<< I want Dual 1.4GHz Tbirds!!!!
SMP ALL THE WAY! >>
way to $$. intel smp is expensive as it is..
SMP for 98% of the home users out there is OVERKILL.
Come on now people, admit it. How many people do you know that use computers are barely literate with the machines, let alone anything with the internet related. Most of the computer users I know know little to nothing about their hardware and know even less about how the internet work. All they know is: turn machine on, click on dialup shortcut, click on email shortcut, click on browswer shortcut.
For these people, SMP really means nothing. Unfortunately, that's over 90%+ of the users who use PCs for their "day to day" use. Nothing wrong with it, I'm not bashing them, but I want to point out that the simplest tasks don't need SMP.
HOWEVER, for the <1% of the users out in world, WE are ones who know more is better when it comes to SMP. I have an SMP rig at home, but its my server and not my gaming box (2x P3-550s). I cannot wait until AMD gets full gear into SMP so that Intel can fight against them for the workstation/server market. More choices for us is a good thing because no single company can screw us on the prices for the hardware.
I believe that SMT+SMP is the real key to getting much more speed for less money. SMT is essentially two processors on the same die (from Intel anyways). Putting two processors on the same die is just like having SMP, but without the added space. So, put two SMT processors on one board and four processors for the price of two! Woohoo! Unfortunately, its an Intel only thing and its not "out" yet (sampling). SMT has more of a chance of being commonplace (ihmo) than SMP because of cost.
Either way, I want a dual AMD rig. But I'd want a SMP+SMT AMD rig. That'll be some bragging rights!
vash
Come on now people, admit it. How many people do you know that use computers are barely literate with the machines, let alone anything with the internet related. Most of the computer users I know know little to nothing about their hardware and know even less about how the internet work. All they know is: turn machine on, click on dialup shortcut, click on email shortcut, click on browswer shortcut.
For these people, SMP really means nothing. Unfortunately, that's over 90%+ of the users who use PCs for their "day to day" use. Nothing wrong with it, I'm not bashing them, but I want to point out that the simplest tasks don't need SMP.
HOWEVER, for the <1% of the users out in world, WE are ones who know more is better when it comes to SMP. I have an SMP rig at home, but its my server and not my gaming box (2x P3-550s). I cannot wait until AMD gets full gear into SMP so that Intel can fight against them for the workstation/server market. More choices for us is a good thing because no single company can screw us on the prices for the hardware.
I believe that SMT+SMP is the real key to getting much more speed for less money. SMT is essentially two processors on the same die (from Intel anyways). Putting two processors on the same die is just like having SMP, but without the added space. So, put two SMT processors on one board and four processors for the price of two! Woohoo! Unfortunately, its an Intel only thing and its not "out" yet (sampling). SMT has more of a chance of being commonplace (ihmo) than SMP because of cost.
Either way, I want a dual AMD rig. But I'd want a SMP+SMT AMD rig. That'll be some bragging rights!
vash
I agree with what that guy said.
I have an SMP rig, as well as two non SMP ones at work.
The SMP rig is a dual P2-350 overclocked to 392, one CuMine/666, and one celery 700(bought by mistake, so i grabbed it).
All are equipped with 256 MB memory.
I like the CuMine the most by far, the SMP rig the second, and the Celery pretty much sucks.
Of course the primary rig has the nicest equipment, but still, the dually P2's just are no match for the P3.
And this is doing excessive amounts of multitasking, only real work.
Of course I do a bit of gaming after office hours as well, but I havent counted that in
I have an SMP rig, as well as two non SMP ones at work.
The SMP rig is a dual P2-350 overclocked to 392, one CuMine/666, and one celery 700(bought by mistake, so i grabbed it
).All are equipped with 256 MB memory.
I like the CuMine the most by far, the SMP rig the second, and the Celery pretty much sucks.
Of course the primary rig has the nicest equipment, but still, the dually P2's just are no match for the P3.
And this is doing excessive amounts of multitasking, only real work.
Of course I do a bit of gaming after office hours as well, but I havent counted that in
Am I the only one who thinks this is the future of the industry, or should be at least looked at...
As it becomes harder and harder to keep ramping these things up without issues of heat and size, why not use several smaller chips working in a parallel process to get the work done faster and generate less heat doing it...
In the long run I think more programs could take advantage of this by becoming coded to be multithreaded...hey and at the same time they can compile all the older programs to sse2 so P4 will actually do what it is supposed to do....first part I am serious last part is joke for fkloster...
As it becomes harder and harder to keep ramping these things up without issues of heat and size, why not use several smaller chips working in a parallel process to get the work done faster and generate less heat doing it...
In the long run I think more programs could take advantage of this by becoming coded to be multithreaded...hey and at the same time they can compile all the older programs to sse2 so P4 will actually do what it is supposed to do....first part I am serious last part is joke for fkloster...
I totally disagree with that. I have been running dual 500 celeries on a bp6 for about 1 year and 7 months. All i can say it is one of the best and most stable computers ive ever used. Even when i go over to a friends house to use there single proc tbird 1 gigahertz computers, i cannot feel the speed difference in most of the everyday apps. When the 760MP board comes out, i will upgrade. Multi Procs rule
rigmah
rigmah
<< You can actually run two apps at the exact same time. >>
If I recall correctly, my Windows 3.1 did that too...
SMP will be the way of the future. Here is why.
It only makes sense to increase the native word size so many times. Since the industrial revolution, the numbers we work with really haven't got that much bigger. We went from millions (10^6) to trillions (10^12). 32 bit integers give us billions (10^9) and 64 bit integers give us numbers as large as 10^19. It will take a huge amount of time until we commonly require fixed-point number with more precision. So going from 4 bits to 8 to 16 then to 32, now to 64 and one day 128 bits makes sense. But we get diminishing returns with each change as so many of the numbers we deal with will not require all these bits. Home CPUs of 64 bits is not too many years away, and 128 bit home computers I expect to see in my lifetime. 256 bits, I don't know. Sure for graphics cards, 256 bit registers for mmx/sse, but not for general-purpose registers.
Smaller and faster transistors. Every year I hear people saying that we have reached the end of the road (or rather it is in sight) to how small and fast a transistor can be made, and every year they find a way to get around it. Don't ask me when the road ends, but it will end.
Unless you are using a computer made back in the 80's your CPU is already implementing parallelism. Remember those pipelines to allow the CPU to perform multiple steps at once? There is AMP at work. Next step SMP.
Eventually manufacturing techniques will be able to place far more transistors on a chip than they can take full advantage of with a single processor. That is when they'll start putting two CPUs onto a single chip. And one day 4 cpus and so on.
It only makes sense to increase the native word size so many times. Since the industrial revolution, the numbers we work with really haven't got that much bigger. We went from millions (10^6) to trillions (10^12). 32 bit integers give us billions (10^9) and 64 bit integers give us numbers as large as 10^19. It will take a huge amount of time until we commonly require fixed-point number with more precision. So going from 4 bits to 8 to 16 then to 32, now to 64 and one day 128 bits makes sense. But we get diminishing returns with each change as so many of the numbers we deal with will not require all these bits. Home CPUs of 64 bits is not too many years away, and 128 bit home computers I expect to see in my lifetime. 256 bits, I don't know. Sure for graphics cards, 256 bit registers for mmx/sse, but not for general-purpose registers.
Smaller and faster transistors. Every year I hear people saying that we have reached the end of the road (or rather it is in sight) to how small and fast a transistor can be made, and every year they find a way to get around it. Don't ask me when the road ends, but it will end.
Unless you are using a computer made back in the 80's your CPU is already implementing parallelism. Remember those pipelines to allow the CPU to perform multiple steps at once? There is AMP at work. Next step SMP.
Eventually manufacturing techniques will be able to place far more transistors on a chip than they can take full advantage of with a single processor. That is when they'll start putting two CPUs onto a single chip. And one day 4 cpus and so on.
SMP is a nerd thing. So is overclocking people spend an absurd amount of money on expensive cooling so that they can overclock something from 1.2 to 1.4 ghz.
Im a nerd
ALL YOUR CHAMPIONSHIPS ARE BELONG TO SHAQ
Im a nerd
ALL YOUR CHAMPIONSHIPS ARE BELONG TO SHAQ
- Oct 14, 1999
- 11,999
- 307
- 126
Sure would be cool to see the DDR-II packaging in the .13 P!!! or next-gen Athlon. Could you imagine the actual CPU packaging the mere width of a quarter or dollar coin? Multiple processors will be commonplace on HDT-compatible motherboards; the smaller packaging would make it easier to fit them on smaller (Micro-ATX) boards.
I can only imagine that one day we will see heatsinks as wide as the motherboard, with cache and cpu making contact to it UNDERNEATH the motherboard. Quick-change clips are on almost every case I've used in the last year, why not on future motherboards too? Heaven forbid our future case manufacturers make a backside-guide to slide into with an easy to clip mount on the front of the board.
I can only imagine that one day we will see heatsinks as wide as the motherboard, with cache and cpu making contact to it UNDERNEATH the motherboard. Quick-change clips are on almost every case I've used in the last year, why not on future motherboards too? Heaven forbid our future case manufacturers make a backside-guide to slide into with an easy to clip mount on the front of the board.
- Oct 14, 1999
- 11,999
- 307
- 126
Check out this case-mount idea: Motherboard Idea
Remember that the CPU and L3 cache would face the Heatsink underneath the Motherboard.
Remember that the CPU and L3 cache would face the Heatsink underneath the Motherboard.
Need to go rackmount soon
yup you NEED to. gimme a break. what do you do thats so fvckin important?
yup you NEED to. gimme a break. what do you do thats so fvckin important?
dude, its not what you do anymore on AT, its what you HAVE!
no one needs anything more than 700mhz [there are exceptions]. im going smp, i dont need it..
no one needs anything more than 700mhz [there are exceptions]. im going smp, i dont need it..
<< But The industry or community itself has too many ppl jumping on smp like a fad and those ppl don't even have Applications or any serious use for dual processor systems. >>
And there are lots of people that have 1.4G T-bird and they really don't do anything but play the occassional game of Q3. so what's your point? You don't need 'serious' use to do anything you want... the entire OCing industry (which this site has named themselves after) is about doing almost extreme thigns to your computer to gain VERY little real life performance.
- Oct 14, 1999
- 11,999
- 307
- 126
<<what's so bad about this article? i agree with every point he made. >>
SMT or SMP, both require MP programming. He advocates the use of single-processor O/S's although there are plenty of SMP-capable O/S's on the market for little or no price.
He took a jab at the videocard as the bottleneck of gaming. Well 640x480 is not the limit of any videocard I've played in the last two years. Again, blowing smoke and showing mirrors. His point was lost over lack of substance.
He attacks the credibility of Sysmark 2000 and Sysmark 2001. We can certainly tell him to blow it out his arse if the website uses the benchmark ever again! Obviously the editor said it isn't credible.
Next was an attack on DDR use. Well DDR gave more than a slight improvement on alot of tasks. Like any memory system, if you don't put a strain on it you don't know its ability. He doesn't put a strain on his obviously, so perhaps he should have stayed with EDO RAM?
By the nature of the article being an editorial he is playing the devil's advocate. He wandered pro and con of SMP, taking a jab at a non-existent target of users that "don't need" SMP. (i.e. We don't need MMX, but that doesn't mean we shouldn't have it in the CPU.) The truth lay somewhere, but not necessarily where his opinions lay.
Then to say "you don't have the right"... the right to what? I seem to have lost his point there. As far as his bogus implication about "dragging the community down" he was plain out of line. People that chose SMP are not affecting those that choose single processors. Rather, the truth is the opposite. Microsoft built in the functionality to their last two of three consumer operating systems; Windows 2000 and Windows XP. They disable the SMP/SMT functionality which takes extra work. Truth seems to be that the single processor community is dragging the SMP community down.
SMT or SMP, both require MP programming. He advocates the use of single-processor O/S's although there are plenty of SMP-capable O/S's on the market for little or no price.
He took a jab at the videocard as the bottleneck of gaming. Well 640x480 is not the limit of any videocard I've played in the last two years. Again, blowing smoke and showing mirrors. His point was lost over lack of substance.
He attacks the credibility of Sysmark 2000 and Sysmark 2001. We can certainly tell him to blow it out his arse if the website uses the benchmark ever again! Obviously the editor said it isn't credible.
Next was an attack on DDR use. Well DDR gave more than a slight improvement on alot of tasks. Like any memory system, if you don't put a strain on it you don't know its ability. He doesn't put a strain on his obviously, so perhaps he should have stayed with EDO RAM?
By the nature of the article being an editorial he is playing the devil's advocate. He wandered pro and con of SMP, taking a jab at a non-existent target of users that "don't need" SMP. (i.e. We don't need MMX, but that doesn't mean we shouldn't have it in the CPU.) The truth lay somewhere, but not necessarily where his opinions lay.
Then to say "you don't have the right"... the right to what? I seem to have lost his point there. As far as his bogus implication about "dragging the community down" he was plain out of line. People that chose SMP are not affecting those that choose single processors. Rather, the truth is the opposite. Microsoft built in the functionality to their last two of three consumer operating systems; Windows 2000 and Windows XP. They disable the SMP/SMT functionality which takes extra work. Truth seems to be that the single processor community is dragging the SMP community down.
what are you talking about? the only point he is really making is that SMP is impractical for most people. He wants SMP supporters to stop promoting SMP as necessary, because it is not. If you carefully read the article, you would've noticed that. I agree with him. The only reason you should get SMP is if you're running a server, or if you just like wasting superfluous amounts of money as a hobby or on your ego. It's like telling someone who doesn't race that he needs to turbo charge his engine and get nitrous boosts for his car. Sure, it's nice, but only if you know what its for, know how to use it, and most importantly, if you have a use for it. If that guy only needs it to get from work and back, the sup-ups don't mean jack. Likewise, for the average home user, SMP doesn't mean jack, and is a waste a money.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter