The Tyan SMP board uses an 8 layer PCB design!! The rumor is it will cost $500usd or more. I haven't seen anything about the board that would lead me to suspect that it doesn't comply to at least the extended-ATX standard. All of Tyan's high end dual CPU boards do.
Werd to that. Both Athlons and Durons are SMP capable, we're just waiting for AMD to release their 760MP chipset this december for boards to be avilable in January/February 2001. There's nothing to get alarmed about, AMD has never said it would happen before then and so far they seem to be keeping their promisses so I don't see why 760MP won't be on track.
1-The P4 isn't really an incremental step forward as it's a completly new architecture. The first since the Pentium Pro.
That said, in terms of performance it will be only marginaly as fast as a 1GHz P3 or a 1GHz Athlon when it debuts at 1.4 and 1.5GHz.
Despite the delays, the P4 is still a very much rushed product. The P4 is moot already to everyone but investors. Intel will be moving from 423-pin packaging to a 478-pin packaging with the 0.13um Northwood that will replace the current P4. That will be the real next generation CPU intel wants to release.
2-Regular people do not need 64bit CPUs, nor will they for another half decade at least.
"I'll take my 700-mhz Athlon at normal speed thanks. Tried the overclocking thing -- didn't like it. The money, time and effort is equivalent to buying a faster rated CPU. Plus, can we say chip burnout?"
Old Athlons don't OC past 10% of their FSB. So you could only realistically get 770MHz at best out of it unless you used a Peltier, but I would still take the extra 70MHz.
Burning out a chip is harder than you think, unless you do something really boneheaded without knowing what the cpu you use is at least theoretically capable of.
I have no idea where you're getting your Overclocking How-To info from, but if you think overclocking is using water blocks, radiators or liquid nitrogen, then you're way off.
Like many thousands of others, I run my Celeron 300A at 450MHz and have done so for the past 2 years. It was as simple as going into the bios at boot-up and uping the FSB from 66MHZ to 100MHz (in incremental steps!)... It didn't cost a dime and took about 10 minutes to do safely.
Of course not all CPUs Overclock this well but that's the trick behind Overclocking; knowing what cpu to buy and what needs to be done to overclock it to a level you're comfortable with.
These new boards form Abit and Asus make teh whole process even more fool proof.
But that wouldn't fix memory bandwidth and speed issues. That's the real crux of current computing. We have 1GHz chips but memory running at 133MHz... Most modern CPUs have a theoretical peak execution rate over 1 GFLOPS, but only enough memory bandwidth to do a small fraction of that. I didn't say that this was for tomorrow, and I totally agree that 128MB RAM would make the die way too big and hot in the foreseeable future. But, I never meant for all of the system memory to be on the die. A 16MB buffer on-die would be a big help for most applications. You can't argue with Moore's law, and in 6 years, fabrication processes WILL be down 4x, which is about 0.05 microns and from some of the research that I've read about technology exists that can achieve this. So I expect that in half a decade it will be perfected enough to be a viable mass market solution. If it doesn't happen in 5 years it will then in 10, but whatever the time frame I believe it will happen.
Expansion cards would only makes performance/compatibility matters worse IMO. I think the real solution is a single chip with integrated CPU (possible multiple CPU's), RAM (L1, L2, + as much system memory as possible), North/South Bridge Controllers, LAN, Graphics controller, sound, etc... Yeah, that would pretty much mean that everyone's choices of computer configuration would dwindle down to a few combo models, but the tradeoff is well worth it IMO: The price would be lower (once we can get down to sub 0.1 micron BICMOS of course) there would be hardly any OS related hardware setup issues since every OS could have 100% support for every single Single-Chip-PC config available (since at the most, there would be maybe a dozen different ones available from different manufacturers), and the pesky memory bandwidth/speed issues of current systems would be all but gone, since everything would have direct access to memory and the CPU/GPU. Apparently AMD's Sledgehammer, which is supposed to eventually have multiple CPUs on the same die, will also integrate the North Bridge controller. This could be a case of trickle-up economics, as future high-end systems would be come more integrated like "low end" systems (I815 etc...) increasingly are today
That's why/. posted it under the AMD logo, it's reeally news about AMD t-bird chips not so much the motherboards. But obviously hvaing an unlocked CPU is useless if mobos don't have manual multiplier selection possible
Slashdot Mirror
The Tyan SMP board uses an 8 layer PCB design!! The rumor is it will cost $500usd or more. I haven't seen anything about the board that would lead me to suspect that it doesn't comply to at least the extended-ATX standard. All of Tyan's high end dual CPU boards do.
Werd to that. Both Athlons and Durons are SMP capable, we're just waiting for AMD to release their 760MP chipset this december for boards to be avilable in January/February 2001. There's nothing to get alarmed about, AMD has never said it would happen before then and so far they seem to be keeping their promisses so I don't see why 760MP won't be on track.
All the heatsinks made for the P4 so far are Aluminium with a single run of the mill fan. Not 2lbs Copper monsters.
1-The P4 isn't really an incremental step forward as it's a completly new architecture. The first since the Pentium Pro. That said, in terms of performance it will be only marginaly as fast as a 1GHz P3 or a 1GHz Athlon when it debuts at 1.4 and 1.5GHz. Despite the delays, the P4 is still a very much rushed product. The P4 is moot already to everyone but investors. Intel will be moving from 423-pin packaging to a 478-pin packaging with the 0.13um Northwood that will replace the current P4. That will be the real next generation CPU intel wants to release. 2-Regular people do not need 64bit CPUs, nor will they for another half decade at least.
Old Athlons don't OC past 10% of their FSB. So you could only realistically get 770MHz at best out of it unless you used a Peltier, but I would still take the extra 70MHz.
Burning out a chip is harder than you think, unless you do something really boneheaded without knowing what the cpu you use is at least theoretically capable of.
I have no idea where you're getting your Overclocking How-To info from, but if you think overclocking is using water blocks, radiators or liquid nitrogen, then you're way off.
Like many thousands of others, I run my Celeron 300A at 450MHz and have done so for the past 2 years. It was as simple as going into the bios at boot-up and uping the FSB from 66MHZ to 100MHz (in incremental steps!)... It didn't cost a dime and took about 10 minutes to do safely.
Of course not all CPUs Overclock this well but that's the trick behind Overclocking; knowing what cpu to buy and what needs to be done to overclock it to a level you're comfortable with.
These new boards form Abit and Asus make teh whole process even more fool proof.
But that wouldn't fix memory bandwidth and speed issues. That's the real crux of current computing. We have 1GHz chips but memory running at 133MHz... Most modern CPUs have a theoretical peak execution rate over 1 GFLOPS, but only enough memory bandwidth to do a small fraction of that.
I didn't say that this was for tomorrow, and I totally agree that 128MB RAM would make the die way too big and hot in the foreseeable future. But, I never meant for all of the system memory to be on the die. A 16MB buffer on-die would be a big help for most applications.
You can't argue with Moore's law, and in 6 years, fabrication processes WILL be down 4x, which is about 0.05 microns and from some of the research that I've read about technology exists that can achieve this. So I expect that in half a decade it will be perfected enough to be a viable mass market solution. If it doesn't happen in 5 years it will then in 10, but whatever the time frame I believe it will happen.
Expansion cards would only makes performance/compatibility matters worse IMO. I think the real solution is a single chip with integrated CPU (possible multiple CPU's), RAM (L1, L2, + as much system memory as possible), North/South Bridge Controllers, LAN, Graphics controller, sound, etc... Yeah, that would pretty much mean that everyone's choices of computer configuration would dwindle down to a few combo models, but the tradeoff is well worth it IMO: The price would be lower (once we can get down to sub 0.1 micron BICMOS of course) there would be hardly any OS related hardware setup issues since every OS could have 100% support for every single Single-Chip-PC config available (since at the most, there would be maybe a dozen different ones available from different manufacturers), and the pesky memory bandwidth/speed issues of current systems would be all but gone, since everything would have direct access to memory and the CPU/GPU. Apparently AMD's Sledgehammer, which is supposed to eventually have multiple CPUs on the same die, will also integrate the North Bridge controller. This could be a case of trickle-up economics, as future high-end systems would be come more integrated like "low end" systems (I815 etc...) increasingly are today
That's why /. posted it under the AMD logo, it's reeally news about AMD t-bird chips not so much the motherboards. But obviously hvaing an unlocked CPU is useless if mobos don't have manual multiplier selection possible