Personal computing is about to undergo a fundamental transformation, if industry cheerleaders are to be believed — and to stunning effect. Rebounding basketballs or ricocheting bullets in today’s computer games, shown only as rough approximations of reality, will become more true-to-life. Voice recognition, now so error-prone as to be scarcely usable, will morph into a dependable tool as computers become able to understand and execute complex verbal commands.
64-bit processor can handle more games, DVD copying
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Marketing will be marketing. It is designed to fluff up a technology to be larger than life to make it sell!
Sounds like the romp that M$ always goes on – “The new windows will make life easier, drive space shuttles, make a cup of tea and tell your horoscope” – Yeah right!!!
…it can already tell your fortune
I’m waiting for a 64-bit iBook, no hurry here, they’ll only get cheaper and 128-bit (if ever!) is a long long ways away.
PS- I believe the older version of KDE had a Tea Cooker Timer or something.
The word bullsh*t doesn’t go far enough.
The only *** major *** difference between 32 and 64 bit CPUs is how much memory they can access in one chunk. This does nothing to help graphics directly. That is the job of the graphics card.
I think the much bigger deal would be if a computer have 4 CPUs instead of one or two. Then the game makers would have a lot more distributed horsepower to use. Assuming the OS is good at mutithreading.
Doesn’t sse2 allow operations on 128bit ints or two 64bit floats at once already? We could do 64int operations with mmx since 1997.
What advantages are there to 64bit processors that cannot be done with mmx,sse,sse2 etc?
I’m talking strictly multimedia here, where the article seems to claim most of the gains are to be had.
Right now most game benchmarks are faster on 32 bits then 64 due to lack of optimizations. After Windows XP64 is released there will be more 64 bit native code and people will have a reason to migrate. Things will probably get interesting around christmas. Intel is planning to hit 4 Ghz by the end of the year so we should see some Athlon 64 3800+’s or 4000+’s around that time.
Oh yes, and we will have PCI express cards then also.
What? How does a 64 bit processor make any of this happen? I have been using 64 bit computers for years and there is NO difference except how much memory can be accessed. Don’t believe it.
KDE still comes with a tea cooker app.
Question though, he states that 64 bit processors can pull more information at once, this is true and false. They do pull 64 bits at a time instead of 32 (for integer operations), but this is only an advantage if you have a number greater than 4 billion. Of course for mathematics research 32 bits is a measely pittance, and I’d say even 64 bit doesn’t cut it for some future operations. But for the gamer, 64 bit isn’y gonna crank more speed out of your game. It may someday mean a greater accuracy. 64 bits is slower for example if you have a number like 4, because you don’t need to pull as many bits.
I don’t think personal computing is up to super computing, since Virginia Tech built an 1100 Node dual G5 cluster. The average person couldn’t even afford the space requirements for so many computers, even less the power and maintenance costs.
If you want more realistic physics in games then you’re going to need a boatload of memory….much more than what can be addressed in the 32-bit world. But yeah, 64-bit doesn’t necessarily mean speed gains. Aside from more memory in the 16-bit to 32-bit transition, a big gain was because it takes a lot more instructions when you want to do arithmetic operations on something bigger than 65535 which is pretty important.
… 64 bit address space is big enough to include RAM + HDD + backups, so it is perfectly feasible to create an OS that allows applications to forget about files and all related stuff.
Though I have no idea as what it could be.
Howdy
I love articles like this as it kinda gets you used to the marketing crap your going to see for the next few years :0)
I think the design of the processor is WAY more important to the speed of the apps then the 32 vs 64 bit crap they keep feeding us.
If half the things were true about 64 bit CPUs we all would of been running server hardware for years!
I`m not saying they are better but please a dose of reality every now and then is nice too.
i`m not saying they are
Errr aren`t*
The article is deffinatly worthless marking crap, but you guys are WAY downplaying 64-bit computing. Its not just about memory. 64-bit CPUs are deffinatly capable of doing complex calculations faster. They can do much more per clock then any 32-bit CPU and thats a strait fact.
It will be a while before software takes full advantage of the new CPUs, but there will deffinatly be benefits for gaming (more complex AI and physics modeling), media editing, and voice recognition. Big benefits.
About as fluffy a piece as ever.
Well I expect to get an AMD64 sometime but I am not expecting the huge increase in throughput, only about 20% or so for same clock speed for chip design SW. Thats not really very impressive is it since that 64b cpu will probably use 2x the power of my XP2400.
I am much more excited about the prospect of parallel computing again since I am building the kind of cpu the industry dare not build (but once did). The funny thing is that although it will be 10x slower than Athlon, its only a few $ per copy so I can explore possibly putting 256 or so together with out too much expense.
The other funny part is I can choose to use RLDRAM instead of regular DDRAM that is 4x faster and I can call it L2 cache. For some reason the industry wont’t use the fastest RLDRAM to effectively take L3 cache to 32Mbytes pushing all its weight on just 1 very hot cpu.
oh well, the future won’t be predicted from these fluff pieces.
If half the things were true about 64 bit CPUs we all would of been running server hardware for years!
I think the price was a little prohibitive before. The biggest advantage 64-bit computing will bring is the amount of memory it can handle. Not many people need 4GB on the desktop now but it won’t be too long before it’s not enough. 1-2GB is not all that uncommon now.
What an utter load of crap. That author should be relieved of duty.
umm…yeah ok.
I don’t think it’s much of a big deal. 64-bit computers have existed since forever, if you could afford them. Personally I think it’s a bit overkill, maybe 48-bit would be a neat compromise. That’s 262,144 GB (256 TB) of addressing and more than you need for anything else you might want to do (in fact 16-bit integers, for most uses, are more than enough). 17,179,869,184 GB is a bit silly but 4 GB is not enough if you want to implement something cool like persistence on mass storage devices.
For advanced calculations… well that’s what special extensions like MMX and SSE are there for. We already have 128-bit registers in Pentiums!
Of course the problem with these big word sizes is that integers and pointer get bigger. That nice struct just doubled in size! But functionality did not increase. And think every machine instruction with an encoded literal is now twice as big too. Eek. Wasted memory – it soon adds up.
The other cool thing about 48-bit is that I could fit 8 characters of this 6-bit minimalist character set I invented one rainy day nicely into a word. With 32 and 64, there’s wastage because they are not multiples of 6.
😉
OK, I guess that doesn’t apply to anyone else (!) Sry.
repnz stosq
i didnt read the article but there is a grain of truth in the fact that with a 64-bit cpu you can shuffle in one job what the 32-bit cpu would need 2 jobs for, that means again that you can have bigger numbers and therefor more accurate info on stuff like physics, 3d graphics and so on. basicly the games and videos will be just as fast but they will look way better and realistic doing so. sure you would need a 64-bit gpu to handle the info but thats not a problem as i think most allready are atleast that but now they are working with a 64-bit cpu and that means that they talk the same language, no need waiting for the cpu to do its just so you can take the data and use it to generate the next image (the gpu still needs to be told by the game where to put stuff in the game remeber?). most desktop office suites and so on dont need 64-bit but thats not the selling point. its for the power gamers, and those are the ones that have been pushing the hardware sales for the last 10-20years anyways. hell the reason i have a ms os around and keep updateing my hardware is just that, games. and with powerpci (like someone have allready stated) we have the next gen transport bus to keep all this ticking! watch out sone and ms x-box, the pc is far from dead as a gameing hog:)
As if the 64 bit CPU were just invented yesterday… The commodity marketplace and its commentators are at it again.
I’m writing this email on a Sun Ultra 60 with two 64 bit Ultrasparc 450MHz RISC CPUs and about 1.5GB of system memory. It is a nice usable machine and I like it– but it’s no speedster. My other machine is an AMD 32 bit AthlonMP 2200+ 2 CPU box with about the same amount of memory. It is much faster, for example crunching Seti@home work units about 4x faster than the Sun as a computational benchmark. Reason: not 32 vs 64 bit CPU or vice versa; clock speed and overall system/bus/memory design. Older vs. new.
The longer data word may help only if bus designs accomodate high bandwidth memory access, as is the commonplace strategy in most video card design.
Sure, the 64 bit CPU relieves the memory constraints but the real speed and expanded capability on the desktop will come from widespread use of efficient parallel multiple CPU – multithreaded designs supported by multithreaded software, just as it has in larger scale computing.
“Voice recognition, now so error-prone as to be scarcely usable, will morph into a dependable tool as computers become able to understand and execute complex verbal commands.”
Ummmm, bullshit. I’m a radiology resident, and at work I use voice recognition sostware (based off the Dragon engine) for all my dictations. I routinely dictate at around 98% accuracy. It doesn’t make mistakes with words like “appendicitis” or “intercostal”. I occasionaly have problems with monosyllabic words that I tend to rush.
So, I’m not sure what the hell this person is talking about. It the software was any better it could probably take my lunch order, and remind me when it’s my girlfriend’s birthday two days in advance (I have GOT to start remembering it).
I see a lot of the same stuff about how the 64 bit is not really faster then the 32 bit CPUs. Lets take a look down memery lane and look at what was said when the first practical 32 bit machines were coming to the market from memory ( which is failing more and more 🙂 I hate to say it but the realities of the trends of CPUs will show that 64 bit machines will open doors no one here can foresee.
In the X86 space, the biggest improvement is the additional CPU registers, since that’s a x86 weak spot. Recompiling code for AMD64 boosts speed based on that alone.
Fortunately, the AMD64, even in 64-bit mode, sets the default operand size to 32-bit, avoiding unnecessary code bloat. And as has been mentioned, unless you are already using integers larger than 32-bit, or are bumping into the 4GB memory limit, there is no inherent *immediate* benefit in the 64-bitness alone. I’m not talking about increased registers or other improvements, just the fact the you now have 64-bit addressing and integers. The 32-bit x86 CPUs already had support for 64-bit floating point, and 128-bit SIMD. So AMD64 really just completes the circle.
Can a 64-bit CPU grab more data at one time from memory? Not necessarily- The P3, P4, and Athlon XP already had a 64-bit data bus. How much data you can grab at one time is really more a function of your memory bus design. The Opteron has a 128-bit wide memory bus, the Athlon64 (non-FX) has a 64-bit wide one. So the data bus capacity is not really dependent on a chip being called 32 or 64 bit.
Doesn’t sse2 allow operations on 128bit ints or two 64bit floats at once already? We could do 64int operations with mmx since 1997.
What advantages are there to 64bit processors that cannot be done with mmx,sse,sse2 etc?
I’m talking strictly multimedia here, where the article seems to claim most of the gains are to be had.
SSE2 and similar extensions have up to 128 bit registers but that doesn’t mean they operate on 128-bit operands. That 128 bit register is actually 4 x 32-bit registers. What you do is you load 4 32-bit pieces of data into the register and execute one instruction on all elements at once, hence the term Single Instruction Multiple Data (SIMD).
On a 64 bit platform your int is 64 bits wide by default, so it can reach higher numbers. This does tax the memory bandwidth and takes up more memory in data/instruction cache.
The article also neglected to mention that although we can stuff 8 GB of RAM into a G5 Mac, applications still can’t access all of that 8 GB, still limited to 4 GB. But it’s still a benefit. Suppose you do put 8 GB of RAM in there, your apps with be using up to 4 GB or real memory and not HD swap.
“The article is deffinatly worthless marking crap, but you guys are WAY downplaying 64-bit computing. Its not just about memory. 64-bit CPUs are deffinatly capable of doing complex calculations faster. They can do much more per clock then any 32-bit CPU and thats a strait fact.”
Actually, if all else was equal, the 64-bit chip would be slower for most things.
“It will be a while before software takes full advantage of the new CPUs, but there will deffinatly be benefits for gaming (more complex AI and physics modeling), media editing, and voice recognition. Big benefits.”
That’ll have little to do with the “bitness” of the chip, unless those tasks require over 4GB per process. You need to understand that 64-bit chips aren’t new.
“I see a lot of the same stuff about how the 64 bit is not really faster then the 32 bit CPUs. Lets take a look down memery lane and look at what was said when the first practical 32 bit machines were coming to the market from memory ( which is failing more and more 🙂 I hate to say it but the realities of the trends of CPUs will show that 64 bit machines will open doors no one here can foresee. ”
They haven’t opened those doors in the decade they’ve been available. The transition from 16-bit to 32-bit addressing was a bigger deal – considering software at the time was already limited by it.
I’m not talking about increased registers or other improvements, just the fact the you now have 64-bit addressing and integers. The 32-bit x86 CPUs already had support for 64-bit floating point, and 128-bit SIMD. So AMD64 really just completes the circle.
That’s a moot point. Like I said in my above post, the SIMD extensions on current x86 chips still operate on 32 bit operands, but in multiples of them at once. You were right about the benefits of more registers, biggers integers. There’s no point in bringing up SIMD extensions, comparing SIMD to 64 bit computing is vector vs. scalar. Using the vector unit has performance benefits but the programmer/compiler has to take into consideration the limitations of the vector unit, and the application has to be designed well to take advantage of SIMD. Not every app can benefit from it, we can say the same about 64 bit CPUs. Not every app will benefit (but not necessarily suffer) from 64 bitness.
Can a 64-bit CPU grab more data at one time from memory? Not necessarily- The P3, P4, and Athlon XP already had a 64-bit data bus. How much data you can grab at one time is really more a function of your memory bus design. The Opteron has a 128-bit wide memory bus, the Athlon64 (non-FX) has a 64-bit wide one. So the data bus capacity is not really dependent on a chip being called 32 or 64 bit.
I’m a little confused by what you’re saying here. The width of the memory bus would dictate bandwidth, in relation to frequency of data transmission. But accessing chunks of memory at a particular address isn’t related to memory bus width, at aleast I don’t believe it is. More importantly is that if the code loads a memory address (a pointer) into a register, that register better be big enough to hold that data. In C++ if you did sizeof(void*) it would return 4 (bytes) which means 32 bits. Do the same on a 64 bit platform (in 64 bit mode) and I bet it would return 8.
I can’t help but chuckle at the idea of “the jump from 32 bits to 64 bits will be like the jump from 16 bit to 32 bit.”
People who believe that are forgetting something: the software development industry is in a drought of innovation. Back then, when computers went from 16 bit CPUs to 32 bit CPUs I don’t think the mouse was invented yet, so everything was mostly keyboard and command line interface. Then the PC became more visual and multimedia oriented.
Someone will have to come up with “the next big thing” either an alternative to window-based GUI or new input devices, basically change the way we interact with the OS. Maybe then we can justify having more bits and more RAM.
“People who believe that are forgetting something: the software development industry is in a drought of innovation. ”
That’s what happens when an industry matures.
— “Actually, if all else was equal, the 64-bit chip would be slower for most things.” —
Technically thats true, but most of the things a CPU does would also go just as quick on a 16-bit cpu, all else being equal. There are many calculations that are much faster on a 64-bit CPU however, and the large speed increase for those far outpaces any small speed decrease for the majority of simple intiger operations a CPU does during most of its operation.
Not to mention, all is NOT equal anyway. These new 64-bit cpus are also significantly more powerful and have superior architectures.
— “That’ll have little to do with the “bitness” of the chip, unless those tasks require over 4GB per process. You need to understand that 64-bit chips aren’t new.” —
I very much understand that, but you need to understand what 64-bit means. Its not just some extension to increase memory addressing. In fact, the increase in max system memory is technically just a side-effect, though a much desired one. The examples I listed were no mistake. They all will DEFFINATLY benifit from a 64-bit CPU, even if all else is equal. AI, physics modeling, video editing, and voice recognition all do extreamly complex operations and the better the precision of the numbers, the better the result. 64-bit means that programers can go to greater levels of precision with less of a drop in performance.
At the same time!!!
This reminds me of the old scene with kids arguing about their favorite consoles.
Seriously though, in the console world, bitness is heavily used as marketing ploy. Refer to PS2, which claimed a 128-bit CPU. Surely, if that was as helpful as described in the article, the PS2 would be comparable to a massive supercomputer. Not so.
In the early 90’s, the console wars were based on bitness. That’s why Nintendo put out the N64. 64 bits. Automatic ‘edge’ over the competition. Whereas the PS1 had a 32 bit processor, but still produced better visuals.
But, the general public still goes by the notion of ‘bigger is better’. They are the ones that would choose a product over another based on the bigger number in the description of one of them. Sadly, this ploy is effective.
You know, the Sega Dreamcast was the first 128-bit console. Can’t wait until SPARC, IA-64, and POWER release 128bit versions. hehehehe.. Then again the dreamcast was ahead of its time. It failed and is no longer made
blah blah blah new technology blah blah better life, peace in the world blah blah blah.
faster computer = heavier use = in the end the same, too slow, machine
Doubling the word size does loads of things, but improving speed isn’t one of them. When making the transition from 32 to 64 bit years ago with the early Alpha RISC chips we were surprised by the utter LACK of performance improvement. After some basic research at the time I formed the opinion that application performance improves in SPITE of 64 bits, not because of it.
With every memory address now double the size, most executables bloat right up. Almost twice the data to load from disk, shuffle round memory & clog the bus.
The real difference between 32 and 64 bit processors is the register size and the maximum manageable memory.
While having 64 bit of memory may seem useful no chipset (except if you’re the NSA) can handle a full 64 bit allocation space, 40 bit is the most you can expect these days.
Going forward while 64 bit means “wow, a lot of bits in a row” it’s almost useless for consumer level applications, only databases and some server application can really get some advantage on this…
The real difference could be a new instruction set (no more BCD opcodes into P4 core for example) but it’s unlikely to happen on XP64 and Xeon/P4 ME (same instruction set) while Itanium was almost thrown away by Intel as “server processor” (IE “not the x86 successor”).
3D doesn’t need 64 bit CPU (well, 3D works almost alone), voice recognition has nothing to do with it and probably, you’ll notice LESS speed on 64 bit environment rather than with 32 bit ones.
While write back caches makes almost everything better having to move and process 64 bit registers requires:
1- more memory bandwidth
2- more CPU work to manage 32/16/8 bit operands
3- more TLB
4- a completely different (and less efficient) CPU page caching/handling mechanism (you’ll see when total amount of ram will reach 16-32 Tb)
5- more RAM
Those are hardware related problems that will slow down 64bit CPU (vs 32 bit ones).
There’s no evidence that a 64 bit environment and processor works better than a 32 bit one with common applications, it’s the opposite right now and, as far as I know, no one really needs a 64 bit processor for desktop or workstation use (well’except if you’re going to work with Photoshop with 10.000×10.000 true color images, but that’s the need of RAM!).
AMD itself promote XP64 as a 32bit processor with 64bit instruction rather than a 64bit processor…
The real difference between 32 and 64 bit processors is the register size and the maximum manageable memory.
While having 64 bit of memory may seem useful no chipset (except if you’re the NSA) can handle a full 64 bit allocation space, 40 bit is the most you can expect these days.
Going forward while 64 bit means “wow, a lot of bits in a row” it’s almost useless for consumer level applications, only databases and some server application can really get some advantage on this…
The real difference could be a new instruction set (no more BCD opcodes into P4 core for example) but it’s unlikely to happen on XP64 and Xeon/P4 ME (same instruction set) while Itanium was almost thrown away by Intel as “server processor” (IE “not the x86 successor”).
3D doesn’t need 64 bit CPU (well, 3D works almost alone), voice recognition has nothing to do with it and probably, you’ll notice LESS speed on 64 bit environment rather than with 32 bit ones.
While write back caches makes almost everything better having to move and process 64 bit registers requires:
1- more memory bandwidth
2- more CPU work to manage 32/16/8 bit operands
3- more TLB
4- a completely different (and less efficient) CPU page caching/handling mechanism (you’ll see when total amount of ram will reach 16-32 Tb)
5- more RAM
Those are hardware related problems that will slow down 64bit CPU (vs 32 bit ones).
There’s no evidence that a 64 bit environment and processor works better than a 32 bit one with common applications, it’s the opposite right now and, as far as I know, no one really needs a 64 bit processor for desktop or workstation use (well’except if you’re going to work with Photoshop with 10.000×10.000 true color images, but that’s the need of RAM!).
AMD itself promote XP64 as a 32bit processor with 64bit instruction rather than a 64bit processor…
“Not to mention, all is NOT equal anyway. These new 64-bit cpus are also significantly more powerful and have superior architectures.”
Not really, no. They aren’t “significantly” more powerful. The chip that *does* offer a big improvement is the AMD64 series, since they have an increased register count. The other 64-bit chips on the market at this level aren’t significantly more powerful – and in fact less so for most things – compared to a P4.
“I very much understand that, but you need to understand what 64-bit means. Its not just some extension to increase memory addressing. In fact, the increase in max system ”
If gives you a hardware long long type.
“memory is technically just a side-effect, though a much desired one. The examples I listed were no mistake. They all will DEFFINATLY benifit from a 64-bit CPU, even if all else is equal. AI, physics modeling, video editing, and voice recognition all do extreamly complex operations and the better the precision of the numbers, the better the result. 64-bit means that programers can go to greater”
Most of what you describe doesn’t benefit from a hardware long long type. Floating point is already 64-bit on 32-bit hardware (80-bit internally on the P4).
“levels of precision with less of a drop in performance.”
Right. Sure. I’ll say it again – 64-bit chips have been available for a decade. They aren’t magic, nor do they provide what you think they provide.
if 64-bit is so useless then why dont we switch back to 16-bit chips? So what you all are saying is obviously we shouldnt have even upgraded from 16, well maybe 62 bit enterprise chips are pointless too? oh negativity negativity you all will probably harass my posts too then? saying how stupid i am or something?
if 64-bit is so useless then why dont we switch back to 16-bit chips? So what you all are saying is obviously we shouldnt have even upgraded from 16, well maybe 62 bit enterprise chips are pointless too? oh negativity negativity you all will probably harass my posts too then? saying how stupid i am or something?
No. Why do you need to go back to 16 bit chips if you think 64 bits is useless? Just stick with 32 bits for now.
There’s a reason for our negativity. We get sick of hearing people over hype something. Just reading the article in the news post, I can tell they really don’t have a grasp of what they’re talking about (the freep.com site I’m talking about). It just sounded like they took the press release materials and rehashed it into something resemble “news”, basically doing someone else’s marketing.
A lot of people in here understand 64 bit computing. We’re not out to stope the world from 64 bit computing, we’re just trying to water it down. History has shown that most things in the tech industry that get too hyped up end up being a flop.
*cough* .Net *cough*
everything will work.
I am glad the people who do create new processors are not like negative posters here who seem like to critisize enthusiasts just for laughs or to show off there knowledge.
otherwise there would be no pc/internet/linux/etc at all.
Well I wouldn’t mind having the good stuff as long as it’s affordable. I’m considering an AMD64 Chip as long as I can run Linux and UT2004 on it. I think I’ll wait until Microsoft gets a true 64-Bit windows on the 64-Bit chip but that may be a bit off too. I wouldn’t mind optimized programs, greater MP3/MP4 encoding, DVD burning, and all of that jazz.
I still see that eventually 64-Bit chips will replace 32-bit ones before the end of this decade.
Isn’t this just like reprint of an article from when the Nintendo 64 came out? I mean they’ve been pushing ’64-bit’ computing on us since the Atari Jaguar came out. Are these new chips relly going to change anything? Nope. Are they going to make our lives easier? Nope, that’s a job for the software designers and hardware designers, not the chip designers.
Most of the computers that could be used to make our lives easier are 4, 8, 16 or 32 bit. They are the microcontrollers in every little device we use or own. From your microwave to your cell phone to your camera. Those are the processors that should be more exciting but they are not. I guess it’s just not exciting to many people.
Personally I would rather have a cell phone that only needed to be recharged once a week rather then every other day. And yes, processors do still take up lots of power. Only the display uses more in most devices.
— “Not really, no. They aren’t “significantly” more powerful. The chip that *does* offer a big improvement is the AMD64 series, since they have an increased register count. The other 64-bit chips on the market at this level aren’t significantly more powerful – and in fact less so for most things – compared to a P4.” —
I specifically said ‘these new…’ reffering to the new AMD64 and PPC970 which ARE superior in many ways, with which you seem to agree. Do you really think this article is about Alphas or UltraSPARCs??? Stay on topic!
— “If gives you a hardware long long type.” —
Of course, the ability to handle 64-bit longs and pointers and do 64-bit arithmatic more efficiently is not at all beneficial right? Oh, wait, it certainly is! Primarily in… well, you know. See below.
— “Most of what you describe doesn’t benefit from a hardware long long type. Floating point is already 64-bit on 32-bit hardware (80-bit internally on the P4).” —
See below.
— “Right. Sure. I’ll say it again – 64-bit chips have been available for a decade. They aren’t magic, nor do they provide what you think they provide.” —
Right sure. I’ll say it again – I know that they have been available for decades and they are not magic. They do provide what I think because I only think they provide what they in fact do provide. The only disagreement here is that you think AI, physics modeling, video editing, and voice recognition would not benefit signifcantly from that, and I do. I’d also like to add cryptography to that list as well. I didn’t think of it before, but its certainly becoming more important these days.
Honestly, I don’t know how to convince you that the above applications would benefit, I thought it was obvious. I give up.
Right sure. I’ll say it again – I know that they have been available for decades and they are not magic. They do provide what I think because I only think they provide what they in fact do provide. The only disagreement here is that you think AI, physics modeling, video editing, and voice recognition would not benefit signifcantly from that, and I do. I’d also like to add cryptography to that list as well. I didn’t think of it before, but its certainly becoming more important these days.
How will AI benefit? Neural nets make extensive use of floating point data. Vision also uses floating point data. Same goes for physics modelling. How will 64 bits benefit those applications?
Cryptography doesn’t really benefit from the move to 64 bits. It benefits much more from vector processing, e.g. Altivec. Check out the distributed.net client. Run the benchmark on your top of the line P4/Opteron. Then run it on any G4/G5 computer. Get shocked by the results 🙂
I don’t know anything about video editing or voice recognition, so I’ll leave that to someone else.
As someone worked in the field of speech recognition, I completely cannot understand the statement about the improvement of recognition accuracy with 64-bit CPUs. It is too obvious that someone is using this application to hype 64-bit CPUs, while not understanding the underlying science at all. And I am also dissatisfied by the fact that Eugenia verbally copies this completely unfounded claim.
I think everyone who worked on this field do realise, that the major challenge for speech recognition is not the computing complexity, but the ability to build stochastic models.
Let me break this into to areas: The recognition itself and the training of the model.
In term of recognition, what the CPU have to do is classification by (typically Gaussian mixture) distributions. In fact, the speed of a Pentium II (~ 300 MHz) is just fine to decode most Hidden Markov Models in real time using a reasonable – the so called – A*-search range in Viterbi decoding. The decoding precision is just limited by this A*-search range, but you can easily show experimentally that increasing the A*-search range beyond this (I just called it “reasonable”) value that allow real time decoding on a Pentium II 300 MHz does not improve the speech recognition significantly.
And in fact, the Intel MMX extention already have specialised instructions for Viterbi decoding for years. And I am not aware that AMD64 or Itanium do have better instructions for Viterbi decoding. – But this is not the point. Viterbi decoding is definitively not the limiting factor.
In term of training, one should distinguish between the initial training and adaption. Both algorithms depend on iterative approaches, such as the so called EM (Expectation Maximisation) algorithm. The adaption is not critical, as you only run a few step of the algorithm, and just with the few new voice signals you just acquired.
And for initial traning, this is typically something done by those developers of speech recognition and not what end-users do. And again, people train their model until the iterative algorithm converges. The convergence will be faster on faster CPUs (but this is not such a big deal, as people can use distributed computing for training if the CPU is too slow), but yet the quality of the model is still limited by the model itself.
In short, without a change in the approach and therefore the science we use to attach this problem, the “magical” increase of recognition accuracy is just a dream dreamed by those peoble abusing it for hyping faster CPUs. And beside this, I know a lot of people using DEC Alpha systems for years to develop speech recognisers. How does it come that they do not publish significantly better results in peer-reviewed journals than those 32-bit CPU users publish?
> The only disagreement here is that you think AI, physics
> modeling, video editing, and voice recognition would not
> benefit signifcantly from that, and I do.
OK. I have some background in AI and speech recognition. Would you mind to explain to me how AI and voice recognition benefits from those 64-bit CPUs?
> Honestly, I don’t know how to convince you that the above
> applications would benefit, I thought it was obvious.
Maybe it is very useful to be able to convince people before you make any claims.
And no, even with my AI and speech recognition background and with peer-reviewed publications, it is at least not obvious to me.
> AI, physics modeling, video editing, and voice
> recognition all do extreamly complex operations
> and the better the precision of the numbers, the
> better the result. 64-bit means that programers
> can go to greater levels of precision with less
> of a drop in performance.
OK let be answer to each field I have some background with:
AI:
There are essentially two areas of AI. Those AI fields with non-fuzzy approaches definitively DO NOT benefit from faster IEEE double or quadro precision (does AMD64 have IEEE quadros?). This should be really obvious: You works with graphs and other information that is purely integer. And 32-bit integer is big enough for even the biggest problems. I do not imagine anyone still have a glimps what he is doing if the graph is bigger than 2^31 ~ 10^9.
Those “fuzzy” AI deals with a lot of NP complete optimisations. The biggest challenge is in fact the optimisation itself. The convergence of a high dimensional problem is already as bad as imaginable. Most people working there would be rather thankful for a very fast IEEE single precision.
Speech recognition:
The discussion of the “fuzzy” AI, very close to patter recognition, brings one closer to the problem: What kind of data precision are we actually dealing with?
For speech, most people would never have or even see a audio frequency digitaliser with more than 16 bit precision. Most signal processing algorithms like FFT are completely stable for IEEE float, time-domain filtering can be even realised to good accuracy on integer fixed digit.
So maybe someone ought to explain to me where do you need the precision of a 64-bit floating point, or more importantly, where do you actually have data as precise as 18 decimal digits in speech recognition in the first place?
Video processing:
The same problem. I have never seen sofar CCD with 16-bit precision and beyond in astronomical applications. And as I know the price tag, you honestly do not want to build a consumer video camera with that thing. In reality 12-bit is really the peak level of consumer CCD today, film compositing uses 16-bit to the most…
Again: Where do you get data that need precision as much as 18 decimal digits?
You might imagine that there are fancy effect filters that could hypothetically have some use with IEEE double, but I do not see why such specialised usage would be the killer application, or if at all will be used by more than few hundred people in the world.
and beyond … except … in astronomical applications.
> I am glad the people who do create new processors
> are not like negative posters here who seem like
> to critisize enthusiasts just for laughs or to show
> off there knowledge.
I do not think this is the point. The complains are because someone comes with a claim for certain better-suitability in the first place. Certainly there will be very useful applications of 64-bit CPUs, but
1. it will not be a brand new world. High performance computing is reality and we are talking about performances much higher than any consumer payable AMD64 computing power, even desktop 64-bit computing is reality (you just have to buy an Alpha, UltraSPARC or POWER3 workstation).
2. it will not occur in those fields the author thinks so easily.
Companies like Intel do not develop CPUs for a specific field of use, maybe some extensions like MMX, but not the core architecture itself. In fact, those purely-multimedia and non IEEE complaint MMX and 3DNow! are pretty much outdated. SSE-SSE3 are not specialised vector units, as they work fully IEEE complaint, and you can use them for scientific simulations, etc.
What I do not understand, is why anyone uses specialised applications, without knowing specifically or having experience with the details of algorithms used there, and stating that it is going to benefit from 64-bit architectures.
> otherwise there would be no pc/internet/linux/etc
> at all.
I do not remember those people involved in the very technical development of these things what you mention, did not do it with scientific and technical precision, but rather walk around and hyping their product marketing-like without even know what they are saying.
The problem is, that the author of this article bases his assumption that those applications will benefit, because he THINKS they will benefit, he does not KNOW they will benefit. And in at least one application, I know quite for sure it won’t.
Isn’t this the old reduced instruction set versus the bloated instruction set arguement? If a fast computer can handle less instructions faster or a slow computer can handle many more efficient commands. Now here we get to the question of split words too. Does it actually use a 32 bit command along with 32 bit’s of data. That could be fast.
Good post.
“I specifically said ‘these new…’ reffering to the new AMD64 and PPC970 which ARE superior in many ways, with ”
AMD64 has extra registers. The PPC970 doesn’t benefit similarly when running 64-bit code.
“which you seem to agree. Do you really think this article is about Alphas or UltraSPARCs??? Stay on topic!”
This is about *64-bit* chips! 64-bit chips did not give us these things in the past, either. This IS THE TOPIC.
“Of course, the ability to handle 64-bit longs and pointers and do 64-bit arithmatic more efficiently is not at all beneficial right? Oh, wait, it certainly is! Primarily in… well, you know. See below. ”
It isn’t going to revolutionize anything as this article implies – and for many users, they won’t see a benefit for years.
”
Right sure. I’ll say it again – I know that they have been available for decades and they are not magic. They do provide what I think because I only think they provide what they in fact do provide. The only disagreement here”
No.
“is that you think AI, physics modeling, video editing, and voice recognition would not benefit signifcantly from”
Not, really.
“Honestly, I don’t know how to convince you that the above applications would benefit, I thought it was obvious. I give up.
”
You’re wrong, so there’s not anything you could say to convince me.
listen guys, what’s your problem? why are you so pissed off?
have you read the article? have you noticed that it is “Detroit Free Press”, not a “Microprocessor Report” or smth like that?
HAVE YOU NOTICED THAT? we all (ie comp relates persons) should be very grateful that those kind of people are enthusiastic about 64-bits even though they have no idea what they are talking about.
and you behave like small kids in a movie theater who thinks that the action is for real and cry that it is against law of physics for a bruce lee to jump 10 meters high.
grow up!
The difference is that in movies, you *know* the action isn’t real. This article is trying to pass off rubbish as fact. People expect this article to provide real facts, regardless of whether its Detroit Free Press or an Intel White paper. There’s nothing in the article to indicate that the information isn’t accurate. Its marketing disguised as something technical.
After reading the article, I felt like I must be in Play School, or something. Thank you JULIO OJEDA-ZAPATA, for that fascinating insight into the coming revolution.
I hope you are getting some gifts from the companies you mentioned because I wouldn’t want to be marketing manager and not get paid for it. Umm, I mean journalist.
“we all (ie comp relates persons) should be very grateful that those kind of people are enthusiastic about 64-bits even though they have no idea what they are talking about.”
Sorry, I do not share your point:
1. Your opinion is that a technically false statement should kept uncorrected, if just the intention is – subjectively – just? I do not think so.
2. Neither do I think that it is wise to advertise any product based on false hope. People will – sooner or later – notice that those promises are not fulfilled. I see any indication that this is the case for AI and speech recognition, why should I keep my opinion for myself?
3. Why do not they just state plainly that 64-bit CPU will solve all the problems of our civilisation? May be like: “With AMD64, we will get a cure for all kinds of cancer, close the ozone hole, tabletop cold fusion, … world peace, …”
And no, even that my work would benefit a lot if we have cheap 64-bit technology, I am not grateful for these people. There are two reasons:
1. It is plainly wrong to give other false hope.
2. The AMD 64-bit for masses is a cheap compromise. People really in need of computing power need architectures like Itanium. But, oh yeah, backwards compatibility is for the “masses” so important, that this has become the key feature. Just go again over the words: backwards … key feature.
Oh, and 64-bit technology is not new. If you need it, just have to go to Sun and buy an UltraSPARC (their entry level workstation are not that expensive). Well, if you really need it… But obviously the mass are not storming to Sun and say “Intel/AMD 32-bit CPU is crap, we need UltraSPARC”. People obviously do not need it.
You say:
“Your opinion is that a technically false statement should kept uncorrected, if just the intention is” positive ” I do not think so. ”
Ok – imaging a kid comes up to you, bristling with enthusiasm and tell you hey tomorrow i will get new computer – amd64 – it is 64 bit ! it will allow me to do many new things !!!
Will you tell her: “you are idiot, 64 bits won’t get you anywhere.” ?
you say:
“Neither do I think that it is wise to advertise any product based on false hope. People will – sooner or later – notice that those promises are not fulfilled. ”
well you forgot to mention that it is just your opinion. my opinion is that the promises will be fullfilled, though it _may_ be other promises and/or in unexpected ways
you say:
“Why do not they just state plainly that 64-bit CPU will solve all the problems of our civilisation?”
all people are naive when it goes beyond their areas of expertise, but people are not idiots.
you say:
“The AMD 64-bit for masses is a cheap compromise. People really in need of computing power need architectures like Itanium”
and then you say:
“But obviously the mass are not storming to Sun and say “Intel/AMD 32-bit CPU is crap, we need UltraSPARC”. People obviously do not need it.”
have you noticed that people are not sorming intel to get itanium as well. so they do not need it, per your logic? but you have just said that they need itanium. please deside for yourself, whether they need it or not.
i tell you, if they buy sun, sgi, etc – they need it. Or may be you are the ONE who knows better ?? (sarcasm)
also please read my next message
if i understand you correctly, in your messages you have expressed disdain for laymen who enthusiastically hail 64 bits without real understanding what they are talking about. also you are refusing to be grateful to those people.
please let me remind you, that if it was not for all those _consumers_ who are gladly buying all those new pcs and programs thus creating a huge mass-market, if not for them – the pc would not evolve into consumer device – it would be prohibitivly expesive – and YOU won’t afford to have one (or a couple) for personal use.
you are biting the feeding hand.
dixi.
“if i understand you correctly, in your messages you have expressed disdain for laymen who enthusiastically hail 64 bits without real understanding what they are talking about. also you are refusing to be grateful to those people. ”
Why be grateful for hype that’s not even correct hype?
“please let me remind you, that if it was not for all those _consumers_ who are gladly buying all those new pcs and programs thus creating a huge mass-market, if not for ”
*As* a consumer, I would hope people aren’t being misled.
“them – the pc would not evolve into consumer device – it would be prohibitivly expesive – and YOU won’t afford to have one (or a couple) for personal use.”
Oh, please.
“you are biting the feeding hand.”
Ignorance is not something worthy of praise.
“Why be grateful for hype that’s not even correct hype?”
i am sorry, this is not what i said.
i said we should be grateful for user’s enthusiastic buying IT products, thus funding r&d and driving down costs.
“Oh, please”
terrific argument.
as for last 2 line of your message, i am sorry, but i cannot understand how does your reply relate to my words you quoted
“*As* a consumer, I would hope people aren’t being misled”
i deduct that you think of that article being misleading. ok. that is your opinion. my opinion is that by that article people are not being misled.
“i am sorry, this is not what i said. i said we should be grateful for user’s enthusiastic buying IT products, thus funding r&d and driving down costs. ”
You think we should be grateful to people who are *misled* into buying things they don’t need so YOU, ssme, can buy things cheaper. Pretty selfish of you.
“terrific argument”
It matched your post.
“i deduct that you think of that article being misleading. ok. that is your opinion. my opinion is that by that article people are not being misled.”
The article *is* misleading. This isn’t just my opinion.
[typed on a 64-bit machine that didn’t revolutionize anything the guy says it will just because it’s a 64-bit machine.]
“The article *is* misleading. This isn’t just my opinion.”
The article *is not* misleading. This isn’t just my opinion.
“The article *is not* misleading. This isn’t just my opinion.”
It’s not only your opinion, but it’s incorrect as well.
if i understand you correctly, in your messages you have expressed disdain for laymen who enthusiastically hail 64 bits without real understanding what they are talking about. also you are refusing to be grateful to those people.
———————-
Not for the laymen who can’t be expected to know more than they are meant to know. Computers are out of the realm of expertise for many. Which is why the article in DFP is far more damaging than helpful. It gives false information and raises false hopes. The average layperson will not be able to spot the inaccuracies of the article and will just bite it, hook, line and sinker. Then when they find that AMD64 didn’t revolutionize anything, how do you think they’ll feel? What do you think their response is going to be, when the next *real* revolutionary thing comes along?
please let me remind you, that if it was not for all those _consumers_ who are gladly buying all those new pcs and programs thus creating a huge mass-market, if not for them – the pc would not evolve into consumer device – it would be prohibitivly expesive – and YOU won’t afford to have one (or a couple) for personal use.
————————-
How does this defend the inaccuracies of the article?
you are biting the feeding hand.
——————
Biting the feeding hand? Did you read the article? Do you even understand what the article is saying? How will 64-bits make my ” icons and screensavers might evolve, becoming eerily akin to real-life people and landscapes”?
The ricocheting bullets, converting DVDs faster, better voice recognition(?!?) have nothing to do with 64 bits! But you seem to miss that.
So tell me, were you the one who wrote that article? Because you seem to be defending all the blatant mistakes in the article. Oh, and the “Its just a newspaper, not some 1337 hackers paper so it doesn’t need to be accurate” line of reasoning is just BS. Inaccuracy is inaccuracy. The public needs to know the truth, not some BS that is more suited to an advertisement.
i think i start understanding how you ticks.
“Inaccuracy is inaccuracy”
it is not a good case for pursuing scientific truths. the article has nothing to do with accuracies,etc. it just expresses the general public goodwill toward and positive fascination with comp progress. argueing with it about scientific facts is like argueing with a kid. people who read and write those articles (ie “normal” people) would take your calling it BS as attack on their goodwill. thus my rant about feeding hand.
also with all due respect, it is only your opinion that you can provide in relation to the subject. nothing like hard proof or facts. and let me remind you that opinion is like an ass, everybody has one, but it is not always good to flash it in public.
and i think you selected too easy target for your criticism – really, criticising dfp article abouts 64bits pros and cons! want to criticise – pick up a Microprocessor Report or better write an article like “64-bits are useless” and send it to them.
“it is not a good case for pursuing scientific truths. the article has nothing to do with accuracies, etc.”
Oh, people with less training/knowledge in IT *deserves* less accuracy?
“it just expresses the general public goodwill toward and positive fascination with comp progress. argueing with it about scientific facts is like argueing with a kid.”
Fine, as I said, why don’t you just start writing an article (best in the New York Times) that AMD64 is going to solve every thinkable problem of your civilisation?
No, this article expresses the authors possible good will, not one of the public. In fact, it is the public opinion this author (and you) are trying to manipulate.
And sorry, I do not even want to imagine what you tell your kids of science. Obviously the truth is not so important there…
“also with all due respect, it is only your opinion that you can provide in relation to the subject. nothing like hard proof or facts.”
I worked in research of speech recognition, and specifically in measurement of origin of the recognition errors. Beside a quite wide consent in the scientific community about what leads to recognition error that I know as matter of fact, I know the problem this field is facing by real-world measurement. And it is definitively not AMD64 that will solve the problem.
I think it is quite arrogant of you to say that there is no hard proof or facts, or do you have research results that demonstrate the superiority of “64-bit recognisers”?
“and let me remind you that opinion is like an ass, everybody has one, but it is not always good to flash it in public.”
Sorry, if I say this in a direct way. I would be grateful if you are not working in science or engineering. Your attitude is exactly those people, wasting tax money, by misleading funding agencies and the public.
Your problem is simply that you cannot shift your point of view. You argue that this misleading is OK, because the outcome suits you. But hey, other newspaper can mislead you in other ways, maybe not in your best interest. By your logic, it is like to say, lying in a court is OK, if it just serves the purpose. And I do not think that this is the consensus in this society.
Well, let me just say this: I acknowledge that there are such people around in a society. But do not expect my compassion for them.
*sigh*
OK, let’s look at the claims one by one.
1) More realistic icons & screensavers
What’s this got to do with 64 bits? Honestly, tell me. Unless you have an icon that more than 2GB in size when loaded in memory, or a screensaver that guzzles more than 2GB of memory, 64 bits gives you nothing.
2) Richocheting bullets and other physical modelling becoming more accurate
I would hazard a guess that anyone who wants to accurately model a physical system would use floating points. Most of the time anyway. 64 bits does not bring any benefits, since it only affects integers. Programs like MATLAB deal with 64 bit *floats* which are supported on current hardware.
Faster DVD and Video encoding/decoding
Do a search on the internet for benchmarks. You’ll see that such tasks benefit hugely from vector processing, along the lines of Altivec and SSE. The move to 64 bits doesn’t bring much benefit. It may benefit those professionals who need loads or RAM.
and i think you selected too easy target for your criticism – really, criticising dfp article abouts 64bits pros and cons! want to criticise – pick up a Microprocessor Report or better write an article like “64-bits are useless” and send it to them.
Microprocessor Report is generally accurate, and they target people who generally know what it is they are reading. DFP on the other hand, targets the layperson. The audience of MR can smell a lie from a mile away. The average layperson hasn’t got a clue about processors (and they shouldn’t have to!!). The DFP article has a responsibility to the public to be honest and accurate with their information. Its a newspaper!! Its where the public go to get informed about current events and (perhaps) technologies.
The article in DFP is misleading at best. Come on, if a newspaper is publishing false information in a field you’re familiar with, would you trust articles from that newspaper regarding fiels you’re *not* familiar with? How would you know you aren’t being misled?
“I know the problem this field is facing by real-world measurement. And it is definitively not AMD64 that will solve the problem.”
again just an opinion. and sorry, your experience can only add weight to it, but cannot change opinion into proof or fact.
” I would be grateful if you are not working in science or engineering.”
sorry to disappoint you
“You argue that this misleading is OK, because the outcome suits you”
this is the problem of our discussion. i said nothing of that. but you just see what you want to see.
that dsp article is a small artifact from big outside world that found its way into your small IDE-al world of programming. and it looks like you failed to recognise it as a foreign object, smashing its naive far-fetched predictions with all your knowledgeable might.
waiting for applause ? not from me.
that poor Julio, if confronted with your kind of criticism, would be just hurt and would be convinced one more time that “those IT people are total jerks”.
it is attitude like yours that alienate people towards us, and every time i work with users to give them the programs they want i see those scars.
i can only hope that programming you do will compensate for alienation you produce, but somehow i doubt it
“” I would be grateful if you are not working in science or engineering.””
“sorry to disappoint you”
I bet. Sure you are.
“it is attitude like yours that alienate people towards us, and every time i work with users to give them the programs they want i see those scars. ”
What’s this “us” now? Regardless, you’d rather LIE TO THEM. That’s exactly what this article is – a lie.
i am just trying to make you understand that the article is a kind of “great-future-of-technology” one, you can find one in nearly every popular paper or magazine – hailing progress in biotech, aerospace, etc.
argueing with such article about technical details is … plain ridiculous. you are only making fool of yourself, like a person who argues that a bruce lee cannot jump 10m.
“The DFP article has a responsibility to the public to be honest and accurate with their information.”
they do – to the best of their knowledge.
and that one is not false or misleading, because (even other things being even) new 64-bit processors _do_ give more power into hands of programmers, which would translate into progress in software.
what’s misleading here ?
i would add – if programmers are good enough and we are lucky (especially when it comes to AI, those guys definitely hit the wall 20 years ago).
and , yes!, we may see “More realistic icons & screensavers” and “Richocheting bullets”. and even more.
as for 64-bit floats – P4 expands floating point regs into 128. so it is already better than you think.
“That’s exactly what this article is – a lie”
poor JCS, you think the life/world works mechanically like your favorite compiler and is described by simple formulaes.
well, i see you saying: (a little out of context) “there’s not anything you could say to convince me”
so i pull the plug. bye.
“poor JCS, you think the life/world works mechanically like your favorite compiler and is described by simple formulaes.”
What on Earth are you talking about?
“well, i see you saying: (a little out of context) “there’s not anything you could say to convince me””
You can’t convince me, since you are WRONG.
“so i pull the plug. bye.”
Good riddance.
You don’t get it, do you?
A lie is still a lie. You’re completely convinced that the public should be fed lies? As long as those lies are sugar coated? You are making a fool of yourself, not anyone else who points out the flaws in the articles.
and , yes!, we may see “More realistic icons & screensavers” and “Richocheting bullets”. and even more.
And how exactly will that happen? Like I said before, a 2GB icon isn’t something desirable. Neither is a screensaver that requires more than 2GB of memory to run. You haven’t provided a shred of evidence to back up that ridiculous claim.
as for 64-bit floats – P4 expands floating point regs into 128. so it is already better than you think.
No, its not. The FP registers are *still* 80 bits wide. They’ve been like that for a long time. Those 128 bit registers you talk about are the XMM registers, used by SSE/SSE2 instructions.
P4 registers:
yes, of course. and there are FP operations with SSE regs. are they unusable for “common” fp calcs? – i never did mmx/sse myself…
ok, back to business:
you say: “A lie is still a lie.”
“lie”-is a VERY strong word. actually it is an accusation. lie would be if the author _knows_ that,say, 64-bit won’t get voice recognition, and still says so in the article. Have you got proof (or at least evidence) of the author’s lying ?
please never ever accuse anyone of lying until you have a solid proof.(and even then…one must be very careful). this forum is harmless useless ranting, but in real life you may (will!) find yourself in deep-deep trouble.
why do i insist that the article is _not_ misleading ?
A (high school) student research paper and a submission to “Nature” are being judged according to different standards. The same is true for mass-media article vs article in a computer magazine.
Look at the article:
the author could have say that you just need to buy new computer to be entered into 64-bit world. (that would be misleading).No, he explained the need of software.
Good!
the author could have limited it to recommendation to buy new Windows64. (that would be misleading).No, he say you would need os _and_ applications.
Very good!
the author could have said that 64-bit processors can run programs that 32-bit processors cannot. (that would be very misleading, though technically it is a correct statement). He does not say so.
Excellent!
the author knows that it is not his area of expertise, and he starts by saying that he bases his article on words of “industry cheerleaders” – he & his readers know about marketing.
Kudos!
The main line of the article is that 64-bit processors are more powerful than same 32-bit ones. (very true!). more powerful processors can result in fundamentally (possibly!) better (true!) software.
i ask you again: what’s misleading?
comparing to other mass-media publications – this article is very, very good. probably that’s why it found way into osnews.
you say:”Then when they find that AMD64 didn’t revolutionize anything, how do you think they’ll feel? What do you think their response is going to be, when the next *real* revolutionary thing comes along? ”
and what would it be? look, the article mentions voice recognition. i remember it being promised from…er, beginning of 90s. So, do “they” pound on table requesting vr? nope. Was “their” belief in comp technology shaken? not a bit. Any other big unfulfilled promise? AI. for 30 years. Any PR problem? none.
“they” just do not behave the way you think “they” do. and “they” are much smarter than you think. even though they have no idea about 64bits.
the article promises more realistic icons and screensavers. Will this promise be fullfilled ? You bet! Because of 64 bits ? doubtedly. Because of more powerfull processors? very much.
Maybe a suggestion: Just forget ssme. I think we must accept that people like him exist in our society. And otherwise those ethics commission of IEEE, NSF, … would be quite boring.
@ssme: And no, ssme. You obviously don’t even care to read some articles in the three leading journals of speech recognition (“IEEE Trans. Speech and Audio Processing”, “Speech Comm.”, “Computer Speech & Language”), otherwise, you would not give such a infatile argumentation here. It is not just my “one” opinion as you call. Also people like Hermansky have a similar opinion (but you probably not even know who he is).
Come on, if you have the boldness, say AMD64 will revolutionise speech recognition on comp.speech.research, or better, write a paper and submit it to ICASSP (the deadline for 2004 is over, however).
you are not understanding _what_ is being discussed here.
hint: it is _not_ speech recognition.
1. Oh, yes, I do understand what is discussed here: Your sense of “truth”.
2. You started labeling the arguments brought by me, WJG, JCS as “personal opinion, not a matter of fact or provable” in the first place. Let me quote:
“also with all due respect, it is only your opinion that you can provide in relation to the subject. nothing like hard proof or facts.”
Your sudden change of topic is pathetic.
You might be right if you call this like arguing with a kid, though in a different respect: Right now I feel like arguing with with a kid that keeps claiming the Earth is flat, and even after realising that this might not be true, starts to claim that it does not matter.
And I do not wish to waste the OSNews.com resources on this anymore.