Announcement

Collapse
No announcement yet.

Why are multicore processors faster

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Why are multicore processors faster

    Is there any page on the web that explains in a simple manner why multicore systems are faster? Let us a ssume a single core and a dualcore system both have the same architecture and clock speed, why would the dualcore be faster when it is still technically one CPU with the same clock speed?

  • #2
    With one core you can do one task at a time.

    With dual cores you can do 2 different tasks at once. Or if you are a smart programmer, split a single task into 2 pieces and work on both pieces at the same time to complete it at almost twice the speed.

    Comment


    • #3
      But the processor is still running at the same clock speed, doing 2 things at once means youll be doing them twice as slower than if you did them one at a time. How exactly does dividing the CPU into many subprocessors speed things up?

      And are you saying that single core processors cant run two programs at once? I think we can positively confirm that good old pre-multicore computers were able to run more than just one process simultaneously.

      Comment


      • #4
        doing 2 things at once means youll be doing them twice as slower
        No. Each core runs at full speed. (ignoring the newTurbo mode features in some CPUs)

        And are you saying that single core processors cant run two programs at once?
        Yes, this is exactly what I am saying. But you don't see this as a user. Your single core CPU does task switching and scheduling, so maybe every 30 milliseconds it switches it's focus between active tasks. This is called a context switch. See,
        http://msdn.microsoft.com/en-us/libr...8VS.85%29.aspx
        http://en.wikipedia.org/wiki/Context_switch
        http://en.wikipedia.org/wiki/Scheduling_%28computing%29
        So as a user you are tricked into thinking the single core CPU is working on more than a single task at once.

        Comment


        • #5
          Thanks for the explanation about context switching, I didn't know about that.

          No. Each core runs at full speed. (ignoring the newTurbo mode features in some CPUs)
          But multicore isn't multi-CPU, it's still one CPU running at the same clock speed consuming the same amount of power as a single core, so how would it be able to do 2 tasks at once without needing to divide the speed to finish the 2 tasks?

          Comment


          • #6
            Originally posted by 2006 View Post
            Thanks for the explanation about context switching, I didn't know about that.



            But multicore isn't multi-CPU, it's still one CPU running at the same clock speed consuming the same amount of power as a single core, so how would it be able to do 2 tasks at once without needing to divide the speed to finish the 2 tasks?
            A multicore CPU is more than one CPU on the same package. So a multicore CPU is a multi-CPU system its just that the CPU's are in the same package as apposed to being in two or more separate packages.
            Main Box*AMD Ryzen 7 5800X*ASUS ROG STRIX B550-F GAMING*G.SKILL 32GB 2X16 D4 3600 TRZ RGB*Geforce GTX 1070Ti*Samsung 980 Pro 1 TB*Samsung 860 EVO 1 TB*Samsung 860 EVO 2 TB*Asus DRW-24B3LT*LG HL-DT-ST BD-RE WH14NS40*Windows 10 Pro 21H2

            Comment


            • #7
              So they do consume more power then, is that why we arent seeing clock speeds going any higher than the famous P4 prescott 3.8 ghz?

              Comment


              • #8
                So they do consume more power then
                In general no. Technology has got better, reducing power consumption over time.

                CPUs are vastly more complex that you might at first suspect.

                The Prescott ran at a high clock speed because of it's long pipeline. But while long pipelines allow for high clock speeds, they are hard to keep full, leading to poor efficiency. There is a discussion of it here,
                http://www.anandtech.com/print/1230
                http://en.wikipedia.org/wiki/Instruction_pipeline
                "A deeper pipeline means that there are more stages in the pipeline, and therefore, fewer logic gates in each stage. This generally means that the processor's frequency can be increased as the cycle time is lowered. This happens because there are fewer components in each stage of the pipeline, so the propagation delay is decreased for the overall stage."

                There is also a nice table on the Wikipedia page showing pipeline lengths for a collection of CPUs.

                Comment


                • #9
                  CPUs have been taking less power because transistors kept getting smaller. If two CPUs with the same architecture at 45 nm, with one being single core the other dual core, would the dualcore consume the same amount of power? If so, then i'm confused, because if it is TWO CPUs running at the same clockspeed, would this not logically consume more power?

                  The Prescott ran at a high clock speed because of it's long pipeline. But while long pipelines allow for high clock speeds, they are hard to keep full, leading to poor efficiency.
                  I read that very long page that details this. I have a question, why are longer pipelines necessary for higher clock speeds? Can they not make 3.8 ghz with short pipelines?

                  Comment


                  • #10
                    Why are longer pipelines necessary for higher clock speeds?
                    This was already answered in the article & my post above. The clock speed is limited by the slowest part of the pipeline. Short pipelines have more complexity at each step, making for a lower clock speed, but increasing efficency (in terms of work done per clock cycle).

                    CPUs have been taking less power because transistors kept getting smaller.
                    Attributing everthing to the size of the transistor is too simplistic. Voltage levels have come down, 5.5V to 3.3V to 2.5V to 1.3V..., materials have improved, and chips have got smarter about saving power by underclocking and shutting down unused parts of the chip, etc...

                    Comment


                    • #11
                      All I know is that people overclock these CPUs to speeds greater than they are sold for so I didnt think longer pipelines were a necessary feature if people can just increase the clock speed to their hearts content.

                      Attributing everthing to the size of the transistor is too simplistic. Voltage levels have come down, 5.5V to 3.3V to 2.5V to 1.3V..., materials have improved, and chips have got smarter about saving power by underclocking and shutting down unused parts of the chip, etc...
                      Well, do multicore processors consume more power or not?

                      Comment


                      • #12
                        Well, do multicore processors consume more power or not?
                        More power than what? In what cirsumstance? Idle, under full load, with a single thead? What type of work load? Games, video encoding, word processing, etc.. Are you talking about TDP, or actual power usage?......

                        I think you are looking for a simple black and white answer. But things are more complex than that.

                        Comment


                        • #13
                          TDP.
                          (stupid character limit)

                          Comment

                          Working...
                          X