An replacement

Posted by ALB42 on 29. Juni 2017One Comment

My Blizzard 1260 card is broken a while before, I sent it to someone to repair, but this will need time. That means my A1200 is rather useless currently. Which is not a nice situation. I decided to buy a Blizzard 1230 IV with 68882 FPU. Later when my 1260 is back maybe I can revive my other Tower A1200 to use that card. The card arrived today amazing it is even with the original package and manual, I did not expect that, very close experience to when I bought my Blizzards back in the days.

Blizzard 1230 IV in Original package

Now I can make some additional tests about FPU speed and compatibility of FreePascal. First again the Mandelbrot test again together with the old Values:

Mandelbrot results (Runtimes, shorter is better)

Test 68060/50 MHz FPU 68060/50Mhz SoftFPU Vampire SoftFPU 68030 68882/50 Mhz FPU 68030 SoftFPU
Mandelbrot single precision 0.12 s 9.53 s 3.81 s 2.14 s 38.03 s

Mandelbrot double precision 0.15 s 23.72 s 13.37 s 2.31 s 71.87 s

and again the Scimark test:

SciMark2 results (MFlops, higher is better)


Vampire V600 V2+ 128 MB SoftFPU code
** ** ** SciMark2a Numeric Benchmark, see http://math.nist.gov/scimark ** ** ** ** Delphi Port, see http://code.google.com/p/scimark-delphi/ ** ** ** Mininum running time = 2.00 seconds Composite Score MFlops: 0.06 FFT Mflops: 0.03 (N=1024) SOR Mflops: 0.12 (100 x 100) MonteCarlo: Mflops: 0.03 Sparse matmult Mflops: 0.08 (N=1000, nz=5000) LU Mflops: 0.02 (M=100, N=100)

Amiga1200 68060/50 FPU code
Mininum running time = 2.00 seconds Composite Score MFlops: 2.26 FFT Mflops: 1.18 (N=1024) SOR Mflops: 5.05 (100 x 100) MonteCarlo: Mflops: 0.86 Sparse matmult Mflops: 1.81 (N=1000, nz=5000) LU Mflops: 2.41 (M=100, N=100)

Amiga1200 68030+68882/50 FPU code
Mininum running time = 2.00 seconds Composite Score MFlops: 0.29 FFT Mflops: 0.15 (N=1024) SOR Mflops: 0.56 (100 x 100) MonteCarlo: Mflops: 0.13 Sparse matmult Mflops: 0.29 (N=1000, nz=5000) LU Mflops: 0.34 (M=100, N=100)

The Result stays very bad for SoftFPU even the 68882 is around 5 time faster than the SoftFPU on a Vampire. That fits also to the results I saw in the apollo forum, someone showed a
SoftFPU emulation result of AIBB FMath Test resulting in a runtime of 9.73 s, my new 68882 solve this test in 1.51 s. That means the 68882 is 6.5 times faster than the Vampire with SoftFPU emulation and this time its independent from FPCs SoftFPU implementation, seems the FPC SoftFPU is not that bad actually.

One comment to "An replacement"

  1. Roy Gillotti sagt:

    The SoftFPU is not optimized yet, a proof of concept done by someone as their first assembly language project. Lets see what a fully optimized for the vampire SoftFPU brings to the table.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.