wall-rt@cscosl.ncsu.edu (Richard Todd Wall) (04/10/90)
I was working on a 386 SX machine with 2 meg. of memory and
running Quarterdeck Expanded Memory Manager. The machine's memory was
set up with 640k base memory, 512k extended, and 512k expanded. When
we ran PC Lab's Benchmark test of the memory the rusult was that
the 640K memory was faster than the expanded, and the expanded was
faster than the extended. Does any one know why this is??? Or was
benchmark wrong????
o o ______ _ _____ _ _ o o mcnc!bacchus!beretta!todd
' ' / / / ' ) / // // ' ' or wall-rt@cscosl.ncsu.edu
/ ______/ __/ / / / __. // //
_/ (_) (_/_(_/_ (_(_/ (_/|_</_</_ Durham, North Carolinahd7x@vax5.cit.cornell.edu (04/13/90)
In article <1990Apr9.234033.7308@ncsuvx.ncsu.edu>, wall-rt@cscosl.ncsu.edu (Richard Todd Wall) writes: > I was working on a 386 SX machine with 2 meg. of memory and > running Quarterdeck Expanded Memory Manager. The machine's memory was > set up with 640k base memory, 512k extended, and 512k expanded. When > we ran PC Lab's Benchmark test of the memory the rusult was that > the 640K memory was faster than the expanded, and the expanded was > faster than the extended. Does any one know why this is??? Or was > benchmark wrong???? The benchmark is wrong. If you read the documentation that comes with the PC Labs Benchmark 5.1, it says that the extended memory test was designed to work on 80286 machines, and thus uses the BIOS to switch to protected mode. This is a slow method and is in practice not used on 80386 machines, according to the documentation. Thus, they say not to rely on the extended memory test on 80386 machines. I can guess that the reason the expanded memory tested was faster was that your memory was being accessed by QEMM, which of course is specific to the 80386 and thus is much more efficient at switching to protected mode. -Sanjay Aiyagari (HD7X@vax5.cit.cornell.edu)