xeenrecoil
Distinguished
- Jan 7, 2003
- 842
- 0
- 18,990
This is really good news, this also means the home user 64-bit version of the microsoft OS will be out, which means the world can now shift to 64-bit computing at an affordable, consumer ease of use type level, as most people use windows and arent familiar with the Linux OS, also the prohibative cost of previous 64-bit solutions coupled with that unfamilliarity of Other Operating systems have stopped people in their tracks in the past.
Ahhhh progress you gotta love it, world meet 64-bit...64-bit meet world.
good stuff...good stuff
ps. apple doesnt count, cuz like who uses an apple? /me chuckles
XeeN
Ahhhh progress you gotta love it, world meet 64-bit...64-bit meet world.
good stuff...good stuff
ps. apple doesnt count, cuz like who uses an apple? /me chuckles
XeeN
Mephistopheles
Distinguished
- Feb 10, 2003
- 2,444
- 0
- 19,780
Yes, well, that was a bit exaggerated. Most of those who know what going from 32-bit to 64-bit means also know about existing 64-bit machines...
64-bit is a hype. Yesterday a friend of mine actually believed that it was going to double performance over 32-bit...
All it does is enable memory accesses over 4 GB with linear addresses. But first of all, 4 GB is a lot of memory no matter how you look at it and is sufficient for almost anything. Secondly, every process can have its own 4 GB of addressing space, so you don't have to share it with other programs. Thirdly, you <i>can</i> address more than 4 GB on a 32-bit processor, by changing segments. Since 4 GB is a lot of memory this shouldn't occur often. All we need is operating system support.
So, 64-bit is actually slower. It takes longer to calculate a 64-bit address than a 32-bit address. It also requires an extra prefix byte which takes extra memory throughput. 64-bit calculations can already be done with MMX so that's not giving an advantage either.
What <i>will</i> speed things up is the integrated memory controller, and the extended register set. But it won't double performance. Intel on the other hand will have Prescott ready soon, which thanks to Enhanced Hyper-Threading, could double performance in the best case.
No, I'm not an Intel fanboy, but I've had enough of these AMD fanboys thinking Athlon64 will be revolutionary...
All it does is enable memory accesses over 4 GB with linear addresses. But first of all, 4 GB is a lot of memory no matter how you look at it and is sufficient for almost anything. Secondly, every process can have its own 4 GB of addressing space, so you don't have to share it with other programs. Thirdly, you <i>can</i> address more than 4 GB on a 32-bit processor, by changing segments. Since 4 GB is a lot of memory this shouldn't occur often. All we need is operating system support.
So, 64-bit is actually slower. It takes longer to calculate a 64-bit address than a 32-bit address. It also requires an extra prefix byte which takes extra memory throughput. 64-bit calculations can already be done with MMX so that's not giving an advantage either.
What <i>will</i> speed things up is the integrated memory controller, and the extended register set. But it won't double performance. Intel on the other hand will have Prescott ready soon, which thanks to Enhanced Hyper-Threading, could double performance in the best case.
No, I'm not an Intel fanboy, but I've had enough of these AMD fanboys thinking Athlon64 will be revolutionary...
xeenrecoil
Distinguished
- Jan 7, 2003
- 842
- 0
- 18,990
heya Mephistopheles;
hehe i simply meant that it is a good thing that the 64-bit solution is comming to the home user, from this point forth software companies will start moving to the 64-bit platform slowely but surely, as more people buy 64-bit machines, and 64-bit machines will be in the mainstream, this is a good thing it really is, progress like this is always good, it keeps us moving forward.
XeeN
hehe i simply meant that it is a good thing that the 64-bit solution is comming to the home user, from this point forth software companies will start moving to the 64-bit platform slowely but surely, as more people buy 64-bit machines, and 64-bit machines will be in the mainstream, this is a good thing it really is, progress like this is always good, it keeps us moving forward.
XeeN
xeenrecoil
Distinguished
- Jan 7, 2003
- 842
- 0
- 18,990
heya c0d1f1ed;
I would like to clear something up, i have never owned an AMD machine in my life so i am hardly an AMD fanboy, that being said, taking my post in context as it should be, i merely stated that we are moving forward in progress and IT IS GOOD, progress of this type is always good.
XeeN
I would like to clear something up, i have never owned an AMD machine in my life so i am hardly an AMD fanboy, that being said, taking my post in context as it should be, i merely stated that we are moving forward in progress and IT IS GOOD, progress of this type is always good.
XeeN
If I knew that x86-64 was the future, I'd agree almost whole-heartedly that the shift is welcomed. However, that's hardly the case. With Intel's adamant support of IA-64 and willingness to drop x86 (them being a giant whos influence on the market is very great and that's an understatement), having 2 64-bit ISA's doesn't improve mainstream product compatibility and availability, it decreases it.
If and when Intel moves IA-64 to mainstream desktop (and this is most probable, knowing how bullheaded Intel is), and x86-64 has entrenched itself into the market. We will have a platform compatibility issue on our hands. And it won't be something as minor as SSE or non-SSE application either. We're talking an issue on par with the Mac or PC issue. Could you imagine the current mainstream PC market splitting into two?
There are only 3 possible outcomes, none of them favorable:
1. AMD looses out completely and x86-64 dies. Considering how much money AMD has poured into this thing, the company will probably go under. Intel regains total market control and we'll see progress come to a halt.
2. IA-64 never works. Knowing how bullheaded Intel is, they will continue to push it. They will continue and continue until either they go under to the entire software/windows platform is left a mess. Even if nobody writes for IA-64, they will be hessitant to write for x86-64 as that means you're cutting off the Intel market (which, even if we see a steady decline in, will still remain a good size for at least 10 years I think). Adoption will not only be slow, it'll be horrific in between and we'll run into the 4 GB memory limitation but have no clear cut way to circumvent it.
3. Both ISA's are adopted. Software vendors now provide two versions of each software title. An IA-64 version and an x86-64 version. We'll see two probably entirely different markets similar to Mac vs PC. IA-64 software won't work on x86-64 platforms and vice versa, you'll have to buy separate software titles for either.
None of these possibilities sound pleasing at all and frankly I'm pretty mad at both companies for not comming to an agreement on this matter.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
If and when Intel moves IA-64 to mainstream desktop (and this is most probable, knowing how bullheaded Intel is), and x86-64 has entrenched itself into the market. We will have a platform compatibility issue on our hands. And it won't be something as minor as SSE or non-SSE application either. We're talking an issue on par with the Mac or PC issue. Could you imagine the current mainstream PC market splitting into two?
There are only 3 possible outcomes, none of them favorable:
1. AMD looses out completely and x86-64 dies. Considering how much money AMD has poured into this thing, the company will probably go under. Intel regains total market control and we'll see progress come to a halt.
2. IA-64 never works. Knowing how bullheaded Intel is, they will continue to push it. They will continue and continue until either they go under to the entire software/windows platform is left a mess. Even if nobody writes for IA-64, they will be hessitant to write for x86-64 as that means you're cutting off the Intel market (which, even if we see a steady decline in, will still remain a good size for at least 10 years I think). Adoption will not only be slow, it'll be horrific in between and we'll run into the 4 GB memory limitation but have no clear cut way to circumvent it.
3. Both ISA's are adopted. Software vendors now provide two versions of each software title. An IA-64 version and an x86-64 version. We'll see two probably entirely different markets similar to Mac vs PC. IA-64 software won't work on x86-64 platforms and vice versa, you'll have to buy separate software titles for either.
None of these possibilities sound pleasing at all and frankly I'm pretty mad at both companies for not comming to an agreement on this matter.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
eden
Champion
- Oct 3, 2001
- 26,541
- 0
- 50,780
They can only if you have the need or are using 64-bit integers, basically anything above 32-bit size.
--
<A HREF="http://www.lochel.com/THGC/html/news.html" target="_new"><font color=red><b>Join the Tom's Hardware Guide Community Photo album, send us your pics!</font color=red></b></A>
--
<A HREF="http://www.lochel.com/THGC/html/news.html" target="_new"><font color=red><b>Join the Tom's Hardware Guide Community Photo album, send us your pics!</font color=red></b></A>
The pure definition of going "64-bit" has nothing to do with increasing the average IPC. Now, x86-64 does have a few improvements *other* than just adding 64-bit capabilities. Among those are the integrated memory controller, a micro-ops packing function to help execution efficiency, double the GPR's and SSE registers, etc. The latter requiring optimization in order to take effect.
Of course, all of these things could be done on a 32-bit processor as well, or other types to speed up applications. There's really no "need" for the processor to be 64-bit capable.
Even if you're in the situation in which you need to do 64-bit integer math, MMX, SSE and SSE2 offers instructions for doing this.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
Spitfire_x86
Splendid
- Jun 26, 2002
- 7,220
- 0
- 25,780
We've already seen so called 70% performance boost from HT (Northwood version)
----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new"> My Website</A></b>
<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new"> My Rig</A></b>
I like this quote
The chip also comes as AMD continues to suffer financial losses and lose market share. From the end of 1986 through the first quarter of this year, AMD's cumulative net earnings, including sell-offs and acquisition expenses, come to about $221 million, about the same amount of money Intel makes every three weeks. AMD is expected to lose money in the second quarter as well.
The rest it just bla bla AMD have allready slowing production of opteron as there were having trouble sold them
I dont like french test
The chip also comes as AMD continues to suffer financial losses and lose market share. From the end of 1986 through the first quarter of this year, AMD's cumulative net earnings, including sell-offs and acquisition expenses, come to about $221 million, about the same amount of money Intel makes every three weeks. AMD is expected to lose money in the second quarter as well.
The rest it just bla bla AMD have allready slowing production of opteron as there were having trouble sold them
I dont like french test
not that anyone really cares...but the a64 would be very good for long mathimatical/scientific calculations that in 32bit mode require double percision....the a64 would not need to run double percision and therefore would be much faster than its 32bit competition but again this is ONLY in 64bit mode.
For scientific calculations you need floating-point numbers. Athlon64 will feature SSE2 which is going to help in that area. But the Pentium 4 had this since the beginning, and the 64-bit part of A64 isn't going to help here.
That's what I meant with hype. The Athlon is just catching up with the Pentium and they call it revolutionary...
That's what I meant with hype. The Athlon is just catching up with the Pentium and they call it revolutionary...
That's what they're planning and have been all along. The question isn't whether they'll move IA-64 to desktop, it's when they'll do it. Certainly they're not in a position to do it now, but in 2006-2008, I think they'll be in a position to sell IA-64 processors for the price of desktop processors (hell, Deerfield, a 1 GHz Low-voltage Itanium 2 with 1.5MB of L3 cache is suppose to be $700 and released within weeks).
As for running backwards compatible x86 code. That I have no idea how they're going to do it. I'd say that that marvelous Alpha developement team they have will probably be working on something similar to Alpha's FX32! software emulation of x86 code. Hey, if Transmeta can do it, why couldn't Intel.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
Mephistopheles
Distinguished
- Feb 10, 2003
- 2,444
- 0
- 19,780
That is not true. Mathematical and Scientific calculations require almost exclusively high-precision floating point operands. I should know, because I work with those things - and mostly, noone really cares about 64-bit integers, only about the extra adressing. Also, all Pentium 4s work default with 64-bit precision floats, and max out at 80-bit FP precision. Plus, Pentium 4s chew through floats incredibly quickly if you compile your software with the latest HT-enabled, SSE2-enabled Intel compilers.
64-bit precision in itself is not a tangible advantage of x86-64 at all.
Mephistopheles
Distinguished
- Feb 10, 2003
- 2,444
- 0
- 19,780
Itanium as a desktop part?... Oooh, I hope I see that in my lifetime. It looks like a great processor.
Also, I really want to see Deerfield reviews around the web, or I'll get really irritated. If they review $700 Opterons at THG, why not review a $700 Deerfield, which is actually, at least in paper, considerably faster in FP?
And about 32-bit emulation, well, I think that that was a clear departure from current standarts, unlike Opteron, and I appreciate that very much. I don't know that much about this architecture, and that's why I was interested in Deerfield reviews or something... 'cause it seems as if noone is interested in IA64!
But really, the problem that intel faces is competition from IA64 to IA32. What if they made a chipset available that supported AGP and SATA and conventional dual channel DDR400 and Deerfield?... And what if people actually find out that that is a good alternative? It would be great...
(can you imagine us running full-blown Itaniums as desktop CPUs and playing games with its FP capabilities? 😎 )
Also, I really want to see Deerfield reviews around the web, or I'll get really irritated. If they review $700 Opterons at THG, why not review a $700 Deerfield, which is actually, at least in paper, considerably faster in FP?
And about 32-bit emulation, well, I think that that was a clear departure from current standarts, unlike Opteron, and I appreciate that very much. I don't know that much about this architecture, and that's why I was interested in Deerfield reviews or something... 'cause it seems as if noone is interested in IA64!
But really, the problem that intel faces is competition from IA64 to IA32. What if they made a chipset available that supported AGP and SATA and conventional dual channel DDR400 and Deerfield?... And what if people actually find out that that is a good alternative? It would be great...
(can you imagine us running full-blown Itaniums as desktop CPUs and playing games with its FP capabilities? 😎 )
I don't think IA-64 is a miracle ISA. In fact, all evidence seems to indicate that it's not. It's better than x86, but then again, that's like saying something is better than a pile of smelly dog s@#@. I wouldn't jump up and down over running a desktop IA-64 processor. When used correctly (i.e. software is properly optimized either by using good compilers or good hand-code), it can be incredibly fast. When not used correctly, it can be amazingly slow. We'll just have to see how Intel does on the software side.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
"We are Microsoft, resistance is futile." - Bill Gates, 2015.
Kelledin
Distinguished
- Mar 1, 2001
- 2,183
- 0
- 19,780
AFAIK SSE2 isn't able to manage 80-bit FPs. Even if it is, it loses a lot of the SIMD advantages, because it can't pack more than one in a single XMM register. If it can't store 80-bit reals in XMM registers, then it's in a tight spot on 80-bit FP numbers--it has to resort to the old x87 FPU that got castrated on the P4.
I seem to recall this being a major hangup in porting 3DSMax to SSE2. 3DSMax previously liked to use a lot of 80-bit reals, so it needed a significant rewrite before it was ready to benefit from SSE2.
<i>I can love my fellow man...but I'm damned if I'll love yours.</i>
eden
Champion
- Oct 3, 2001
- 26,541
- 0
- 50,780
Would it explain a bit why the SSE2 optimizations on 3dSMAX have shown rather lax performance(increase)?
--
<A HREF="http://www.lochel.com/THGC/html/news.html" target="_new"><font color=red><b>Join the Tom's Hardware Guide Community Photo album, send us your pics!</font color=red></b></A>
--
<A HREF="http://www.lochel.com/THGC/html/news.html" target="_new"><font color=red><b>Join the Tom's Hardware Guide Community Photo album, send us your pics!</font color=red></b></A>
What happens if AMD was to die by say the end of 2004...Do you guys think Intel will be loyal to they're customer's by going into the price/performnce ratio as they're going (copying AMD) now or would they go into sleep mode?
Maybe Tejas may never see daylight especially when one must cosider that a monopoly like intel at that time will suck as much money as possible i can guarantee you that. No doubt however i will have to prove afterall it's chipzilla's character.
I have been reading threads here for a while @ THG. A lot of strong points have been made about prescott, but saying it will out perform A64 because of plain paper spec's is what i don't understand. In fact some of you went out to get yourselves one of those p4's with springdale/canterwood mobo's. I can't say it's stupidity but rather rare.
No one has actually seen benchmark's with A64 running @ 64bit mode instead some of you went out and bought a system just speculating "it will outperform A64" which is very sad, beacuse not only can the prescott not run x86-64 bit WIN XP but also if the prescott fails to surpass A64 which most are speculating it not to be then that means that buying your new systems will be invalid rather in others crippled to run anything very fast but only @ 32bit mode....
Maybe Tejas may never see daylight especially when one must cosider that a monopoly like intel at that time will suck as much money as possible i can guarantee you that. No doubt however i will have to prove afterall it's chipzilla's character.
I have been reading threads here for a while @ THG. A lot of strong points have been made about prescott, but saying it will out perform A64 because of plain paper spec's is what i don't understand. In fact some of you went out to get yourselves one of those p4's with springdale/canterwood mobo's. I can't say it's stupidity but rather rare.
No one has actually seen benchmark's with A64 running @ 64bit mode instead some of you went out and bought a system just speculating "it will outperform A64" which is very sad, beacuse not only can the prescott not run x86-64 bit WIN XP but also if the prescott fails to surpass A64 which most are speculating it not to be then that means that buying your new systems will be invalid rather in others crippled to run anything very fast but only @ 32bit mode....
Doesn't anyone remember that we had 32BIT CPUs when our software was only 16BIT. We need 64BIT to move foward in the industry. Not just for performance but for progress.
And as for AMD going belly up that would be a sad day for even Intel users. If AMD wasn't around to offer price and performance competition then Intel would most likely slack off and riase prices.
And as for AMD going belly up that would be a sad day for even Intel users. If AMD wasn't around to offer price and performance competition then Intel would most likely slack off and riase prices.
TRENDING THREADS
-
-
Question Effect of USB 3 cable length at high speed- Started by Pimpom
- Replies: 5
-
-
Discussion What's your favourite video game you've been playing?- Started by amdfangirl
- Replies: 4K
-
Facebook
Twitter
Instagram