Announcement

Collapse
No announcement yet.

How An Old PowerMac G5 Compares To Modern Intel CPUs

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

  • #11
    Originally posted by Min1123 View Post

    …, it doesn't have support for controlling the fans, so it makes the machine sound like a chorus of vacuum cleaners).
    Fan controls on Apple dual-core G5 work since about a decade. I know because I have been there, done that: http://t2-project.org/hardware/workstation/Apple/G5/

    Still under my desk, does not make much noise.

    Comment


    • #12
      Originally posted by cRaZy-bisCuiT View Post
      Is this just click bate? I mean, come on, no one would have expected something different. What about quality instead of quantity when it comes down to the number of articles?
      It is legit because a lot of people still measure performance by clock.

      Comment


      • #13
        Originally posted by rene View Post

        Fan controls on Apple dual-core G5 work since about a decade. I know because I have been there, done that: http://t2-project.org/hardware/workstation/Apple/G5/

        Still under my desk, does not make much noise.
        Your (personal) block size is set too low, please re-read original.

        Comment


        • #14
          Originally posted by AllLinux View Post
          Comparison to ARM would be more interesting.
          I came here to say this. PPC and ARM are both RISCs; just the additional instructions alone in x86 (including x86-64) would give it an edge in many of these modern-day tests. Something like the Pandaboard (which from what I recall, Michael has) would make for a very interesting test, since it is also a 32-bit dual core but clocked lower. I have no doubt a Cortex A9 would still outperform a G5, but it wouldn't absolutely destroy it in every test.
          Last edited by schmidtbag; 14 February 2017, 11:31 AM.

          Comment


          • #15
            Originally posted by cRaZy-bisCuiT View Post
            Is this just click bate? I mean, come on, no one would have expected something different. What about quality instead of quantity when it comes down to the number of articles?
            Uh, clickbait? what else would you want the title to be?

            Comment


            • #16
              Michael, do you have an old i7 920 lying around? I'm still running one of those, but now with 16 Gb of RAM and a Radeon r290 and 2 SSD's, remembering how expensive it was to get at the time. Especially the motherboard, my GA-EX58-UD4 X58 cost me $250 and the CPU about $300.

              Comment


              • #17
                Unfortunately with no access to any PowerPC hardware, I'm not able to deliver any results showing how modern POWER hardware is performing.
                PowerPC != POWER. :P If you do want to check out some sort of modern PowerPC hardware, you can look over on the AmigaOS website.

                The biggest loss with dropping the old PPC Macs will probably be the native big-endian testing environment. Even the ARM chips are pretty much always run little-endian now, so there isn't a lot of testing for endian-neutralness anymore.
                Last edited by mulenmar; 14 February 2017, 12:15 PM.

                Comment


                • #18
                  Sorry, but the article and comparison is just clearly wrong. How is it fair that on x86 you use pretty hefty optimizations and on G5 you use just -O3 ? It's shown particularly in Himeno benchmark. Guy, this G5 is equipped with AltiVec if you ever heard about the term.

                  Comment


                  • #19
                    Originally posted by mulenmar View Post

                    PowerPC != POWER. :P If you do want to check out some sort of modern PowerPC hardware, you can look over on the AmigaOS website.
                    Not sure, but if you talk about X1000, then it may barely come to G5 power and if you talk about good old Sam 440x then no way, absolutely no way. I.e. G5 crushes this SoC completely.

                    Comment


                    • #20
                      Originally posted by kgardas View Post

                      Not sure, but if you talk about X1000, then it may barely come to G5 power and if you talk about good old Sam 440x then no way, absolutely no way. I.e. G5 crushes this SoC completely.
                      Well, the life is not that easy after all. I do have nbench2 benchmark which I run on PPC 970 (G5) @ 1.8 GHz and on AMC PPC460 -- running inside the WD MyBook @ 800 MHz. G5 crushes really, but if recounted to same frequency the picture is different:

                      G5/PPC970:

                      Code:
                      $ cat powerpc_ppc970_1.8GHz_debian50_gcc432.txt
                      
                      BYTEmark* Native Mode Benchmark ver. 2 (10/95)
                      Index-split by Andrew D. Balsa (11/97)
                      Linux/Unix* port by Uwe F. Mayer (12/96,11/97)
                      
                      TEST                : Iterations/sec.  : Old Index   : New Index
                                          :                  : Pentium 90* : AMD K6/233*
                      --------------------:------------------:-------------:------------
                      NUMERIC SORT        :          775.36  :      19.88  :       6.53
                      STRING SORT         :           99.68  :      44.54  :       6.89
                      BITFIELD            :      1.2477e+08  :      21.40  :       4.47
                      FP EMULATION        :          137.92  :      66.18  :      15.27
                      FOURIER             :           10268  :      11.68  :       6.56
                      ASSIGNMENT          :          20.183  :      76.80  :      19.92
                      IDEA                :          2827.6  :      43.25  :      12.84
                      HUFFMAN             :          1136.8  :      31.52  :      10.07
                      NEURAL NET          :          25.489  :      40.95  :      17.22
                      LU DECOMPOSITION    :          764.64  :      39.61  :      28.60
                      ==========================ORIGINAL BYTEMARK RESULTS==========================
                      INTEGER INDEX       : 38.756
                      FLOATING-POINT INDEX: 26.656
                      Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
                      ==============================LINUX DATA BELOW===============================
                      CPU                 : 1.8GHz PPC970
                      L2 Cache            : 512K unified
                      OS                  : Linux 2.6.26-2-powerpc64
                      C compiler          : gcc version 4.3.2 (Debian 4.3.2-1.1)
                      libc                : libc-2.7.so
                      MEMORY INDEX        : 8.499  (1GHz == 4.72)
                      INTEGER INDEX       : 10.655 (1GHz == 5.92)
                      FLOATING-POINT INDEX: 14.784 (1GHz == 8.21)
                      Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
                      * Trademarks are property of their respective holder.
                      AMC PPC460:

                      Code:
                      $ cat mybook-ppc-apm82181.txt
                      
                      BYTEmark* Native Mode Benchmark ver. 2 (10/95)
                      Index-split by Andrew D. Balsa (11/97)
                      Linux/Unix* port by Uwe F. Mayer (12/96,11/97)
                      
                      TEST                : Iterations/sec.  : Old Index   : New Index
                                          :                  : Pentium 90* : AMD K6/233*
                      --------------------:------------------:-------------:------------
                      NUMERIC SORT        :           366.6  :       9.40  :       3.09
                      STRING SORT         :          33.706  :      15.06  :       2.33
                      BITFIELD            :      1.1814e+08  :      20.26  :       4.23
                      FP EMULATION        :          71.943  :      34.52  :       7.97
                      FOURIER             :          4907.7  :       5.58  :       3.13
                      ASSIGNMENT          :          6.1533  :      23.41  :       6.07
                      IDEA                :          972.25  :      14.87  :       4.42
                      HUFFMAN             :          607.72  :      16.85  :       5.38
                      NEURAL NET          :          8.2772  :      13.30  :       5.59
                      LU DECOMPOSITION    :          190.48  :       9.87  :       7.13
                      ==========================ORIGINAL BYTEMARK RESULTS==========================
                      INTEGER INDEX       : 17.867
                      FLOATING-POINT INDEX: 9.014
                      Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
                      ==============================LINUX DATA BELOW===============================
                      CPU                 : AMC APM82181 800MHz (PPC460 core including FPU)
                      L2 Cache            :
                      OS                  : Linux 2.6.32.11-svn70860
                      C compiler          : gcc version 4.4.5 (Debian 4.4.5-8)
                      libc                : libc-2.11.3.so
                      MEMORY INDEX        : 3.913  (1GHz == 4.89)
                      INTEGER INDEX       : 4.917  (1GHz == 6.15)
                      FLOATING-POINT INDEX: 4.999  (1GHz == 6.25)
                      Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
                      * Trademarks are property of their respective holder.
                      of course slight variation may be in different GCC version in use... Well, would you also like to see POWER7 or POWER8 or Xeon E5 SandyBridge or E3 Haswell? :-) -- Michael, you should definitely add nbench2 to your performance testing...

                      Comment

                      Working...
                      X