Quad-Core "Barcelona" to "Outperform Clovertown by 40%"

ZuntarZuntar North Carolina Icrontian
edited February 2007 in Science & Tech
When it comes to quad-core processors for the desktop and server arenas, Intel has pretty much had the market to itself since November 2006 (if you don't count AMD's Quad FX platform). Intel's quad-core processors were officially announced on November 14 in the form of the desktop-oriented Core 2 Extreme QX6700 and the server-oriented Clovertown Xeon 5300 series.
More here!

Comments

  • edited February 2007
    I don't know much how Dual Core CPU's work but there is talk in the future you may have 3 or 4 CPU's in one .I think when it is executing a program part of the CPU is executing this and the other part is executing that and than puts it together at the end.
  • edited February 2007
    Dual core cpus, is like saying 2 cpu in one. For example intel core 2 duo E6400, has 2 cores that work at the speed of 2.13ghz each and each has 1mb cache. So basically is 2 cpus. When talking about quad cores, is the same thing but with 4 cores.
  • ZuntarZuntar North Carolina Icrontian
    edited February 2007
    airoh69 wrote: »
    Dual core cpus, is like saying 2 cpu in one. For example intel core 2 duo E6400, has 2 cores that work at the speed of 2.13ghz each and each has 1mb cache. So basically is 2 cpus. When talking about quad cores, is the same thing but with 4 cores.

    I believe that Intel's quad core is two dual cores, whereas AMD's quad core is four fully functioning cores. :cool:
  • kryystkryyst Ontario, Canada
    edited February 2007
    zergpc209 wrote: »
    I don't know much how Dual Core CPU's work but there is talk in the future you may have 3 or 4 CPU's in one .I think when it is executing a program part of the CPU is executing this and the other part is executing that and than puts it together at the end.

    Not so much. They way it works in theory is that every time a process is started it's assigned to the CPU with the least amount of load. So if you have two apps running each app would be running on a separate core. So it should keep both cores running as lean as possible. Currently Unix/linux/OSX do this really well windows XP doesn't handle it quite as well though. However that's not the fault of the OS it's more to do with windows apps not being designed to support multiple processors.
  • edited February 2007
    Not so much. They way it works in theory is that every time a process is started it's assigned to the CPU with the least amount of load

    I believe this is what Windows XP does when you are multitasking and one program will run on one core and the other program on the other core .

    Well the multithreaded games and software should take all the work load and thread it in 2 so one core is working on one work well the other is working on the other.

    Well keep in mind a dual core running at 2.0 GHz will be 2 x fast than a one core running at 2.0 GHz because it is like 2 CPU's in one.The problem is they can't make core run faster than 3.0 GHz.
  • kryystkryyst Ontario, Canada
    edited February 2007
    zergpc209 wrote: »
    I believe this is what Windows XP does when you are multitasking and one program will run on one core and the other program on the other core .

    Well the multithreaded games and software should take all the work load and thread it in 2 so one core is working on one work well the other is working on the other.

    Well keep in mind a dual core running at 2.0 GHz will be 2 x fast than a one core running at 2.0 GHz because it is like 2 CPU's in one.The problem is they can't make core run faster than 3.0 GHz.


    That's how linux and OS X work. Windows XP is a different beast entirely. The issue is that XP supports it but not all the software does and even when everything is working as it should it still doesn't work as efficiently as Linux or OS X handle multi-cpu'd systems.

    Also a dual core system at 2.0 ghz will not run faster then a single cored system. They are both still running at 2.0 ghz. The difference is that once the load reaches about 75% the dual core system will continue to run unaffected while the single core system will bog down.

    Now comparing a dual core 2.0 ghz system to a single core 3.0 ghz system. The 3.0ghz system will run faster as long as the load is below 75% (that % actually can very). But as soon as it hits 75% a dual core 2.0ghz system should run better because each core is running at less then 50%. This is assuming it's not a single application thread that is hogging a majority of the load.
  • edited February 2007
    That's how linux and OS X work. Windows XP is a different beast entirely. The issue is that XP supports it but not all the software does and even when everything is working as it should it still doesn't work as efficiently as Linux or OS X handle multi-cpu'd systems.

    So Windows XP does not say one program for this core or one program for that core or divide the work 2 the 2 cores? It is on work load if one core is more than 75% it moves some of the work load to the other core ?

    Also a dual core system at 2.0 ghz will not run faster then a single cored system. They are both still running at 2.0 ghz. The difference is that once the load reaches about 75% the dual core system will continue to run unaffected while the single core system will bog down.

    I thought 2 CPU's running at 2.0 GHz would be fater than single core CPU running at 2.0 Ghz not that the GHz would be 4 Ghz but that you have 2 CPU's in one.
    Now comparing a dual core 2.0 ghz system to a single core 3.0 ghz system. The 3.0ghz system will run faster as long as the load is below 75% (that % actually can very). But as soon as it hits 75% a dual core 2.0ghz system should run better because each core is running at less then 50%. This is assuming it's not a single application thread that is hogging a majority of the load.


    So it is base on load and not one program for this core and one program for that core or divide the work load.

    And only multithreaded games and software would take a task and one load for one core and one load for the other core ?

    If Windows xp does not really support dual cores than I should get Windows vista when all the patches and fixes come out like sp1 .
  • kryystkryyst Ontario, Canada
    edited February 2007
    zergpc209 wrote: »
    So Windows XP does not say one program for this core or one program for that core or divide the work 2 the 2 cores? It is on work load if one core is more than 75% it moves some of the work load to the other core ?

    Ideally that's how it works. But some software won't recognize the 2nd core and will always try to operate on the primary core so that's why often in a dual core XP system you'll see one core at 100% and the other core idling. The load balancing at the OS level is still at the whim of the applications trying to run. Also because many applications don't actually run in a protected mode they can crash and suddenly over run your system maxing out the cpu to 100%. Other OS's including windows 2003 server use a protected mode so that no application can get out of hand and bring down the system.
    I thought 2 CPU's running at 2.0 GHz would be fater than single core CPU running at 2.0 Ghz not that the GHz would be 4 Ghz but that you have 2 CPU's in one.

    So it is base on load and not one program for this core and one program for that core or divide the work load.

    And only multithreaded games and software would take a task and one load for one core and one load for the other core ?

    That's basically it. The CPU speed is how fast your computer runs. Adding more cpu's won't make your computer run faster they handle the load better. The less load on a particular CPU the better it'll perform. But it still won't be running faster then it's Speed can allow. Think of your system currently. If you reboot and load 1 application that application will load normally however if you start up a 2nd application, or a 3rd, 4th etc.... they will make the whole system run slower because they are using up CPU time. More CPU's less load more programs can run at the full speed of your CPU.
    If Windows xp does not really support dual cores than I should get Windows vista when all the patches and fixes come out like sp1 .

    I wouldn't suggest getting Vista just for the dual core support.
  • edited February 2007
    That's basically it. The CPU speed is how fast your computer runs. Adding more cpu's won't make your computer run faster they handle the load better.

    The clock speed is how fast the CPU runs not how fast the work gets done .AMD does more work for clock speed there is other things that speed up the CPU like fast FSB or more cache and more work for clock cycle .
    The less load on a particular CPU the better it'll perform. But it still won't be running faster then it's Speed can allow.

    If you have a car and have to take 2 boxes from house A to house B but can only go 100 KB on the road in hour it would take you 1 hour for one box well having 2 cars going at the same speed it take 1 hour for both box thus- it is faster .

    But if a car is not full and only both cars are at 20% full there would be no point as one car must be full or at least each car 60% full to be faster.So having dual core is only good if one core is full or at least no less than both cores at 60%.

    Think of your system currently. If you reboot and load 1 application that application will load normally however if you start up a 2nd application, or a 3rd, 4th etc.... they will make the whole system run slower because they are using up CPU time. More CPU's less load more programs can run at the full speed of your CPU.

    More programs taking up CPU time will slow the computer down but if one CPU can work on one program and other CPU other program it will be faster.If I go to store and there is only one person working and 8 people it will take long time!! Now I go to store 8 people working and 8 people that is 8 X faster but if there is 8 people working and only 5 people than there is 3 people who are doing nothing and this is a waste.:(
  • Your-Amish-DaddyYour-Amish-Daddy The heart of Texas
    edited February 2007
    Here. Lemme offload some tidbits from my notes. Good ol' copy-pasta.

    "I remember back when the Athlon MP's came out and I fought as hard as I could just to SIT at a machine that sported those hot-blooded quantispeed enabled monsters. I never really got a chance until someone blew the resistors on their board and dumped it on me for an XP-2200 Tbird-A that I'd overclocked slightly. Man, those were the days. 1024MB DDR266, ATI 9600XT, Two 80GB drives. I was the toast of the Klan, and our entire brigade. (I'll explain that privately, but it's still a bit sensitive.) I could hit CS in less than 4 seconds if I had run it recently. The only real problem I had was the heat issue. Man they ran hot. Two AMD MP 2200's they were. Throughbreds. 1.8ghz a piece. Oh but they could crank out some heat. I had read somewhere that they were by no means anything to play games on, but they held their own. Their biggest benefit I noticed while on IRC one night. I was switching between IRC and my TV tuner card. I had remembered on my previous XP-2200, this took about a second, but I realized that it was almost instantaneous. I tried it several times, random spans. It was indeed much faster. It seems the two processors
    worked independently, AND together on multiple tasks, based on the actual processor time demand. This blew my mind. I remembered the dual P3's that could work simultaneously to increase overall performance, but I've yet to see anything like this. Double processor design is the way to go. I stare into my glass of vodka, wondering if there will ever be a day that this design could be condensed, or at least made simple so that people could experience this for themselves."

    Further down the page I have a diagram of the MP bus schematic and several other things scribbled out that are of no relevance. But I appended them further on with more relative date.

    "P4 Dual cores, or the D line came out today. They were giving a demonstration at Circuit City. Man, was I ever stupid for wishing for a simple design in this complex system. Processor ran at 2ghz, yet I swear it took more time for application switching than my old MP system. I could go between Firearms and Ventrilo in the time it took my video card to realize I had changed programs and to change my resolution. I sometimes wonder if the people at Intel knew good ol' Tim Leary on a very intimate level. While they managed to pack a core inside of a core, all they really did was make a partition for the processor, inside of a partition. That's just asking for a failure somewhere. However the advent of DDR2 coming along soon may just be what Intel needs to manage to get this dodo off the endangered species list. I still really don't trust DDR as it is, but it seems the new standard of it, and PCI-Express, which is literally a beefed up version of the 32-bit, 33mhz bus we've all come to hate due to it's limitations. But I can't be too judgmental. Intel's been the innovator of the bus system. AMD's yet still the top of the crop with their 64-bit processors. I've never seen anything run that fast."

    I go on about the Winchester core and it's benefits, how awesome the 64-bit bus and Hypertransport are, but I come back to the whole Dual Core crap.

    "Well. AMD's X2 processors have finally made their way to my garage. I have to say that they are no better than the Intel's. It's still a logical core inside a physical core. I doubt their will ever be a new Dual-CPU system that'll be affordable to the public any time in the next ten years. However the Opterons seem to be promising, but I'll never be able to afford the board, much less the eight-hundred dollar processors. Not to mention the Error-correcting memory I'd have to have. PCI-X has become the standard. 16-lanes, 32-bit. It's the same thing as PCI but it seems to work like AGP in some ways by making a dedicated line to the processor, but it doesn't seem to be as efficient. I wished I had more money to experiment with, but insanity only makes so much these days."

    Basically, I still stand by my point that two, or four physical processors will always be more efficient than two cores, or four on one die.
  • kryystkryyst Ontario, Canada
    edited February 2007
    zergpc209 wrote: »
    The clock speed is how fast the CPU runs not how fast the work gets done .AMD does more work for clock speed there is other things that speed up the CPU like fast FSB or more cache and more work for clock cycle .

    You are talking about two different things. How fast the work gets done is a factor of many things. Here we are only talking about processing power and that is a direct relation of the speed of the CPU. Faster the CPU the faster that machine can process. Adding more cores won't speed up how fast a machine can process until the load on a system increases beyond about 75% and then only if the software is designed to work with understands and knows that it has access to multiple CPU's. A machine with multiple cores never runs faster then it's CPU speed allows it can just do more with what it has.

    So from a users point of view. Having a single core 2ghz machine and going to a dual core 2ghz machine. Some things will run faster other things won't. It's all a question of load. So currently in a windows system if you have a 2ghz single core machine. You'll get a far more noticeable speed increase if you were to switch to a single core 4ghz machine instead of a dual core 3ghz machine under most circumstances. The exception being is if you are running many independent applications. In that case the 3ghz dual core machine will handle the load better.
  • yaggayagga Havn't you heard? ... New
    edited February 2007
    This is why I leave my P4 Hyperthreading turned off, it just feels faster most of the time, which corresponds exactly to what Kryyst has been saying for quite a while.
  • kryystkryyst Ontario, Canada
    edited February 2007
    Hyperthreading is a slightly different beast all together though many similarities exist. One thing though is that Hyperthreading has been around now for longer so there are more applications using it. I personally question though the significance of it one way or another. It's an extra instruction set that helps when the software puts it into consideration, but if not it's just wasted overhead.

    Dual core chips though won't make things slower that's for sure. The worse case is you just don't notice the benefits.
  • edited February 2007
    You are talking about two different things. How fast the work gets done is a factor of many things. Here we are only talking about processing power and that is a direct relation of the speed of the CPU.

    Yes the processing power will not run faster than 2.0 GHz but if you have 2 CPU's running at 2.0 GHz that is 2 x faster than one CPU at 2.0 Ghz if the work load is no less than 60% each.

    Look I'm trying not to argue I'm using logic here if there is some thing in the engineering that does not work like this than explain.


    Basic logic if 2 man taking boxes from house A to house B and can only walk 3 KM in hour it be faster than one man taking one box than making other trip.But if some how the engineering works different I don't understand.
    Faster the CPU the faster that machine can process.

    Faster it can eexecute the machine code not how fast the computer is .
    Adding more cores won't speed up how fast a machine can process until the load on a system increases beyond about 75% and then only if the software is designed to work with understands and knows that it has access to multiple CPU's.

    And the same logic adding more staff working in a store will not speed thinks up if you don't have the people.It is waste to have 8 staff working and only 3 people in the store.

    If you have 3 cores and only one core is used it is waste .You have to be using enough CPU power will it will use both cores if the CPU only need one core it is a waste .

    A machine with multiple cores never runs faster then it's CPU speed allows it can just do more with what it has.

    That is right the more people walking taking the boxes from house A to house B will not walk faster than 3 KM but work will be done faster.
    So from a users point of view. Having a single core 2ghz machine and going to a dual core 2ghz machine. Some things will run faster other things won't.

    That is right if you want it to run faster both cores must be use .
    It's all a question of load.

    And the more load is more work think of a washing machine there is 2 solution to get work done faster adding more washing machine or make one big washing machine.Or think of a space ship if you have to take 3 trips you have 2 solution you can have 3 space ships or one big space ship.
    So currently in a windows system if you have a 2ghz single core machine. You'll get a far more noticeable speed increase if you were to switch to a single core 4ghz machine instead of a dual core 3ghz machine under most circumstances. The exception >being is if you are running many independent

    Well yes or if the software is not programed for dual core.
  • Your-Amish-DaddyYour-Amish-Daddy The heart of Texas
    edited February 2007
    Man, I don't know how to reply to this. He's got a point for being a moron, but he still really doesn't bring anything to the table. More cores+More load=Short lead time, but More Cores+More Load=Long Cycle Time. Lemme put it like this.

    Yeah. Single cores are great for heavy loads that don't change much. Dual cores are great for changing loads that aren't too heavy (Talkin' 50% here, since the cores have to be condensed to fit on the die.). Quads won't be too much better since it's one die, four cores. It's either gonna be huge, or very inefficient in the long run vs single cores.

    I think I read somewhere that the second core was on a separate die, but were it, it would have to have a TDP at least twice it's original design with a single die, since the processor would have double heat sources outputting the same heat, at the same time. And I also wonder how the dual-core, dual die would work since they'd both have to be connected to all the pins to function. Bottlenecking much? I'm a single-core guy. My FX-55 does everything I need it to, and if I could use DDR, I wouldn't change anything.

    Remember boys. It's a Logical core. A pretend core. A slave. A secondary. They're not in unison. It's not a true dual design unless they're separate and can work together, and separately.
  • kryystkryyst Ontario, Canada
    edited February 2007
    Zerg you make my head hurt with your broken logic. You have glimpses of getting it but then you seem to lose what your talking about.

    If you want to talk about end user performance then then having a dual 2ghz machine will run faster then a single core 2ghz machine while under load. However when they aren't under load there is no speed benefit. In either case a 3ghz machine will be equal if not faster then a a dual core 2ghz machine in most cases. The exception being a the dual core machine in a an OS that's using it correctly will be much better at handling many smaller apps along side a couple big ones. So if you were say browsing, answering emails, burning a dvd and using skype. The dual core 2ghz machine would probably handle that better then the 3ghz single core processor.

    However adding more cores doesn't change the speed that process take place at it just allows more process to happen under a given load.
  • Your-Amish-DaddyYour-Amish-Daddy The heart of Texas
    edited February 2007
    Yeah. And in best-case scenario, with a dual core 2ghz chip vs a 2ghz single, the performance gain won't even be 50%, no matter what the software is designed for. Just like saying X64 is better than X86, when X64 only has what, two more instructions that replace the memory controller, and is 32bits wider? Oh sure, the SOFTWARE makes it wo--no it doesn't. Software doesn't make any difference unless it's just poorly put together. Were it software limitations, then the dual cores would have to have a separate operating system designed specifically for them, and same with the X64's if they weren't designed the same. I'm not saying there's no difference, but by god it won't be night and day. Kryyst is right. It's gonna be such a subtle change, if any noticed. Now verses single processor designs and double processor designs, it is almost night and day. From Experience, two physical processors will always be better than dual core, because dual cores have to work one way. Together. Double processor design lets them work independently AND together. One chip can do one thing, the other can do something totally different or they can split the load betwixt them and work in a method rarely thought of to this day. A beautiful example of the real charge of technology--Nay, magic. I have to say that the first time I saw a 3d video game, I honestly thought there was a little world in my computer. I don't remember what game it was, but it was some time ago. I guess I'm rambling now. Zerg, I wanna hit you with this. AMD and Intel release whitepages, and manuals for all their products. Might wanna dig'em up and get some facts brother.
  • edited February 2007
    Some how we are having a communication problem so that take it slow.Do you agree that the CPU speed is how fast the CPU execute the machine code not how fast the job gets done or how fast computer is .

    And do you agree adding more cores does not make the CPU run faster but only the work is done faster..
  • Your-Amish-DaddyYour-Amish-Daddy The heart of Texas
    edited February 2007
    No I do not agree. Adding more cores makes the bus longer. Thus everything has to wait for everything to finish. Even with synchronous instructions, they will take longer to execute. A single core full die will execute code faster than a half-sized dual core. They are LITERALLY cramming two smaller processors on one normal sized die. More cores means more cycle time. That's like taking a car with a sweet 351 in it and cramming two four cylinder engines in it. If you could get it to work right, It'd be hell getting it to run synchronous and manage the transmission or the drive train, it wouldn't be any more efficient than the original 351 that was in there. I'm not all about this dual core crap. Give me a single core, double processor design and I'll show you the most efficient setup in the world. Single core, double processor design is how it's supposed to work. Not this physical core/logical core setup crap that's all the rage now.
  • edited February 2007
    kryyst wrote: »
    I wouldn't suggest getting Vista just for the dual core support.

    I have a vista cd but I am waiting to install it hearing horror stories about nvidias lack of good drivers and the sparsity of other drivers for some of my hardware. Has there been much press on Vistas support for multicore CPU's? It would be nice to jump over in year or so when things are more nailed down. DirectX 10 and better support for multicore cpu's would be nice.

    My main focus is games. I think multithreaded games could benefit widely from this.
  • kryystkryyst Ontario, Canada
    edited February 2007
    Vista has better support for multiple cores. XP's wasn't bad but Vista's is more integrated again it has the same support as 2003 server. The issue is still that most software isn't built to take full use of it. In a year that'll be change since multi-core is the processors are going since they can't make them much faster anymore.
Sign In or Register to comment.