There are a couple of different sorts of people that might use Linux. The first are the developers, the techies. They're the ones that built GNU/Linux. A Linux company that offends these as a group does so at their own peril. A couple of companies tried cashing in on these. No good. Not enough money here, too much resistance, and it's like biting the hand that feeds you. Red Hat hasn't irritated these at all. As a matter of fact, it tends to coddle them. RH expanded the range of packages they offer by adding Fedora to their lineup (hopefully adopting yum as well) with frequent updates. RH has traditionally been extremely pro-open source, and anal about getting rid of non-free source packages in their distribution. Xv went away, xanim went away, netscape navigator went away (probably before it was a good idea to do so, as a matter of fact). They're less so than Debian, but more so than almost any of the other "corporatish" types. RH donates lots of money and developer time into core Linux products, so that all the hackers benefit. Gcc owes a lot to RH.
The second group are the mom-and-pop types. Joe Sixpack. Ordinary old home users. These generally haven't had much interest in Linux so far. RH has put a lot of money into developing GNOME to make things more palatable, but there's a long ways to go. RH hasn't really hurt this group at all.
The third group are business desktop users. This is a potentially growing market. I don't know exactly why Red Hat isn't as interested as they could be -- presumably because users tend to resist change a bit, and because businesses have balked at retrainting costs. However, RH (and Mandrake and SuSE) have all put a lot of resources into projects that will benefit these folks. They're slowly but steadily trickling into the fold -- there are migrations to Linux, but not away.
Last of all, there are the server folks. These are the ones that want Red Hat Enterprise Server. All they care about is supported servers. They want support contracts and someone to call if things break. They want very occasional updates, and don't care about the latest-and-greatest browser plugin. They want very heavy QA. For them, $699 is very attractive, especially if it lets them migrate from Solaris and AIX and the associated (expensive) hardware. If you don't fall into this group, you don't want RH Enterprise. Incidently, Debian Stable is probably a good alternative for techie admins that don't want support contracts. There are IT managers that just aren't *comfortable* dealing with most Linux companies, and want a familiar old contract and guarantees -- just with less vendor lock-in and cheaper prices.
The problem is that you're thinking that RH is trying to get you on RHE. Nope. RH doesn't market RHE to you and doesn't have an interest in doing so. Maybe one day, when they absorb enough generic suits and drop enough of the Linux folks, they'll do something unutterably stupid like trying to sock techies up for cash, but not today. About the only way they get money from you is if you want pressed CDs (worth it if you're installing on a bunch of machines at work and don't want to worry about a burned CD going bad) or if you want to buy their up2date service (frankly, not worth it -- even aside from being free, yum blows up2date out of the water).
So don't bash RH. This isn't a case of Red Hat going evil. They aren't transitioning you to give you a worse deal. They're just expanding into the server market as fast as possible.
You're talking about RH not including a web server. Ridiculous. RH still has a boxed version that it sells. It still has a web server. It comes with *installation support*, same as it always has. You can buy a support contract if you want for usage support.
Fedora is a group of developers that got together and packaged a lot more software for Red Hat than Red Hat did themselves. Red Hat realized that a lot of RH users really liked Fedora. Now, Fedora is becoming an officia
Darn, I was just about to write "he's probably not an SCO apologist", when you beat me to it with a more authoritative answer.
Oh, well.
SCO resellers have been some of the folks that are somewhat screwed by the situation. There's not really any benefit to SCO starting a bunch of lawsuits and building up bad PR. As a matter of fact, to SCO resellers -- who, as far as I can tell, tend to sell to fairly boring, conservative companies -- are a big losing player. They want SCO to stay in business, but SCO playing a risky game isn't great for them or their customers, who want boring and safe. If SCO makes tons of money, that translates into little benefit to SCO resellers. If SCO starts to focus on Linux licensing rather than their own Unix (as they said they would in their SEC filing), SCO resellers get screwed.
Basically, SCO resellers get screwed by this, pretty much no matter what happens. They aren't thrilled. Linux may technically be a competitor, but it's not as if they couldn't move to reselling and supporting Linux (well, at least more easily then to Windows).
As an aside, I suspect that a lot of folks at SCO -- not upper management -- are quite irritated as well. Their jobs are being used in a big throw of the dice that seems to primarily benefit upper management and a holding company.
Finally, SCO retailers, from exerpts from the newsgroup and blogs, seem to be a pretty decent source of levelheaded information on the whole debacle. They've been listening to what's been going on for a while. Some of this may be the fact that they tend to be older than most of the folks on Slashdot, but they aren't full of Linux fanaticism. On the other hand, they're in touch with reality, unlike McBride's PR releases.
Hard Disk speed is important! Take a deep breath, and think about how long it takes to load all the 4 layer textures you use playing modern games. Those do NOT store completely in RAM and have to be loaded dynamically.
That's ridiculous. Your system is paging, or the game engine you're using is rather poorly written.
A system with a gig of memory easily has space for the OS, working space for the game, and enough room to store a full CD worth of textures. Most games I know of ship on CDs, and it's unusual for a game to take up more than three CDs. As a general rule of thumb, most of the space in a modern game is devoted to FMV. Audio data comes in second. Finally, it's quite rare for a game to be using *all* the textures in the game anywhere near each other.
Last of all, game designers are not stupid. If they're trickle-loading levels, they allow for a pretty healthy margin of safety. You don't wait until the first time you need a texture for the current frame to start loading it. You are not going to be in a deathmatch and have the game suddenly decide that it needs a texture, because that game is going to have been playtested and testers are going to have complained about something like this.
Finally, the best speedup you can possibly see is cutting such a delay in perhaps half. If you're hitting the disk at all, you're going to have a noticeable delay. Throw that same money at RAM. It's a much better investment.
2. Older Nvidia cards are NOT Dx9 compliant, which will limit their long-term viability.
Apparently we have a different meaning of "long-term viability". No existing video card is going to be a fancy card three years from now. Furthermore, OGL and DX are designed to make it awfully easy to gracefully degrade if the hardware doesn't handle a favorite feature. Furthermore, game developers design around what hardware exists, not around what the APIs can support. I remember when Matrox sold the first cards with hardware environmental bump-mapping support. There were a tiny handful of games that supported it. Sure, the API was out there, but nobody was interested in blowing time on a feature that most people wouldn't get the most out of.
I know this because I benchmark my purchases to make sure I get the intended result, not from idle speculation...
Let me guess. You use a dedicated benchmarking program, *not* an actual game. 3dmarkwhatever. And yes, it's designed to very carefully exercise the entire feature set of the API. This is *not* what actual games do.
AMD processors have a long history of problems with certain game vendors, most notably Sony/Verant
The primary issue with AMD has been supporting chipset problems with hardware, not with the CPUs themselves.
1GB of ram is WAY too small to cache all the textures of the newer games. WAY WAY too small. Ask Tim Sweeny why he's so hot on 64-bit architecture, and you'll find it's mostly the RAM ceiling.
Sweeny's an optimistic guy. With the most influential chip vendor pushing 32 bit hardware to consumers, with existing 64 bit hardware inducing a 32-bit performance hit, and with zero existing installed base, Tim can require a 64 bit architecture for his next game if he wants. Of course, it'll flop in the market, but he's welcome to do so. Apparently he's forgotten how (not) smoothly the transition to 32 bits went. Remember that for prices to come down and a product to take off as a non-luxury item, it has to sell to businesses and to a lesser degree, to Joe Sixpack home users (remember the fate of Aureal, who forgot that).
Headphones get uncomfortable after awhile to me.
Fair enough.
I went with the klipsch promedia system and have enjoyed it for over 3 years now. This is coming from a trance DJ who has 2 seperate sets of Professional quality studio 'phones to try as well.
(A) You're a DJ and you use computer speakers instead of stereo speakers?
> Hard disk - I know SCSI is expensive so get a drive that does at least 7200rpm
Why on earth do you consider a high speed hard drive important? Games are one of the few things that generally *don't* hit the hard drive while running.
Get a regular ol' hard drive (granted, it's hard to *find* 5400 RPM anymore, though I tend to prefer 'em for the sound levels and the lack of heat.).
Get at least 512MB of RAM to avoid paging instead.
Graphics card is next, get a OEM version of one of the later Nvidia chipsets and you save a boat load of money and still have good performance, you should get at least 128MB of RAM on the video card
Here I agree with you.
CPU speed doesn't have to be the latest one out, right now your best bet is to get at least a 2.4GHz with a 800MHz front side bus, that way you will have hyperthreading and the operating system will see 1 physical processor and say that you have 2 CPU's
2.4GHz AMD or Intel? Bit of a difference.
System Memory should be at least 1 GB nowadays @ PC133 MHz
I disagree -- I think half of this is acceptable. memory is quite easy to upgrade, and it's cheaper the longer you wait. If you're extremely adverse to upgrading RAM, then perhaps buying all now is a good idea.
Things that I consider important that the poster didn't:
* Decent set of headphones, unless you're hooking your computer to your nice stereo system. Headphones are *far* cheaper (I'd say roughly order-of-magnitude) for equivalent quality than speakers. They also give a better stereo effect. Aim for at least $80, and listen to 'em. You will lose some bass, unfortunately, which a lot of people like -- but if you live in a college dorm or play games at night, you aren't going to be able to shake the neighborhood anyway.
* Consider a CRT. LCDs are insanely popular right now, but have a lower refresh rate, look more jagged (due to their nice, sharp pixels) on things that aren't supposed to look jagged (like edges of objects), don't have as intense colors, aren't as bright, and cost more. LCDs *are* nicer for reading text, though.
* Consider a gamepad and/or joystick (for emulation). If you'll have friends playing, get a couple. The majority of PC games focus on the mouse/keyboard, but not all, and for games that can be played with these, it's awfully pleasant to do so.
* If you like FPSes, get a mouse with at least four buttons.
* Dual processors, lights, case mods, fancy sound cards, luxury input devices, wireless keyboards, etc are a waste of money. If you want 'em, fine, but there's no point in getting swept up in the "I'm spending $n, so I might as well spend $n+m" syndrome.
Most people usually don't have the $ for the 'latest and greatest' hardware. And by the time they can afford the lastest whiz-bang video card, it's already outdated.
I agree. Don't buy bleeding edge hardware *or* software. Software that's been out 18 months or more hasn't gotten worse in quality, and it's a whole lot cheaper to run and more reliable (from bugfixes). Sometimes expansion packs start getting bundled.
Best platform for games, hands down, and you can do just as much work with it as you can on any other platform.
[sigh] An ever-so-true comment. Well, the state of Linux gaming is slowly being improved from a global standpoint (though the setback of having Loki go under definitely sucked).
I'm not sure that Maxwell's demon is a true perpetual motion machine. It couldn't power an engine forever. It just does a very good job of harnessing available energy. If you're using it to do work, it must feed off of kinetic energy in each molecule. Once all the molecules have been slowed to absolute zero, it stops.
The whole idea of "harnessing entropy" is only a useful concept because we're too big and clumsy to deal with individual molecules. Maxwell's demon drops down to the molecular level.
I believe that the second law of thermodynamics is not an absolute law, just because the concept of "order" is pretty meaningless at the level we're talking about. It's simply true for all practical purposes.
The problem is that a good amount of our basic physics assumes that the Second Law holds. If it breaks, then any energy-producing devices that rely on it holding will not function either, contrary to the first impression assumptions. So you can't decrease entropy while holding enthalpy constant and then obtain enthalpy by feeding off of the lack of entropy in the system. You can, however, in *at least in theory, though perhaps not practically* decouple enthalpy and entropy, and use enthalpy as your only form of energy store.
The way I see it, computer gaming isn't ready for the masses either.
True. It tends to be palatable more to the smarter, geekier types.
Since aftermarket patches are possible, intial releases are frequently of just-barely-playable quality. Drivers for your computer have to be carefully controlled and balanced, or everything will explode. And the system upgrade cycle is much more frequent, and vastly more expensive.
Both of the last two complaints are only present if you buy the bleeding edge. You're under no constraint to do so -- I certainly don't.
Yup. And the people evaluating these purchases are folks with an MIS degree. They don't have a fucking clue when it comes to evaluating low-level technical benefits of one or another. They do understand bullet points aimed at checkboxes ("open protocols...check!").
Frankly, I think even "open source" is two vague, and a specific set of requirements should be put in place -- i.e. no NDA for source access.
As effectively as Windows, which falls into the I-wish-she-knew-more-but-Windows-aint-gonna-help.
The overwhelming majority of home PC work is done by a friendly guru or someone following step-by-step directions. If Linux get step-by-step directions from vendors (which it will with enough market share) there will be no problems.
The standard ABI on Windows is a big deal, though. It means that "supporting Linux" means supporting Red Hat.
I contend that the reason behind this "spec" is largely a grudge on the part of MA against MS.
And I contend that it's not. Amazing as it sounds, MS is not the only software vendor that supplies governments. This also is likely to have a major impact on Oracle (and may be fallout from the California-Oracle debacle). Lots of contractors that do custom jobs may not provide source. This now provides a soft requirement that they do so.
The problem is that you're assuming a decision maker that is honest, perfectly logical, and has the company's interests conjoined with his own. This just isn't the case. If purchasers now have to *justify* each non-OSS purchase (i.e. checkbox is not filled in), it acts as a check on ignoring the drawbacks of proprietary software.
I'm not saying absolutely that you are wrong about this, but I don't think that you support your assumption that open source spawns lesser capital investment and less R&D.
There is one, but it runs counter to the result of his assumptions. If you assume that closed-source software allows the producing of barriers to entry and that not doing so allows for a perfectly efficient market, then profits may only be obtained by producing closed-source software. Profits are required for capital, and capital is required for R&D. Hence, closed-source is the only system that provides for R&D.
Of course, the grandparent was also pushing open markets, so I really don't think that was where he was going.
I'd rather see state-sponsored research than corporate research, anyway, which tends to be aimed at building up patent portfolios than producing a good new product. Corporate R&D (this is big company stuff, not dinky research lab stuff) tends to consist of acquiring new companies that did something correctly and are now making money, using 90% of in-house research to build up patent portfolios to build a barrier to entry, and doing a minimal amount of actual work on new features and products in-house.
It is is no way the responsible of the executive branch of Mass. to punish anti-competitive behavior.
The government is the elected voice of the people. How in the hell do you expect the people to *do* anything if they can't even add requirements that might exclude some current offerings?
This is about a lot more than Microsoft. It's about moving their purchasing to open-source based systems. This covers a vast amount of software.
Furthermore, in an environment where OSS authors get paid, OSS is significantly more economically efficient. Right now, closed-source vendors are provided with an incentive, in that if they produce a closed-source piece of software, someone has to reverse-engineer compatibility or a full clone for it to be replaced. In an OSS environment, this is no longer the case. This just provides a counterbenefit to OSS companies, producing a vaguely level playing field, perhaps in the benefit of OSS, which is in the Mass. govt long term benefit.
Obviously, it might not occur in so many words, but what if they mandated "Commercially developed software, including full support, verification of IP contained within, etc. etc."
Then some company would provide "verification guarantees" and resell Linux. Christ, it's not like closed-source products haven't contained other folks' code before.
a) Being open source and conforming to open standards is only viewed as a highly desirable pair of attributes, not mandatory. (I wonder whether purchasers will do anything with the second other than looking for an "XML" checkmark, but we'll see...)
b) There's plenty of OSS that isn't free-as-in-beer. There's shareware (xv), and commercial (Qt) OSS software -- and Qt is even GPLed, though I expect most OSS wouldn't be free.
I hope Microsoft doesn't manage to sell "shared source" as open source, because, frankly, it's not.
The main problem is that OSS's (well, any truly open-standards software, and OSS to a somewhat greater degree) benefits are very much present when it comes to preventing lock-in. This is of extremely high value (it's the main reason Microsoft continues to have a death grip on the office market). In general, at least in the corporate world (and I would assume the bureaucratic one), purchasing decisions are generally made with an eye to the next few quarters, and with a tendency to avoid sticking out. Ideally, a purchaser wants to do something that isn't out of the ordinary, but be able to present it in a money-saving way. This minimizes his personal risk for blame, and maximizes his potential for promotion. The latest "money-saving" plan-of-the-month from the regular old vendor does exactly that.
The idea is to produce a divergence between the interests of the purchasing agent and organization being purchased for. If your product has vaguely plausible bar graph attached showing that it saves lots of money that a purchaser can hand upwards when he makes his decision (and after a sufficient amount of wining and dining), it acts as a huge incentive for the purchasing agent. It does not, however, benefit the organization being purchased for.
Furthermore, I disagree with you about the anti-trust bit. Not only is Microsoft the *least* likely to be impacted by this (it's *still* difficult to move away from them), but considering OSS first primarily tends to ward off future monopolies. It keeps lock-in from getting *established*. It's less good at fighting it once present. Clearly, organizations have paid through the nose in the past when it comes to purchasing.
That's not necessarily what he said. He didn't say it was lacking features that were specifically required, just that it didn't have the feature set. It may be that someone wanted flexibility to use some not-currently-used-feature in the future.
Frankly, asking product A to literally do everything product B does -- a perfect feature superset -- is crazy, when it comes to apps the size of what we're talking about. I'll settle for needing to use a different approach to particular problems, as long as the app reproduces all the important features on way or another.
While that's a reasonable complaint if you're arguing for large-business customers, he wasn't. It's significant for small-business folks. Furthermore, I see little reason why his argument wouldn't scale. If anything, it's more useful if more computers are present, since (a) he pointed out that the majority of the conversion costs were essentially constant (him figuring out conversion processes) and (b) maintenance, a major cost that normally scales linearly with the number of machines, has been eliminated, and (c) traditionally, per-user costs increase superlinearly -- enterprise customers are herded towards more expensve solutions. All three of these points are in favor of large-scale deployment working more successfully than small-scale deployment.
This *does* sound like an unusually good conversion, IMHO, though. There were no oddball custom apps that were written exclusively for Windows aside from a few macros and VB. There were no oddball elderly hardware devices that needed to be supported. This wouldn't be the case for a bank or a physics research lab, respectively.
In general, I've had pretty good luck with set-it-up-and-forget-it Linux servers. They really are more suited to this than Windows servers. Better designed for remote administration, cheaper, less stuff masking what's actually happening and making troubleshooting difficult.
Slashdot Mirror
Look, mate.
There are a couple of different sorts of people that might use Linux. The first are the developers, the techies. They're the ones that built GNU/Linux. A Linux company that offends these as a group does so at their own peril. A couple of companies tried cashing in on these. No good. Not enough money here, too much resistance, and it's like biting the hand that feeds you. Red Hat hasn't irritated these at all. As a matter of fact, it tends to coddle them. RH expanded the range of packages they offer by adding Fedora to their lineup (hopefully adopting yum as well) with frequent updates. RH has traditionally been extremely pro-open source, and anal about getting rid of non-free source packages in their distribution. Xv went away, xanim went away, netscape navigator went away (probably before it was a good idea to do so, as a matter of fact). They're less so than Debian, but more so than almost any of the other "corporatish" types. RH donates lots of money and developer time into core Linux products, so that all the hackers benefit. Gcc owes a lot to RH.
The second group are the mom-and-pop types. Joe Sixpack. Ordinary old home users. These generally haven't had much interest in Linux so far. RH has put a lot of money into developing GNOME to make things more palatable, but there's a long ways to go. RH hasn't really hurt this group at all.
The third group are business desktop users. This is a potentially growing market. I don't know exactly why Red Hat isn't as interested as they could be -- presumably because users tend to resist change a bit, and because businesses have balked at retrainting costs. However, RH (and Mandrake and SuSE) have all put a lot of resources into projects that will benefit these folks. They're slowly but steadily trickling into the fold -- there are migrations to Linux, but not away.
Last of all, there are the server folks. These are the ones that want Red Hat Enterprise Server. All they care about is supported servers. They want support contracts and someone to call if things break. They want very occasional updates, and don't care about the latest-and-greatest browser plugin. They want very heavy QA. For them, $699 is very attractive, especially if it lets them migrate from Solaris and AIX and the associated (expensive) hardware. If you don't fall into this group, you don't want RH Enterprise. Incidently, Debian Stable is probably a good alternative for techie admins that don't want support contracts. There are IT managers that just aren't *comfortable* dealing with most Linux companies, and want a familiar old contract and guarantees -- just with less vendor lock-in and cheaper prices.
The problem is that you're thinking that RH is trying to get you on RHE. Nope. RH doesn't market RHE to you and doesn't have an interest in doing so. Maybe one day, when they absorb enough generic suits and drop enough of the Linux folks, they'll do something unutterably stupid like trying to sock techies up for cash, but not today. About the only way they get money from you is if you want pressed CDs (worth it if you're installing on a bunch of machines at work and don't want to worry about a burned CD going bad) or if you want to buy their up2date service (frankly, not worth it -- even aside from being free, yum blows up2date out of the water).
So don't bash RH. This isn't a case of Red Hat going evil. They aren't transitioning you to give you a worse deal. They're just expanding into the server market as fast as possible.
You're talking about RH not including a web server. Ridiculous. RH still has a boxed version that it sells. It still has a web server. It comes with *installation support*, same as it always has. You can buy a support contract if you want for usage support.
Fedora is a group of developers that got together and packaged a lot more software for Red Hat than Red Hat did themselves. Red Hat realized that a lot of RH users really liked Fedora. Now, Fedora is becoming an officia
Darn, I was just about to write "he's probably not an SCO apologist", when you beat me to it with a more authoritative answer.
Oh, well.
SCO resellers have been some of the folks that are somewhat screwed by the situation. There's not really any benefit to SCO starting a bunch of lawsuits and building up bad PR. As a matter of fact, to SCO resellers -- who, as far as I can tell, tend to sell to fairly boring, conservative companies -- are a big losing player. They want SCO to stay in business, but SCO playing a risky game isn't great for them or their customers, who want boring and safe. If SCO makes tons of money, that translates into little benefit to SCO resellers. If SCO starts to focus on Linux licensing rather than their own Unix (as they said they would in their SEC filing), SCO resellers get screwed.
Basically, SCO resellers get screwed by this, pretty much no matter what happens. They aren't thrilled. Linux may technically be a competitor, but it's not as if they couldn't move to reselling and supporting Linux (well, at least more easily then to Windows).
As an aside, I suspect that a lot of folks at SCO -- not upper management -- are quite irritated as well. Their jobs are being used in a big throw of the dice that seems to primarily benefit upper management and a holding company.
Finally, SCO retailers, from exerpts from the newsgroup and blogs, seem to be a pretty decent source of levelheaded information on the whole debacle. They've been listening to what's been going on for a while. Some of this may be the fact that they tend to be older than most of the folks on Slashdot, but they aren't full of Linux fanaticism. On the other hand, they're in touch with reality, unlike McBride's PR releases.
I didn't mean a priori. I meant first impression assumptions. The two have nothing whatosever to do with each other.
Your other comment isn't worth responding to.
[shrug] I did skip over the rest when I saw the 2.4 GHz line.
Hard Disk speed is important! Take a deep breath, and think about how long it takes to load all the 4 layer textures you use playing modern games. Those do NOT store completely in RAM and have to be loaded dynamically.
That's ridiculous. Your system is paging, or the game engine you're using is rather poorly written.
A system with a gig of memory easily has space for the OS, working space for the game, and enough room to store a full CD worth of textures. Most games I know of ship on CDs, and it's unusual for a game to take up more than three CDs. As a general rule of thumb, most of the space in a modern game is devoted to FMV. Audio data comes in second. Finally, it's quite rare for a game to be using *all* the textures in the game anywhere near each other.
Last of all, game designers are not stupid. If they're trickle-loading levels, they allow for a pretty healthy margin of safety. You don't wait until the first time you need a texture for the current frame to start loading it. You are not going to be in a deathmatch and have the game suddenly decide that it needs a texture, because that game is going to have been playtested and testers are going to have complained about something like this.
Finally, the best speedup you can possibly see is cutting such a delay in perhaps half. If you're hitting the disk at all, you're going to have a noticeable delay. Throw that same money at RAM. It's a much better investment.
2. Older Nvidia cards are NOT Dx9 compliant, which will limit their long-term viability.
Apparently we have a different meaning of "long-term viability". No existing video card is going to be a fancy card three years from now. Furthermore, OGL and DX are designed to make it awfully easy to gracefully degrade if the hardware doesn't handle a favorite feature. Furthermore, game developers design around what hardware exists, not around what the APIs can support. I remember when Matrox sold the first cards with hardware environmental bump-mapping support. There were a tiny handful of games that supported it. Sure, the API was out there, but nobody was interested in blowing time on a feature that most people wouldn't get the most out of.
I know this because I benchmark my purchases to make sure I get the intended result, not from idle speculation...
Let me guess. You use a dedicated benchmarking program, *not* an actual game. 3dmarkwhatever. And yes, it's designed to very carefully exercise the entire feature set of the API. This is *not* what actual games do.
AMD processors have a long history of problems with certain game vendors, most notably Sony/Verant
The primary issue with AMD has been supporting chipset problems with hardware, not with the CPUs themselves.
1GB of ram is WAY too small to cache all the textures of the newer games. WAY WAY too small. Ask Tim Sweeny why he's so hot on 64-bit architecture, and you'll find it's mostly the RAM ceiling.
Sweeny's an optimistic guy. With the most influential chip vendor pushing 32 bit hardware to consumers, with existing 64 bit hardware inducing a 32-bit performance hit, and with zero existing installed base, Tim can require a 64 bit architecture for his next game if he wants. Of course, it'll flop in the market, but he's welcome to do so. Apparently he's forgotten how (not) smoothly the transition to 32 bits went. Remember that for prices to come down and a product to take off as a non-luxury item, it has to sell to businesses and to a lesser degree, to Joe Sixpack home users (remember the fate of Aureal, who forgot that).
Headphones get uncomfortable after awhile to me.
Fair enough.
I went with the klipsch promedia system and have enjoyed it for over 3 years now. This is coming from a trance DJ who has 2 seperate sets of Professional quality studio 'phones to try as well.
(A) You're a DJ and you use computer speakers instead of stereo speakers?
(
Try Tales Of Middle Earth and Zangband if you like rogue.
> Hard disk - I know SCSI is expensive so get a drive that does at least 7200rpm
Why on earth do you consider a high speed hard drive important? Games are one of the few things that generally *don't* hit the hard drive while running.
Get a regular ol' hard drive (granted, it's hard to *find* 5400 RPM anymore, though I tend to prefer 'em for the sound levels and the lack of heat.).
Get at least 512MB of RAM to avoid paging instead.
Graphics card is next, get a OEM version of one of the later Nvidia chipsets and you save a boat load of money and still have good performance, you should get at least 128MB of RAM on the video card
Here I agree with you.
CPU speed doesn't have to be the latest one out, right now your best bet is to get at least a 2.4GHz with a 800MHz front side bus, that way you will have hyperthreading and the operating system will see 1 physical processor and say that you have 2 CPU's
2.4GHz AMD or Intel? Bit of a difference.
System Memory should be at least 1 GB nowadays @ PC133 MHz
I disagree -- I think half of this is acceptable. memory is quite easy to upgrade, and it's cheaper the longer you wait. If you're extremely adverse to upgrading RAM, then perhaps buying all now is a good idea.
Things that I consider important that the poster didn't:
* Decent set of headphones, unless you're hooking your computer to your nice stereo system. Headphones are *far* cheaper (I'd say roughly order-of-magnitude) for equivalent quality than speakers. They also give a better stereo effect. Aim for at least $80, and listen to 'em. You will lose some bass, unfortunately, which a lot of people like -- but if you live in a college dorm or play games at night, you aren't going to be able to shake the neighborhood anyway.
* Consider a CRT. LCDs are insanely popular right now, but have a lower refresh rate, look more jagged (due to their nice, sharp pixels) on things that aren't supposed to look jagged (like edges of objects), don't have as intense colors, aren't as bright, and cost more. LCDs *are* nicer for reading text, though.
* Consider a gamepad and/or joystick (for emulation). If you'll have friends playing, get a couple. The majority of PC games focus on the mouse/keyboard, but not all, and for games that can be played with these, it's awfully pleasant to do so.
* If you like FPSes, get a mouse with at least four buttons.
* Dual processors, lights, case mods, fancy sound cards, luxury input devices, wireless keyboards, etc are a waste of money. If you want 'em, fine, but there's no point in getting swept up in the "I'm spending $n, so I might as well spend $n+m" syndrome.
Most people usually don't have the $ for the 'latest and greatest' hardware. And by the time they can afford the lastest whiz-bang video card, it's already outdated.
I agree. Don't buy bleeding edge hardware *or* software. Software that's been out 18 months or more hasn't gotten worse in quality, and it's a whole lot cheaper to run and more reliable (from bugfixes). Sometimes expansion packs start getting bundled.
Best platform for games, hands down, and you can do just as much work with it as you can on any other platform.
[sigh] An ever-so-true comment. Well, the state of Linux gaming is slowly being improved from a global standpoint (though the setback of having Loki go under definitely sucked).
I'm not sure that Maxwell's demon is a true perpetual motion machine. It couldn't power an engine forever. It just does a very good job of harnessing available energy. If you're using it to do work, it must feed off of kinetic energy in each molecule. Once all the molecules have been slowed to absolute zero, it stops.
:-)
The whole idea of "harnessing entropy" is only a useful concept because we're too big and clumsy to deal with individual molecules. Maxwell's demon drops down to the molecular level.
I believe that the second law of thermodynamics is not an absolute law, just because the concept of "order" is pretty meaningless at the level we're talking about. It's simply true for all practical purposes.
The problem is that a good amount of our basic physics assumes that the Second Law holds. If it breaks, then any energy-producing devices that rely on it holding will not function either, contrary to the first impression assumptions. So you can't decrease entropy while holding enthalpy constant and then obtain enthalpy by feeding off of the lack of entropy in the system. You can, however, in *at least in theory, though perhaps not practically* decouple enthalpy and entropy, and use enthalpy as your only form of energy store.
It may well be unpatentable.
No, it appears to be a method of *storing* energy, as per the article.
It's not light digging up coal in the wirld.
The way I see it, computer gaming isn't ready for the masses either.
True. It tends to be palatable more to the smarter, geekier types.
Since aftermarket patches are possible, intial releases are frequently of just-barely-playable quality. Drivers for your computer have to be carefully controlled and balanced, or everything will explode. And the system upgrade cycle is much more frequent, and vastly more expensive.
Both of the last two complaints are only present if you buy the bleeding edge. You're under no constraint to do so -- I certainly don't.
Not proportionally as high, however. The US standard of living is still much higher. There tends to be an influx of wealth into western nations.
Some of the problems, like carrying a non-waterproof device, could be fixed by reengineering.
Some of the convenience and spam issues are apparently handled much better in the UK than the US, this is true.
I'm still uncomfortable on the privacy front -- the worst issues aren't handled.
Yup. And the people evaluating these purchases are folks with an MIS degree. They don't have a fucking clue when it comes to evaluating low-level technical benefits of one or another. They do understand bullet points aimed at checkboxes ("open protocols...check!").
Frankly, I think even "open source" is two vague, and a specific set of requirements should be put in place -- i.e. no NDA for source access.
As effectively as Windows, which falls into the I-wish-she-knew-more-but-Windows-aint-gonna-help.
The overwhelming majority of home PC work is done by a friendly guru or someone following step-by-step directions. If Linux get step-by-step directions from vendors (which it will with enough market share) there will be no problems.
The standard ABI on Windows is a big deal, though. It means that "supporting Linux" means supporting Red Hat.
Any open source RDBMS should suffice.
What the hell are you talking about? What feature or specific performance issue do you have with Postgres?
My only guess is that you're trolling, since you group all open source software together.
Are you the same astroturfer that posted the first post I see on the article?
I contend that the reason behind this "spec" is largely a grudge on the part of MA against MS.
And I contend that it's not. Amazing as it sounds, MS is not the only software vendor that supplies governments. This also is likely to have a major impact on Oracle (and may be fallout from the California-Oracle debacle). Lots of contractors that do custom jobs may not provide source. This now provides a soft requirement that they do so.
The problem is that you're assuming a decision maker that is honest, perfectly logical, and has the company's interests conjoined with his own. This just isn't the case. If purchasers now have to *justify* each non-OSS purchase (i.e. checkbox is not filled in), it acts as a check on ignoring the drawbacks of proprietary software.
I'm not saying absolutely that you are wrong about this, but I don't think that you support your assumption that open source spawns lesser capital investment and less R&D.
There is one, but it runs counter to the result of his assumptions. If you assume that closed-source software allows the producing of barriers to entry and that not doing so allows for a perfectly efficient market, then profits may only be obtained by producing closed-source software. Profits are required for capital, and capital is required for R&D. Hence, closed-source is the only system that provides for R&D.
Of course, the grandparent was also pushing open markets, so I really don't think that was where he was going.
I'd rather see state-sponsored research than corporate research, anyway, which tends to be aimed at building up patent portfolios than producing a good new product. Corporate R&D (this is big company stuff, not dinky research lab stuff) tends to consist of acquiring new companies that did something correctly and are now making money, using 90% of in-house research to build up patent portfolios to build a barrier to entry, and doing a minimal amount of actual work on new features and products in-house.
It is is no way the responsible of the executive branch of Mass. to punish anti-competitive behavior.
The government is the elected voice of the people. How in the hell do you expect the people to *do* anything if they can't even add requirements that might exclude some current offerings?
This is about a lot more than Microsoft. It's about moving their purchasing to open-source based systems. This covers a vast amount of software.
Furthermore, in an environment where OSS authors get paid, OSS is significantly more economically efficient. Right now, closed-source vendors are provided with an incentive, in that if they produce a closed-source piece of software, someone has to reverse-engineer compatibility or a full clone for it to be replaced. In an OSS environment, this is no longer the case. This just provides a counterbenefit to OSS companies, producing a vaguely level playing field, perhaps in the benefit of OSS, which is in the Mass. govt long term benefit.
Obviously, it might not occur in so many words, but what if they mandated "Commercially developed software, including full support, verification of IP contained within, etc. etc."
Then some company would provide "verification guarantees" and resell Linux. Christ, it's not like closed-source products haven't contained other folks' code before.
a) Being open source and conforming to open standards is only viewed as a highly desirable pair of attributes, not mandatory. (I wonder whether purchasers will do anything with the second other than looking for an "XML" checkmark, but we'll see...)
b) There's plenty of OSS that isn't free-as-in-beer. There's shareware (xv), and commercial (Qt) OSS software -- and Qt is even GPLed, though I expect most OSS wouldn't be free.
I hope Microsoft doesn't manage to sell "shared source" as open source, because, frankly, it's not.
The main problem is that OSS's (well, any truly open-standards software, and OSS to a somewhat greater degree) benefits are very much present when it comes to preventing lock-in. This is of extremely high value (it's the main reason Microsoft continues to have a death grip on the office market). In general, at least in the corporate world (and I would assume the bureaucratic one), purchasing decisions are generally made with an eye to the next few quarters, and with a tendency to avoid sticking out. Ideally, a purchaser wants to do something that isn't out of the ordinary, but be able to present it in a money-saving way. This minimizes his personal risk for blame, and maximizes his potential for promotion. The latest "money-saving" plan-of-the-month from the regular old vendor does exactly that.
The idea is to produce a divergence between the interests of the purchasing agent and organization being purchased for. If your product has vaguely plausible bar graph attached showing that it saves lots of money that a purchaser can hand upwards when he makes his decision (and after a sufficient amount of wining and dining), it acts as a huge incentive for the purchasing agent. It does not, however, benefit the organization being purchased for.
Furthermore, I disagree with you about the anti-trust bit. Not only is Microsoft the *least* likely to be impacted by this (it's *still* difficult to move away from them), but considering OSS first primarily tends to ward off future monopolies. It keeps lock-in from getting *established*. It's less good at fighting it once present. Clearly, organizations have paid through the nose in the past when it comes to purchasing.
That's not necessarily what he said. He didn't say it was lacking features that were specifically required, just that it didn't have the feature set. It may be that someone wanted flexibility to use some not-currently-used-feature in the future.
Frankly, asking product A to literally do everything product B does -- a perfect feature superset -- is crazy, when it comes to apps the size of what we're talking about. I'll settle for needing to use a different approach to particular problems, as long as the app reproduces all the important features on way or another.
While that's a reasonable complaint if you're arguing for large-business customers, he wasn't. It's significant for small-business folks. Furthermore, I see little reason why his argument wouldn't scale. If anything, it's more useful if more computers are present, since (a) he pointed out that the majority of the conversion costs were essentially constant (him figuring out conversion processes) and (b) maintenance, a major cost that normally scales linearly with the number of machines, has been eliminated, and (c) traditionally, per-user costs increase superlinearly -- enterprise customers are herded towards more expensve solutions. All three of these points are in favor of large-scale deployment working more successfully than small-scale deployment.
This *does* sound like an unusually good conversion, IMHO, though. There were no oddball custom apps that were written exclusively for Windows aside from a few macros and VB. There were no oddball elderly hardware devices that needed to be supported. This wouldn't be the case for a bank or a physics research lab, respectively.
In general, I've had pretty good luck with set-it-up-and-forget-it Linux servers. They really are more suited to this than Windows servers. Better designed for remote administration, cheaper, less stuff masking what's actually happening and making troubleshooting difficult.