dan@rna.UUCP (Dan Ts'o) (07/20/84)
Hi, 1) Does anyone have hard numbers on the performance of the MicroVAX I ? Preferably benchmarks that can also be run on other machines. 2) Can anyone comment on the ease with which one can port 4.2BSD to the MicroVAX I. Certain instructions and data types are missing, as are devices such as the time of year clock, etc. Thanks. Cheers, Dan Ts'o Dept. Neurobiology Rockefeller Univ. 1230 York Ave. NY, NY 10021 212-570-7671 ...cmcl2!rna!dan
kvc@scgvaxd.UUCP (07/25/84)
With regard to porting 4.2 to the MicroVAX I: DEC is offering ULTRIX on the MicroVAX I. Since ULTRIX is 4.2, I suspect that it is not very difficult to port. The instructions that are missing are little used anyway and the processor provides some support for them, like processing operands before taking the fault. On performance: As far as I've ever been able to find out, performance of the uVAX I is somewhere between a 730 and a 750 (probably closer to 730). I suspect that the processor runs at a respectable speed, but I have no idea how the QBUS would affect performance. I have played with the uVAX a bit at DECUS. ULTRIX on it looked just like 4.2 on anything else, and uVMS on it looked just like VMS anywhere else. I was told by Kathleen Morse, a uVMS developer, that they use the same binaries when they build the distribution kit for uVMS and VMS. I suspect you could get UNIX to run as easily if you emulate the missing instructions at some low level in the OS (this is why uVMS runs VMS binaries). I thought ULTRIX on the uVAX seemed lots slower relative to ULTRIX on the 750 than VMS on the uVAX compared to VMS on a 750. I suspect that a little tuning for 4.2 in the small environment may fix it up. This is just the way things seemed to me. (Please no VMS vs. UNIX flames on this but VMS really semed faster on the uVAX. ULTRIX may have just needed some more work.) The diskless systems they had running VAXELN (a stand-alone, message-based, distributed, real-time, [catch-phrase of your choice], etc. OS for VAXes from DEC) ran real fast with all sorts of concurrent processes playing music, moving robot arms that played towers of hanoi, and processing data from a video camera. In that environment the processor sure seemed to hum right along. Remember, this is all what I got from a few hours hands-on. Some real benchmarks would be a lot more useful, but I haven't seen any yet. Does anyone out there have any? /Kevin Carosso allegra!scgvaxd!kvc Hughes Aircraft Co.