But that was done on purpose, so they wouldn't hit the obvious wall that hurts all VLIW architectures: increasing IPC without changing the architecture, and without adding all the complex re-ordering logic seen in RISC-like superscalar processors. Once you get above one VLIW per clock, you have to throw the compiler's assumptions out the window, or you need to re-compile the code.
If you don't have to support the old architecture, you can change it to increase IPC without excessive overhead. This was the concept behind adding an interpreter layer between the chip and the OS. Of course, they didn't realize that they were trading one performance bugaboo for another: instead of making a bigger, more expensive chip, they sapped tons of performance doing x86 instruction transation and re-ordering in software. This cost them tons of performannce, as a lot of the time, their VLIW pipeline was only %50 filled.
Transmeta had the same problem Intel did with Itanium: with the exception of perfectly tailored code, the VLIW compiler couldn't keep processor resource utilization anywhere near %100. Transmeta had one additional problem over Intel: their compiler had to work in REAL TIME, with a tiny 16 or 32MB buffer. It's no wonder they got toasted by the x86 market..Itanium, even with Intel backing, is on the way to a similar fate.
Absolutely, it is. You can use a higher voltage and a process with larger feature sizes to make your electronics more resistent to a bit flip. The larger feature size and voltage means it takes more energy to flip a particular transistor, at the cost of larger circuits and more power consumed.
You have to add in some buffers to handle the sudden power spikes from particles, so your transistors themselves are safe from damage. Sure, you could still have radiation error events, but they're much less probable with the above setup. MUCH cheaper than lifting enough lead to shield the whole damn circuit board.
Well said! And conversely, the video phone has yet to kill the audio-only phone, although the tech has been around (and affordable) for 40-odd years. Picturephone used only three twsted pair wires, which was well within the capabilities of 1960s telephony tech. And sure, Picturephone was expensive, but today the tech is much cheaper, and yet there is little uptake.
About the only place you'll see video phones today are small niche markets (like field reporters, or soldiers on tours of duty phoning home). For most people, video phones are a solution searching for a problem.
They're OEM licenses. Technically, you have to buy hardware in order to purchase the license, and technically, the license is tied to said hardware. Of course, the "hardware" can consist of nothing more than a mouse.
I have to agree entirely. Hilary lost because she insisted on being in the spotlight for years leading up the the campaign; this is the main reason her supporters and haters were so divided. The problem with this was, all the usual campaign hand-waiving and distractions can't change the mind of voters who made up their minds years before.
Sure, this got her the over-protective mom vote, but the rest of the rational people in this country realized how stupid this all was. A person with this kind of "censor everything" agenda would only grab for more if you gave an inch. It was more than enough for me to vote against her.
But that doesn't solve the basic mainboard architecture... you know, the stuff that brings you I/O performance. I hate sitting on my PC waiting for stuff. That is so 1980ies
So, you'd argue that PCIe raid controllers don't have enough bandwidth for high-capacity I/O? A PCIe switched fabric can handle tens of Gigabytes per second, and you can pipe as many RAID controllers on that as you like. You can go cheap and use SATA controllers with NCQ, or you can go all-out and get SAS controllers.
This isn't a Pentium processor on GTL+ bus with PCI interconnect. Today's workstation/server-class mainboards have support for 8 to 24 cores, and enough interconnect bandwidth (if you go i7 or AMD) to move a library of congress every day.
I would disagree that Intel's power consumption has been "higher." In terms of performance per watt on the desktop, Intel has been impressively competitive.
For example, check out this review of Intel's 45nm Core2 procesors. The true consumption of the processors puts AMD's Athlon 64 X2 far in last. Even if you take into account the power consumed by a memory controller (5w or less, typical), the 45nm Core2 processors have very low power consumption, and the 65nm cores are competitive.
Now, let's read into those results. First of all, the Athlon 64 X2 6000+ listed (a typo, it's a 6400+ in the rest of the article) is a 90nm core, so it's not a fair comparison. You can add to the fray the Brisbane 5050e, which has 45w peak power consumption - this is more competitive with the Core 2 power consumption, although it's not amazing.
At the end of the day, the sad fact is this: every AMD dual core on the market will lose in a desktop performance battle with either the E8200 (27w load) or the E6550 (40w load). You need only look at the article I linked above, where an X2 clocked at 3.2 GHz cannot beat either of the above processors. Even ignoring the 45nm E8200, if you just compare 65nm to 65nm, the Intel chip offers similar power consumption (40w peak) with an incredible performance boost to boot.
Perhaps I'm a bit of a purist, but if ECC Ram is actually self-correcting, I would worry about how/why it got corrupted in the first place. I find it much cheaper and easier to buy good quality non-ECC Ram instead.
ECC isn't just really meant to counter hardware memory errors. It's also meant to prevent single bit flip error events.
Cosmic rays and background radiation can cause single bit flip events. With increasing data densities, the likelyhood of a high-energy particle colliding with a memory cell increases. The other with increasing data densities is: the higher the density, the smaller the cell, and therefore the amount of energy required to flip a bit from 0 to 1 is reduced.
It's a similar problem to the RAID5 array issue (posted recently here on Slashdot), where the data densities are so high that you may enccounter a disk read error before your entire array rebuilds. Higher data densities are making once unheard-of errors possible.
And sometimes, you will immediately see the effects of a bit flip. Other times, the bit flip may remain undiagnosed for years, which can be more annoying (you may not have a backup from that long ago).
So, it's no surprise that ECC is becomming more popular, especially for workstations. As for myself, my current desktop doesn't have ECC, but I'm seriously considering the extra cost of ECC in my next computer.
Almost all 17" LCDs are 1280x1024 which is 5:4. Anyone know why they use this and not 1280x960 (4:3)?
Because "back in the day," 1280x1024 was more popular than 1280x960, even on 4:3 displays!
How did this happen?
The popularity of Silicon Graphics workstations was the impetus behind the industry adoption of 1280x1024. At a time when memory was incredibly expensive, you designed the resolution to fit the memory, and wasted nothing. 1280x1024 was the best choice for a 3D architecture because you could fit one 24-bit frame buffer in 4MB of ram, and the resolution was divisible by 32 or 64 in each dimension (faster hardware accesses with 32-bit or 64-bit workstation processors). The closest 4:3 resolution that played that nicely was 1280x960, which gave you lower resolution, and still required 4MB per 24-bit framebuffer.
But why could SGI get away with a 5:4 resolution on a 4:3 CRT screen? Because in a 3D-rendered application or world, the resolution doesn't affect the view window. The view window is fixed by the program, and objects exist as triangles that are resolution-independent. The only effect resolution has is in final rendering of the scene to a framebuffer: more resolution means a sharper final rendering.
When LCDs came along, it was suddenly much easier to change the form-factor to fit the resolution. And since 1280x1024 was so popular, the 5:4 LCD monitor was born.
A fun fact: the SGI workstations without any 3D graphics capability were locked at a 4:3 resolution, 1024x768. See here for more information.
Just an off-the-cuff calculation on my part shows power consumption dropped over %50 over Barcelona, clock-for-clock.
This is good news, because when AMD moved from 90nm to 65nm, their leakage was so bad that the power consumption only dropped around %10 clock-for-clock. Combine this with better cache architecture (larger, and faster), and AMD may have a winner in the server space.
I'm not sure if they're going to take back the desktop anytime soon. Intel doesn't have the FBDIMM downside on desktop systems, and I'm fairly sure that Shanghai didn't add major microarchitecure changes, so a quad-core Core2, let alone an i7, should continue to dominate the desktop.
However, it is nice to know that the market once again will have a choice in processors. AMD's 65nm offerings were spanked in terms of performance and power consumption by Intel's lineup, but Shanghai will at least compete on the power front, if not the performance front. We shall see what happens when AMD releases their desktop version.
I love all these naysayers parroting about on the subject, when the fact is the lander survived to-design spec, and met mission goals. What, you people need every Nasa project to last as long as the rovers, or it's somehow not a success?
Phoenix was destined to die, regardless of the dust on solar panels problem. It's located in a much colder area of Mars than the rovers, and doesn't have radioisotope heaters (the rovers do). It outlasted the design goal of 90 days operation, so I'm quite happy for them.
What's wrong with being designed for a short life?
Yeah, I have a memory of an even older liquidation: Computer City. About ten years ago, they liquidated entire stores, but the discount was pathetic. Complete computers systems with a "retail" price of $1000 were going for $800, and people were kicking and screaming to buy them. It was pathetic, and I left without buying anything.
I'm not even going to check Circuit City's "markdowns."
Which is exactly why I don't shop at commissioned stores when I can help it. I prefer to educate myself from an impartial source rather than have propaganda thrown at me. Those sellers were motivated enough to negotiate ridiculous deals on things like TVs, but that's only relevant to me once every ten years.
So, you still admit that the commissioned salespeople would draw you in for a single big purchase. That's all they need to justify keeping the commission system, since a lot of people will do the same. And hell, it'as more tha you would buy friom them without commissioned salesepeople.
The fact is, I don't buy computers from big-box stores anymore, because I can educate myself online, and because I can get things shipped direct. But back in 1994 when I was buying my first computer (no knowledge), it was refreshing to have the sales staff guide me to the right purchase. I told the man what I wanted in general (a mid-range 486 DX-2 50), and he didn't try to upsell me. I bought several more computers direct from them for myself and family before I started building my own. One of the computers I bought was only $350 (in 1999), but the salseman didn't sneer at that low commission; I got service just the same.
We were really impressed by their service, so myself and my family purchased several other items from Circuit City, like stereos and VCRs. Even outside the computer department, the service was excellent.
Today, I would still consider purchasing a notebook from them, and of course I would buy large items like TVs and stereo equipment, provided they still had these great salespeople.
I really havn't visited Circuit City since they stopped paying the employees commissions. When I used to enter a Circuit City, an employee would FIND ME, and try to sell me something. Many employees were actually well-informed, because they had to justify their attempts to upsell people who came in just looking for "a computer." This made purchases much easier than any other store with apathetic employees.
The last time I stepped into one of their stores, the employees didn't give me so much as a glance. I've known for quite some time that the company was doomed, and tossing the commission structure was the first sign of the end.
I have a cabinet, in which I keep all my CD's. In the computer, I have the lossy version of every one of them, recorded in variable-bit-rate mp3. Until I can tell the difference between them and the original, even if I have enough disc space, I won't use it for lossless audio.
Yup, my CDs are my backup, but I haven't touched the things in years. I use the same Lame VBR V5 encoded mp3s at home (on my computer speakers or on my Grados) an on my car stereo. In all cases, I can't tell the difference.
Do you want to know the REAL reason why there hasn't been a 128k listening test at Hydrogen Audio recently? It's because they already did this three years ago, and the winner(s) were already imperceptible from the source then.
128k is conqured territory. This is mostly an update to see where encoders have gone in the last four years (if anywhere).
FLAC is still not a good option for portable devices because the higher bandwidth required translates to shorter battery life.
As you have stated above, with simple encoding methods, most of the power wasted accessing a pile of bits is the power to access those bits, not the power required to decode them. On portable devices, FLAC will always be at a disadvantage compared to a codec that uses 10x less data. The question is: will power consumption ever get so low that a 10x power difference goes unnoticed?
This is the true power of mp3: an excellent combination of low complexity and high compression.
This is actually one of the things I hate about AAC: the fact that it is NOT guaranteed to "play anywhere." AAC does not have a standardized profile (LC, MAIN, SRS, LTP), and it does not even have a standardized filename (mp4 or m4a for unprotected files). This makes it difficult for embedded players to support the format correctly, because you have to make sure your encoder is producing the right profie and the right file extension.
For example, the AAC on CD capability on my car stereo only works with Itunes-formatted AAC files with m4a file extensions. If you want to use a different encoder (i.e. Nero), you won't have much luck. This pisses me off, so instead I use VBR mp3s on CDs, which play everywhere.
This is why the market is destined to stick with MP3: it's "good enough," and works everywhere. The advertising campaign stating this fact is long overdue.
The funny thing: when I was shopping for my new car last year, I tested out my mp3 CDs on every car I test-drove. They were formatted by/album/1trackname,/album/2trackname, etc. The CDs played each album correctly, with tracks in-order in every car I drove, which means that the format has essentially become an industry standard.
Re:Most humans aren't that smart
on
The State of Game AI
·
· Score: 2, Interesting
Sure, that's fine and good if you limit your character to an "Orc," and limit their exposure to "field of battle." Limiting scope is the easiest way to get AI that "works."
A good example of limited-scope AI is the shock troopers from the original Half-Life: despite poor hardware specs, the troops reacts to player attacks, falling back and covering their movements with fire/grenades. This only "worked" because the troopers had a very limited list of actions to select from, and the paths the troopers could follow were already defined in stone. Just watch them attack or retreat: they will always launch attacks from or fall-back to the same strong points, using one of perhaps three pre-defined paths.
What I'm talking about is truly diversified AI - is your Orc male or female? Does it have an outgoing or introverted personality? Does it start fights, or run away from them? Does it hunger for something, or does it lack drive? Does it try to be well-liked by everyone, or does it care more about others' happiness? Does it pick it's nose? Does it dance and sing? What exactly makes your Orc engaging, interesting, pathetic, predictable and unexpected, all at the same time?
All the above factors and more can be described to create a more well-rounded AI, one that you can use in all sorts of situations. I mean, limited AI in limited situations has it's place (it makes a great background for a movie scene, or a well-scripted game encounter). But ultimately, what people want to see is AI that does something totally unexpected, like sontaneously get into an arguement with superiors, try to lie, etc. Unfortunately, indentifying all these various fragments that make-up the culmination of an AI's tendancies is difficult, and requires a good AI algorithm to put the data to good use.
Yup, I'm completely against the war in Iraq, but I absolutely think that the war in Afghanistan was justified and necessary: a signifncant strike on our soil demanded a counter-attack. Even if the international community had not been supportive I would have still felt justified in the attack.
If Bush has simply handled Afghanistan without getting a hard-on for war with Iraq, his presidency wouldn't have been so vilified. We should have concentrated on one enenmy, and finished the job.
The saddest fact is, the distraction in Iraq means that zero progress has been made in Afghanistan in the last five years. Hopefully, we can slowly drop the Iraq issue, and make Afghanistan a priority.
How does that justify the lack of support? I mean, people above are gushing about how Linux is great because it supports ancient, obscure hardware, but now that a piece of hardware isn't supported it's suddenly crap?
Because as far as mechanical devices go, you get what you pay for. If you pay more for a printer, you expect it to last longer, and you also expect the cartridges to cost only a fraction of the original hardware purchase price.
The parent is saying you wouldn't expect a $100 mechanical printer to still work after five years (or be viable to buy ink for), so why should you expect the drivers to be supported? You didn't pay enough for that kind of support.
One example of the opposite: I purchased a Brother HL-1240 laser printer for $350 in 2001, and it's still running strong. The replacement toner carts are only $60, which makes continued support for the printer economical (although I've never had to purchase toner). The drivers support Win9x and 2000 out of the box, and now support XP, Vista and OS X (via CUPS).
I would not expect the above performance or support from a $99 laser printer, or a $75 inkjet.
Re:Most humans aren't that smart
on
The State of Game AI
·
· Score: 2, Informative
Wouldn't that be something?
Unfortunately, there's no way to produce an AI like this, because each one would be a work of art. The immense amount of time it would take the programmer to construct personalities like the above from the ground-up would be prohibitive, and no amount of tools could streamline this.
Really, this is the hardest part abouut AI design: classifying the entire human existence into easy-to-handle pieces. Unless you can successfully generalize human experiences and tendancies into neat little packages, there's no way you can create such an impressive AI as the above. You would spend too much time just doing each AI by hand.
The WD Greenpower is a 5400 RPM 3.5" desktop drive with high density platters, and all the low-noise and power-saving features you couild previously only find in a notebook drive. The are also easier to find, and much less expensive than laptop drives.
SPCR reviewed the drive, and found it to be the quietest they ever tested. Further, when you compare it to mainstream "quiet" 7200 RPM drives, the performance hit is only about %10. In all the performance tests I've seen, the Greenpower outperforms my old Barricuda 7200.10, and paces well with
I use the drive in my gaming rig, because my last two drives were too loud. The load times are fine - even Fallout 3 loads in seconds. I also use another one of these drives as external backup for my Mac Mini - good performance, and low noise/heat make this an excellent choice.
Why Bethesda needed to bastardize a great franchise that they did not create into a FPS I will never know.
Fallout 3 is not an FPS. Two reasons why:
1. You can play the game in third-person entirely. Just zoom-out to your preferred distance using the mouse wheel, and then you control the camera with the mouse. This feature was available in Oblivion, but combat was clunky in third-person view, so it was hardly ever used.
2. If you use VATS, you never have to aim at a single enemy - pressing the VATS button will INSTANTLY bring up any enemy within the range of your perception, and you can use action points to kill them at your leisure.
3. Even if you want to play the game like an FPS, you still suffer from your skills. Even if you hit an enemy point-blank with your uber-leet mousing skills, the shot can still miss, and the hit location (and thus damage) is determined by your skill level.
Yes, you can play the game as an FPS - you can aim your weapons by hand if you wish and never touch VATS. But as a Fallout old-timer (I bougth Fallout 2 on release), I've actually moved to the third-person view, and of course I use VATS for all combat. The game plays great!
Slashdot Mirror
But that was done on purpose, so they wouldn't hit the obvious wall that hurts all VLIW architectures: increasing IPC without changing the architecture, and without adding all the complex re-ordering logic seen in RISC-like superscalar processors. Once you get above one VLIW per clock, you have to throw the compiler's assumptions out the window, or you need to re-compile the code.
If you don't have to support the old architecture, you can change it to increase IPC without excessive overhead. This was the concept behind adding an interpreter layer between the chip and the OS. Of course, they didn't realize that they were trading one performance bugaboo for another: instead of making a bigger, more expensive chip, they sapped tons of performance doing x86 instruction transation and re-ordering in software. This cost them tons of performannce, as a lot of the time, their VLIW pipeline was only %50 filled.
Transmeta had the same problem Intel did with Itanium: with the exception of perfectly tailored code, the VLIW compiler couldn't keep processor resource utilization anywhere near %100. Transmeta had one additional problem over Intel: their compiler had to work in REAL TIME, with a tiny 16 or 32MB buffer. It's no wonder they got toasted by the x86 market..Itanium, even with Intel backing, is on the way to a similar fate.
Take a look at this wonderful website. Between the in-depth reviews and lively forum communities, they won't steer you wrong.
They even have a great online player comparison tool if you can't make up your mind.
maybe outer space vanishes 50% of the spiders.
Nahh, the spider in question is just a Rogue, gone into hiding.
When they open up that box...SNEAK ATTACK!
Absolutely, it is. You can use a higher voltage and a process with larger feature sizes to make your electronics more resistent to a bit flip. The larger feature size and voltage means it takes more energy to flip a particular transistor, at the cost of larger circuits and more power consumed.
You have to add in some buffers to handle the sudden power spikes from particles, so your transistors themselves are safe from damage. Sure, you could still have radiation error events, but they're much less probable with the above setup. MUCH cheaper than lifting enough lead to shield the whole damn circuit board.
Well said! And conversely, the video phone has yet to kill the audio-only phone, although the tech has been around (and affordable) for 40-odd years. Picturephone used only three twsted pair wires, which was well within the capabilities of 1960s telephony tech. And sure, Picturephone was expensive, but today the tech is much cheaper, and yet there is little uptake.
About the only place you'll see video phones today are small niche markets (like field reporters, or soldiers on tours of duty phoning home). For most people, video phones are a solution searching for a problem.
They're OEM licenses. Technically, you have to buy hardware in order to purchase the license, and technically, the license is tied to said hardware. Of course, the "hardware" can consist of nothing more than a mouse.
I have to agree entirely. Hilary lost because she insisted on being in the spotlight for years leading up the the campaign; this is the main reason her supporters and haters were so divided. The problem with this was, all the usual campaign hand-waiving and distractions can't change the mind of voters who made up their minds years before.
She lost my vote early on when she revealed her true colors as a censorship machine. First she tried to to censor video games for violence at the federal level (thankfully, it failed to garner support). Then she promised if elected that she would protect us from computer-generated porn and violence.
Sure, this got her the over-protective mom vote, but the rest of the rational people in this country realized how stupid this all was. A person with this kind of "censor everything" agenda would only grab for more if you gave an inch. It was more than enough for me to vote against her.
But that doesn't solve the basic mainboard architecture ... you know, the stuff that brings you I/O performance. I hate sitting on my PC waiting for stuff. That is so 1980ies
So, you'd argue that PCIe raid controllers don't have enough bandwidth for high-capacity I/O? A PCIe switched fabric can handle tens of Gigabytes per second, and you can pipe as many RAID controllers on that as you like. You can go cheap and use SATA controllers with NCQ, or you can go all-out and get SAS controllers.
This isn't a Pentium processor on GTL+ bus with PCI interconnect. Today's workstation/server-class mainboards have support for 8 to 24 cores, and enough interconnect bandwidth (if you go i7 or AMD) to move a library of congress every day.
I would disagree that Intel's power consumption has been "higher." In terms of performance per watt on the desktop, Intel has been impressively competitive.
For example, check out this review of Intel's 45nm Core2 procesors. The true consumption of the processors puts AMD's Athlon 64 X2 far in last. Even if you take into account the power consumed by a memory controller (5w or less, typical), the 45nm Core2 processors have very low power consumption, and the 65nm cores are competitive.
Now, let's read into those results. First of all, the Athlon 64 X2 6000+ listed (a typo, it's a 6400+ in the rest of the article) is a 90nm core, so it's not a fair comparison. You can add to the fray the Brisbane 5050e, which has 45w peak power consumption - this is more competitive with the Core 2 power consumption, although it's not amazing.
At the end of the day, the sad fact is this: every AMD dual core on the market will lose in a desktop performance battle with either the E8200 (27w load) or the E6550 (40w load). You need only look at the article I linked above, where an X2 clocked at 3.2 GHz cannot beat either of the above processors. Even ignoring the 45nm E8200, if you just compare 65nm to 65nm, the Intel chip offers similar power consumption (40w peak) with an incredible performance boost to boot.
And this trend isn't limited to dual-core processors. Intel's 65nm quad beats Barcelona in terms of performance per watt. On the desktop, Intel has the performance/watt leadership.
Perhaps I'm a bit of a purist, but if ECC Ram is actually self-correcting, I would worry about how/why it got corrupted in the first place. I find it much cheaper and easier to buy good quality non-ECC Ram instead.
ECC isn't just really meant to counter hardware memory errors. It's also meant to prevent single bit flip error events.
Cosmic rays and background radiation can cause single bit flip events. With increasing data densities, the likelyhood of a high-energy particle colliding with a memory cell increases. The other with increasing data densities is: the higher the density, the smaller the cell, and therefore the amount of energy required to flip a bit from 0 to 1 is reduced.
It's a similar problem to the RAID5 array issue (posted recently here on Slashdot), where the data densities are so high that you may enccounter a disk read error before your entire array rebuilds. Higher data densities are making once unheard-of errors possible.
And sometimes, you will immediately see the effects of a bit flip. Other times, the bit flip may remain undiagnosed for years, which can be more annoying (you may not have a backup from that long ago).
So, it's no surprise that ECC is becomming more popular, especially for workstations. As for myself, my current desktop doesn't have ECC, but I'm seriously considering the extra cost of ECC in my next computer.
Almost all 17" LCDs are 1280x1024 which is 5:4. Anyone know why they use this and not 1280x960 (4:3)?
Because "back in the day," 1280x1024 was more popular than 1280x960, even on 4:3 displays!
How did this happen?
The popularity of Silicon Graphics workstations was the impetus behind the industry adoption of 1280x1024. At a time when memory was incredibly expensive, you designed the resolution to fit the memory, and wasted nothing. 1280x1024 was the best choice for a 3D architecture because you could fit one 24-bit frame buffer in 4MB of ram, and the resolution was divisible by 32 or 64 in each dimension (faster hardware accesses with 32-bit or 64-bit workstation processors). The closest 4:3 resolution that played that nicely was 1280x960, which gave you lower resolution, and still required 4MB per 24-bit framebuffer.
But why could SGI get away with a 5:4 resolution on a 4:3 CRT screen? Because in a 3D-rendered application or world, the resolution doesn't affect the view window. The view window is fixed by the program, and objects exist as triangles that are resolution-independent. The only effect resolution has is in final rendering of the scene to a framebuffer: more resolution means a sharper final rendering.
When LCDs came along, it was suddenly much easier to change the form-factor to fit the resolution. And since 1280x1024 was so popular, the 5:4 LCD monitor was born.
A fun fact: the SGI workstations without any 3D graphics capability were locked at a 4:3 resolution, 1024x768. See here for more information.
Just an off-the-cuff calculation on my part shows power consumption dropped over %50 over Barcelona, clock-for-clock.
This is good news, because when AMD moved from 90nm to 65nm, their leakage was so bad that the power consumption only dropped around %10 clock-for-clock. Combine this with better cache architecture (larger, and faster), and AMD may have a winner in the server space.
I'm not sure if they're going to take back the desktop anytime soon. Intel doesn't have the FBDIMM downside on desktop systems, and I'm fairly sure that Shanghai didn't add major microarchitecure changes, so a quad-core Core2, let alone an i7, should continue to dominate the desktop.
However, it is nice to know that the market once again will have a choice in processors. AMD's 65nm offerings were spanked in terms of performance and power consumption by Intel's lineup, but Shanghai will at least compete on the power front, if not the performance front. We shall see what happens when AMD releases their desktop version.
I love all these naysayers parroting about on the subject, when the fact is the lander survived to-design spec, and met mission goals. What, you people need every Nasa project to last as long as the rovers, or it's somehow not a success?
Phoenix was destined to die, regardless of the dust on solar panels problem. It's located in a much colder area of Mars than the rovers, and doesn't have radioisotope heaters (the rovers do). It outlasted the design goal of 90 days operation, so I'm quite happy for them.
What's wrong with being designed for a short life?
Yeah, I have a memory of an even older liquidation: Computer City. About ten years ago, they liquidated entire stores, but the discount was pathetic. Complete computers systems with a "retail" price of $1000 were going for $800, and people were kicking and screaming to buy them. It was pathetic, and I left without buying anything.
I'm not even going to check Circuit City's "markdowns."
Which is exactly why I don't shop at commissioned stores when I can help it. I prefer to educate myself from an impartial source rather than have propaganda thrown at me. Those sellers were motivated enough to negotiate ridiculous deals on things like TVs, but that's only relevant to me once every ten years.
So, you still admit that the commissioned salespeople would draw you in for a single big purchase. That's all they need to justify keeping the commission system, since a lot of people will do the same. And hell, it'as more tha you would buy friom them without commissioned salesepeople.
The fact is, I don't buy computers from big-box stores anymore, because I can educate myself online, and because I can get things shipped direct. But back in 1994 when I was buying my first computer (no knowledge), it was refreshing to have the sales staff guide me to the right purchase. I told the man what I wanted in general (a mid-range 486 DX-2 50), and he didn't try to upsell me. I bought several more computers direct from them for myself and family before I started building my own. One of the computers I bought was only $350 (in 1999), but the salseman didn't sneer at that low commission; I got service just the same.
We were really impressed by their service, so myself and my family purchased several other items from Circuit City, like stereos and VCRs. Even outside the computer department, the service was excellent.
Today, I would still consider purchasing a notebook from them, and of course I would buy large items like TVs and stereo equipment, provided they still had these great salespeople.
I really havn't visited Circuit City since they stopped paying the employees commissions. When I used to enter a Circuit City, an employee would FIND ME, and try to sell me something. Many employees were actually well-informed, because they had to justify their attempts to upsell people who came in just looking for "a computer." This made purchases much easier than any other store with apathetic employees.
The last time I stepped into one of their stores, the employees didn't give me so much as a glance. I've known for quite some time that the company was doomed, and tossing the commission structure was the first sign of the end.
I have a cabinet, in which I keep all my CD's. In the computer, I have the lossy version of every one of them, recorded in variable-bit-rate mp3. Until I can tell the difference between them and the original, even if I have enough disc space, I won't use it for lossless audio.
Yup, my CDs are my backup, but I haven't touched the things in years. I use the same Lame VBR V5 encoded mp3s at home (on my computer speakers or on my Grados) an on my car stereo. In all cases, I can't tell the difference.
Do you want to know the REAL reason why there hasn't been a 128k listening test at Hydrogen Audio recently? It's because they already did this three years ago, and the winner(s) were already imperceptible from the source then.
128k is conqured territory. This is mostly an update to see where encoders have gone in the last four years (if anywhere).
FLAC is still not a good option for portable devices because the higher bandwidth required translates to shorter battery life.
As you have stated above, with simple encoding methods, most of the power wasted accessing a pile of bits is the power to access those bits, not the power required to decode them. On portable devices, FLAC will always be at a disadvantage compared to a codec that uses 10x less data. The question is: will power consumption ever get so low that a 10x power difference goes unnoticed?
This is the true power of mp3: an excellent combination of low complexity and high compression.
This is actually one of the things I hate about AAC: the fact that it is NOT guaranteed to "play anywhere." AAC does not have a standardized profile (LC, MAIN, SRS, LTP), and it does not even have a standardized filename (mp4 or m4a for unprotected files). This makes it difficult for embedded players to support the format correctly, because you have to make sure your encoder is producing the right profie and the right file extension.
For example, the AAC on CD capability on my car stereo only works with Itunes-formatted AAC files with m4a file extensions. If you want to use a different encoder (i.e. Nero), you won't have much luck. This pisses me off, so instead I use VBR mp3s on CDs, which play everywhere.
This is why the market is destined to stick with MP3: it's "good enough," and works everywhere. The advertising campaign stating this fact is long overdue.
The funny thing: when I was shopping for my new car last year, I tested out my mp3 CDs on every car I test-drove. They were formatted by /album/1trackname, /album/2trackname, etc. The CDs played each album correctly, with tracks in-order in every car I drove, which means that the format has essentially become an industry standard.
Sure, that's fine and good if you limit your character to an "Orc," and limit their exposure to "field of battle." Limiting scope is the easiest way to get AI that "works."
A good example of limited-scope AI is the shock troopers from the original Half-Life: despite poor hardware specs, the troops reacts to player attacks, falling back and covering their movements with fire/grenades. This only "worked" because the troopers had a very limited list of actions to select from, and the paths the troopers could follow were already defined in stone. Just watch them attack or retreat: they will always launch attacks from or fall-back to the same strong points, using one of perhaps three pre-defined paths.
What I'm talking about is truly diversified AI - is your Orc male or female? Does it have an outgoing or introverted personality? Does it start fights, or run away from them? Does it hunger for something, or does it lack drive? Does it try to be well-liked by everyone, or does it care more about others' happiness? Does it pick it's nose? Does it dance and sing? What exactly makes your Orc engaging, interesting, pathetic, predictable and unexpected, all at the same time?
All the above factors and more can be described to create a more well-rounded AI, one that you can use in all sorts of situations. I mean, limited AI in limited situations has it's place (it makes a great background for a movie scene, or a well-scripted game encounter). But ultimately, what people want to see is AI that does something totally unexpected, like sontaneously get into an arguement with superiors, try to lie, etc. Unfortunately, indentifying all these various fragments that make-up the culmination of an AI's tendancies is difficult, and requires a good AI algorithm to put the data to good use.
Yup, I'm completely against the war in Iraq, but I absolutely think that the war in Afghanistan was justified and necessary: a signifncant strike on our soil demanded a counter-attack. Even if the international community had not been supportive I would have still felt justified in the attack.
If Bush has simply handled Afghanistan without getting a hard-on for war with Iraq, his presidency wouldn't have been so vilified. We should have concentrated on one enenmy, and finished the job.
The saddest fact is, the distraction in Iraq means that zero progress has been made in Afghanistan in the last five years. Hopefully, we can slowly drop the Iraq issue, and make Afghanistan a priority.
How does that justify the lack of support? I mean, people above are gushing about how Linux is great because it supports ancient, obscure hardware, but now that a piece of hardware isn't supported it's suddenly crap?
Because as far as mechanical devices go, you get what you pay for. If you pay more for a printer, you expect it to last longer, and you also expect the cartridges to cost only a fraction of the original hardware purchase price.
The parent is saying you wouldn't expect a $100 mechanical printer to still work after five years (or be viable to buy ink for), so why should you expect the drivers to be supported? You didn't pay enough for that kind of support.
One example of the opposite: I purchased a Brother HL-1240 laser printer for $350 in 2001, and it's still running strong. The replacement toner carts are only $60, which makes continued support for the printer economical (although I've never had to purchase toner). The drivers support Win9x and 2000 out of the box, and now support XP, Vista and OS X (via CUPS).
I would not expect the above performance or support from a $99 laser printer, or a $75 inkjet.
Wouldn't that be something?
Unfortunately, there's no way to produce an AI like this, because each one would be a work of art. The immense amount of time it would take the programmer to construct personalities like the above from the ground-up would be prohibitive, and no amount of tools could streamline this.
Really, this is the hardest part abouut AI design: classifying the entire human existence into easy-to-handle pieces. Unless you can successfully generalize human experiences and tendancies into neat little packages, there's no way you can create such an impressive AI as the above. You would spend too much time just doing each AI by hand.
The WD Greenpower is a 5400 RPM 3.5" desktop drive with high density platters, and all the low-noise and power-saving features you couild previously only find in a notebook drive. The are also easier to find, and much less expensive than laptop drives.
SPCR reviewed the drive, and found it to be the quietest they ever tested. Further, when you compare it to mainstream "quiet" 7200 RPM drives, the performance hit is only about %10. In all the performance tests I've seen, the Greenpower outperforms my old Barricuda 7200.10, and paces well with
I use the drive in my gaming rig, because my last two drives were too loud. The load times are fine - even Fallout 3 loads in seconds. I also use another one of these drives as external backup for my Mac Mini - good performance, and low noise/heat make this an excellent choice.
Why Bethesda needed to bastardize a great franchise that they did not create into a FPS I will never know.
Fallout 3 is not an FPS. Two reasons why:
1. You can play the game in third-person entirely. Just zoom-out to your preferred distance using the mouse wheel, and then you control the camera with the mouse. This feature was available in Oblivion, but combat was clunky in third-person view, so it was hardly ever used.
2. If you use VATS, you never have to aim at a single enemy - pressing the VATS button will INSTANTLY bring up any enemy within the range of your perception, and you can use action points to kill them at your leisure.
3. Even if you want to play the game like an FPS, you still suffer from your skills. Even if you hit an enemy point-blank with your uber-leet mousing skills, the shot can still miss, and the hit location (and thus damage) is determined by your skill level.
Yes, you can play the game as an FPS - you can aim your weapons by hand if you wish and never touch VATS. But as a Fallout old-timer (I bougth Fallout 2 on release), I've actually moved to the third-person view, and of course I use VATS for all combat. The game plays great!