Looptest output for SGI Octane, single CPU results.
Os:irix 6.4
>>uname -a
IRIX64 marr 6.4 02121744 IP30 mips
---- Hardware: 2 195 MHZ IP30 Processors CPU: MIPS R10000 Processor Chip Revision: 2.7 FPU: MIPS R10010 Floating Point Chip Revision: 0.0 Main memory size: 896 Mbytes Instruction cache size: 32 Kbytes Data cache size: 32 Kbytes Secondary unified instruction/data cache size: 1 Mbyte ---- Compiler: egcs snapshot-19980608 which seems to be in quite bad shape for SGI irix 6.4. The compiler works much better for e.g. sparc-sun-solaris2.5.1 Compiler flags: -O>>g++ -v Reading specs from /usr/cvap/petern/egcs/egcs-19980608/lib/gcc-lib/mips-sgi-irix6.4/egcs-2.91.40/specs gcc version egcs-2.91.40 19980608 (gcc2 ss-980502 experimental)
--------------- >>looptest This program measures the performance of DAXPY operations using various C loop structures.
In-cache: Mflops/s Description 76.049 for, indirection, unit stride 78.011 for, indirection, unit stride, no += 78.61 for, indirection, unit stride, backwards loops 120.42 for, unroll=4, unit stride, constants loaded into temps 122.9 for, unroll=4, unit stride, constants loaded into temps, 4 read then 4 write 120.46 for, unroll=4, unit stride, constants loaded into temps, no += 89.111 for, unroll=4, unit stride, constants loaded into temps, CSE for index offsets 122.85 for, unroll=4, unit stride, constants loaded into temps, backwards 123.31 for, unroll=8, unit stride, constants loaded into temps 74.136 for, indirection, unit stride, constants into temps 75.784 for, indirection, non-unit stride 75.747 for, indirection, non-unit stride, constants loaded into temps 79.971 while, pointer increment, unit stride 122.85 while, pointer increment, unit stride, constants loaded into temps 43.518 while, pointer increment, non-unit stride 81.343 while, pointer increment, unroll=4, non-unit stride, constants loaded into temps 91.14 for, unroll=4, unit stride, constants loaded into temps, prefetching 92.554 interlaced, for, indirection, unit stride 113.64 for, unroll=4, unit stride, interlaced, constants loaded into temps
Out of cache: Mflops/s Description 18.803 for, indirection, unit stride 18.812 for, indirection, unit stride, no += 20.643 for, indirection, unit stride, backwards loops 21.191 for, unroll=4, unit stride, constants loaded into temps 21.871 for, unroll=4, unit stride, constants loaded into temps, 4 read then 4 write 21.26 for, unroll=4, unit stride, constants loaded into temps, no += 19.906 for, unroll=4, unit stride, constants loaded into temps, CSE for index offsets 23.291 for, unroll=4, unit stride, constants loaded into temps, backwards 21.33 for, unroll=8, unit stride, constants loaded into temps 18.807 for, indirection, unit stride, constants into temps 18.812 for, indirection, non-unit stride 18.797 for, indirection, non-unit stride, constants loaded into temps 19.783 while, pointer increment, unit stride 21.682 while, pointer increment, unit stride, constants loaded into temps 18.263 while, pointer increment, non-unit stride 20.985 while, pointer increment, unroll=4, non-unit stride, constants loaded into temps 20.766 for, unroll=4, unit stride, constants loaded into temps, prefetching 12.813 interlaced, for, indirection, unit stride 14.411 for, unroll=4, unit stride, interlaced, constants loaded into temps --------------------- blitz-dev list -------------------------------- * To subscribe/unsubscribe: mail to majordomo@oonumerics.org, with "subscribe blitz-dev" or "unsubscribe blitz-dev" in the body of the message * Blitz++ web page: http://oonumerics.org/blitz/
This archive was generated by hypermail 2b29 : Wed Feb 20 2002 - 04:30:04 EST