Again, could be limited to GPL/GPL-compatible licensed code. I mean, since nVidia/ATI already have the code under license, how many trade secrets could there possibly be left to protect?
There are no other trade secrets; or any at all in all likelihood. Patents are not trade secrets, nor is copyrighted information a trade secret. Sigh.
Anyway, could they? Yes. Will they? Pretty obviously not. People (again, including Alan Cox, who is pretty well respected both inside and outside of the OSS community) have been asking for them to do this for years. And they've certainly brought up the concept of dual licensing, the successful implementations of it, and so forth. But S3 (or, rather, its various corporate overlords -- currently Via) isn't biting.
And no, you can't reimplement it in another way -- the S3TC algorithm is pretty basic and simple. People have been looking at another way to do it that's not utterly absurd and failing for some time now. It's not to say it can't be done -- but it's pretty unlikely at this point. And, yes, if you want to use texture compression (and you do) then you have to use an algorithm that results in S3TC compliant textures since that's what the hardware is expecting. Otherwise you're stuck with lower quality textures that use more memory space.
First they compare a $115 card to cards costing $125 and $129
The ATI Radeon 1600Pro can be had for $99. The GF6600GT is $115.
they didn't pan it for performance, but for basic flaws?
Where'd you get that. In their conclusion they very clearly pan it for performance. It's not even the 2nd best card in its price range -- it's third best. By a large margin.
Ok, AA doesn't appear to work for GL, that is bad but will almost certainly get fixed in the drivers pretty soon.
Well that'd be new and different -- S3 actually fixing their drivers. I wouldn't hold your breath.
And it's worse than that, if you bother to read the review (or even the conclusion) -- AA/AF doesn't work in a number of other games, and when it does it generally causes performance to drop into the useless category.
That would be because SLI mode doesn't work? What sort of idiot would buy a $99 card for SLI work?
They state, clearly, that SLI isn't common at this price point, but that's irrelevant. This is S3's own implementation of it and it doesn't work worth a damn. It's a selling point on their card, so it should work.
It looks like S3 is trying something interesting, throw high speed but dumb hardware at the problem of 3D instead of trying to put more compute power than a P4 on a board
What on earth is that statement based on? They appear to have as much hardware as the competition. In fact, more than the competition does in the same price range. And they appear to have similar hardware algorithms (fast Z-clear, occlusion culling, etc.) as the competition. Whoever modded you up not didn't read the article, they don't understand graphics hardware in the slightest.
I know I'd certainly switch from ATI Radeon 9250 (most current 3D with Open drivers) to this new S3 tech if it had an open driver.
Better hope that 9250 doesn't die then, because that's not happening anytime soon. Go read one of my other posts in this article if you'd care to know why.
No, you're thinking copyrights, which keep getting extended. Patents do not. Patent term has not substantially changed in the US in over 100 years, with only a minor (and very good) change in 1994. And while there is some international controversy on what should be patentable, there's not really any on term.
Worst case, patents expire 20 years from the earliest claimed filing date (design patents are different, as are patents issued prior to June 8, 1995, but neither is relevant here).
Supply full GPL/BSD licensed source code to the X.org and kernel.org for inclusion in mainline.
They're not going to do that. If for no other reason than their own texture compression technology (S3TC) which they license to other video card makers (namely ATI and Nvidia, as well as MS for DirectX drivers).
Even if they were to release the souce you probably couldn't use it unless they granted some kind of license to use the patented algorithms freely. And they haven't done that to date despite lobbying by various people (including Alan Cox).
Of course, people who actually know this have been saying it everytime someone says "open up the source!" to video card makers, and most people still don't get it. Sigh.
2002 Camry, had repeated mechanical problems with the driver's side window, which wouldn't allow it to close, and had to be fixed, 3 times, under warranty
No lemon laws in your state?
And, FWIW, the 2002 Camry was the first year of a redesign. There's ample evidence that first year vehicles have far more problems than later year models based on the same chassis. And that goes for all auto makers.
The only thing I can come up with, is that Toyota is making a business out of trading money for good reviews of their vehicles.
So they're bribing the entire European auto press, the entire American auto press, and every single consumer oriented group out there? Including, as you mention, Consumer Reports which has a well known and documented history of buying goods "off-the-shelf" like any other consumer instead of testing vehicles specifically sent to reviewers? And, of course, then there's the issue of all those used Toyota owners out there who aren't screaming bloody murder...
Sorry, doesn't fly.
Oh, and I know a number of people with Toyotas who have had relatively few problems with them. There, anecdotal evidence to counter your own.
That is recently, comparing to my Phoenix/Firebird/Firefox usage history.
Since it was fixed in 1.0.3, which came out not long after 1.0 (a hair over 5 months), then you're complaining about it being busted in beta versions as well? Are you serious?
Yes, it should've been fixed in 1.0 release, but that doesn't make your statement any less silly.
Here we go again; the claim that a better Windows box costs half as much. But, as usual, no specifics are provided, mainly because it's untrue.
No, it's not untrue. I've done the comparisons numerous times, with very specific builds and including the pluses and minuses of each system (there are distinct +'s for the Apple, and distinct +'s for the non-Apple). Here's a link to the last one I did on this site, and it was almost exactly "half the price" of the equivalent Apple:
Of course, die hard Apple users will ignore specific system comparisons and keep telling themselves that "its untrue" and that "nobody provides specifics"... without bothering to either look for someone who does or trying to do so themselves.
But here, since you want to be spoon fed: Dell Inspiron 1705 (under Home): Upgrade CPU to 2.16 GHz ($480), select re-install CD ($10), remote control ($23), WUXGA screen (1920x1200; $119; higher res), 1GB 667 MHz memory ($140), 120 GB HD ($116), 8x DL DVD burner ($48), 802.11 a/b/g wireless ($23), Bluetooth ($39), 256 MB Nvidia GeForce Go 7800 ($239 -- much, much more powerful than the ATI X1600 for once).
The base price is $2844. The default discount is 20% off, for $2275, but you can apply a $750 off coupon instead making it $2094.
Alternately, drop the memory down to 533 MHz, get a 2 year warranty, and pay $2829/$2122/$2079.
Not half price this time, but $750 is still quite a chunk of change.
And, no, I don't expect it to convince you. But I'm really sick of Apple zealots claiming that Apple doesn't charge a huge premium on their systems.
Try compiling millions of lines of software all day long, then you will realize how slow hard drive technology is buddy.
Hrm. At my current job a full build from scratch, single threaded, takes roughly an hour on Windows. On some of our archaic Unix boxes (HP being the worst) it takes ~24 hours for an optimized build, including unittests -- if nobody else is using the system (on a newer system w/ 16 CPUs it only takes 1.5 hours).
At my previous job it took roughly 6 hours to build the code. We had utter crap systems there too. Deeply CPU and memory starved.
I think it is more because of Windows reliance of virtual memory, when you need to swap a gigabyte of ram to the hard drive when it is already saturated, the you get the system responding very slowly.
If you're having to swap to disk that much then you need more memory. I saw a vast improvement in my compile times moving from 1 GB of RAM to 2 GB under XP.
I mean something completely new. 80mb/s sustained transfer rate is rather pathetic, especially if your generating gigabytes of compiled code.
Uh... ok, so let's say your source is 5 million lines (which I doubt). That means that you have The moment the capacity barrier is reached
Uh... ok. Get back to me in 50 years then. Everytime someone thinks we're about to hit the capacity wall R&D finds a way around it. It's happened no less than 3 times in the past 5 years. It's very similar to the continued advance of transistors in spite of the doomsayers.
Before mouthing off about how little someone knows about things, get a clue first and try to figure out where they are coming from. I experience how slow hard drives are every day, and simply desire something better in the industry, a position many are repeating in this thread because they also know what they are talking about and have actual experience
No, you've merely demonstrated your abysmal lack of knowledge about hardware and how your system actually works. It's sad really -- something I see far too much in fellow programmers.
Either a new I/O standard needs to be invented, something that doesn't tax your system when excessive hard drive transfers are made, or the frigging hard drives just need to start getting up to gigabyte transfer rates.
We have new I/O standards all the time. In the desktop realm the latest is SATA, with transfer rates of up to 3 Gb/s (375 MB/s, and yes that's base10). Of course, the fastest SATA drives can't even come close to a third of that transfer rate. I/O standards have always vastly outpaced the drives they're connected to. That's a good thing really -- you wouldn't want to be limited by your interconnect -- but it's rather annoying to keep explaining to people that upgrading from (for example) ATA-66 to ATA-133 isn't going to double their HD speed.
As for them "tax[ing] your system" -- since when? DMA has been standard on every transport known to man for nearly a decade. PIO is long gone, and thankfully so. If your computer spends more than a couple percent of its time on HD data transfers then there is something hideously wrong with your system.
As for transfer rates -- there's the rub. HD manufacturers are always wanting to increase them, but there's very little they can do -- it's pure physics. With a given rotational speed there's only so much data that passes under the head in a given time. You can do one of two things to increase that transfer rate -- spin the platter faster or make the data denser. If you increase the rotational speed then you have to deal with more heat, and that generally means using smaller or less dense platters which cuts down your transfer rate. If you increase data density (which is what this does) then you increase heat slightly, but it's nothing in comparison. So if you can double your data density you can double your transfer rate.
For example, the server-oriented version of this Seagate drive is only 300 GB, but that's a doubling in capacity for 15K rpm drives. And it's expected that it'll have sustained transfer rates of around 125 MB/s, roughly 1/3 higher than its nearest competitor.
Similarly I'd expect this drive (or at least a drive in this lineup) to have the highest transfer rate for 7200 rpm drives -- probably in the 80 MB/s range (the fastest are currently around 60 MB/s). That's really how drives have been getting faster and faster for the past few years. A brand new drive 5 years ago would've been fast if it was getting 25-30 MB/s. Nowadays 50-55 is standard. With perpendicular recording we're looking at upping that to the 75-80 range.
In any case, I could care less about hard drives doubling or tripling in size, until they show significant improvements in performance, or move to solid state, then I am apathetic about the whole industry.
Then you're just showing how little you know about things. I'd love to see solid state persistant storage take off too, but so far every technology that's promised that has fallen flat. So far the only one that's gotten anywhere is Flash -- which isn't all that much faster (probably due to I/O restrictions), has write issues (largely solved w/ write balancing), and cost (20-50x HDs, and only improving slightly faster). I can buy a 250 GB HD for ~$70 right now. I don't even want to think about how much it would cost to build an array of Flash devices of equivalent size.
I'd suggest that terms like giga are mainly used in computing.
What, you've never heard of gigawatts? Even mispronounced ones?
And, quite frankly, the use of base2 vs base10 in computing is hardly uniform. In fact, the only place it's used is in storage units (memory and hard drives), and even then HDs stopped using it years ago (only the OS still reports in base2). And some memory is sold in base10 units now as well (in particular Flash memory).
Networking has always used base10 (10/100/1000 Mbit ethernet is base10), processor speeds are base10, bus speeds are base10 (there's some confusion for you -- memory capacity is base2, speed is base10), and so forth.
Not by fiat, but by agreement.
Yup, and the world has agreed that HD sizes are measured base10 now. It's just the OS vendors who are slow to catch up -- and MS even displays both numbers in some places.
That was an option of course (as was short-term disability, which would be a lot more sane), but I'm the sole provider for the household (my wife decided that 80+ hour work weeks at Sprint didn't play nicely with having a family, and I support her in that). We have a good bit of savings, but since we're hoping to move in the next year or two I'd rather not touch that unless necessary.
and new parents have to take their parental leave within 6 months instead of 12
Gads... I currently work for the 2nd largest software company (behind MS) currently and I'd rejoice at this option if we had it. We had our second kid two weeks ago, she was in the NICU for a week due to an infection, and I've only taken 3 days off (and have precisely 1 day of vacation left) because we're not given any such thing and only get 2 weeks/year vacation (accrued over the year). If I had to take an extended period of time off then I would have to go on short term disability.
Fortunately the kiddo is doing great now, so I haven't had to do that.
At least we get free soft drinks too... and I'm nowhere even vaguely near the main campus (other side of the country).
Agreed. Writing an enterprise class OS is hard too. As is an enterprise class file system. yet there are open source examples of both. And there are dozens of other examples.
Frankly, I think he was simply trying to downplay the real issue (which ATI outright stated) -- intellectual property issues.
No matter how much the hue and cry (which, as he points out, is pretty damn small right now), neither company could release OSS drivers even if they wanted to because they're constrained by IP agreements. At the very least both use S3's texture compression technology. There have been rumors of Nvidia (via its 3Dfx legacy and later acquisition) being beholden to SGI for years. And there's probably some cross-licensing agreements between the companies, and to other companies.
Which means that when it comes down to it the current state of affairs is going to continue. If you want 3D support (that is, support for vaguely modern hardware) then you'll have to use a proprietary kernel module. It's far from perfect though, and it's pretty much going to doom any hopes for more gaming on Linux or for serious 3D desktop support (ala OS X's Aqua or Vista's Aero) since major distros won't ship 3D enabled kernels due to the legal issues.
The only potential way out is for there to be a stable kernel ABI -- something Linus has long opposed. But that doesn't really resolve the legal issues -- it just makes it easier for closed source kernel modules to be developed (and harder for the kernel team as a whole). And it doesn't solve the issue of fixing kernel bugs that occur when you have closed source modules loaded (the kernel devs won't touch them, for obvious reasons).
I'm so tired of people saying this. Have you even bothered looking? If you have an ATI or Nvidia card then you have multiple desktop capability built into your video drivers. They're a damn sight faster than any software based virtual desktops.
There's also software solutions, in case you're stuck with an archaic video card or a crappy Intel video card. My preferred one is VirtuaWin, but I've also used DeskWin. They're both OSS under the GPL.
Yes, MS has a PowerToy that does it as well (but it sucks). And even if you have ATI/Nvidia cards you may want to take a look at VirtuaWin -- it's extremely configurable and has a bunch of plugins available.
f you already have a good digital camera and you want to digitize prints, my advice is to photograph them
I've seen a couple of people say this now, and I have to scratch my head at it.
Even a basic scanner can do 600 dpi. For a 4x6 picture that's 8,640,000 pixels, or roughly 8.6 MP. You'll have to have a pretty high end camera to exceed that -- and that's discounting the likelihood of photographing something besides just the picture. Also realize that with a scanner you get even higher resolutions with larger pictures. With a digital camera you're stuck at whatever its resolution is, so with larger source pics you end up with worse results, not better ones!
And if you look at low cost scanners they are generally in the 2400x1200 dpi area, so now you're talking about ~70 MP for a lowly 4x6 now -- probably overkill, but I'd rather have overkill than underkill for a situation like this -- you really don't want to have to do it more than once.
Re:Apple is going to make a killing...
on
Going To Boot Camp
·
· Score: 1
I started from the leftmost one, the supposedly best one.
Then you have an upgraded warranty, additional software, and various other crap.
Start with the 2nd from the right -- it's the lowest end Core Duo processor. You can then upgrade it to your heart's content (beyond the leftmost one if you want).
Oh, and if you got $2150 as the before-discounts-price then you're eligible for the $750 off coupon -- which is quite a bit more than anything else you can possibly use.
unless you hunt for those rare deals
Thing is, they're not rare. Hell, Dell advertised these deals in our local Sunday paper. And they have similar deals running all the time. And "hunting" for the deal consists of putting "Dell coupon" into Google and hitting "I feel lucky". If you don't want to use them, fine, but then you may as well start ignoring prices altogether because it's apparantly not a priority for you.
Re:Apple is going to make a killing...
on
Going To Boot Camp
·
· Score: 1
If you got a different price today then they may have changed their pricing or deals. It was $1501 yesterday (and the day before, and last week).
The $450 coupon is from one of the deals they are currently running. You can click on "Save up to $750 on Inspirons" in the Dell Home section, or just go read up on it here: http://www.xpbargains.com/dell_coupons/
Re:Apple is going to make a killing...
on
Going To Boot Camp
·
· Score: 1
It seems to me that you are still missing some important points
No. I said it wasn't a full list.
On the Mac side, you didn't include MagSafe, Gigabit Ethernet, optical audio in/out (and I'm not sure if the Dell has built in speakers and microphone), ambient light sensors, and DVI port.
MagSafe is cool, I grant you that. Gigabit ethernet? Uh, you realize that neither system has enough bus bandwidth (not to mention hard drive speed) to even come close to handling that data rate, right? Optical in-out, sure, but I don't consider that a huge thing. Yes, the Dell has built in speakers and microphone -- were you serious? Ambient light sensor is only relevant for the backlit keyboard, which I mentioned. I am surprised the Dell doesn't have a DVI port, but hey, the Apple doesn't have a VGA port.
On the Dell side, it has two more USB connectors (but they are Type B !!), a flash memory slot (not sure what it's used for), S-Video port (additional $20 adaptor for the Mac), and integrated modem (ditto).
Type B? Who cares? That's standard USB port. It has a 5-in-1 memory reader, not a "flash memory slot" -- and I should've listed that as a big feature because, IMO, it is. Saves me a USB port and a dongle for reading various camera card formats. S-Video is "eh" to me, just as digital audio out is. I mentioned the integrated modem.
You left out (on the Dell) 802.11a support, a non-slot loading drive (so it can handle small CD/DVDs), and the instant-media support (don't boot the OS for music/movies). None of which I consider important, but you seem to want a full list.
Also, for many people like me, Win XP MCE is not a good alternative for XP Pro, reducing the price gap.
Exactly what features do you need in Pro that aren't in MCE? There's exactly one difference you know. Admittedly, it can be a fairly important one to some business users, but the gap in features between XP Pro and OS X is much larger, so it's more fair IMO to compare the systems with those two.
Re:You get what you pay for...
on
Going To Boot Camp
·
· Score: 2, Insightful
How much do hobbyist quality video production, audio production, photo editing/cataloging and DVD authoring suites cost for Windows?
$99. Adobe Photoshop/Primere Elements Bundle. Yes, that's street price.
You can cobble stuff together from OSS/Free, but it's nowhere near the quality and ease of use of either iLife or Adobe, so I simply don't recommend trying. Particularly for the video editing/DVD authoring bit (although, on that front, Nero is decent and can be had for $40).
Both downloadable for free, from either MS, Cygwin, or MinGW. But I do Unix development, so it's not of much interest to me. A decent shell is, but that's what Cygwin's for.
A full suite of enterprise capable network daemons such as http, ftp, telnet, ftp, ssh?
IIS is included in Pro (but not installed by default). As is a telnet daemon (not enabled by default under SP2).
FTP and ssh daemons are freely available online if you wanted them. And XP's remote desktop is superior to VNC (admittedly, one of the rare cases of XP being better), so I guess that's why you didn't mention it.
Are you seriously trying to say all of that is worth $1000+? I can replace the software for under $200, as I mentioned. Most of the extra functionality can be downloaded for free.
Re:Apple is going to make a killing...
on
Going To Boot Camp
·
· Score: 1
There are software differences (which are solved for $1000 for what essentially boiled down to running OS X. I suspect most other people look at it the same way.
Damnit, I should've previewed.
What that originally said was:
... (which are solved for <$200) and other hardware differences, but they're pretty minor.
I just bought a new laptop and while I looked at the Macbooks I couldn't justify spending >$1000 for what essentially...
Re:Apple is going to make a killing...
on
Going To Boot Camp
·
· Score: 4, Insightful
was going to buy a Dell M90 loaded = $3500 or so
MacBook Pro with all options = $3400
Huh, I get $3600 for the MacBook... or did you forget to include the 3 year warranty that's standard on the Dell M90?
But even so, you're comparing a 17" laptop with 1920x1200 resolution and an insanely powerful graphics card (Nvidia Quadro FX 2500M w/ 512 MB dedicated memory) to a 15" laptop with 1440x900 resolution and a very powerful graphics card (ATI Mobility 1600 w/ 256 MB dedicated memory). Not to mention the vast raft of other differences between the two. And the Dell is more expensive? Imagine that! I bet you didn't even try to use free Dell coupons that cut the price down, now did you? Hell, I can't even guess what screwed up way you configured the Dell to make such an absurd comparison -- because, unsurprisingly, you didn't give details.
Hey, let's try for a more fair comparison. Base level MacBook Pro: $1999 Dell Inspiron E1505 upgraded appropriately: $1501 - $450 = $1051 (upgrades: T2400 CPU, XP MCE install CD, remote control, 15.4" WSXGA screen, 512MB single DIMM, 80 GB 5400 rpm HD, 8X DVD/CD DL burner, 3945 Wireless NIC + Bluetooth, ATI Radeon X1400)
Now the Dell is not a 100% replica of the Apple, but neither is the Apple completely superior to the Dell. This isn't a full list of differences, but it hits the high points:
Dell advantages: Higher resolution screen (1680x1050), dual layer burner, built in modem, two mouse buttons on the laptop itself, $950 in your pocket. (Yesterday you could've gotten 1 GB of RAM for the same price, but that deal expired)
Mac advantages: Much more powerful video card, slightly faster RAM (the Dell can take the RAM, they're just being cheap asses), integrated web cam, backlit keyboard, can boot both OS X and XP.
Again, that's not a full list. There are software differences (which are solved for $1000 for what essentially boiled down to running OS X. I suspect most other people look at it the same way.
The problems lie in GNOME and KDE using far too much memory (and probably CPU cycles as well). That's GNOME and KDE, though -- not X11. Try any alternative window manager/desktop environment, and you're likely to see a vast speed improvement.
Agreed, but that's not a valid answer. Ditch KDE and Gnome and you're left with a desktop environment that is nowhere near as friendly to most people. Sure, advanced users can use it with no issues, but that doesn't help people who are trying to get into Linux, or get others into it.
And the KDE/Gnome issues come back if you want to use a program compiled with their libraries -- at that point you have to load the libs into memory and you're now worse off than you would be if you were running KDE or Gnome already (longer load time due to loading the shared libs, possibly a larger overall memory footprint). Don't use those programs? Uh... ok. There goes most of the recent GUI programs for Linux.
The point I'm trying to make is that XP and OS X can deliver all of the eye candy and usability in the given memory footprint while still being faster than X (w/ KDE or Gnome). That's fairly damning. Both KDE and Gnome are bloated projects (as you note) and could benefit from exactly the same kind of total revamp that Windows needs. Hell, just moving to the STL (instead of recreating it from scratch, or attempting to do OO in C) would help reduce memory footprint, improve speed, and (most importantly) increase development speed.
Here are the requirements for slackware 3.4: 3.4 (Kernel 2.0.33)
* Intel 8086
Uh, I'm sorry, but Slackware is simply not going to run on an 8086, 80186, or 80286.
And while it may boot on an 8 MB RAM system, don't try to use X on that sucker. Or much else for that matter.
Here's the kicker, if you actually READ the article
But this is/.! Nobody reads the article!
It is a rather horribly named article though -- I've got to imagine the name was foisted upon it due to column width/length restrictions or other dead tree issues. It's missing the word "Development" after "Windows" (and even then, that could be misleading, but it's at least better).
And guess what the requirements for Vista is going to be?
Awfully similar to those for other modern OS's, including Mac OS X and modern desktop Linux systems. And, as with any piece of software, I wouldn't want to use it at anywhere near the "minimum" spec.
In NT and all its children, DOS is a separate application provided as a compatability layer running on a kernel which has nothing to do with DOS.
I know that. But why even have it? Remove it and you reduce support costs -- even if that branch of code is stable and rarely requires fixes, it's a section that needs testing whenever you go to QA. Same for the entire Windows-on-Windows emulation layer.
As another poster said, I've heard that it's gone in the 64-bit versions, but I haven't confirmed that myself.
As regards the WMF issue
And you missed the entire point. I know the issues with WMF, including why it was designed that way, but it's an archaic format that is rarely used in modern computing. So ditch it. Ditch support for any archaic data file formats -- there are certainly plenty of them out there.
Note that all of this will create a huge amount of work in the short term. WMF is actually used by some printer drivers, so they'll have to be reworked. That's fine, and it's to be expected. But in the long run you'll save money in development, QA, and support time by shedding tons of dead code. And yes, you'll have some customers scream about breaking their legacy app. So offer to help them with development resources -- that will cost you less than supporting the code for entire userbase.
And yes, I'm oversimplifing the issue, but it's increasingly clear that the current codebase is unsupportable. It's collapsing under its own weight. Can you imagine if glibc on Linux had to support every function call from 1990 forward? That's basically what MS is trying to do. It's a nightmare.
Slashdot Mirror
Again, could be limited to GPL/GPL-compatible licensed code. I mean, since nVidia/ATI already have the code under license, how many trade secrets could there possibly be left to protect?
There are no other trade secrets; or any at all in all likelihood. Patents are not trade secrets, nor is copyrighted information a trade secret. Sigh.
Anyway, could they? Yes. Will they? Pretty obviously not. People (again, including Alan Cox, who is pretty well respected both inside and outside of the OSS community) have been asking for them to do this for years. And they've certainly brought up the concept of dual licensing, the successful implementations of it, and so forth. But S3 (or, rather, its various corporate overlords -- currently Via) isn't biting.
And no, you can't reimplement it in another way -- the S3TC algorithm is pretty basic and simple. People have been looking at another way to do it that's not utterly absurd and failing for some time now. It's not to say it can't be done -- but it's pretty unlikely at this point. And, yes, if you want to use texture compression (and you do) then you have to use an algorithm that results in S3TC compliant textures since that's what the hardware is expecting. Otherwise you're stuck with lower quality textures that use more memory space.
First they compare a $115 card to cards costing $125 and $129
The ATI Radeon 1600Pro can be had for $99. The GF6600GT is $115.
they didn't pan it for performance, but for basic flaws?
Where'd you get that. In their conclusion they very clearly pan it for performance. It's not even the 2nd best card in its price range -- it's third best. By a large margin.
Ok, AA doesn't appear to work for GL, that is bad but will almost certainly get fixed in the drivers pretty soon.
Well that'd be new and different -- S3 actually fixing their drivers. I wouldn't hold your breath.
And it's worse than that, if you bother to read the review (or even the conclusion) -- AA/AF doesn't work in a number of other games, and when it does it generally causes performance to drop into the useless category.
That would be because SLI mode doesn't work? What sort of idiot would buy a $99 card for SLI work?
They state, clearly, that SLI isn't common at this price point, but that's irrelevant. This is S3's own implementation of it and it doesn't work worth a damn. It's a selling point on their card, so it should work.
It looks like S3 is trying something interesting, throw high speed but dumb hardware at the problem of 3D instead of trying to put more compute power than a P4 on a board
What on earth is that statement based on? They appear to have as much hardware as the competition. In fact, more than the competition does in the same price range. And they appear to have similar hardware algorithms (fast Z-clear, occlusion culling, etc.) as the competition. Whoever modded you up not didn't read the article, they don't understand graphics hardware in the slightest.
I know I'd certainly switch from ATI Radeon 9250 (most current 3D with Open drivers) to this new S3 tech if it had an open driver.
Better hope that 9250 doesn't die then, because that's not happening anytime soon. Go read one of my other posts in this article if you'd care to know why.
Here in the US, effectively never.
No, you're thinking copyrights, which keep getting extended. Patents do not. Patent term has not substantially changed in the US in over 100 years, with only a minor (and very good) change in 1994. And while there is some international controversy on what should be patentable, there's not really any on term.
Worst case, patents expire 20 years from the earliest claimed filing date (design patents are different, as are patents issued prior to June 8, 1995, but neither is relevant here).
Supply full GPL/BSD licensed source code to the X.org and kernel.org for inclusion in mainline.
They're not going to do that. If for no other reason than their own texture compression technology (S3TC) which they license to other video card makers (namely ATI and Nvidia, as well as MS for DirectX drivers).
Even if they were to release the souce you probably couldn't use it unless they granted some kind of license to use the patented algorithms freely. And they haven't done that to date despite lobbying by various people (including Alan Cox).
Of course, people who actually know this have been saying it everytime someone says "open up the source!" to video card makers, and most people still don't get it. Sigh.
2002 Camry, had repeated mechanical problems with the driver's side window, which wouldn't allow it to close, and had to be fixed, 3 times, under warranty
No lemon laws in your state?
And, FWIW, the 2002 Camry was the first year of a redesign. There's ample evidence that first year vehicles have far more problems than later year models based on the same chassis. And that goes for all auto makers.
The only thing I can come up with, is that Toyota is making a business out of trading money for good reviews of their vehicles.
So they're bribing the entire European auto press, the entire American auto press, and every single consumer oriented group out there? Including, as you mention, Consumer Reports which has a well known and documented history of buying goods "off-the-shelf" like any other consumer instead of testing vehicles specifically sent to reviewers? And, of course, then there's the issue of all those used Toyota owners out there who aren't screaming bloody murder...
Sorry, doesn't fly.
Oh, and I know a number of people with Toyotas who have had relatively few problems with them. There, anecdotal evidence to counter your own.
That is recently, comparing to my Phoenix/Firebird/Firefox usage history.
Since it was fixed in 1.0.3, which came out not long after 1.0 (a hair over 5 months), then you're complaining about it being busted in beta versions as well? Are you serious?
Yes, it should've been fixed in 1.0 release, but that doesn't make your statement any less silly.
Here we go again; the claim that a better Windows box costs half as much. But, as usual, no specifics are provided, mainly because it's untrue.
7 9&cid=15077483
No, it's not untrue. I've done the comparisons numerous times, with very specific builds and including the pluses and minuses of each system (there are distinct +'s for the Apple, and distinct +'s for the non-Apple). Here's a link to the last one I did on this site, and it was almost exactly "half the price" of the equivalent Apple:
http://hardware.slashdot.org/comments.pl?sid=1823
Of course, die hard Apple users will ignore specific system comparisons and keep telling themselves that "its untrue" and that "nobody provides specifics"... without bothering to either look for someone who does or trying to do so themselves.
But here, since you want to be spoon fed:
Dell Inspiron 1705 (under Home): Upgrade CPU to 2.16 GHz ($480), select re-install CD ($10), remote control ($23), WUXGA screen (1920x1200; $119; higher res), 1GB 667 MHz memory ($140), 120 GB HD ($116), 8x DL DVD burner ($48), 802.11 a/b/g wireless ($23), Bluetooth ($39), 256 MB Nvidia GeForce Go 7800 ($239 -- much, much more powerful than the ATI X1600 for once).
The base price is $2844. The default discount is 20% off, for $2275, but you can apply a $750 off coupon instead making it $2094.
Alternately, drop the memory down to 533 MHz, get a 2 year warranty, and pay $2829/$2122/$2079.
Not half price this time, but $750 is still quite a chunk of change.
And, no, I don't expect it to convince you. But I'm really sick of Apple zealots claiming that Apple doesn't charge a huge premium on their systems.
Try compiling millions of lines of software all day long, then you will realize how slow hard drive technology is buddy.
Hrm. At my current job a full build from scratch, single threaded, takes roughly an hour on Windows. On some of our archaic Unix boxes (HP being the worst) it takes ~24 hours for an optimized build, including unittests -- if nobody else is using the system (on a newer system w/ 16 CPUs it only takes 1.5 hours).
At my previous job it took roughly 6 hours to build the code. We had utter crap systems there too. Deeply CPU and memory starved.
I think it is more because of Windows reliance of virtual memory, when you need to swap a gigabyte of ram to the hard drive when it is already saturated, the you get the system responding very slowly.
If you're having to swap to disk that much then you need more memory. I saw a vast improvement in my compile times moving from 1 GB of RAM to 2 GB under XP.
I mean something completely new. 80mb/s sustained transfer rate is rather pathetic, especially if your generating gigabytes of compiled code.
Uh... ok, so let's say your source is 5 million lines (which I doubt). That means that you have The moment the capacity barrier is reached
Uh... ok. Get back to me in 50 years then. Everytime someone thinks we're about to hit the capacity wall R&D finds a way around it. It's happened no less than 3 times in the past 5 years. It's very similar to the continued advance of transistors in spite of the doomsayers.
Before mouthing off about how little someone knows about things, get a clue first and try to figure out where they are coming from. I experience how slow hard drives are every day, and simply desire something better in the industry, a position many are repeating in this thread because they also know what they are talking about and have actual experience
No, you've merely demonstrated your abysmal lack of knowledge about hardware and how your system actually works. It's sad really -- something I see far too much in fellow programmers.
Either a new I/O standard needs to be invented, something that doesn't tax your system when excessive hard drive transfers are made, or the frigging hard drives just need to start getting up to gigabyte transfer rates.
We have new I/O standards all the time. In the desktop realm the latest is SATA, with transfer rates of up to 3 Gb/s (375 MB/s, and yes that's base10). Of course, the fastest SATA drives can't even come close to a third of that transfer rate. I/O standards have always vastly outpaced the drives they're connected to. That's a good thing really -- you wouldn't want to be limited by your interconnect -- but it's rather annoying to keep explaining to people that upgrading from (for example) ATA-66 to ATA-133 isn't going to double their HD speed.
As for them "tax[ing] your system" -- since when? DMA has been standard on every transport known to man for nearly a decade. PIO is long gone, and thankfully so. If your computer spends more than a couple percent of its time on HD data transfers then there is something hideously wrong with your system.
As for transfer rates -- there's the rub. HD manufacturers are always wanting to increase them, but there's very little they can do -- it's pure physics. With a given rotational speed there's only so much data that passes under the head in a given time. You can do one of two things to increase that transfer rate -- spin the platter faster or make the data denser. If you increase the rotational speed then you have to deal with more heat, and that generally means using smaller or less dense platters which cuts down your transfer rate. If you increase data density (which is what this does) then you increase heat slightly, but it's nothing in comparison. So if you can double your data density you can double your transfer rate.
For example, the server-oriented version of this Seagate drive is only 300 GB, but that's a doubling in capacity for 15K rpm drives. And it's expected that it'll have sustained transfer rates of around 125 MB/s, roughly 1/3 higher than its nearest competitor.
Similarly I'd expect this drive (or at least a drive in this lineup) to have the highest transfer rate for 7200 rpm drives -- probably in the 80 MB/s range (the fastest are currently around 60 MB/s). That's really how drives have been getting faster and faster for the past few years. A brand new drive 5 years ago would've been fast if it was getting 25-30 MB/s. Nowadays 50-55 is standard. With perpendicular recording we're looking at upping that to the 75-80 range.
In any case, I could care less about hard drives doubling or tripling in size, until they show significant improvements in performance, or move to solid state, then I am apathetic about the whole industry.
Then you're just showing how little you know about things. I'd love to see solid state persistant storage take off too, but so far every technology that's promised that has fallen flat. So far the only one that's gotten anywhere is Flash -- which isn't all that much faster (probably due to I/O restrictions), has write issues (largely solved w/ write balancing), and cost (20-50x HDs, and only improving slightly faster). I can buy a 250 GB HD for ~$70 right now. I don't even want to think about how much it would cost to build an array of Flash devices of equivalent size.
I'd suggest that terms like giga are mainly used in computing.
What, you've never heard of gigawatts? Even mispronounced ones?
And, quite frankly, the use of base2 vs base10 in computing is hardly uniform. In fact, the only place it's used is in storage units (memory and hard drives), and even then HDs stopped using it years ago (only the OS still reports in base2). And some memory is sold in base10 units now as well (in particular Flash memory).
Networking has always used base10 (10/100/1000 Mbit ethernet is base10), processor speeds are base10, bus speeds are base10 (there's some confusion for you -- memory capacity is base2, speed is base10), and so forth.
Not by fiat, but by agreement.
Yup, and the world has agreed that HD sizes are measured base10 now. It's just the OS vendors who are slow to catch up -- and MS even displays both numbers in some places.
Could you not afford to take unpaid leave?
That was an option of course (as was short-term disability, which would be a lot more sane), but I'm the sole provider for the household (my wife decided that 80+ hour work weeks at Sprint didn't play nicely with having a family, and I support her in that). We have a good bit of savings, but since we're hoping to move in the next year or two I'd rather not touch that unless necessary.
and new parents have to take their parental leave within 6 months instead of 12
Gads... I currently work for the 2nd largest software company (behind MS) currently and I'd rejoice at this option if we had it. We had our second kid two weeks ago, she was in the NICU for a week due to an infection, and I've only taken 3 days off (and have precisely 1 day of vacation left) because we're not given any such thing and only get 2 weeks/year vacation (accrued over the year). If I had to take an extended period of time off then I would have to go on short term disability.
Fortunately the kiddo is doing great now, so I haven't had to do that.
At least we get free soft drinks too... and I'm nowhere even vaguely near the main campus (other side of the country).
Firstly that is a very arrogant approach
Agreed. Writing an enterprise class OS is hard too. As is an enterprise class file system. yet there are open source examples of both. And there are dozens of other examples.
Frankly, I think he was simply trying to downplay the real issue (which ATI outright stated) -- intellectual property issues.
No matter how much the hue and cry (which, as he points out, is pretty damn small right now), neither company could release OSS drivers even if they wanted to because they're constrained by IP agreements. At the very least both use S3's texture compression technology. There have been rumors of Nvidia (via its 3Dfx legacy and later acquisition) being beholden to SGI for years. And there's probably some cross-licensing agreements between the companies, and to other companies.
Which means that when it comes down to it the current state of affairs is going to continue. If you want 3D support (that is, support for vaguely modern hardware) then you'll have to use a proprietary kernel module. It's far from perfect though, and it's pretty much going to doom any hopes for more gaming on Linux or for serious 3D desktop support (ala OS X's Aqua or Vista's Aero) since major distros won't ship 3D enabled kernels due to the legal issues.
The only potential way out is for there to be a stable kernel ABI -- something Linus has long opposed. But that doesn't really resolve the legal issues -- it just makes it easier for closed source kernel modules to be developed (and harder for the kernel team as a whole). And it doesn't solve the issue of fixing kernel bugs that occur when you have closed source modules loaded (the kernel devs won't touch them, for obvious reasons).
lack of multiple desktops
I'm so tired of people saying this. Have you even bothered looking? If you have an ATI or Nvidia card then you have multiple desktop capability built into your video drivers. They're a damn sight faster than any software based virtual desktops.
There's also software solutions, in case you're stuck with an archaic video card or a crappy Intel video card. My preferred one is VirtuaWin, but I've also used DeskWin. They're both OSS under the GPL.
Yes, MS has a PowerToy that does it as well (but it sucks). And even if you have ATI/Nvidia cards you may want to take a look at VirtuaWin -- it's extremely configurable and has a bunch of plugins available.
f you already have a good digital camera and you want to digitize prints, my advice is to photograph them
I've seen a couple of people say this now, and I have to scratch my head at it.
Even a basic scanner can do 600 dpi. For a 4x6 picture that's 8,640,000 pixels, or roughly 8.6 MP. You'll have to have a pretty high end camera to exceed that -- and that's discounting the likelihood of photographing something besides just the picture. Also realize that with a scanner you get even higher resolutions with larger pictures. With a digital camera you're stuck at whatever its resolution is, so with larger source pics you end up with worse results, not better ones!
And if you look at low cost scanners they are generally in the 2400x1200 dpi area, so now you're talking about ~70 MP for a lowly 4x6 now -- probably overkill, but I'd rather have overkill than underkill for a situation like this -- you really don't want to have to do it more than once.
I started from the leftmost one, the supposedly best one.
Then you have an upgraded warranty, additional software, and various other crap.
Start with the 2nd from the right -- it's the lowest end Core Duo processor. You can then upgrade it to your heart's content (beyond the leftmost one if you want).
Oh, and if you got $2150 as the before-discounts-price then you're eligible for the $750 off coupon -- which is quite a bit more than anything else you can possibly use.
unless you hunt for those rare deals
Thing is, they're not rare. Hell, Dell advertised these deals in our local Sunday paper. And they have similar deals running all the time. And "hunting" for the deal consists of putting "Dell coupon" into Google and hitting "I feel lucky". If you don't want to use them, fine, but then you may as well start ignoring prices altogether because it's apparantly not a priority for you.
If you got a different price today then they may have changed their pricing or deals. It was $1501 yesterday (and the day before, and last week).
The $450 coupon is from one of the deals they are currently running. You can click on "Save up to $750 on Inspirons" in the Dell Home section, or just go read up on it here: http://www.xpbargains.com/dell_coupons/
It seems to me that you are still missing some important points
No. I said it wasn't a full list.
On the Mac side, you didn't include MagSafe, Gigabit Ethernet, optical audio in/out (and I'm not sure if the Dell has built in speakers and microphone), ambient light sensors, and DVI port.
MagSafe is cool, I grant you that. Gigabit ethernet? Uh, you realize that neither system has enough bus bandwidth (not to mention hard drive speed) to even come close to handling that data rate, right? Optical in-out, sure, but I don't consider that a huge thing. Yes, the Dell has built in speakers and microphone -- were you serious? Ambient light sensor is only relevant for the backlit keyboard, which I mentioned. I am surprised the Dell doesn't have a DVI port, but hey, the Apple doesn't have a VGA port.
On the Dell side, it has two more USB connectors (but they are Type B !!), a flash memory slot (not sure what it's used for), S-Video port (additional $20 adaptor for the Mac), and integrated modem (ditto).
Type B? Who cares? That's standard USB port. It has a 5-in-1 memory reader, not a "flash memory slot" -- and I should've listed that as a big feature because, IMO, it is. Saves me a USB port and a dongle for reading various camera card formats. S-Video is "eh" to me, just as digital audio out is. I mentioned the integrated modem.
You left out (on the Dell) 802.11a support, a non-slot loading drive (so it can handle small CD/DVDs), and the instant-media support (don't boot the OS for music/movies). None of which I consider important, but you seem to want a full list.
Also, for many people like me, Win XP MCE is not a good alternative for XP Pro, reducing the price gap.
Exactly what features do you need in Pro that aren't in MCE? There's exactly one difference you know. Admittedly, it can be a fairly important one to some business users, but the gap in features between XP Pro and OS X is much larger, so it's more fair IMO to compare the systems with those two.
How much do hobbyist quality video production, audio production, photo editing/cataloging and DVD authoring suites cost for Windows?
$99. Adobe Photoshop/Primere Elements Bundle. Yes, that's street price.
You can cobble stuff together from OSS/Free, but it's nowhere near the quality and ease of use of either iLife or Adobe, so I simply don't recommend trying. Particularly for the video editing/DVD authoring bit (although, on that front, Nero is decent and can be had for $40).
How much does a PDF creator cost for Windows?
Uh... free?
DevTools? A compiler?
Both downloadable for free, from either MS, Cygwin, or MinGW. But I do Unix development, so it's not of much interest to me. A decent shell is, but that's what Cygwin's for.
A full suite of enterprise capable network daemons such as http, ftp, telnet, ftp, ssh?
IIS is included in Pro (but not installed by default). As is a telnet daemon (not enabled by default under SP2).
FTP and ssh daemons are freely available online if you wanted them. And XP's remote desktop is superior to VNC (admittedly, one of the rare cases of XP being better), so I guess that's why you didn't mention it.
Are you seriously trying to say all of that is worth $1000+? I can replace the software for under $200, as I mentioned. Most of the extra functionality can be downloaded for free.
Damnit, I should've previewed.
What that originally said was:
I just bought a new laptop and while I looked at the Macbooks I couldn't justify spending >$1000 for what essentially...
was going to buy a Dell M90 loaded = $3500 or so
MacBook Pro with all options = $3400
Huh, I get $3600 for the MacBook... or did you forget to include the 3 year warranty that's standard on the Dell M90?
But even so, you're comparing a 17" laptop with 1920x1200 resolution and an insanely powerful graphics card (Nvidia Quadro FX 2500M w/ 512 MB dedicated memory) to a 15" laptop with 1440x900 resolution and a very powerful graphics card (ATI Mobility 1600 w/ 256 MB dedicated memory). Not to mention the vast raft of other differences between the two. And the Dell is more expensive? Imagine that! I bet you didn't even try to use free Dell coupons that cut the price down, now did you? Hell, I can't even guess what screwed up way you configured the Dell to make such an absurd comparison -- because, unsurprisingly, you didn't give details.
Hey, let's try for a more fair comparison.
Base level MacBook Pro: $1999
Dell Inspiron E1505 upgraded appropriately: $1501 - $450 = $1051 (upgrades: T2400 CPU, XP MCE install CD, remote control, 15.4" WSXGA screen, 512MB single DIMM, 80 GB 5400 rpm HD, 8X DVD/CD DL burner, 3945 Wireless NIC + Bluetooth, ATI Radeon X1400)
Now the Dell is not a 100% replica of the Apple, but neither is the Apple completely superior to the Dell. This isn't a full list of differences, but it hits the high points:
Dell advantages: Higher resolution screen (1680x1050), dual layer burner, built in modem, two mouse buttons on the laptop itself, $950 in your pocket. (Yesterday you could've gotten 1 GB of RAM for the same price, but that deal expired)
Mac advantages: Much more powerful video card, slightly faster RAM (the Dell can take the RAM, they're just being cheap asses), integrated web cam, backlit keyboard, can boot both OS X and XP.
Again, that's not a full list. There are software differences (which are solved for $1000 for what essentially boiled down to running OS X. I suspect most other people look at it the same way.
Glibc DOES support every function call from 1990 onward.
:)
I should've made that more clear -- support them without recompiling. That certainly isn't true.
And that's the boat that Windows is in -- they have to support all these old binaries. Neither Linux nor Mac OS do that.
IOW, Microsoft could do this. But they won't because it will make evaluating alternatives all the more interesting.
Which is pretty much what I said in my original post.
The problems lie in GNOME and KDE using far too much memory (and probably CPU cycles as well). That's GNOME and KDE, though -- not X11. Try any alternative window manager/desktop environment, and you're likely to see a vast speed improvement.
Agreed, but that's not a valid answer. Ditch KDE and Gnome and you're left with a desktop environment that is nowhere near as friendly to most people. Sure, advanced users can use it with no issues, but that doesn't help people who are trying to get into Linux, or get others into it.
And the KDE/Gnome issues come back if you want to use a program compiled with their libraries -- at that point you have to load the libs into memory and you're now worse off than you would be if you were running KDE or Gnome already (longer load time due to loading the shared libs, possibly a larger overall memory footprint). Don't use those programs? Uh... ok. There goes most of the recent GUI programs for Linux.
The point I'm trying to make is that XP and OS X can deliver all of the eye candy and usability in the given memory footprint while still being faster than X (w/ KDE or Gnome). That's fairly damning. Both KDE and Gnome are bloated projects (as you note) and could benefit from exactly the same kind of total revamp that Windows needs. Hell, just moving to the STL (instead of recreating it from scratch, or attempting to do OO in C) would help reduce memory footprint, improve speed, and (most importantly) increase development speed.
Here are the requirements for slackware 3.4:
/.! Nobody reads the article!
3.4 (Kernel 2.0.33)
* Intel 8086
Uh, I'm sorry, but Slackware is simply not going to run on an 8086, 80186, or 80286.
And while it may boot on an 8 MB RAM system, don't try to use X on that sucker. Or much else for that matter.
Here's the kicker, if you actually READ the article
But this is
It is a rather horribly named article though -- I've got to imagine the name was foisted upon it due to column width/length restrictions or other dead tree issues. It's missing the word "Development" after "Windows" (and even then, that could be misleading, but it's at least better).
And guess what the requirements for Vista is going to be?
Awfully similar to those for other modern OS's, including Mac OS X and modern desktop Linux systems. And, as with any piece of software, I wouldn't want to use it at anywhere near the "minimum" spec.
In NT and all its children, DOS is a separate application provided as a compatability layer running on a kernel which has nothing to do with DOS.
I know that. But why even have it? Remove it and you reduce support costs -- even if that branch of code is stable and rarely requires fixes, it's a section that needs testing whenever you go to QA. Same for the entire Windows-on-Windows emulation layer.
As another poster said, I've heard that it's gone in the 64-bit versions, but I haven't confirmed that myself.
As regards the WMF issue
And you missed the entire point. I know the issues with WMF, including why it was designed that way, but it's an archaic format that is rarely used in modern computing. So ditch it. Ditch support for any archaic data file formats -- there are certainly plenty of them out there.
Note that all of this will create a huge amount of work in the short term. WMF is actually used by some printer drivers, so they'll have to be reworked. That's fine, and it's to be expected. But in the long run you'll save money in development, QA, and support time by shedding tons of dead code. And yes, you'll have some customers scream about breaking their legacy app. So offer to help them with development resources -- that will cost you less than supporting the code for entire userbase.
And yes, I'm oversimplifing the issue, but it's increasingly clear that the current codebase is unsupportable. It's collapsing under its own weight. Can you imagine if glibc on Linux had to support every function call from 1990 forward? That's basically what MS is trying to do. It's a nightmare.