Related articles |
---|
New assembly language instructions to support OO languages? tony@my.net (Tony) (2008-12-04) |
Re: New assembly language instructions to support OO languages? tmk@netvision.net.il (Michael Tiomkin) (2008-12-04) |
Re: New assembly language instructions to support OO languages? DrDiettrich1@aol.com (Hans-Peter Diettrich) (2008-12-05) |
Re: New assembly language instructions to support OO languages? David.Schroth@unisys.com (David W Schroth) (2008-12-05) |
Re: Burroughs/Unisys architecture, was New assembly language instructi DrDiettrich1@aol.com (Hans-Peter Diettrich) (2008-12-07) |
From: | Hans-Peter Diettrich <DrDiettrich1@aol.com> |
Newsgroups: | comp.compilers,comp.arch |
Followup-To: | comp.arch |
Date: | Sun, 07 Dec 2008 05:21:08 +0100 |
Organization: | Compilers Central |
References: | 08-12-014 08-12-017 08-12-021 08-12-032 |
Keywords: | architecture |
Posted-Date: | 07 Dec 2008 08:34:22 EST |
David W Schroth schrieb:
> The system I work on is both segmented and paged. The Page Table
> (note singular) consumes only a small fraction of memory. Being a
> global Page Table, there is no penalty incurred when switching
> threads.
How does your system avoid an performance loss of 50% or more, when all
memory references go through an address mapping table that again resides
in memory?
> The time to load a segment register is a few cycles, and
> doesn't seem to normally impact performance.
>
> I can't make any sense of your last sentence no matter how I try.
While a small page table can reside in a separate MMU, a limited
number of segment registers or descriptors in a fixed-size memory area
doesn't make much sense to me. How does your model differ from
multi-level indirect references, with regards to execution speed?
What's the benefit of using segment registers, instead of using base
addresses in any other address register?
DoDi
Return to the
comp.compilers page.
Search the
comp.compilers archives again.