Announcement

Collapse
No announcement yet.

The Incredible Performance & Power Efficiency Of AMD Zen 1 vs. Zen 4C

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

  • #21
    Originally posted by rene View Post
    That would be a shame. But maybe Zen5 design was also already further along in the design state. At some point in time I believe we will see Zc cores on desktop. If not w/ Zen 5, then Zen 6. It just makes so much more sense to only include N high perf optimized cores if only N can turbo boost to the highest freq at a given time.
    no really not.... i can give you the source of my information:



    if you read this then it becomes clear they will only use Zen4c and Zen5c on notebook and mobile devices and cloud-native server CPUs.

    for Desktop AMD5 socket we will see 8core Zen5+3Dcache+ 8core Zen5 means only 16 cores.

    "That would be a shame." "It just makes so much more sense"

    i also think that for some people a 32core Zen5c AM5 socket cpu would make sense. means 2 chiplets of 16core ZEN5c

    also for some a combination of Zen5 8core + 16core Zen5C could also make sense the zen5 8core could also have 3D cache.

    but such a 32core or 24core system would bring desktop AM5 into the server space and companies like it to have market segmentation.

    in the source i linked amd claims that it is because of RAM speed is to slow to satisfy all the cpu cores.
    Phantom circuit Sequence Reducer Dyslexia

    Comment


    • #22
      Originally posted by drakonas777 View Post
      It's not about believing in something. AMD just don't have to use dense cores yet, because they are competitive in MT with the current core config. That's all. Why to invest extra money into a new product series when you are already competitive with Intel hybrid garbage.
      this source claims it is the ram speed whey they dont do more cores: https://www-igorslab-de.translate.go..._x_tr_pto=wapp
      Phantom circuit Sequence Reducer Dyslexia

      Comment


      • #23
        Originally posted by numacross View Post
        The upcoming (availability starting at the end of this month) Ryzens 8000 series APUs already do mixed core types on the desktop AM5 platform:
        this is cool but i think he is not so much in the low-end spectrum he wants highend desktop cpus.
        Phantom circuit Sequence Reducer Dyslexia

        Comment


        • #24
          Originally posted by numacross View Post
          ... Through-Silicon Vias (TSV) used to connect the cache chiplet were removed from its design[/URL].
          That's why I said theoretical. I wasn't implying it could be done with existing parts.

          Comment


          • #25
            Originally posted by rene View Post

            not really, so far on AMD's cpus only a handful of cores single core boos to the highest freq. So optimizing more than a handful for highest freq is a waste of silicon space unless you want a handful for process variation yield issues.
            First as you said, you need more cores if at least 4 should boost to the highest clock (look at CPPC Preferred Cores). And second, take a look at the 8500G:
            Code:
            Max. Boost Clock  Up to 5.0GHz
            Max Zen4c Clock  Up to 3.7GHz
            Zen4 Base Clock  4.1GHz
            Zen4c Base Clock 3.2GHz​
            If you consider that current 7000 CPUs already run with 4.5 Ghz all core load then 3.2 GHz is a heavy loss in performance. I think that AMD will probably combine 8 fat cores on one die with 16c cores on a second die, because those dies are already produced for server CPUs and could easily be combined that way.

            For the Notebook market there will be more individual monolithic variations which will come to desktop as ****G APUs. But those are usually low to mid tier.

            Comment


            • #26
              And I still rent a OVH-Kimsufi server rocking a 15y old Atom N2800...

              Comment


              • #27
                Originally posted by pegasus View Post
                It would flop. Especially on the desktop where you have multitude of single threaded tasks that are latency sensitive.
                See what Sun did with their UltraSparc T1 "Niagara" processors back in the day. They were only usable for a very niche use cases in the server area.
                i take it you don't realize i basically described a GPU, they are fully programmable, lots of small cores, the high end models have thousands of cores, NVIDIA uses a proprietary hardware thread scheduler, I'm pretty sure AMD's and Intel's GPUs also have their own proprietary hardware thread scheduler and they have low clocks.

                I want the CPU to be designed the same way, I think it would do wonders for responsiveness and smoothness and with only 100 or so cores, the power consumption would be very low.

                Comment


                • #28
                  Originally posted by sophisticles View Post
                  I want the CPU to be designed the same way
                  Right. There are further differences between cpus and gpus that go way beyond core size and count. Try to write a simple high-school project operating system for a GPU and you'll see what I mean. It won't work.

                  Closest thing to gpu performance you can get today with cpu flexibility are amd epyc-X chips with lots of L3 cache. They do wonders for memory bound problems that were being ported to GPUs for the past 5 years or so but it now looks like they will all jump back to CPU as it is just much more programmer friendly and has decades of polished tooling behind it.

                  Comment


                  • #29
                    Originally posted by geerge View Post
                    We were astonished that AMD basically pulled performance parity with intel out of seemingly nowhere, at a time when they looked to be struggling to tread water. Zen2 and intel's response of just adding more cores power be damned is when things really started heating up.
                    They were eight full fat cores. That was a surprise at the time. Intel was stuck on the two/four core arrangement for mainstream, with six cores for HEDT. Only Xeon platforms offered more.

                    Comment


                    • #30
                      Originally posted by pegasus View Post
                      Right. There are further differences between cpus and gpus that go way beyond core size and count. Try to write a simple high-school project operating system for a GPU and you'll see what I mean. It won't work.

                      Closest thing to gpu performance you can get today with cpu flexibility are amd epyc-X chips with lots of L3 cache. They do wonders for memory bound problems that were being ported to GPUs for the past 5 years or so but it now looks like they will all jump back to CPU as it is just much more programmer friendly and has decades of polished tooling behind it.
                      Windows 11 running entirely on a GPU:



                      Try that with Linux.

                      Oh wait, you can't.

                      Comment

                      Working...
                      X