referring to HTML as code puts you in the same catagory IMHO as Al Gore's "Open Source" website.
Speaking as a coder, I use LaTeX for all my text formatting needs. Wouldn't the web be much better if it was all LaTeX based?
Who's with me? Who wants to bring the glorious coders' revolution? We have nothing to lose but our chains!
by the level of animosity here. It's not professional, and it's clear most of you are running your mouths without having a clue what you are talking about.
/. posting. Meanwhile, let's demonstrate that Open Drivers is a viable option. That's the best way to convince companies like 3Dfx and nVidia to open their specs.
1. Glide doesn't suck
It's not a perfect API, but at least it's fast, easy and flexible. In fact, as far as we can tell (using VTune to benchmark) it's actually physically impossible to make a D3D card faster than a V3-2000 running Glide -- M$'s code is slower than Glide, so even if the card was infinitely fast, it the program would run only as fast as Glide on a V3 (it turns out D3D runs code in two chunks, the M$ part and the card driver part).
2. Yes releasing full specs would be better
Matrox has done it. Which means that 3Dfx can be convinced to do it. We do need to fight the culture of hardware secrecy that permeates many parts of the electronics industry
3. But binary only is fine for now
At least they're making an effort at OS support. It's their hardware, and Glide is their code. They should have a choice about how they want it used. Why should we turn down binary support for an important API?
4. Writing our own drivers, even from the specs, it hard
I know they're doing it for Matrox with GLX. It remains to be seen whether it will ever be a usably fast GL tho. 3DCards are about the most complex peripheral you could put in your computer. Writing drivers for them isn't trivial, and the drivers need to be really good to be usable -- right now the G200 drivers aren't even near there yet. I would like to see the Open Driver movement prove itself with truly excellent G200 drivers before we go demanding the right to write drivers for other hardware.
The way I see it, the ball's in our court, not theirs. I'm sure they'll find the right person with all the resumes they'll get after a
will suck 5 years down the line, it's just that there will still be games produced for it (like N64 now).
I would just consider $100-$150 (no need to spend more than that on 3D hardware) a part of the price of a new computer every 18-24 mos. Granted you may end up spending more, but that's because computer games continually take advantage of the latest technology that's available. A good thing rather than a bad thing IMHO.
than 4-way N64 games. Goldeneye and MarioKart look terrible in little tiny windows, and the TV is a lousy display anyhow.
...
Andy, if you read this, we're gonna have to have NetFest IV this summer
It's not clear to me that pushing more GPixels is meaningful.
the highest end Voodoo3 can fill 1600x1200 at 60fps. That's _plenty_ of fill rate to go around. And the next generation will be faster.
You have to take these tech spec numbers carefully. they get big very fast, but they don't always mean much.
Don't get me wrong, I don't work for 3Dfx. But I'm a game developer. The PC 3D card companies work hard to support game developers, while devkits for consoles run like $50k, so obviously there's no love lost.
it's really only FP performance now, there's plenty of bandwidth left on the AGP bus.
:)
Napalm will have mad geometry acceleration, 3Dfx isn't stupid. Their line on the whole issue is "well, we guess you can use the CPU for physics or something unimportant like that"
for several reasons.
:). But they've certainly carved out some hard work for themselves.
a) Performance: the numbers I've seen on these things suggest 20million polygons per second. But this has to be taken with a grain of salt. The current leader is almost definitely the Voodoo3, which claims to be capable of 6-8million polys/sec. But our benchmarks of a V3-2000 indicate it gets about 500k best case in the real world. But fine, PsxII will be extremely impressive if it can actually do 5-10million polys/sec.
b) Obsolescence: this thing will be out in Q1'00 in Japan, and not in the States till probably Q4'00. And that's the introductory price -- they will have to keep selling this thing for years (as Nintendo still sells the N64, which came out in like '95?'96? only now is Nintendo making money on the hardware...). But that's an absurdly long time to keep a computer -- we expect them to be obsolete at the best every 2 years.
c) Competition: this thing goes head to head with the current leaders in game performance, nVidia and 3Dfx. But 3Dfx will have their new Napalm boards for this Xmas -- 1st completely new architecture board yet. All kinds of spiffy features (they talked about it at their developers conference at GDC this year). Cycle time in 3D hardware is 9 months, forget Moore's law. N64 & Psx are pretty pathetic now, and they didn't have any 3D hardware competition till partway thru their product lifespan. PsxII will enter a madly competitive, established field.
d) Price: Sure, new computers appear more expensive than Sony's announced price targets for PsxII. But remember, computers right now have a lot more functionality. The wholesale (or pricewatch.com) price of a computer without HD or monitor starts to push Sony's targets, and when you consider that OEM's always have to be making a profit on their machines, since they can't sell the same configuration for very long, a computer is decently price competitive.
This is not to discount what Sony is doing. It's interesting, and I'd be fascinated to know more about the hardware in there (yeah right! Japanese technology...we may never know. Be easier to get an account on the NSA's supercomputer
that the biggest corporation in the industry can't do any better than to let me search their website haphazardly. I search for "performance tuning" on www.microsoft.com. Yes, I found some articles. Some were good. Some were idiotic. They weren't particularly well organized.
Does Linux have this problem? To some degree, for those uninitiates who don't know whom/how to ask. When I had a question about video hardware on Suns, I mailed Jim Mintha of UltraLinux. When I had a question about SMBfs support in the kernel, I mailed the maintainer. The documentation that is there is easily organized in howto's on common subjects. And there's a whole community of USENET junkies ready to answer questions as soon as you ask them.
Repositories with papers on tuning specific packages, organized to show you exactly what's available, will help a lot.