1) While dual core processors will not be completely promising until software is multithreaded to take advantage of this within one application, they will be functional right out of the box.
Like Hyperthreading, you can use Windows 2k and XP to designate what software has which quanta available to it, and it will be threaded accordingly.
This will allow you to divide processing resources among applications, which is typically known as load balancing... a method that dual and quad processor solutions have had for years.
2) Dual core solutions are only recieving leverage from one side- processor manufacturers.
Extended 64bit(remember, they're not pure, full 64bit archs) architectures and instruction sets have support from processor manufacturers and software developers alike.
While we might see the two technologies take off together, they aren't going to be sister systems for another year. They won't fully walk hand in hand until Longhorn is released, which has full native support for both 64bit extensions and dual core processors.
For anyone looking to upgrade, I ask you to heed my advice:
Unlike the Extended 64bit architecture AMD's Athlon has been sitting on for three years, dual core technologies will be available for use out of the box.
However, if you're not using specific software or trying to load balance two CPU intensive tasks(such as, playing a game while burning a DVD), it won't be in your best interest to upgrade right away.
I definitely agree with Sander in saying that everyone shouldn't get their hopes up.
Want to enjoy fewer advertisements and more features? Click here to become a Hardware Analysis registered user.
I agree. The major change will also have to come from the software side. Programmers will have to re-write the stuff they already made in order to utilize the dual core processor. The Idea is sound enought, but, I think it would be more headache than being usefull.
I agree - like SMP and Hyperthreading, Dual-Core will only be useful in applications that support it. Even though we have had SMP for a decade and HT for a few years now, Neither has any real use in mainstream.
Disclaimer: I have used SMP in the past, and one of my current machines is a Dual-PIII rig. While no speed demon, I have dound that the "thread balancing" results in a noticeably "smoother" machine, but for gaming this is not much of an issue.
I ran a Dual 166mmx for many years (Retired it about 2 years ago) with windows 2000 and Linux. While it didn't speed up the applications I was using beyond a 166mmx, it did allow for much smoother multitasking. I could run more things on my machine without the GUI becomming bogged down and unresponsive. It really does help when you start up something in the background (like compressing a video, raytracing, games) and still want to use the computer.
Another thing he was wrong on was the 64 bit version of Windows. There has been an AMD only version around for a year now, and you can get it for free for 6 months. I downloaded it in anticipation of getting an AMD system, but it never happened, so I burned it to CD and have held onto it just in case it can be useful. If you want it, it shouldn't be too hard to find on MS's site (I lost the link).
After hearing about dual core cpu's I've been wondering why we just can't use dual cpu's instead. Much like the current xeon, opteron server boards but using mainstream (lower cost) cpu's instead. Of course you still need load balancing but couldn't it be developed more like the SLI tech being used by Nvidia? Where your not really running two threads at once but merely distributing the load evenly? I'm a newb to the server board/dual cpu concept so bear with me here. Also, if the "right" chipset is not currently available for my concept then is it possible it may become so once dual core cpu's hit the market? If I"m not mistaken, most server chipsets are not really designed for home use functionality. I'd much rather pay for two lower priced existing tech cpu's on a board than one high priced piece lacking application/software support.
Ideally I'd love to see Asus, Abit, or any other quality mobo maker design a board for dual mainstream cpu's that would allow simultaneous overclocking and either combined processing power and functioning as one or at least functioning as two independent cpu's with true Hyperthreading capability. The ability to use current P4's with HT on a dual cpu mobo might be very interesting and would be nice especially if supporting software is forthcoming. Any thoughts on this??
The reason you can't just add more sockets is because there needs to be a chipset specifically designed for it.
While I'd like to see some updated chipsets for quad CPU motherboards, I doubt it's going to happen.
Dual CPU chipsets are updated as much as single CPU chipsets, so that's not a problem.
Dual and quad CPU solutions already do load balancing. The difference, however, in dual core solutions is that the processor can share threads in a very high speed, low latency interconnect, opposed to having to do it through software and a chipset.
"The reason you can't just add more sockets is because there needs to be a chipset specifically designed for it."
Pretty much knew that. Hence my reference to dual cpu server boards.
"Dual CPU chipsets are updated as much as single CPU chipsets, so that's not a problem."
I understand that but they lack the same functionality of mainstream chipsets for the consumer market. They are also more demanding regarding ECC, registered, buffered which is understandable for servers but not for general consumer use. Forget about the price too.
"Dual and quad CPU solutions already do load balancing."
I understand that according to the other posts. SMP is it? I'm not familiar with it but get the jist. What about my reference to SLI by Nvidia? Doable? It is load balancing also but perhaps of a different flavor.
"The difference, however, in dual core solutions is that the processor can share threads in a very high speed, low latency interconnect, opposed to having to do it through software and a chipset."
Interesting point I hadn't considered. While I agree it would be faster than dual cpu's, would the difference really be noticable though? Can't say that I'd would think so, at least not in real world terms anyway, but perhaps.
I guess my whole point is, wouldn't it be nice to see a dual cpu rig with a load balancing chipset that gave us all the features that we have with the Intel 925XE or at least 915 and able to run off the shelf ram?? If you really wanted dual core finctionality then perhaps the new dual core chipset would work with dual cpu's.
It just seems to me that we could work with the cpu's available for a lot less $$.
They also have chipsets with all of the features of desktop boards.
SMP is symmetric multiprocessing, which is typically what SLi does.
And yes, the difference between dual core and dual processor is significant. Dual core processors have privileges to both cores' stacks of memory that don't need to be addressed to both cores to be shared, whereas dual processors have their own stacks of memory that they must address to each processor in order to share.
Well, I guess my final questions would be, why don't we see SMP desktops boards for P4's A64's readily available since the performance improvements would be substantial if SLi is any indicator?
Specifically which dual cpu chipsets now support Intels HD audio?
From what I've seen, the server boards do not have a lot of the features of desktop boards or not alot of each feature. usb ports for one.
What about buffered,registered ram? Still required?
And last but not least, what about cost of these dual core cpu's/mobo's and is it really worth the difference for price vs. performance when compared to my "concept" dual cpu desktop board? I understand your explanation of performance differences but do you really think these new dual core cpu's are gonna be the next best thing to sliced bread if their supported properly or at least better that the A64's?
"Well, I guess my final questions would be, why don't we see SMP desktops boards for P4's A64's readily available since the performance improvements would be substantial if SLi is any indicator?"
SLi is an indicator of how easily influenced people can be into thinking more is better. The technology itself isn't practical because it's not cost effective.
I guarantee you the people who have it don't notice a difference in anything but benchmarks. It will never be apparent in games, because while they have two GPUs running together, once a new GPU comes out, they'll have two outdated GPUs instead of one.
SLi can't replace new graphical instruction sets... that's only something that can be done by getting a new GPU.
If Nvidia was smart, they would offer a flip-chip design, so that you wouldn't have to buy a new PCB, just a new GPU.
"Specifically which dual cpu chipsets now support Intels HD audio?"
Who cares about onboard audio? If you need audio performance, you should always look to Creative's Audigy series. While the quality might be improving, the CPU usage is also increasing. Workstation boards don't have onboard audio because of this.
"From what I've seen, the server boards do not have a lot of the features of desktop boards or not alot of each feature. usb ports for one."
USB2.0 is available on workstation boards. I very rarely use USB, unless it's required by the peripheral.
"What about buffered,registered ram? Still required?"
On some boards. Some boards allow you to use anything. The only problem now is supporting frequencies faster than DDR266.
"And last but not least, what about cost of these dual core cpu's/mobo's and is it really worth the difference for price vs. performance when compared to my "concept" dual cpu desktop board?"
Yes. For one, they cost twice as less. Two, dual CPU boards are usually E-ATX, extended ATX. They're like 13" x 12". With dual core solutions, you don't waste board realestate on a second CPU.
"I understand your explanation of performance differences but do you really think these new dual core cpu's
are gonna be the next best thing to sliced bread if their supported properly or at least better that the A64's?"
Both AMD and Intel are developing dual core processors... this isn't Intel's venture to one up the Athlon 64.
While I respect some of the designs of the Athlon, I find it no better than the Pentium in any real world performance tests, whereas, the Pentium does show improvement in multimedia tasks, such as rendering and audio mixing.
I think that until processor manufacturing technology gets better, dual core processors are a viable solution to get some more performance out of desktop computing.
The military and a bunch of organizations are trying to build a synthetic diamond via catalysts. If this synthetic diamond gets produced, it would have a lot of benefits. Among them would be that it would be more thermally advantageous than silicon, while providing all of the other neccessary properties to completely replace silicon on semiconductors.
I don't know who said it, namely because I just forgot the name, but whoever said that Microsoft needs to "get off their ass and come out with a 64bit OS," or something along those lines needs to know something: Windows XP, which won't be nearly as advanced as Windows Longhorn, which will be 64-bit compatable (I'm excluding WinXP64 as a "new" OS, although it will be 64-bit compatable), was roughly 40 million lines of code long.
It takes a long time just to develop a program, a simple program, that's about 10,000 lines long. Think about coding 40 million lines, using more than one person, who don't all share the same experience, and trying to get it to mesh. Now think about how Windows Longhorn will have to apply to both 32-bit AND 64-bit architecture, and tell me how long you think it would take you.
Now, WinXP64 is supposed to be out by May or June, if I remember correctly, but I'm not going to count it as a "new and improved" operating system. It's just revamping Win XP to be 64-bit compatable. I'm sure it will take a lot of work, but if Microsoft took the focus away from WinXP64 (I'm guessing at LEAST 150 programers) and focused all of their Operating System Development center's attention on Longhorn, which is projected, I believe for a Q1/Q2 2006 release, they could have Longhorn out by the end of 2005.
The "extended 64-bit instructions," or however you called it, is why I think it's a complete waste of time to make WinXP64. Why release just an updated 64-bit compatable operating system 6-10 months before your latest and greatest operating system is going to be released? It just seems like a bad marketing move.