VMS hardware

Categories:

• VAX workstations
• small VAXes (MicroVAX I, II, 3000, 4000)
• medium VAXes (VAX 700, VAX 6000, VAX 7000, VAX 8000)
• big VAXes (VAX 9000, VAX 10000)
• ft VAXes
• DEC ALPHAs (DEC 2000,3000,4000,7000,10000)
• AlphaStation ALPHAs
• AlphaServer ALPHAs
• AlphaStation XP/DS/ES
• AlphaServer DS/ES/GS
• Integrity

General note:

Note that the information about hardware is based on numerous source and are not guaranteed to be correct. It is most likely that there are several incorrect informations present.

Special note on benchmarks:

Meaning:

• VUPS is a geometric mean of a benchmark suite of 99 benchmarks with VAX 780 = 1.0 (only used for VAX systems)
• SPECmark89 is a geometric mean of a suite of 10 benchmarks with VAX 780 = 1.0
• SPECint92 and SPECfp92 are geometric mean of respectively a integer and floating point benchmark suite with VAX 780 = 1.0
• SPECint95 and SPECfp95 are geometric mean of respectively a integer and floating point benchmark suite with SparcStation 10 (40 MHz SuperSparc) using conservative optimization = 1.0
• SPECint2000 and SPECfp2000 are geometric mean of respectively a integer and floating point benchmark suite with Ultra 10 300 MHz using conservative optimization = 100

Since both VUPS and SPECmark89/SPECint92/SPECfp92 is based on VAX 780, then 1.0 VUPS = 1.0 SPECmark89 = 1.0 SPECint92 = 1.0 SPECfp92 is an obvious assumption.

Since a SparcStation 10 is rated at 50.2 SPECint92 / 60.2 SPECfp92 and 1.13 SPECmark95 / 1.38 SPECmark95 (note that those are with aggressive optimization not conservative and therefore <> 1.0), then an assumption would be 44.0 SPECint92 = 44.0 SPECfp92 = 1.00 SPECint95 = 1.00 SPECfp95.

Since an Ultra 10 300 MHz is rated at 12.1 SPECint95 / 12.9 SPECfp95 and 133 SPECint2000 / 126 SPECfp2000 (note that those are with aggressive optimization not conservative and therefore <> 1.0), then an assumption would be 10 SPECint2000 = 10 SPECfp2000 = 1 SPECint95 = 1 SPECfp95.

But do not use such primitive bencmarks as an indication for how your application will perform. Those bencmarks are very CPU intensive, while real world applications often also depends on disk speed, net speed, graphics speed etc.etc..

Especially do not compare a VAX and an Alpha and conclude from this bencmarks, that the Alpha is f.ex. 25 times faster. The CPU may be 25 times faster for doing arithmetic, but memory access is not 25 times faster and neither is disk access.

VMS page:

Link to VMS page.

Suggestions/reports:

Please email suggestions about new topics and error reports to Arne Vajhøj (arne@vajhoej.dk)

Trademarks:

VMS, OpenVMS, VAX, AXP, DECUS and Encompass are trademarks of HP.