gilgalad@caen.engin.umich.edu (Ralph Seguin) (11/16/90)
Linpack results
-----------------
Results of the LINPACK program on various architectures.
System Architecture KFLOPs/second
------- ------------------------- ---------------------
Amiga 2500/30 25 MHz 68030/68882, 205
4 megs 32 bit RAM,
1 meg 16 bit CHIP RAM,
40 meg disk.
Apollo DN3500 25 MHz 68030/68882, 79
8 megs 32 bit RAM,
3.2 gigabytes local
disk space.
Apollo DN4000 25 MHz 68020/68881, 70
8 megs 32 bit RAM,
660 meg disk.
Apollo DN10000 RISC proprietary, 2527
SV80 32 megs 32 bit RAM,
1.3 gigabytes local
disk space.
Stellar GS10000 RISC proprietary, 910
16 megs 32 bit RAM,
1.3 gigabytes local
disk space.
SPARCstation 1 16 MHz SPARC, 1023
8 megs 32 bit RAM,
660 meg disk.
RIOS 320 20 MHz POWER6000, 7305
8 megs 32 bit RAM,
660 meg disk.
DECstation 5000 25 MHz MIPS R2000, 2405
8 megs 32 bit RAM,
1.3 gigabytes
local disk space.
DECStation 3100 16.67 MHz MIPS R2000, 1266
8 megs 32 bit RAM,
660 and 200 meg
local disks.
Sun 4/280 16.67 MHz SPARC, 727
32 megs 32 bit RAM,
2.0 gigabytes
local disk space.
LINPACK should be seen as a much better benchmark of floating point
performance of a system (although, it is still not very good 8-). A GOOD
indicator is a SPECmark. LINPACK provides a better instruction mix
and actually does something. Remember, always take benchmark results
with a grain of salt. Many things contribute to the overall performance
of a system.
See ya, Ralph
Ralph Seguin gilgalad@dip.eecs.umich.edu
536 South Forest Apt. #915 gilgalad@caen.engin.umich.edu
Ann Arbor, MI 48104 (313) 662-4805rjtg0209@uxa.cso.uiuc.edu (11/17/90)
/* Written 1:28 am Nov 16, 1990 by gilgalad@caen.engin.umich.edu in uxa.cso.uiuc.edu:comp.sys.next */
/* ---------- "Linpack results" ---------- */
Linpack results
-----------------
Results of the LINPACK program on various architectures.
System Architecture KFLOPs/second
------- ------------------------- ---------------------
Amiga 2500/30 25 MHz 68030/68882, 205
4 megs 32 bit RAM,
1 meg 16 bit CHIP RAM,
40 meg disk.
Apollo DN3500 25 MHz 68030/68882, 79
8 megs 32 bit RAM,
3.2 gigabytes local
disk space.
Apollo DN4000 25 MHz 68020/68881, 70
8 megs 32 bit RAM,
660 meg disk.
Apollo DN10000 RISC proprietary, 2527
SV80 32 megs 32 bit RAM,
1.3 gigabytes local
disk space.
Stellar GS10000 RISC proprietary, 910
16 megs 32 bit RAM,
1.3 gigabytes local
disk space.
SPARCstation 1 16 MHz SPARC, 1023
8 megs 32 bit RAM,
660 meg disk.
RIOS 320 20 MHz POWER6000, 7305
8 megs 32 bit RAM,
660 meg disk.
DECstation 5000 25 MHz MIPS R2000, 2405
8 megs 32 bit RAM,
1.3 gigabytes
local disk space.
DECStation 3100 16.67 MHz MIPS R2000, 1266
8 megs 32 bit RAM,
660 and 200 meg
local disks.
Sun 4/280 16.67 MHz SPARC, 727
32 megs 32 bit RAM,
2.0 gigabytes
local disk space.
LINPACK should be seen as a much better benchmark of floating point
performance of a system (although, it is still not very good 8-). A GOOD
indicator is a SPECmark. LINPACK provides a better instruction mix
and actually does something. Remember, always take benchmark results
with a grain of salt. Many things contribute to the overall performance
of a system.
See ya, Ralph
Ralph Seguin gilgalad@dip.eecs.umich.edu
536 South Forest Apt. #915 gilgalad@caen.engin.umich.edu
Ann Arbor, MI 48104 (313) 662-4805
/* End of text from uxa.cso.uiuc.edu:comp.sys.next */