KOSH [Kommunity Orientated Software Hardware] Weekly Summary Week: 1st March 1999 Number: 008 Mailing List: kosh-hardware-o In the mailing list this week, the following items were discussed. Please do not email the scribe regarding any of these topics, it is not his job to answer these questions, but merely to report the topics of conversation. If you have any queries about this summary, please email summaries@kosh.convergence.org, stating the Summary Number, and Mailing List Name and he will try to answer your queries. a) Subject: SCISC model of hardware level CPU independence Summary of Debate: As a superset of other instruction sets, the SCISC set would present a wide range of instructions, possibly even including instructions that have yet to be implemented in any piece of hardware. In general it is felt that this method will not be efficient. By using a pseudo machine level code, much higher abstraction would be lost. This higher abstraction would allow a compiler to make much more intelligent optimization decisions based upon the capabilities of the CPU in question. b) Subject: Other examples of code translation. Summary of Debate: It is observed that recent x86 chips are translating x86 instructions into simpler instructions which are passed to the core, in effect imitating RISC while maintaining x86 compatability. Under FX!32 x86 code is run on an alpha processor, however the results are less than native code on the alpha would be. One case where native code performance may be less of an issue would be the chip from Transmeta, as it is being designed to run as an emulator. c) Subject: News from outside the list. Summary of Debate: An article on the Russian Elbrus project was posted, mentioning their chip that is said to outperform Merced. In addition an article mentioning that HP is joining other major computer companies in showing an interest in linux.