This program is very like the FP one in presentation - it shares the same instruction sheet - but it restricts itself to arithmetic using whole numbers (IS stands for Integer and String, whereas FP stood for Floating Point). This restrictions makes IS about ten times faster than its stablemate. Again, compilation is fast and you can compile several programs into different areas of memory.
The compiled code is slower than that generated by Mcoder, and quite a lot slower than Zip, but the IS compiler has the bonus of support for very simple sprites. The lack of array-handling is annoying, although not too hard to get around if you're prepared to use PEEK and POKE or string-slicing instead.
BENCHMARK TESTS _ COMPARING THE COMPILERS
Eight standard BenchMark programs were used in the comparison; timings for the execution of each benchmark program are given in seconds, with the speedup ratios achieved by each of the compilers printed on a grey background.ZX BASICBM1: 4.9BM2: 9.0BM3: 21.9BM4: 20.7BM5: 25.2BM6: 62.8BM7: 90.0BM8: 25.0MEHMOODBM1: * (0x)BM2: 0.065 (138x)BM3: 9.0 (2.4x)BM4: 4.2 (4.9x)BM5: 4.2 (6.0x)BM6: * (0x)BM7: * (0x)BM8: * (0x)SOFTEK FPBM1: 1.75 (2.8x)BM2: 2.1 (4.3x)BM3: 8.7 (2.5x)BM4: 9.4 (2.2x)BM5: 9.4 (2.7x)BM6: 19.7 (3.2x)BM7: 24.0 (3.8x)BM8: 22.5 (1.1x)SOFTEK ISBM1: 0.058 (84x)BM2: 0.076 (118x)BM3: 0.57 (38x)BM4: 0.98 (21x)BM5: 0.99 (25x)BM6: 1.32 (48x)BM7: * (0x)BM8: * (0x)MCODER2BM1: 0.043 (113x)BM2: 0.097 (93x)BM3: 0.62 (35x)BM4: 0.90 (23x)BM5: 0.92 (27x)BM6: 1.17 (54x)BM7: 1.47 (61x)BM8: * (0x)ZIP 1.5BM1: 0.031 (158x)BM2: 0.064 (141x)BM3: 0.194 (113x)BM4: 0.108 (192x)BM5: 0.115 (219x)BM6: 0.29 (235x)BM7: 0.46 (191x)BM8: * (0x)