With Apple NetRestore/NetBoot, you can boot off another image straight over the network just by holding down option and selecting what image you want to boot off of. No preinstalled software, no login, no special hardware, (okay, it has to be within 5 years of so old.)
This means that, assuming you're not putting non-public info on these images, you can net boot a netinstall image, boot an image that has a OS tailored to image your local drive off a simple file server full of images, or skip the imaging thing altogether and just operate off the net image.
In regards to security, you'll have to log in once the image is booted, unless set to autologin. It's all up to the admin.
So uh, that covers everything you can and suggests a few more.
Hold on man. Sure it's great for clustering applications but think about how many apps you have that can actually use this level of thread concurrency.
Since we actually don't know how many cores they're going to have, we'll have to eval it from both the "tons of cores" and the "not that many cores" viewpoint.
I'll concede number 3 since that's obvious no matter what the product.
However, with a lot of cores, say 100. Most people run one app at a time while they're working. If their one major app isn't very parallel, you're just wasting the other cores. Since each core isn't going to run faster than 1ghz nor have the complexity of a PPC750, we're talking something like an old PPC603e or Pentium per thread. On paper, having 100 500mhz Pentiums sure beats out whatever's on th market now, but just having the capibility doesn't mean it can all be used.
That's the "tons of cores scenerio". The other is just that we have lower power consumption multiprocessing on more computers. That's not revolutionary at all. Sure it makes Intel's life harder cuz their procs have neither the efficiency or sheer power, but that doesn't mean they automatically lose.
Finally, who said Sony's going to make a high end workstation using Cell? Don't just speculate, back that up. I can't say Sony doesn't have the engineers to do it, cuz they do. But for what reason would they? It'll be quite the divergence from what they do now which is consumer electronics. A high end workstation's also gotta have an OS and application software. Despite the power, it'll totally lose out to Macs simply because Apple can leverage the same processor technology while they already have apps.
In all seriousness, CE has always been better written than NT and also more portable. Moving it to a PowerPC shouldn't be that hard considering they've worked with it before.
And it not being an x86 just means it'll be faster, cooler, and probably cheaper and easier to design around.
And yeah, it's pretty much confirmed that they're using a PowerPC. I recall seeing somebody mentioning they're seeding developers using Apple G5s.
The Cell processor is getting majorly overhyped. Basically, it reduces down to getting a bunch of cores probably based off the Power5 series into one die and basically doing the same thing as Sun's throughput computing efforts or the direction of current GPUs.
On the software side, it'll most likely abstract out into very efficient threading, although each thread might not be that fast (but still fast enough). Kinda like having 100 233mhz processors in a UNIX box. It is far from being a new idea or disruptive. In fact, you might as well say Sega's Saturn, a quad processor programming nightmare would be where this stuff's evolved from, except now we can fit magnitudes more cores into a console than we could 10 years ago.
Anyhow, I just love the fact that all three major consoles are using PowerPC architectures.:)
Ah, Nokia phones are nearly unknown in Japan BECAUSE JAPAN DOESN'T USE GSM. Well, there might be one provider of GSM in Japan....maybe. But because of the population density, GSM will get inferior quality of service no matter how you slice it. And hence, nobody will care.
The problem with why the Xbox might need Japan is because the majority of games that sell the best, so far have come from Japan. DDR, Final Fantasy series, Mario, Gran Turismo, etc.
Well, consider it this way. The reason Xserves and Power Macs are shipping as well as they are (which might still be a bit hindered) is because none of them are being diverted to iMacs for a while. Getting the initial orders of the pro lines out gives them more room as sales slow a bit from the 2.5ghz intro. This little bit more slack ought to let them ship the iMacs after the delay.
Hey! My Acura had the same situation, except I think I only had 4 speakers. Bose head unit with adying filter capacitor that causes it to make loud whiny noises at the right frequency to drive all people insane.
You know what REALLY sucks? That the Good Guys charged me to have the amps disabled even though I later read the headunit I had them install had alternate outputs for amped speakers.
Kinda like the IT guys out there who choose Windows when they don't need to because it keeps them employed.
Ahh screw you all, I just want a Nissan Skyline R34 GTR...... The only thing your Volvo and any BMW will be seeing is each other drooling at back of it fading into the distance:)
Re:Yeah, but I'd still toss it, Maya uses 3 button
on
Apple Delays New iMac
·
· Score: 1
Technically, you're using three. One to click, then thumb and opposite to actually steer the mouse.
Adding two buttons either makes it so that one finger uses two buttons, or you got a full hand.
If you have more than three buttons, you risk having some people squeeze too hard when moving and triggering accidentally.
Your logic, as well as the links you're referring to, are outdated. The specific reason is that the automotive industry shifted from carburators to electronic fuel injection.
All that you mentioned only applies to engines with a carb.
The electronic fuel injection systems currently arn't intelligent enough to modify all of their datapoints according to realtime data. In most cases, there's just a large chart that tells the system what to do, and in the event that something goes off the chart, it reverts to one that sacrifices power and efficiency for safety.
It is in this case that you appear to gain power and efficiency when using high octane gasoline. The problem is, this happens much more often than you think it would. In my old Acura, switching from 87 to 89 and then to 91 would get me increments of an additional 25 miles out of the same amount of fuel. In the times I've seen an additional 50 miles out of a tank and have this situation reproducable solely by changing the gas, I'm fairly convinced despite understanding what you're saying.
I'm sorry but I'm going to say this as gently as I can.....
YOU'RE A FRIGGIN' MORON.
Sure, in the strictest terms, you're right. A text editor and a word processor when working on html are equivalent. However, the context of the parent post was not of the typical unix geek, html webguru, etc sort of user. The whole article was about Word 5.1, and in the context of Word 5.1 we're talking about people who prefer the click-format-text-bold to make their words bold, not .
Most people do not use html to make their documents. Certainly not a simple essay. Definetely not their letter to grandma.
Okay, after reading your reply and links, as well as the original post, yeah, you're right, it's a troll.
Nevertheless, if it was cleaned up to only include facts, the troll has a grain of truth to it. Here I am trying to install Gentoo on an old Mac (PM6500/225 as a server) wondering if I should give up after all the problems I've run into and install a LinuxPPC 2000 distro 5 years old with no security patches (since that distro is long gone)..... and hope nobody's got a PPC rootkit. (Debian woody crashes before I can get finish booting the installer, YDL feels unbelieveably slow, and Mandrake crashes in the installer.)
Anyhow, I'm impressed that War3 does run now since it actually is the only thing I use my PC for anymore, however it certainly doesn't look like a easy thing to do.
Just because some games run, doesn't mean that it qualifies as a gaming platform since it's obvious that not all people play your selected games.
Where's my Warcraft 3? My Windows box's got it. My Mac's got it.
Where's Command and Conquer Generals? Windows' got it. Mac's got it.
Realistically, you post is just the flip side of: "Where's Final Fantasy X? My PS2's got it. Nobody else does, so they're all inferior. nyahh!" One game doesn't make a gaming platform. One genre doesn't make a gaming platform either. Wide selection (especially the most popular) matched to many different target audiences does.
Er.... Mac Zealots, Intel Engineers, PC users WITH brains.... they all say the same thing:
A P4 is less efficient than a P3.
Proof? Look at the early Tom's Hardware, Ars Technica articles when the P3 1ghz and P4 1.4/1.8ghz units were on the market. The P4 was slower despite the higher clock rate. The only thing it could do faster was cook my breakfast faster because of the increased heat output. (refer to the AMD 1800XP frying an egg article for fun)
Shortly after, the P3, was discontinued because it's shorter pipeline wasn't able to be clocked much faster and the P4 was. Plus, marketing would easily want the higher clock speed over the higher real speed because it's easier to convince stupid people to buy it with a bigger number.
The P4's at a dead end now. Intel's increasing use of the Pentium M is proof of that. And in case you didn't know, the Pentium M is just a P3 with a improved bus and cache. The core design is still basically unchanged. And just so you know, THIS IS A GOOD THING.
Just because it isn't open source doesn't mean it isn't useful. This mentality is just as bad as the PHB 'if it isn't from microsoft/intel and compatible with Office, I can't trust it' mentality.
Where are the BeOS-like filesystem queries on Linux? Where are the Baldur's Gate clones? And, most of all, where is the stuff that, once and for all, asserts the superiority of the open source community, the proof that we can invent, rather than wait for the corporations to do it for us?
Er...... cloning/replicating a BeOS-like filesystem and cloning Balder's Gate....... you call this "proof that we can invent"?
The best way to promote the opensource community would be to actually make new ideas rather than new implementations. Or lay the groundwork for those. For example, yet another window manager or OSX skin is just being a poser. Yet another interface paradigm with studies on usability to back them up will make a good impression. Laying the groundwork for a properly user friendly OS would also be beneficial, like moving away from this monolithic kernel.
Er.... Estes rocket engines were solid black powder engines. They burned for a few seconds and then had a ejection charge (except for booster stages) which opened your parachute. I have the F15 jet model as well which deployed the glider wings using the ejection charge and rubber band.
Glow engines are a completely different thing. They're 2 stroke internal combustion engines. Cox was the purveyor of the little cheap ones you could get in the local Toys-r-us store. The displacement was often the.049/Class 1/2A engines that ran off glow fuel, a fuel made up of alcohol, castor oil, and nitromethane (%25). To make it easier to start, I found that adding %5 of WD-40 into the mix when doing the initial burn-in helped quite a bit. The glow refers to the glow-plug which was the equivalent of the spark plugs in cars. They glow by the use of a battery when starting, and continue to glow powered by the combustion of fuel, which then leads to having enough heat to set off the next combustion cycle.
If only I figured out how to build the models correctly; I wouldn't have made the flying cruise missile of a plane by sticking on an oversized engine and letting it loose with only a rudder to control it and the wrong elevator angle:P
Hey, the Kayaks and Vectras were not common HPs machines. They were what HP considered "workstation class" and obviously had better designs then a consumer class machine. The first time I saw a Kayak was at school with a little programmable LCD+buttons in front. The machine POSTed with a smiley.
Anyhow, the comparison you're making is sorta like this: Would you like a HP Pavilion versus a HP Kayak? Would you like a Compaq Presario versus a Compaq ProLiant?
It's not hard to see that whether it's an HP or a Compaq, it can still be either good or shitty.
Personally, with the new Compaq Evo laptops, Compaq's name has the upper hand in "seems to be better designed" compared to HP's bright-blinding-blue-light-always-shining-directly -in-your-eyes laptops.
Highly doubtful. Throwing better hardware at the problem (GPUs) versus using what can be reduced to a tiny coprocessor add-on inside the CPU (SSE*) is pretty much a no-contest scenerio.
In some of the results, the GPU was found to be over 16X faster. There's no way that SSE would have helped since the parallelism in the GPU is way beyond what you can get out of SSE. Sure, it might have made it closer, but having the CPU lose by a 4X multiplier still grants their study credit as they wanted to see that GPUs can do stuff better in certain cases.
Like I said before, I could have gone on longer. But now it's more like the effects of bad engineering in other areas rather than focusing on the bad engineering itself.
Addaon is pretty much right. Now Intel's got to keep supporting the wrongly designed earlier versions while Motorola/IBM just have one to handle. It's like having to release beta products and still support them later. Sure they have some real world advantages, but it hasn't been as clear cut as Altivec.
There's two(well, part two is sorta two parts) problems I see in terms of "real world advantages" in regards to MMX/SSE*.
1. Future design impact. (This is an engineering issue) They released it, they have to support it until the end of the x86 series. MMX for the most part is now just making the microcode engine bigger which does put a limiter (although small) on how well they can optimize the core later. Eventually SSE* will face the same problem. Simply put, less layers of logic is faster.
2. Developer support. This is not really an engineering issue. Intel has to compete with AMD who sometimes does not have support for the same instructions. This impacts the optimization for the programs people write since you'll have to write for the common denominator or have to take care of two set of routines: one for P4s, one for everything else. At that point, it's easier to search for algorithmic improvements that benefit your app as a whole. Sure compilers do make that a bit easier, but it's not as simple as tossing in a few flags. You need to make sure what you have is parallizable in the first place. Kinda like how the old "ooh, my dual 1ghz is faster than your 1.5ghz" arguments only hold when you got more than one thread working towards a common goal.
3. More developer support. MMX/SSE* most likely overlap across versions. At least functionally. Meaning, if I were to optimize my code and try to select the best instructions to do so, I'd have to research exactly which version works better for each case. If I pick the wrong one, there an unknown impact. With huge instruction sets like x86, we're either going to go through a lot of trouble of researching because of Intel's eagerness to release an incomplete optimization, or leave it up to the compiler to take it's best guess at how your code works. I do however concede that less and less people are doing x86 assembly since compilers are pretty good, but nevertheless, a developer knows his own code best.
Now (we're pretty much out of the engineering and into the long term outlook part), I can also concede that IBM and Motorola suffer from the same problems. Except the scale is so much more smaller. The future design impact is mitigated since Altivec wasn't designed to be a permanent addition into the ISA. Whether or not Intel intended MMX/SSE* to be permanent was decided for them when they promoted to the extent that they did. AMD being pretty much forced to follow demonstrates that. Sure, IBM/Motorola/Apple promoted Altivec a lot too. But PowerPCs are used in much more than just desktops and usually without Altivec. And variations between how IBM and Motorola implemented each instruction do exist, but it's certainly not as severe as waiting for the Athlon 64 to maybe support SSE3 in the next version. (By the way, when I mean not severe, there's a certain instruction on the G5 which takes twice as long as the same instruction on the G4. Nevermind that it hasn't shown up in any benchmarks I've seen, but it does show that IBM did put enough thought into supporting everything Motorola did since the only program I've seen broken was Virtual PC.) As on the developer side. Who uses Altivec? A few embedded DSP systems custom designed to each chip and Apple. Apple's at least making great use of Altivec in the OS and developer tools. There's no "oh what instructions should we use" sort of thing, it's either using Altivec or not. Apple's advantage is that also most of it is through library functions meaning it can be easily switched in and out of large amounts of code sinc
I don't know about the rest of Slashdot, but here goes my opinion:
For starters, I recently graduated from UC Berkeley. My last class was CS152 where the students form their own groups and design a CPU of their own. http://www-inst.eecs.berkeley.edu/~cs152
I like good designs. I like tech where people put a lot of work into making it work better and more elegantly than is actually necessary. I admire it when a design just speaks about it's designers and says "we know our shit and we know how to do it good." And to top it off, I'm willing to pay for it....but I can't complain when it's cheap:)
In short, my problem with Intel is that they're no longer using good engineering to sell. They've switched to marketing bullshit.
The Pentium 4, from an architecture standpoint, is a brute force method of pushing the clock speed for no reason other than to let the marketing guys say "ooh, 3 ghz." The long pipeline is what gives it that ability. The power consumption from the extra flipflops needed to form each pipelining stage and from fliping them so fast makes those huge ass heat sinks necessary. An Intel P4 uses up more physical resources to run simply so they can try to sell more for less performance.
Take a look at the PowerPC and the Athlons. AMD's marketing-speak numbers actually reflect a generally equivalent (yeah yeah, there are always variations, but it's still generally close enough) to the P4's computation power. Sure the Athlons give out ALMOST as much heat, but they're still less than the P4. The lower clock rate helps too in making the system stable since it's easier to have gates that latch/hold correctly at slower speeds (duh.) Less radio interference. You don't need to overclock everything on the board or rely on perfectly placed wires to avoid interference. Heck, it's easier and cheaper to design almost everything else on the board and chipset because of this AND STILL GET THE SAME OR BETTER PERFORMANCE. And the PowerPCs? Looking at Apple's machines, most of them still don't have direct CPU fans. They can use the same chips in the laptops and the desktops. No "mobile" version necessary. Heck, in certain cases, a 500Mhz G4 will actually beat out a 3Ghz P4, although those cases are specialized. (like distributed.net)
All of their "innovations" which they advertise don't really mean anything. MMX was a joke even as far back as the P200s. To execute a MMX instruction, you have to do a bunch of stuff to clear out your registers before you can use the instruction. Unless you had a lot to do, you might have just wasted the gains of the instruction on the prep time since you'll probably have to commit a bunch of stuff to memory which takes a while. NetBurst (the super duper long pipeline) doesn't make your internet experience better, it just increases your instruction latency (which I admit doesn't matter much when you're clocking 3ghz) and cause your CPU to croak on a branch prediction oopsies.
There's a good reason why the current Centrino/PentiumM chipsets are based off a P3. It's because it was a more respectable design than the P4 in almost every way except marketing speak. It was more power efficient. It also had a higher IPC count. (instruction per cycle) The 1ghz P3 beat out all the 1.6-1.8ghz P4s on introduction and so they just scrapped the P3s soon after because it didn't help their marketing. What kind of company scraps the better product?
The P3, by the way, should have been named the P2a or something since all that was changed was a few instructions and the cache size. But hey, 3 sounds better than 2.
Finally, the majority of developers no longer do optimizations by hand. So the "slight processor/OS incompatibility" shouldn't ever happen since almost every compiler for x86 is aware of both processors. The instruction set is the same. If the output is the same, it doesn't matter who's processor did it as long as 1+1=2 and 0xffffffff + 0x00000001 = (either 0 or an exception or whatever was specified by the ISA) Though,
Slashdot Mirror
With Apple NetRestore/NetBoot, you can boot off another image straight over the network just by holding down option and selecting what image you want to boot off of. No preinstalled software, no login, no special hardware, (okay, it has to be within 5 years of so old.)
This means that, assuming you're not putting non-public info on these images, you can net boot a netinstall image, boot an image that has a OS tailored to image your local drive off a simple file server full of images, or skip the imaging thing altogether and just operate off the net image.
In regards to security, you'll have to log in once the image is booted, unless set to autologin. It's all up to the admin.
So uh, that covers everything you can and suggests a few more.
Hold on man. Sure it's great for clustering applications but think about how many apps you have that can actually use this level of thread concurrency.
Since we actually don't know how many cores they're going to have, we'll have to eval it from both the "tons of cores" and the "not that many cores" viewpoint.
I'll concede number 3 since that's obvious no matter what the product.
However, with a lot of cores, say 100. Most people run one app at a time while they're working. If their one major app isn't very parallel, you're just wasting the other cores. Since each core isn't going to run faster than 1ghz nor have the complexity of a PPC750, we're talking something like an old PPC603e or Pentium per thread. On paper, having 100 500mhz Pentiums sure beats out whatever's on th market now, but just having the capibility doesn't mean it can all be used.
That's the "tons of cores scenerio". The other is just that we have lower power consumption multiprocessing on more computers. That's not revolutionary at all. Sure it makes Intel's life harder cuz their procs have neither the efficiency or sheer power, but that doesn't mean they automatically lose.
Finally, who said Sony's going to make a high end workstation using Cell? Don't just speculate, back that up. I can't say Sony doesn't have the engineers to do it, cuz they do. But for what reason would they? It'll be quite the divergence from what they do now which is consumer electronics. A high end workstation's also gotta have an OS and application software. Despite the power, it'll totally lose out to Macs simply because Apple can leverage the same processor technology while they already have apps.
And what's wrong with using Windows CE/Embedded?
In all seriousness, CE has always been better written than NT and also more portable. Moving it to a PowerPC shouldn't be that hard considering they've worked with it before.
And it not being an x86 just means it'll be faster, cooler, and probably cheaper and easier to design around.
And yeah, it's pretty much confirmed that they're using a PowerPC. I recall seeing somebody mentioning they're seeding developers using Apple G5s.
The Cell processor is getting majorly overhyped. Basically, it reduces down to getting a bunch of cores probably based off the Power5 series into one die and basically doing the same thing as Sun's throughput computing efforts or the direction of current GPUs.
:)
On the software side, it'll most likely abstract out into very efficient threading, although each thread might not be that fast (but still fast enough). Kinda like having 100 233mhz processors in a UNIX box. It is far from being a new idea or disruptive. In fact, you might as well say Sega's Saturn, a quad processor programming nightmare would be where this stuff's evolved from, except now we can fit magnitudes more cores into a console than we could 10 years ago.
Anyhow, I just love the fact that all three major consoles are using PowerPC architectures.
Ah, Nokia phones are nearly unknown in Japan BECAUSE JAPAN DOESN'T USE GSM. Well, there might be one provider of GSM in Japan....maybe. But because of the population density, GSM will get inferior quality of service no matter how you slice it. And hence, nobody will care.
The problem with why the Xbox might need Japan is because the majority of games that sell the best, so far have come from Japan. DDR, Final Fantasy series, Mario, Gran Turismo, etc.
Well, consider it this way. The reason Xserves and Power Macs are shipping as well as they are (which might still be a bit hindered) is because none of them are being diverted to iMacs for a while. Getting the initial orders of the pro lines out gives them more room as sales slow a bit from the 2.5ghz intro. This little bit more slack ought to let them ship the iMacs after the delay.
Oh, and Sumit (the guy who started that project) has been posting at www.thexcar.com/forums
:)
The project's at a point where it might move over to a database backend. If you're a coder, start coding
http://www.sourceforge.net/projects/tmrs
Hey! My Acura had the same situation, except I think I only had 4 speakers. Bose head unit with adying filter capacitor that causes it to make loud whiny noises at the right frequency to drive all people insane.
You know what REALLY sucks? That the Good Guys charged me to have the amps disabled even though I later read the headunit I had them install had alternate outputs for amped speakers.
Kinda like the IT guys out there who choose Windows when they don't need to because it keeps them employed.
Ahh screw you all, I just want a Nissan Skyline R34 GTR...... The only thing your Volvo and any BMW will be seeing is each other drooling at back of it fading into the distance :)
Technically, you're using three. One to click, then thumb and opposite to actually steer the mouse.
Adding two buttons either makes it so that one finger uses two buttons, or you got a full hand.
If you have more than three buttons, you risk having some people squeeze too hard when moving and triggering accidentally.
Your logic, as well as the links you're referring to, are outdated. The specific reason is that the automotive industry shifted from carburators to electronic fuel injection.
All that you mentioned only applies to engines with a carb.
The electronic fuel injection systems currently arn't intelligent enough to modify all of their datapoints according to realtime data. In most cases, there's just a large chart that tells the system what to do, and in the event that something goes off the chart, it reverts to one that sacrifices power and efficiency for safety.
It is in this case that you appear to gain power and efficiency when using high octane gasoline. The problem is, this happens much more often than you think it would. In my old Acura, switching from 87 to 89 and then to 91 would get me increments of an additional 25 miles out of the same amount of fuel. In the times I've seen an additional 50 miles out of a tank and have this situation reproducable solely by changing the gas, I'm fairly convinced despite understanding what you're saying.
So if you look at rank 81 through 92, there's a lot of machines using Xeon 2.8s with gig-e. Rank 81 has 1030 processors, rank 92 has 650.
Yeah, that's a difference of 380 processors, fairly close to 30%....... and Rmax is 2026 for all of them.
Sure, this is one benchmark only, but damn, that must look bad when your extra 380 procs doesn't get any improvement.
I'm sorry but I'm going to say this as gently as I can.....
.
YOU'RE A FRIGGIN' MORON.
Sure, in the strictest terms, you're right. A text editor and a word processor when working on html are equivalent. However, the context of the parent post was not of the typical unix geek, html webguru, etc sort of user. The whole article was about Word 5.1, and in the context of Word 5.1 we're talking about people who prefer the click-format-text-bold to make their words bold, not
Most people do not use html to make their documents. Certainly not a simple essay. Definetely not their letter to grandma.
Okay, after reading your reply and links, as well as the original post, yeah, you're right, it's a troll.
Nevertheless, if it was cleaned up to only include facts, the troll has a grain of truth to it. Here I am trying to install Gentoo on an old Mac (PM6500/225 as a server) wondering if I should give up after all the problems I've run into and install a LinuxPPC 2000 distro 5 years old with no security patches (since that distro is long gone)..... and hope nobody's got a PPC rootkit. (Debian woody crashes before I can get finish booting the installer, YDL feels unbelieveably slow, and Mandrake crashes in the installer.)
Anyhow, I'm impressed that War3 does run now since it actually is the only thing I use my PC for anymore, however it certainly doesn't look like a easy thing to do.
Just because some games run, doesn't mean that it qualifies as a gaming platform since it's obvious that not all people play your selected games.
Where's my Warcraft 3? My Windows box's got it. My Mac's got it.
Where's Command and Conquer Generals? Windows' got it. Mac's got it.
Realistically, you post is just the flip side of: "Where's Final Fantasy X? My PS2's got it. Nobody else does, so they're all inferior. nyahh!" One game doesn't make a gaming platform. One genre doesn't make a gaming platform either. Wide selection (especially the most popular) matched to many different target audiences does.
Er.... Mac Zealots, Intel Engineers, PC users WITH brains.... they all say the same thing:
A P4 is less efficient than a P3.
Proof? Look at the early Tom's Hardware, Ars Technica articles when the P3 1ghz and P4 1.4/1.8ghz units were on the market. The P4 was slower despite the higher clock rate. The only thing it could do faster was cook my breakfast faster because of the increased heat output. (refer to the AMD 1800XP frying an egg article for fun)
Shortly after, the P3, was discontinued because it's shorter pipeline wasn't able to be clocked much faster and the P4 was. Plus, marketing would easily want the higher clock speed over the higher real speed because it's easier to convince stupid people to buy it with a bigger number.
The P4's at a dead end now. Intel's increasing use of the Pentium M is proof of that. And in case you didn't know, the Pentium M is just a P3 with a improved bus and cache. The core design is still basically unchanged. And just so you know, THIS IS A GOOD THING.
Just because it isn't open source doesn't mean it isn't useful. This mentality is just as bad as the PHB 'if it isn't from microsoft/intel and compatible with Office, I can't trust it' mentality.
Where are the BeOS-like filesystem queries on Linux? Where are the Baldur's Gate clones? And, most of all, where is the stuff that, once and for all, asserts the superiority of the open source community, the proof that we can invent, rather than wait for the corporations to do it for us?
Er...... cloning/replicating a BeOS-like filesystem and cloning Balder's Gate....... you call this "proof that we can invent"?
The best way to promote the opensource community would be to actually make new ideas rather than new implementations. Or lay the groundwork for those. For example, yet another window manager or OSX skin is just being a poser. Yet another interface paradigm with studies on usability to back them up will make a good impression. Laying the groundwork for a properly user friendly OS would also be beneficial, like moving away from this monolithic kernel.
Er.... Estes rocket engines were solid black powder engines. They burned for a few seconds and then had a ejection charge (except for booster stages) which opened your parachute. I have the F15 jet model as well which deployed the glider wings using the ejection charge and rubber band.
.049/Class 1/2A engines that ran off glow fuel, a fuel made up of alcohol, castor oil, and nitromethane (%25). To make it easier to start, I found that adding %5 of WD-40 into the mix when doing the initial burn-in helped quite a bit. The glow refers to the glow-plug which was the equivalent of the spark plugs in cars. They glow by the use of a battery when starting, and continue to glow powered by the combustion of fuel, which then leads to having enough heat to set off the next combustion cycle.
:P
Glow engines are a completely different thing. They're 2 stroke internal combustion engines. Cox was the purveyor of the little cheap ones you could get in the local Toys-r-us store. The displacement was often the
If only I figured out how to build the models correctly; I wouldn't have made the flying cruise missile of a plane by sticking on an oversized engine and letting it loose with only a rudder to control it and the wrong elevator angle
AWD using gasoline and electric assist?
/ ar ticles/48694/article.html
The Honda Dual-Note AWD Prototype does it. Too bad it's not for sale. Supposedly it's based off the Acura NSX frame from what a friend first told me.
http://www.edmunds.com/news/conceptcarspotlight
There's a possibility that the Lexus RX400H which is coming out next year does it too.
Hey, the Kayaks and Vectras were not common HPs machines. They were what HP considered "workstation class" and obviously had better designs then a consumer class machine. The first time I saw a Kayak was at school with a little programmable LCD+buttons in front. The machine POSTed with a smiley.
y -in-your-eyes laptops.
Anyhow, the comparison you're making is sorta like this:
Would you like a HP Pavilion versus a HP Kayak?
Would you like a Compaq Presario versus a Compaq ProLiant?
It's not hard to see that whether it's an HP or a Compaq, it can still be either good or shitty.
Personally, with the new Compaq Evo laptops, Compaq's name has the upper hand in "seems to be better designed" compared to HP's bright-blinding-blue-light-always-shining-directl
Highly doubtful. Throwing better hardware at the problem (GPUs) versus using what can be reduced to a tiny coprocessor add-on inside the CPU (SSE*) is pretty much a no-contest scenerio.
In some of the results, the GPU was found to be over 16X faster. There's no way that SSE would have helped since the parallelism in the GPU is way beyond what you can get out of SSE. Sure, it might have made it closer, but having the CPU lose by a 4X multiplier still grants their study credit as they wanted to see that GPUs can do stuff better in certain cases.
Like I said before, I could have gone on longer. But now it's more like the effects of bad engineering in other areas rather than focusing on the bad engineering itself.
Addaon is pretty much right. Now Intel's got to keep supporting the wrongly designed earlier versions while Motorola/IBM just have one to handle. It's like having to release beta products and still support them later. Sure they have some real world advantages, but it hasn't been as clear cut as Altivec.
There's two(well, part two is sorta two parts) problems I see in terms of "real world advantages" in regards to MMX/SSE*.
1. Future design impact. (This is an engineering issue) They released it, they have to support it until the end of the x86 series. MMX for the most part is now just making the microcode engine bigger which does put a limiter (although small) on how well they can optimize the core later. Eventually SSE* will face the same problem. Simply put, less layers of logic is faster.
2. Developer support. This is not really an engineering issue. Intel has to compete with AMD who sometimes does not have support for the same instructions. This impacts the optimization for the programs people write since you'll have to write for the common denominator or have to take care of two set of routines: one for P4s, one for everything else. At that point, it's easier to search for algorithmic improvements that benefit your app as a whole. Sure compilers do make that a bit easier, but it's not as simple as tossing in a few flags. You need to make sure what you have is parallizable in the first place. Kinda like how the old "ooh, my dual 1ghz is faster than your 1.5ghz" arguments only hold when you got more than one thread working towards a common goal.
3. More developer support. MMX/SSE* most likely overlap across versions. At least functionally. Meaning, if I were to optimize my code and try to select the best instructions to do so, I'd have to research exactly which version works better for each case. If I pick the wrong one, there an unknown impact. With huge instruction sets like x86, we're either going to go through a lot of trouble of researching because of Intel's eagerness to release an incomplete optimization, or leave it up to the compiler to take it's best guess at how your code works. I do however concede that less and less people are doing x86 assembly since compilers are pretty good, but nevertheless, a developer knows his own code best.
Now (we're pretty much out of the engineering and into the long term outlook part), I can also concede that IBM and Motorola suffer from the same problems. Except the scale is so much more smaller. The future design impact is mitigated since Altivec wasn't designed to be a permanent addition into the ISA. Whether or not Intel intended MMX/SSE* to be permanent was decided for them when they promoted to the extent that they did. AMD being pretty much forced to follow demonstrates that. Sure, IBM/Motorola/Apple promoted Altivec a lot too. But PowerPCs are used in much more than just desktops and usually without Altivec. And variations between how IBM and Motorola implemented each instruction do exist, but it's certainly not as severe as waiting for the Athlon 64 to maybe support SSE3 in the next version. (By the way, when I mean not severe, there's a certain instruction on the G5 which takes twice as long as the same instruction on the G4. Nevermind that it hasn't shown up in any benchmarks I've seen, but it does show that IBM did put enough thought into supporting everything Motorola did since the only program I've seen broken was Virtual PC.) As on the developer side. Who uses Altivec? A few embedded DSP systems custom designed to each chip and Apple. Apple's at least making great use of Altivec in the OS and developer tools. There's no "oh what instructions should we use" sort of thing, it's either using Altivec or not. Apple's advantage is that also most of it is through library functions meaning it can be easily switched in and out of large amounts of code sinc
I don't know about the rest of Slashdot, but here goes my opinion:
:)
For starters, I recently graduated from UC Berkeley. My last class was CS152 where the students form their own groups and design a CPU of their own. http://www-inst.eecs.berkeley.edu/~cs152
I like good designs. I like tech where people put a lot of work into making it work better and more elegantly than is actually necessary. I admire it when a design just speaks about it's designers and says "we know our shit and we know how to do it good." And to top it off, I'm willing to pay for it....but I can't complain when it's cheap
In short, my problem with Intel is that they're no longer using good engineering to sell. They've switched to marketing bullshit.
The Pentium 4, from an architecture standpoint, is a brute force method of pushing the clock speed for no reason other than to let the marketing guys say "ooh, 3 ghz." The long pipeline is what gives it that ability. The power consumption from the extra flipflops needed to form each pipelining stage and from fliping them so fast makes those huge ass heat sinks necessary. An Intel P4 uses up more physical resources to run simply so they can try to sell more for less performance.
Take a look at the PowerPC and the Athlons. AMD's marketing-speak numbers actually reflect a generally equivalent (yeah yeah, there are always variations, but it's still generally close enough) to the P4's computation power. Sure the Athlons give out ALMOST as much heat, but they're still less than the P4. The lower clock rate helps too in making the system stable since it's easier to have gates that latch/hold correctly at slower speeds (duh.) Less radio interference. You don't need to overclock everything on the board or rely on perfectly placed wires to avoid interference. Heck, it's easier and cheaper to design almost everything else on the board and chipset because of this AND STILL GET THE SAME OR BETTER PERFORMANCE. And the PowerPCs? Looking at Apple's machines, most of them still don't have direct CPU fans. They can use the same chips in the laptops and the desktops. No "mobile" version necessary. Heck, in certain cases, a 500Mhz G4 will actually beat out a 3Ghz P4, although those cases are specialized. (like distributed.net)
All of their "innovations" which they advertise don't really mean anything. MMX was a joke even as far back as the P200s. To execute a MMX instruction, you have to do a bunch of stuff to clear out your registers before you can use the instruction. Unless you had a lot to do, you might have just wasted the gains of the instruction on the prep time since you'll probably have to commit a bunch of stuff to memory which takes a while. NetBurst (the super duper long pipeline) doesn't make your internet experience better, it just increases your instruction latency (which I admit doesn't matter much when you're clocking 3ghz) and cause your CPU to croak on a branch prediction oopsies.
There's a good reason why the current Centrino/PentiumM chipsets are based off a P3. It's because it was a more respectable design than the P4 in almost every way except marketing speak. It was more power efficient. It also had a higher IPC count. (instruction per cycle) The 1ghz P3 beat out all the 1.6-1.8ghz P4s on introduction and so they just scrapped the P3s soon after because it didn't help their marketing. What kind of company scraps the better product?
The P3, by the way, should have been named the P2a or something since all that was changed was a few instructions and the cache size. But hey, 3 sounds better than 2.
Finally, the majority of developers no longer do optimizations by hand. So the "slight processor/OS incompatibility" shouldn't ever happen since almost every compiler for x86 is aware of both processors. The instruction set is the same. If the output is the same, it doesn't matter who's processor did it as long as 1+1=2 and 0xffffffff + 0x00000001 = (either 0 or an exception or whatever was specified by the ISA) Though,