Herewith as promised the latest set of benchmarks published by the Systems Performance Evaluation Co-operative, carefully rearranged to create an across the page table in descending order of performance. The Co-op asked us to include the results of the full 10 set of benchmarks, saying that users will then recieve a more complete assement of performance by looking at the results of individual benchmarks. Small statistical differences between machines are considered nominal it adds. One helpful little test that benchmarks really do work is to compare the results of machines that are essentially the same – the Control Data 4680 60MHz Beta and the MIPS RC6280 60MHz Beta – and lo and behold, the results are identical. The same applies with the Silicon Graphics and Control Data machines in the second table, which shows multiprocessor performance – Control Data buys the MIPS RISC-based machines OEM from Silicon Graphics. The strong showing for MIPS is good news for DEC when it gets around to using the latest versions.

April 1990 SPECmarks

CDC MIPS IBM Stardent IBM IBM IBM IBM IBM DEC 4680 RC6280 RS/ 3010 RS/ RS/ RS/ RS/ RS/ STATION 60MHz 60MHz 6000 6000 6000 6000 6000 6000 5000 beta 540 730 930 530 520 320 200

gcc 44.5 44.5 21.0 17.8 17.5 17.5 17.8 13.6 13.7 17.3 (a) (a) (a) (a) (a) (a) esprso 43.7 43.7 24.9 20.3 20.7 20.7 20.7 16.3 16.3 18.5 (a) spice 37.7 37.7 33.2 14.7 27.6 27.6 27.6 19.8 19.8 13.7 2g6 (a) (a) (a) (a) (a) (a) (a) doduc 38.3 38.3 33.2 14.7 27.6 27.6 27.6 21.5 21.5 18.2 (a) (a) (a) (a) (a) (a) (a) nasa7 39.6 39.6 43.4 62.9 36.6 35.5 35.5 26.7 26.6 22.6 II 44.9 44.9 23.7 18.1 19.8 19.8 19.8 15.6 15.6 21.8 (a) eqntott 35.6 35.6 26.7 18.2 22.5 23.0 23.0 18.0 17.7 18.4 (a) mtrx300 52.4 52.4 26.5 108.5 22.4 21.8 21.8 17.5 17.5 17.0 tpppp 52.2 52.2 65.8 29.1 54.7 54.7 54.7 42.7 42.9 22.0 tomcatv 36.8 36.8 91.0 61.9 76.1 76.1 75.7 56.7 56.5 17.3 SPEC 42.2 42.2 34.7 29.0 29.0 28.9 28.9 22.3 22.3 18.5 mark (a) (a) (a) (a) (a) (a) (a)

CDC CDC MIPS HEWLETT CDC SI SI CDC CDC 4380 4360 M/ DN10000 4340 GRAPHICS GRAPHICS 910B- 920B 2000 4D/210S 4D/25S 621 450

gcc 18.8 19.2 19.0 12.8 15.3 18.1 13.4 18.1 13.4 esprso 18.1 18.0 18.3 12.9 17.3 16.7 13.9 16.7 13.9 spice 13.5 13.5 12.1 11.9 13.6 10.0 10.6 10.0 10.6 2g6 (a) doduc 17.4 17.1 17.6 23.0 15.8 14.7 11.1 14.7 11.1 nasa7 18.4 18.4 18.4 20.5 18.1 14.4 12.6 14.4 12.6 II 23.9 23.9 23.8 11.1 20.3 16.3 15.5 16.3 15.5 eqntott 18.4 18.4 18.4 11.1 17.2 16.2 13.3 16.2 13.3 mtrx300 13.4 13.3 13.3 22.0 13.6 9.7 9.9 9.7 9.9 tppp 20.1 20.0 20.4 32.0 17.5 16.1 12.2 16.1 12.2 (a) (a) (a) tomcatv 17.1 16.9 17.7 19.9 14.2 10.8 10.7 10.8 10.7 (a) (a) (a) SPEC 17.7 17.6 17.6 16.6 16.2 14.0 12.2 14.0 12.2 mark (a) (a) (a) (a)

DECsystem DECstation MIPS DATA DEC 5400 3100 RC2030 GENERAL STATION av 410 2100

gcc 11.0 10.9 8.6 9.6 7.6 esprso 14.1 12.0 11.8 12.8 8.9 spice 9.1 9.5 9.0 8.2 7.0 2g6 doduc 12.8 11.3 9.9 6.7 8.3 nasa7 12.8 13.2 13.3 8.9 9.7 II 12.2 13.1 14.2 12.9 9.7 eqntott 13.6 11.2 11.5 10.3 8.8 mtrx300 10.4 9.8 9.6 10.3 7.2 tpppp 13.4 12.5 4.9 8.0 8.9 tomcatv 9.9 9.9 11.0 8.0 7.3 SPEC 11.8 11.3 10.0 9.4 8.3 mark

The other tables, below, measures multi-processor and parallel processing performance in the simplest way the co-operative knows how – by running multiple copies of the Release 1.0 benchmark. Once again, the MIPS chips come out on top, but the Hewlett-Packard Co Apollo DN Series 10000 really begins to show its paces in a multiprocessor configuration.

SPEC CDC Si Hwlt Solbrne Star- Hwlt Mtrla thruput 910B Gr DN- 5/804 dent DN10000 188K -634 4D/ 10000 3020 -QP 340S

gcc 4~18.6 4~18.6 4~11.5 4~12.4 2~16.7 3~11.8 4~11.2 espresso 4~23.1 4~23.1 4~12.7 4~15.5 2~19.4 3~12.7 4~16.1 (a) spice 4~12.7 4~12.7 4~11.2 4~13.6 2~13.7 3~11.3 4~8.7 (a) (a) (a) 2g6 doduc 4~19.1 4~19.1 4~22.3 4~11.5 2~18.4 3~22.4 4~8.7 nasa7 4~12.1 4~12.1 4~17.1 4~15.5 2~51.7 3~18.5 4~7.1 II 4~25.7 4~25.7 4~10.7 4~7.0 2~17.2 3~10.8 4~16.7 (a) eqntott 4~21.0 4~21.0 4~10.7 4~15.6 2~17.4 3~10.8 4~15.0 (a) matrix3 4~10.0 4~10.0 4~15.3 4~20.7 2~70.4 3~17.4 4~7.4 tpppp 4~23.8 4~23.8 4~24.5 4~16.0 2~27

.0 3~28.9 4~8.2 (a) (a) tomcatv 4~11.3 4~11.3 4~17.4 4~11.7 2~48.0 3~18.3 4~8.8 SPEC 67.6 67.6 58.8 54.0 51.0 46.2 41.2 thrpt

SPEC Solbrne DECsys VAX CDC Si Hwlt Solbrne thrupt 5/803 5840 6000 910B Gr DN10000 5/802 -460 -632 4D/ 320S

gcc 3~14.5 4~7.6 6~5.7 2~19.8 2~19.8 2~12.0 2~17.3 (a) espresso 3~15.8 4~14.5 6~5.9 2~23.1 2~23.1 2~12.7 2~15.9 spice 3~13.4 4~7.8 4~6.0 2~13.6 2~13.6 2~11.5 2~14.2 2g6 (a) doduc 3~10.9 4~12.5 6~7.4 2~20.4 2~20.4 2~22.4 2~11.6 nasa7 3~14.9 4~8.9 6~6.6 2~14.3 2~14.3 2~19.3 2~16.8 II 3~8.4 4~9.3 6~5.6 2~26.4 2~26.4 2~10.8 2~12.9 eqntott 3~16.0 4~11.7 6~6.6 2~21.6 2~21.6 2~10.8 2~16.6 (a) matrix3 3~20.8 4~7.4 6~5.9 2~13.0 2~13.0 2~19.5 2~21.4 tpppp 3~14.8 4~11.8 6~6.5 2~23.6 2~23.6 2~31.0 2~15.7 (a) tomcatv 3~11.2 4~8.5 6~6.5 2~25.1 2~15.1 2~18.9 2~12.1 SPEC 14.1 38.8 37.8 37.0 37.0 31.8 30.4 thrpt

SPEC Mtrla D Gen D Gen Solbrne D Gen DECsys VAX thrupt 188K- AV6220S AV412W 5/801 AV4020S 5810 6000 DP -410

gcc 2~16.1 2~12.8 2~8.5 1~19.5 2~7.4 1~8.7 1~6.6 (a) espresso 2~18.1 2~16.8 2~12.2 1~16.0 2~10.2 1~15.0 1~6.4 spice 2~11.5 2~10.3 2~7.2 1~15.0 2~6.0 1~8.0 1~6.4 2g6 doduc 2~9.6 2~8.9 2~6.7 1~11.9 2~5.6 1~13.3 1~7.4 nasa7 2~12.1 2~10.3 2~.67 1~17.8 2~5.8 1~9.3 1~8.2 II 2~18.7 2~16.7 2~12.6 1~17.8 2~10.5 1~10.9 1~7.4 eqntott 2~15.6 2~13.5 2~10.2 1~16.9 2~8.6 1~13.4 1~6.7 (a) matrix3 2~13.0 2~12.5 2~8.6 1~22.7 2~7.8 1~8.1 1~6.4 tpppp 2~12.6 2~10.9 2~7.5 1~17.0 2~6.5 1~14.2 1~7.5 tomcatv 2~11.0 2~10.2 2~7.1 1~12.4 2~6.1 1~8.8 1~7.2 SPEC 27.0 24.0 17.0 16.4 14.6 10.7 7.0 thrpt

The as mean that the code was modified to enable it to run but not to enhance performance!