Ahem. 128MB L3 cache (on the POWER4 in the benchmark)? Daaaamn. I'm not saying that a fat L3 cache has anything to do with SPEC benchmarks (I'm guessing it doesn't), I'm just making an observation: that's a lot of cache! And it's probably bloody expensive to get 128MB of cache-speed memory. HP's comments allude to that but it also has 64GB of RAM so it's sort of a straw man ("let's overconfigure a system and then make fun of how overpriced it is").
I think it's quite silly of HP to say that "IBM's Power4 architecture is outclassed in performance". Really? A 10% difference qualifies as outclassed? I don't agree. And the POWER4's SPECint score is better. "Outclassed?" Hah.
Of course the proof is in the pudding. Let's see what actually hits the streets. Apple has now been "just around the corner from really kicking Wintel's butt" performance-wise for about 8 or 9 years, but it has yet to happen. We were all led to believe that the PPC would blow away x86 and that never happened. With luck, IBM will actually deliver a really kick-ass CPU at clock speeds close to the x86 family, and the superior per-clock performance will actually make it faster. But there would still be the question of price/performance. If Joe PC Buyer can buy a faster PC for the price of a Mac, it doesn't matter that the Mac runs cooler, or at a lower clock speed, or in 64-bit mode. Joe will just say "my $500 PC is faster than your $1500 Mac, end of story". And he will have a good point. Until that changes, the only people who care are the people who are willing to pay a premium for OS X and the Mac experience, and people who need something faster than the fastest desktop PC but still want a user-friendly mainstream desktop OS. The folks who use Office and Outlook all day won't be able to justify the extra $1000 or whatever it would cost to get a Mac that performs similarly.
I'd also like to remind everybody that benchmarks don't necessarily reflect real-world performance. This is a very synthetic benchmark that is great for telling you what the best-case raw CPU performance of these CPUs can be, but it doesn't prove that $REAL_APP will see those performance gains over older CPUs.
In particular it's not clear what the performance cost would be of using code compiled for a PPC604 would be vs. using code compiled with the very best compiler for the POWER4. I'm sure that Steve Jobs will crow about another highly-optimized Photoshop benchmark that we can all wish represented overall system performance, but it doesn't. That said, I imagine that the really important professional creative apps (you know, the ones that cost thousands of dollars per seat and really beat the @%$%@$% out of the CPU) will be quickly updated for the new CPUs because their customers will demand it. (To be fair, the same is true for the Itanium.)
>You can run the same binaries on a 80486 that you can on a Pentium 4.
Not true. If you use MMX or SSE2 instructions then it'll barf on the 486. I imagine there are other new things on the P4 that code *can* use that aren't available on the 486.
Of course a common workaround is to ship both kinds of code (with new features, and without) in your binary and let it decide at runtime / install time which to use. I imagine that some folks also ship source and decide at runtime which code to compile.
I think the fact that the card costs $50 is pretty clearly the reason why the software can't be any good. It's not like they could have spent a lot on the development of the PVR software, so it's junk.
All somebody has to do is to actually write a software-only PVR product that is compatible with some of these cards and all of a sudden these PC+TV card combos might have a chance in the high-end, geek customer segment of the PVR market. But try getting a VC to fund that when there are already 2 semi-established companies struggling in this space and Microsoft even failed and pulled out...
Of course this could be done as an open source app but c'mon... I'm still waiting to see a single open source app with a good UI that has undergone usability testing, or one that does anything multimedia-related and actually works & isn't 2 years behind the closed source stuff. This would require both: up to date multimedia technology AND a UI that was designed by someone who actually cared about what users thought.
There's a race condition in that process between steps 1 and 3, that I think you are partly aware of since you have a step 4.
On small to medium-small projects CVS works OK. (It's why my employer uses; despite CVS's hideous usability problems, it's free and once you learn all its quirks and limitations you can get work done and it doesn't get in your way much.)
But on a medium project, you can get into problems where you do an update, merge, test, and then commit, and (because commits aren't atomic) still you end up with a half-checkin that stopped due to a conflict, and suddenly the head won't build. So you return to step 1 and fix it. Whew, that was scary. But for a moment, it *was* broken, even though you followed the process.
Now all you have to do is scale up x10 or x100 developers (and maybe x10 or x100 lines of code) and you can imagine that it would be broken frequently. The better other developers are about updating often, the greater the chance that they'll grab the broken code, try to work with it, build, and wonder WTF they did that broke it. Super loss of productivity, all because of one stupid feature that nobody even cares about unless they're on a large project.
And then there must be some project size (probably OS scale or something similar) where this problem is so bad that you basically can't get any work done at all because you're just treading water trying to get the changes merged and working. At that point CVS becomes totally out of the question.
I try not to be a grammar nazi, but seriously... your spelling, grammar, and punctuation are so bad that it's really hard to read your post at all. The whole second paragraph is such a mess that I had to read it three times and I still don't know what your point is.
I didn't know that being teased by groups of small children making fun of its name in singsong was a requirement, but with a name like that, it must have been.
Paging Doctor Branding... you're finally needed again in the O.R.! We've got a droid in critical condition from severe taunting trauma to the nomenclaturus maximus.
Why does the military buy the HMMWV when the Camaro Z28 is so much faster and so much cheaper? Morons!
The standard "cheapest dual-P4 I could throw together" PCs that people compare against everything else just aren't as well designed and manufactured as Sun hardware, or for that matter, Compaq hardware. It sounds like your comparison hardware is a bit better than that, but still, I doubt it's as nice as a mid-range server from Sun, IBM, HP, Compaq, etc.
If all you care about is $/MIPS, that's fine, but other hardware buyers have other requirements, like "must keep running during hardware failure", "must boot the whole OS off the network", "must be able to run diagnostics over the network even if the machine fails to boot", "must be made from the exact same parts as the other 150 servers in the cluster", etc. Yes, you can pay for this kind of stuff in a PC, but it costs a hell of a lot more than if you get the bare-bones 10/100 NIC, the el cheapo UltraDMA hard drive, etc.
If your cost model includes stuff like cost of downtime, maintenance, etc. and those things are really expensive, it can make sense. Even in a corporate desktop PC world, getting the same hardware each time is worth the extra cost because you can save money on the support costs.
Cars definitely have recurring revenue. Lots of car owners (myself included) take their ever-more-complex cars to the dealer to be serviced even when they're out of warranty. Auto manufacturers are also notorious for making it hard for third parties to make replacement parts for mechanical failure and especially for accident repair.
Also, VW's US sales figures are not a very good example. GM sold more cars than that - 465,843, according to this press release - in July 2002! (I bet VW sold a hell of a lot more cars in the EU than in the USA.) And remember, any little turd that GM feels like making is going to sell to the "buy American" dorks, corporate fleets, and rental car companies. Of course, they could sell even more if they would sell electric cars instead of crushing them despite their owner's desire to buy them but that's even further OT...
Some supporting comments regarding the parent post...
An open source project doesn't even have to be maintained to be successful. Netatalk (an AppleShare file server package for Unix) was basically dead in the water for years, with numerous bugs and missing important features, but it was so useful that admins stuck with it anyway. Eventually some folks picked it up and ran with it and have made some major improvements to it. Point: as long as the source is freely available and it's useful, somebody will download it, and it's possible for someone to take it and start maintaining it again. An unsupported source tarball can't "go out of business." Corollary: if the source had been lost and all there was was a binary, it probably would have disappeared never to be seen again. (However there are still some non-free binaries that are out there that people refuse to give up on.)
From a corporate adoption standpoint, zero cost is much, much better than cheap, because cheap things still have to go through purchasing. Setting up a new server with a bunch of commercial closed-source software can be a $25000 undertaking, and there is a strong chance the boss will say no. Even getting a 21" monitor can be an uphill battle, and if the boss says no, you can either buy it yourself or live with a teeny screen. But if you can find a server lying around that can run a free OS, you can just skip the hassle of getting bids, making a P.O., and pleading with the boss to approve it. If a portable music gizmo developer said "I already got ogg working last weekend, why not leave it in there" that's a lot different from "I already got WMA in there, here's the number for Microsoft's licensing department". Also, if it's free there's no need to track report units sold in order to know what the royalty payment should be, etc.
Simpler solution: don't install a C compiler on your public-facing production servers. It makes it a lot harder to build a rootkit or other such post-shellcode payload on the target machine.
This may not be reasonable for all servers (and my guess is, most folks running a Linux box aren't going to buy a second one to build stuff on just so that the first one can be stripped down) but it's worth mentioning anwyay.
>I can get a no-OS Intel box for $300 at Fry's and put Linux on it >for free. [...] There isn't any Mac that's that cheap.
Have you heard of eBay? There are older iMacs on there for $300-$600. True, there isn't any *new* Mac that costs $300, but that's not what you said. If you want a Mac and you have $300 you can get one. The true question is, would you want a $300 Mac?
>My Linux box is about comparable to an iMac, >which you can't get for less than about $900.
Which iMac? The new one? Does your Linux box have that built-in LCD thingy? OK, so maybe you don't want to pay for that, but to make a fair price comparison, compare your PC to a similarly configured Mac, not the lowest-end brand new one.
I'd be interested to see the configuration of your $300 Linux box... are you including RAM, HD, keyboard, mouse, modem, ethernet card, CD-RW, nice video card, firewire interface, and a year warranty? An iMac comes with all of that.
>This is all assuming you already have a monitor you can reuse.
Again, not a valid comparison... if you are pricing a PC without a monitor and an iMac which has an integrated monitor, the iMac is going to cost a couple hundred bucks more because you're getting more.
It sounds like maybe you're complaining about bundling (you have a monitor and don't want to pay for an LCD or swivel arm or wireless or a new KB/mouse or firewire etc.). Fair enough; I own a G4 and I don't want an iMac either (at least until they come with 1600x1200 displays). Like I said, consider used Macs for a fair comparison, or add a bunch of peripherals to your PC to give it the same features that the iMac has.
Anyone who can afford to turn away business in this economy, please, do everyone else a favor and do it! Seriously. There are plenty of folks who need work, and if you are not interested in coding for NN4, somebody else may be. If nobody is, great; you have forced that non-customer (since you're turning them away) to move on to a better browser, and they may even become a customer again when they have a browser you do feel like supporting.
This should be a simple economic issue. If it's really that much of a pain in the butt to support NN4, price that extra work at a point where you're OK with having to do it. If it's worth that much to your customer, then you have no excuse complaining; just do the work and take your money Lots of other system-requirements / target platform decisions work like this (do we port to MacOS, do we port to MS SQL Server, do we port to Linux, do we port to iPlanet Web Server, etc.) so this isn't exactly a radical idea. If it's not justifiable from a business sense, just don't bother, but if it is, adjust your prices and STFU.
There are companies out there which have standardized on NN4 and haven't upgraded to NN6.2 or NN7 yet. Bless them. If not for them we'd all be coding in MS-HTML and MS-CSS, or XML and MS-XSL, and wondering why IE 5 was the last browser they released. One of these days they'll upgrade to NN7 (or something similar) and life will suck less. Until then, do your job and separate business logic from presentation, so the only part you have to re-code and QA for NN4 is the presentation layer. XSLT can help with this.
AFAIK nobody uses Wintel for a PVR. ReplayTV uses the VXWorks operating system, on MIPS R3000 if I remember correctly. I agree, it's the "oh my GOD it uses LINUX that's SOO COOOL" angle that made it a headline.
BTW this is quite a parallel to the fact that Linux sucks so utterly as a desktop environment - who cares what the end-user features are or whether anyone would want to use this PVR, it's got Linux in it therefore it must be good. Corollary: if it has Linux in it, everyone should want to use it, full stop - never mind whether it has any worthwhile end-user features. And they wonder why Linux is having such a hard time winning over end-users.
And of course there's the fact that this is Just Another TiVo, which already was Linux-based. What a classic example of sucky/. culture.
TinkerTool will let you show system and hidden files.
How to get .NET benchmarks out there
on
More MS EULA Fun
·
· Score: 2
Step 1: install the software Step 2: do the benchmarks Step 3: write up a detailed report Step 4: put the report on a freshly installed Win2K server sitting on DSL or a cable modem Step 5: announce the fact that your report exists but that you won't disclose it Step 6: wait for someone to disclose it for you
Re:Why do interviewers use "riddles"?
on
Tech-Interview Riddles
·
· Score: 4, Insightful
>You can focus on code centric questions, but will that person still >perform well if they move to writing specs, managing people, or >developing an advertizing plan?
Good, experienced developers should be really good at writing specs since they know what level of detail is needed. In fact I think they stand a better chance of writing an implementable spec than an Analyst who has never coded.
As for managing people or developing an advertising plan, those are totally different skill sets from development. Have you ever worked with a manager who totally sucked as a manager but was a really smart developer? I have, several times. It's painful, and it drives home the point that managing well takes skill.
The idea that you can just dump a bunch of smart people onto any problem and outperform a bunch of experienced people who aren't quite as smart strikes me as terribly naive. Or, in Microsoft's case, conceited: it's pointless to learn from the past because we're all smarter than they were. So we'll get it right the first time and come up with a more clever solution on our own than if we just READ A DAMN BOOK. And so you get badly designed, bug-ridden software that solves problems that were already solved better 10 years beforehand. Oops.
Perhaps if you're truly working on something novel it would clearly be better to have smart people than not smart people, but in that case it's not possible to hire for experience anyway so it's not germane to this discussion.
You want good ads, hire someone who's good at doing ads. You want a good manager, hire a good manager. Or, train someone who is partway there. But don't just throw bright young people at any task and assume that they'll do better than an experienced person. I've worked in companies where that was the explicit philosophy and it's a disaster. After a few nightmare projects that smart manager might figure out some techniques that a less smart manager took a lifetime to develop, but that less smart manager wrote it all down in a book 30 years ago. Don't you wish the smart manager had read it BEFORE their first project as a manager?
Show me a chess club that can beat Marines at paintball and maybe you'll change my mind.
You make a good point about the "how do you learn things you don't already know" process that a person has. What I disagree with is the idea that skills aren't important, and that problem-solving ability is what matters since that will lead to the right solution. (Well, I guess it will, but only after you make all the same mistakes that have been made by your predecessors!)
I suppose that my assumption is that someone with a certain amount of skill and knowledge of industry standard solutions to common problems got that knowledge somehow and so they must have a process for learning. But I like the idea of explicitly interviewing for that.
These tech interview questions are STUPID
on
Tech-Interview Riddles
·
· Score: 5, Insightful
Am I the only one who thinks this interviewing technique is retarded?
Because Microsoft does something most definitely isn't a reason to emulate it. Microsoft isn't exactly known for producing well designed software, nor software that reuses proven patterns or algorithms that solved known problems 20 years ago. Better to hire a bunch of 21 year old college grads who can solve word problems from 8th grade algebra, and pretend that Microsoft invented computers! Whee.
When I hire developers I want them to be good developers, not promising young interns. My interview questions typically involve technology questions, process questions, some theoretical PROGRAMMING questions, and some social / communication questions. I'm not saying that hiring smart people is a bad idea, but ignoring skills and only looking at generic problem solving ability is a recipe for unbelievably bad code. It's like hiring musicians based on measured hearing sensitivity and reflexes. OK, maybe that matters if you want to figure out which 5 year old is going to be a prodigy, but hand them an instrument and the noise that comes out is going to sound like ASS.
Examples of things that "smart" developers I've worked with before have totally missed: - the existence of more efficient data structures than arrays - generalizing code into reusable chunks (functions, objects, whatever) - regular expressions - the difference between "client" and "server" - the reason for using descriptive variable names - collection libraries with built in sorting ("whatcha workin' on?" / "coding up a quicksort algorithm" / "in a J2EE app!?!?")
You can't just get this from reading a book, either, although that definitely helps. You have to have some degree of EXPERIENCE too: at least a few projects, and some awareness of things like performance tuning, security, coding for maintainability, etc.
I would use these "tech interview questions" only for hiring interns or recent college grads where the expectation is zero experience, zero clue, zero skill, and a correspondingly low salary. After all you're investing in someone. But for someone that commands a market rate developer salary in the high five figures, screw the brain teasers - just spend a couple of hours grilling them on skills, experience, discipline, etc. They will respect you big time in return because they know when you extend an offer that they won't be working with a bunch of dumb-asses who can get the explorers across the river without being eaten by the headhunters but who can't code their way out of a soggy paper bag.
Here's the riddle I wish Microsoft would use...
on
Tech-Interview Riddles
·
· Score: 5, Funny
"So there's a programmer writing some code, in C. That programmer needs to use a buffer to store some data. How does the programmer write the code such that an unexpectedly large amount of data doesn't overwrite the stack and result in a remote root exploit?"
This sounds exactly like the system used in San Francisco, except in the last election they added two additional steps: - move all the ballot boxes in the middle of the night for 'security reasons' (remember this was 2 months after 9/11) - throw some of them in the bay.
>I'd really like to know why private business has so much >sway over government in these sorts of things.
See, there's this thing called "bribery". It'a a major factor in this other thing called "corruption". Since you're apparently not aware of either, you should look into these new concepts right away. It'll give you a better understanding of why this "campaign finance reform" thing keeps coming up.
P.S. it's spelled "Holy Moley" in the Captain Marvel comics where (as far as I know) that phrase originated.
Good idea, bad implementation. RTF and TSV are not the best formats. What happens to the formulas, graphs, drawings, and formatting (yes formatting matters) in an Excel document when you save it as.TSV? They disappear.
There need to be open, documented formats for this stuff, that open source apps use [mostly] correctly, for such a switch to work. Otherwise you're left with proprietary = productivity vs. open = time waster. Guess which one makes more business sense in the short term (which is all that bean counters ever look at anyway)?
OK great. Now what about an Oni sequel?
on
Halo for the PC and Mac
·
· Score: 3, Informative
Halo, some FPS on the Xbox, whee. But I really enjoyed Oni, which is available on multiple platforms, but will never have a multiplayer version nor sequels because the same company that made Oni made Halo, and that company is now part of MS.
Go play Oni, it's like $9.00 now new. It's hella fun, but disappointing because as cool as it is, it has no future.:(
Maybe MS will resurrect Oni as a product and keep it alive? OK, I'm just dreaming. But if they are porting Halo maybe they would make sequels / expansion packs to Oni.
Exactly. Apparently "honesty" here is defined as technical inability to break the copy protection AND lack of any friends or contacts who are able to break the copy protection.
Hey MPAA, DVDs already have copy protection, and that doesn't stop them from being swapped on P2P file sharing systems, IRC, FTP servers, and person to person via burner (just like floppies in the old days). Do you really think that this will be better than DVD/CSS security?
BTW I was in Fry's Electronics yesterday and was astonished to see how cheap the various CD-RW and DVD-R/RW/whatever drives are (hadn't bothered to look at prices for a while). Remember the days (late 80s) when the main mode of copying software was floppy swap meets? One person cracked the copy protection, then folks physically went over to someone else's house where they made a copy for each person who then went home and did the same thing. CD-R[W] drives and media are in a similar price range now, and I see the same thing happening with CDs that I saw with floppies years ago. It's totally feasible to go over to your friend's house, bring a PC or two with you, and sit there and burn 100 or more CDs at a time.
This is just a historical reminder about why copy protection is a foolish endeavor. The only difference between nobody being able to copy something and everybody being able to copy something is that one guy who cracks it. If you know anybody who has acquired a cracked copy, then everybody you know can get one.
It's kinda like buying illegal drugs - if you want them, and you know somebody you trust who has them, you can get them. I bet that for every person who is sure they couldn't get them (and thus that the drug war is being won) there is a friend who can get them whom they just haven't bothered to ask yet. Hello, entertainment industry, welcome to your own private drug war. I can't wait to pay taxes for cops to put DVD copiers in jail right next to potheads.
Slashdot Mirror
A possible reason: Sexium, Heptium/Septium, and Octium are really stupid names.
Ahem. 128MB L3 cache (on the POWER4 in the benchmark)? Daaaamn. I'm not saying that a fat L3 cache has anything to do with SPEC benchmarks (I'm guessing it doesn't), I'm just making an observation: that's a lot of cache! And it's probably bloody expensive to get 128MB of cache-speed memory. HP's comments allude to that but it also has 64GB of RAM so it's sort of a straw man ("let's overconfigure a system and then make fun of how overpriced it is").
I think it's quite silly of HP to say that "IBM's Power4 architecture is outclassed in performance". Really? A 10% difference qualifies as outclassed? I don't agree. And the POWER4's SPECint score is better. "Outclassed?" Hah.
Of course the proof is in the pudding. Let's see what actually hits the streets. Apple has now been "just around the corner from really kicking Wintel's butt" performance-wise for about 8 or 9 years, but it has yet to happen. We were all led to believe that the PPC would blow away x86 and that never happened. With luck, IBM will actually deliver a really kick-ass CPU at clock speeds close to the x86 family, and the superior per-clock performance will actually make it faster. But there would still be the question of price/performance. If Joe PC Buyer can buy a faster PC for the price of a Mac, it doesn't matter that the Mac runs cooler, or at a lower clock speed, or in 64-bit mode. Joe will just say "my $500 PC is faster than your $1500 Mac, end of story". And he will have a good point. Until that changes, the only people who care are the people who are willing to pay a premium for OS X and the Mac experience, and people who need something faster than the fastest desktop PC but still want a user-friendly mainstream desktop OS. The folks who use Office and Outlook all day won't be able to justify the extra $1000 or whatever it would cost to get a Mac that performs similarly.
I'd also like to remind everybody that benchmarks don't necessarily reflect real-world performance. This is a very synthetic benchmark that is great for telling you what the best-case raw CPU performance of these CPUs can be, but it doesn't prove that $REAL_APP will see those performance gains over older CPUs.
In particular it's not clear what the performance cost would be of using code compiled for a PPC604 would be vs. using code compiled with the very best compiler for the POWER4. I'm sure that Steve Jobs will crow about another highly-optimized Photoshop benchmark that we can all wish represented overall system performance, but it doesn't. That said, I imagine that the really important professional creative apps (you know, the ones that cost thousands of dollars per seat and really beat the @%$%@$% out of the CPU) will be quickly updated for the new CPUs because their customers will demand it. (To be fair, the same is true for the Itanium.)
>You can run the same binaries on a 80486 that you can on a Pentium 4.
Not true. If you use MMX or SSE2 instructions then it'll barf on the 486. I imagine there are other new things on the P4 that code *can* use that aren't available on the 486.
Of course a common workaround is to ship both kinds of code (with new features, and without) in your binary and let it decide at runtime / install time which to use. I imagine that some folks also ship source and decide at runtime which code to compile.
I think the fact that the card costs $50 is pretty clearly the reason why the software can't be any good. It's not like they could have spent a lot on the development of the PVR software, so it's junk.
All somebody has to do is to actually write a software-only PVR product that is compatible with some of these cards and all of a sudden these PC+TV card combos might have a chance in the high-end, geek customer segment of the PVR market. But try getting a VC to fund that when there are already 2 semi-established companies struggling in this space and Microsoft even failed and pulled out...
Of course this could be done as an open source app but c'mon... I'm still waiting to see a single open source app with a good UI that has undergone usability testing, or one that does anything multimedia-related and actually works & isn't 2 years behind the closed source stuff. This would require both: up to date multimedia technology AND a UI that was designed by someone who actually cared about what users thought.
There's a race condition in that process between steps 1 and 3, that I think you are partly aware of since you have a step 4.
On small to medium-small projects CVS works OK. (It's why my employer uses; despite CVS's hideous usability problems, it's free and once you learn all its quirks and limitations you can get work done and it doesn't get in your way much.)
But on a medium project, you can get into problems where you do an update, merge, test, and then commit, and (because commits aren't atomic) still you end up with a half-checkin that stopped due to a conflict, and suddenly the head won't build. So you return to step 1 and fix it. Whew, that was scary. But for a moment, it *was* broken, even though you followed the process.
Now all you have to do is scale up x10 or x100 developers (and maybe x10 or x100 lines of code) and you can imagine that it would be broken frequently. The better other developers are about updating often, the greater the chance that they'll grab the broken code, try to work with it, build, and wonder WTF they did that broke it. Super loss of productivity, all because of one stupid feature that nobody even cares about unless they're on a large project.
And then there must be some project size (probably OS scale or something similar) where this problem is so bad that you basically can't get any work done at all because you're just treading water trying to get the changes merged and working. At that point CVS becomes totally out of the question.
I try not to be a grammar nazi, but seriously... your spelling, grammar, and punctuation are so bad that it's really hard to read your post at all. The whole second paragraph is such a mess that I had to read it three times and I still don't know what your point is.
I didn't know that being teased by groups of small children making fun of its name in singsong was a requirement, but with a name like that, it must have been.
Paging Doctor Branding... you're finally needed again in the O.R.! We've got a droid in critical condition from severe taunting trauma to the nomenclaturus maximus.
Why does the military buy the HMMWV when the Camaro Z28 is so much faster and so much cheaper? Morons!
The standard "cheapest dual-P4 I could throw together" PCs that people compare against everything else just aren't as well designed and manufactured as Sun hardware, or for that matter, Compaq hardware. It sounds like your comparison hardware is a bit better than that, but still, I doubt it's as nice as a mid-range server from Sun, IBM, HP, Compaq, etc.
If all you care about is $/MIPS, that's fine, but other hardware buyers have other requirements, like "must keep running during hardware failure", "must boot the whole OS off the network", "must be able to run diagnostics over the network even if the machine fails to boot", "must be made from the exact same parts as the other 150 servers in the cluster", etc. Yes, you can pay for this kind of stuff in a PC, but it costs a hell of a lot more than if you get the bare-bones 10/100 NIC, the el cheapo UltraDMA hard drive, etc.
If your cost model includes stuff like cost of downtime, maintenance, etc. and those things are really expensive, it can make sense. Even in a corporate desktop PC world, getting the same hardware each time is worth the extra cost because you can save money on the support costs.
Cars definitely have recurring revenue. Lots of car owners (myself included) take their ever-more-complex cars to the dealer to be serviced even when they're out of warranty. Auto manufacturers are also notorious for making it hard for third parties to make replacement parts for mechanical failure and especially for accident repair.
Also, VW's US sales figures are not a very good example. GM sold more cars than that - 465,843, according to this press release - in July 2002! (I bet VW sold a hell of a lot more cars in the EU than in the USA.) And remember, any little turd that GM feels like making is going to sell to the "buy American" dorks, corporate fleets, and rental car companies. Of course, they could sell even more if they would sell electric cars instead of crushing them despite their owner's desire to buy them but that's even further OT...
Some supporting comments regarding the parent post...
An open source project doesn't even have to be maintained to be successful. Netatalk (an AppleShare file server package for Unix) was basically dead in the water for years, with numerous bugs and missing important features, but it was so useful that admins stuck with it anyway. Eventually some folks picked it up and ran with it and have made some major improvements to it. Point: as long as the source is freely available and it's useful, somebody will download it, and it's possible for someone to take it and start maintaining it again. An unsupported source tarball can't "go out of business." Corollary: if the source had been lost and all there was was a binary, it probably would have disappeared never to be seen again. (However there are still some non-free binaries that are out there that people refuse to give up on.)
From a corporate adoption standpoint, zero cost is much, much better than cheap, because cheap things still have to go through purchasing. Setting up a new server with a bunch of commercial closed-source software can be a $25000 undertaking, and there is a strong chance the boss will say no. Even getting a 21" monitor can be an uphill battle, and if the boss says no, you can either buy it yourself or live with a teeny screen. But if you can find a server lying around that can run a free OS, you can just skip the hassle of getting bids, making a P.O., and pleading with the boss to approve it. If a portable music gizmo developer said "I already got ogg working last weekend, why not leave it in there" that's a lot different from "I already got WMA in there, here's the number for Microsoft's licensing department". Also, if it's free there's no need to track report units sold in order to know what the royalty payment should be, etc.
Simpler solution: don't install a C compiler on your public-facing production servers. It makes it a lot harder to build a rootkit or other such post-shellcode payload on the target machine.
This may not be reasonable for all servers (and my guess is, most folks running a Linux box aren't going to buy a second one to build stuff on just so that the first one can be stripped down) but it's worth mentioning anwyay.
>I can get a no-OS Intel box for $300 at Fry's and put Linux on it
>for free. [...] There isn't any Mac that's that cheap.
Have you heard of eBay? There are older iMacs on there for $300-$600. True, there isn't any *new* Mac that costs $300, but that's not what you said. If you want a Mac and you have $300 you can get one. The true question is, would you want a $300 Mac?
>My Linux box is about comparable to an iMac,
>which you can't get for less than about $900.
Which iMac? The new one? Does your Linux box have that built-in LCD thingy? OK, so maybe you don't want to pay for that, but to make a fair price comparison, compare your PC to a similarly configured Mac, not the lowest-end brand new one.
I'd be interested to see the configuration of your $300 Linux box... are you including RAM, HD, keyboard, mouse, modem, ethernet card, CD-RW, nice video card, firewire interface, and a year warranty? An iMac comes with all of that.
>This is all assuming you already have a monitor you can reuse.
Again, not a valid comparison... if you are pricing a PC without a monitor and an iMac which has an integrated monitor, the iMac is going to cost a couple hundred bucks more because you're getting more.
It sounds like maybe you're complaining about bundling (you have a monitor and don't want to pay for an LCD or swivel arm or wireless or a new KB/mouse or firewire etc.). Fair enough; I own a G4 and I don't want an iMac either (at least until they come with 1600x1200 displays). Like I said, consider used Macs for a fair comparison, or add a bunch of peripherals to your PC to give it the same features that the iMac has.
Anyone who can afford to turn away business in this economy, please, do everyone else a favor and do it! Seriously. There are plenty of folks who need work, and if you are not interested in coding for NN4, somebody else may be. If nobody is, great; you have forced that non-customer (since you're turning them away) to move on to a better browser, and they may even become a customer again when they have a browser you do feel like supporting.
This should be a simple economic issue. If it's really that much of a pain in the butt to support NN4, price that extra work at a point where you're OK with having to do it. If it's worth that much to your customer, then you have no excuse complaining; just do the work and take your money Lots of other system-requirements / target platform decisions work like this (do we port to MacOS, do we port to MS SQL Server, do we port to Linux, do we port to iPlanet Web Server, etc.) so this isn't exactly a radical idea. If it's not justifiable from a business sense, just don't bother, but if it is, adjust your prices and STFU.
There are companies out there which have standardized on NN4 and haven't upgraded to NN6.2 or NN7 yet. Bless them. If not for them we'd all be coding in MS-HTML and MS-CSS, or XML and MS-XSL, and wondering why IE 5 was the last browser they released. One of these days they'll upgrade to NN7 (or something similar) and life will suck less. Until then, do your job and separate business logic from presentation, so the only part you have to re-code and QA for NN4 is the presentation layer. XSLT can help with this.
AFAIK nobody uses Wintel for a PVR. ReplayTV uses the VXWorks operating system, on MIPS R3000 if I remember correctly. I agree, it's the "oh my GOD it uses LINUX that's SOO COOOL" angle that made it a headline.
/. culture.
BTW this is quite a parallel to the fact that Linux sucks so utterly as a desktop environment - who cares what the end-user features are or whether anyone would want to use this PVR, it's got Linux in it therefore it must be good. Corollary: if it has Linux in it, everyone should want to use it, full stop - never mind whether it has any worthwhile end-user features. And they wonder why Linux is having such a hard time winning over end-users.
And of course there's the fact that this is Just Another TiVo, which already was Linux-based. What a classic example of sucky
TinkerTool will let you show system and hidden files.
Step 1: install the software
Step 2: do the benchmarks
Step 3: write up a detailed report
Step 4: put the report on a freshly installed Win2K server sitting on DSL or a cable modem
Step 5: announce the fact that your report exists but that you won't disclose it
Step 6: wait for someone to disclose it for you
>You can focus on code centric questions, but will that person still
>perform well if they move to writing specs, managing people, or
>developing an advertizing plan?
Good, experienced developers should be really good at writing specs since they know what level of detail is needed. In fact I think they stand a better chance of writing an implementable spec than an Analyst who has never coded.
As for managing people or developing an advertising plan, those are totally different skill sets from development. Have you ever worked with a manager who totally sucked as a manager but was a really smart developer? I have, several times. It's painful, and it drives home the point that managing well takes skill.
The idea that you can just dump a bunch of smart people onto any problem and outperform a bunch of experienced people who aren't quite as smart strikes me as terribly naive. Or, in Microsoft's case, conceited: it's pointless to learn from the past because we're all smarter than they were. So we'll get it right the first time and come up with a more clever solution on our own than if we just READ A DAMN BOOK. And so you get badly designed, bug-ridden software that solves problems that were already solved better 10 years beforehand. Oops.
Perhaps if you're truly working on something novel it would clearly be better to have smart people than not smart people, but in that case it's not possible to hire for experience anyway so it's not germane to this discussion.
You want good ads, hire someone who's good at doing ads. You want a good manager, hire a good manager. Or, train someone who is partway there. But don't just throw bright young people at any task and assume that they'll do better than an experienced person. I've worked in companies where that was the explicit philosophy and it's a disaster. After a few nightmare projects that smart manager might figure out some techniques that a less smart manager took a lifetime to develop, but that less smart manager wrote it all down in a book 30 years ago. Don't you wish the smart manager had read it BEFORE their first project as a manager?
Show me a chess club that can beat Marines at paintball and maybe you'll change my mind.
You make a good point about the "how do you learn things you don't already know" process that a person has. What I disagree with is the idea that skills aren't important, and that problem-solving ability is what matters since that will lead to the right solution. (Well, I guess it will, but only after you make all the same mistakes that have been made by your predecessors!)
I suppose that my assumption is that someone with a certain amount of skill and knowledge of industry standard solutions to common problems got that knowledge somehow and so they must have a process for learning. But I like the idea of explicitly interviewing for that.
Am I the only one who thinks this interviewing technique is retarded?
Because Microsoft does something most definitely isn't a reason to emulate it. Microsoft isn't exactly known for producing well designed software, nor software that reuses proven patterns or algorithms that solved known problems 20 years ago. Better to hire a bunch of 21 year old college grads who can solve word problems from 8th grade algebra, and pretend that Microsoft invented computers! Whee.
When I hire developers I want them to be good developers, not promising young interns. My interview questions typically involve technology questions, process questions, some theoretical PROGRAMMING questions, and some social / communication questions. I'm not saying that hiring smart people is a bad idea, but ignoring skills and only looking at generic problem solving ability is a recipe for unbelievably bad code. It's like hiring musicians based on measured hearing sensitivity and reflexes. OK, maybe that matters if you want to figure out which 5 year old is going to be a prodigy, but hand them an instrument and the noise that comes out is going to sound like ASS.
Examples of things that "smart" developers I've worked with before have totally missed:
- the existence of more efficient data structures than arrays
- generalizing code into reusable chunks (functions, objects, whatever)
- regular expressions
- the difference between "client" and "server"
- the reason for using descriptive variable names
- collection libraries with built in sorting ("whatcha workin' on?" / "coding up a quicksort algorithm" / "in a J2EE app!?!?")
You can't just get this from reading a book, either, although that definitely helps. You have to have some degree of EXPERIENCE too: at least a few projects, and some awareness of things like performance tuning, security, coding for maintainability, etc.
I would use these "tech interview questions" only for hiring interns or recent college grads where the expectation is zero experience, zero clue, zero skill, and a correspondingly low salary. After all you're investing in someone. But for someone that commands a market rate developer salary in the high five figures, screw the brain teasers - just spend a couple of hours grilling them on skills, experience, discipline, etc. They will respect you big time in return because they know when you extend an offer that they won't be working with a bunch of dumb-asses who can get the explorers across the river without being eaten by the headhunters but who can't code their way out of a soggy paper bag.
"So there's a programmer writing some code, in C. That programmer needs to use a buffer to store some data. How does the programmer write the code such that an unexpectedly large amount of data doesn't overwrite the stack and result in a remote root exploit?"
This sounds exactly like the system used in San Francisco, except in the last election they added two additional steps:
- move all the ballot boxes in the middle of the night for 'security reasons' (remember this was 2 months after 9/11)
- throw some of them in the bay.
See: http://www.sfbg.com/News/36/09/ogelec.htm
>I'd really like to know why private business has so much
>sway over government in these sorts of things.
See, there's this thing called "bribery". It'a a major factor in this other thing called "corruption". Since you're apparently not aware of either, you should look into these new concepts right away. It'll give you a better understanding of why this "campaign finance reform" thing keeps coming up.
P.S. it's spelled "Holy Moley" in the Captain Marvel comics where (as far as I know) that phrase originated.
Good idea, bad implementation. RTF and TSV are not the best formats. What happens to the formulas, graphs, drawings, and formatting (yes formatting matters) in an Excel document when you save it as .TSV? They disappear.
There need to be open, documented formats for this stuff, that open source apps use [mostly] correctly, for such a switch to work. Otherwise you're left with proprietary = productivity vs. open = time waster. Guess which one makes more business sense in the short term (which is all that bean counters ever look at anyway)?
Halo, some FPS on the Xbox, whee. But I really enjoyed Oni, which is available on multiple platforms, but will never have a multiplayer version nor sequels because the same company that made Oni made Halo, and that company is now part of MS.
:(
Go play Oni, it's like $9.00 now new. It's hella fun, but disappointing because as cool as it is, it has no future.
Maybe MS will resurrect Oni as a product and keep it alive? OK, I'm just dreaming. But if they are porting Halo maybe they would make sequels / expansion packs to Oni.
Exactly. Apparently "honesty" here is defined as technical inability to break the copy protection AND lack of any friends or contacts who are able to break the copy protection.
Hey MPAA, DVDs already have copy protection, and that doesn't stop them from being swapped on P2P file sharing systems, IRC, FTP servers, and person to person via burner (just like floppies in the old days). Do you really think that this will be better than DVD/CSS security?
BTW I was in Fry's Electronics yesterday and was astonished to see how cheap the various CD-RW and DVD-R/RW/whatever drives are (hadn't bothered to look at prices for a while). Remember the days (late 80s) when the main mode of copying software was floppy swap meets? One person cracked the copy protection, then folks physically went over to someone else's house where they made a copy for each person who then went home and did the same thing. CD-R[W] drives and media are in a similar price range now, and I see the same thing happening with CDs that I saw with floppies years ago. It's totally feasible to go over to your friend's house, bring a PC or two with you, and sit there and burn 100 or more CDs at a time.
This is just a historical reminder about why copy protection is a foolish endeavor. The only difference between nobody being able to copy something and everybody being able to copy something is that one guy who cracks it. If you know anybody who has acquired a cracked copy, then everybody you know can get one.
It's kinda like buying illegal drugs - if you want them, and you know somebody you trust who has them, you can get them. I bet that for every person who is sure they couldn't get them (and thus that the drug war is being won) there is a friend who can get them whom they just haven't bothered to ask yet. Hello, entertainment industry, welcome to your own private drug war. I can't wait to pay taxes for cops to put DVD copiers in jail right next to potheads.