sslug-teknik team mailing list archive
-
sslug-teknik team
-
Mailing list archive
-
Message #14627
Re: 1GB på Linux..
Erik Jonsson wrote:
>
> Se http://developer.intel.com/design/chipsets/index.htm
>
> Där finns information om Intels chipset. Jag har ett minne av, att
> där tidigare fanns information om hur mycket RAM ett moderkort med
> visst chipset kunde cacha.
> --
Jeg er enormt glad for at Klavs Klavsen rejste spørgsmålet om memory
og performance, et spørgsmål som viste sig at vedrøre memory og
cache, jeg var netop på vej ud af døren for at købe mere ram. Jeg
har 2*16 MB og ville købe 2*32 MB, evt. 4*32 og kassere 16 MB'erne,
men nu kan jeg se at det havde været at smide 4-5000 kr ud af
vinduet.
Jeg fulgte Erik Jonssons råd og søgte oplysning på Intels webside,
hvor jeg fandt flg.
For Intel® chipsets, we offer a portfolio of solutions supporting
different cacheability limits:
Chipsets for the Intel Pentium Processor Family
430VX = 64MB
430HX = 512MB
430TX = 64MB
og mit chipset er desværre 430VX.
Mit motherboard er et Iwill P55TV Lite med bl. a. flg. features:
Chipset: 430VX
CPU: socket 7 med 166 MHz Pentium
Cache: 512 KB write-back Level 2
Memory: support for indtil 128 MB.
Den webside jeg fandt på Intel har overskriften:
Memory Upgrades: The Relationship Between Cache and Main Memory.
og det væsentligste afsnit er flg.: (Bemærk de sidste linier, som
indeholder ordene 'drop in system performance.)
Operating System Use of Main Memory and Cacheability.
Your operating system, along with the active applications,
determines
the loading of instructions and data into main memory, loading the
main memory from your hard drive, CD ROM drive, disk drive, or
whatever input devices your system has. Most of today's mainstream
operating systems perform this loading of main memory beginning from
the top main memory "location" and working down.
Refer to this example of "loading from the top" for a system with
main
memory within cacheable limits: Suppose your system has a cacheable
main memory limit of 64MB and has 16MB of main memory installed
and you are running an application and operating system that needs
6MB of main memory. If you were to diagnose what main memory was
being utilized, you would most likely find that your system?s 16MB,
15MB,...11MB main memory locations were filled, with your system?s
1MB-10MB main memory remaining empty, ready for more
information that may need to be loaded. In this example, all the
main
memory installed is cacheable and thus this process of loading main
memory "from the top" is typically transparent to you, the user.
But, suppose you loaded your system with main memory above the
cacheable limits. The loading process of main memory -- loading
"from
the top" -- would mean that your non-cacheable main memory would
be loaded first. Once the non-cacheable main memory is filled, the
cacheable main memory would be loaded. Another example illustrates
this point: Suppose your system?s main memory cacheable limit is
64MB again, but now your system has been loaded to 80MB of main
memory (16MB above the cacheable limit) and you were to run that
same application and operating system ran in the previous example
(the
one requiring 6MB of main memory). The load "from the top" process,
would mean that main memory locations 80MB-75MB would be filled.
Since these main memory "locations" are above the main memory
cacheable limit, the information stored at these main memory
locations
are not cacheable, and cannot use the cache memory as a fast
retrieval
storage bin. In this example, you will likely see a drop in system
performance by exceeding the cacheable main memory limits of the
system.
Mvh
Jørgen Heesche
References