Announcement

Collapse
No announcement yet.

Raspberry Pi Zero 2 W Benchmarks - Nice For $15

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

  • Raspberry Pi Zero 2 W Benchmarks - Nice For $15

    Phoronix: Raspberry Pi Zero 2 W Benchmarks - Nice For $15

    At the end of October came the pleasant surprise of the introduction of the Raspberry Pi Zero 2 W. This drop-in replacement to the original Raspberry Pi Zero features a more powerful 1.0GHz quad-core Cortex-A53 compared to the miniscule 1GHz single-core design of the original Pi Zero while boasting 512MB of LPDDR2 RAM. Here are some initial benchmarks of the Raspberry Pi Zero 2 W for those curious about its performance.

    https://www.phoronix.com/vr.php?view=30745

  • #2
    This little sucker has more RAM, more cores, and probably a better scores than my old Pentium 3 and Pentium 4 systems I keep for emergencies in the closet. Truly marvelous to see where we have gone.

    Comment


    • #3
      Would be nice to see some common x86 CPU thrown in the mix as well, e.g. Intel Core i5 2500 to understand what kind of performance we are talking about.

      Comment


      • #4
        Originally posted by kylew77 View Post
        This little sucker has more RAM, more cores, and probably a better scores than my old Pentium 3 and Pentium 4 systems I keep for emergencies in the closet. Truly marvelous to see where we have gone.
        Not to mention, probably uses less power than just the hard drive in those systems too.

        Comment


        • #5
          I currently use a Pi Zero W as a small VPN and Pi-hole server and it works pretty well. It's incredible how awesome those little machines are nowadays.

          Comment


          • #6
            Broadcom means it uses blobs? Augustiner Bräu seems to be more exciting btw..

            Comment


            • #7
              Originally posted by avem View Post
              Would be nice to see some common x86 CPU thrown in the mix as well, e.g. Intel Core i5 2500 to understand what kind of performance we are talking about.
              I have an i3 2100, and it is much, much faster than my Orange Pi Lite 2 (Allwinner H6, armbian, 4xA53 cores @ 1.8 GHz). This rpi won't have a chance.

              Comment


              • #8
                Seems like they've designed another good entry into the Zero line. Very pleased with the release of this device!

                Comment


                • #9
                  Originally posted by avem View Post
                  Would be nice to see some common x86 CPU thrown in the mix as well, e.g. Intel Core i5 2500 to understand what kind of performance we are talking about.
                  I agree it's nice to have something in benchmarks like this to give us a reference point, but that leads into my second point:

                  I'm impressed with the performance we're seeing for that SiFive chip! I'm making assumptions here, but I'm going to guess that most software out there *does not* have any kind of assembly optimizations for those cores. Meanwhile, almost all software for x86 will have some asm in there if there are routines which can be meaningfully optimized. ARM will often have some asm in there as well, but much less than x86.

                  So, what we're seeing here is a bare C/C++ implementation on the SiFive vs a partially optimized ARM vs a heavily optimized x86 (if we do as avem proposes). While assembly optimizations don't cheat in that the chip is still doing all the work--even if it's doing it more efficiently--it's not a fair comparison. Not all sofware will have these optimizations. For example, you may have some strange app that you use that's not popular enough to have attracted the level of development it takes to get optimized assembly code pathways. For you, the relative performance diferences you'd see from avem's proposal wouldn't be respresentative of your code.

                  I guess it always comes down to the basic rule of benchmarking--there is not meaningful benchmark other than your code. Anything else might be vague guidance, but it won't tell you how your code will run.

                  Comment


                  • #10
                    Originally posted by willmore View Post

                    I guess it always comes down to the basic rule of benchmarking--there is not meaningful benchmark other than your code. Anything else might be vague guidance, but it won't tell you how your code will run.
                    Agreed. If we compare the Pi Zero (W) to the Pi Zero 2 (W) then it's a win. If we compare it to anything else, well then the devil is in the details. Even the Pi 400 is very different of you look closely.

                    ASM is a huge deal. Optimizing software will take many years if not decades. Good OS support will take a long time too. For example the Pi Zero 2's Core runs ARMv8-A but the official OS (and these tests) uses 32bit instructions. These devices have various problems in official and 3rd party OSes if you look close enough.

                    For most projects it's good enough and the Pi zero 2 W is very affordable even if you include all the required addons like case, PSU, micro USB to USB cable that does not block the PSU cable, Mini-HDMI to HDMI etc.

                    I ordered one a while ago, but South Africa (ICASA) still needs to approve the device before I can test it.

                    Comment

                    Working...
                    X