Spec2K

From Msim

Jump to: navigation, search

Spec2K is a standardized benchmark testing suite. Here we list all of them and any issues related to their program output. As far as we know, these were compiled using "peak" as described (sort of) here

Contents

Errors

FORRT1

The FORRT1 error is as follows:
forrtl: info: Fortran error message number is 29.
forrtl: warning: Could not open message catalog: for_msg.cat.
forrtl: info: Check environment variable NLSPATH and protection of /usr/lib/nls/msg/en_US.ISO8859-1/for_msg.cat.
forrtl: severe (29): Message not found

This generally occurs because the benchmark failed to get a current working directory due to permissions disabling the read of the current working directory's root directories.

AMMP

ammpNS.1.arg:10800000000 # ./ammp/ammp00.peak.ev6 < ./ammp/ammp.in > ./ammp/ammp.1.out
Full Run: Successful (326547695706 insts)
Reference Output

APPLU

appluNS.1.arg:217900000000 # ./applu/applu00.peak.ev6 < ./applu/applu.in > ./applu/applu.1.out
Full Run: Successful (223883652286 insts)
Reference Output

APSI

apsiNS.1.arg:340800000000 # ./apsi/apsi00.peak.ev6 ./apsi/apsi.in > ./apsi/apsi.1.out
Full Run: Successful (347925355195 insts)
Reference Output

ART

artNS.1.arg:34000000000 # ./art/art00.peak.ev6 -scanfile c756hel.in -trainfile1 a10.img -trainfile2 hc.img -stride 2 -startx 110 -starty 200 -endx 160 -endy 240 -objects 10 > ./art/art.1.out
artNS.2.arg:34000000000 # ./art/art00.peak.ev6 -scanfile c756hel.in -trainfile1 a10.img -trainfile2 hc.img -stride 2 -startx 470 -starty 140 -endx 520 -endy 180 -objects 10 > ./art/art.2.out
Full Run 1: Successful (41795384307 insts)
Reference Output 1
Full Run 2: Successful (45036147233 insts)
Reference Output 2

BZIP2

bzip2NS.1.arg:45800000000 # ./bzip2/bzip200.peak.ev6 ./bzip2/input.source 58 > ./bzip2/bzip2.1.out
bzip2NS.2.arg:45800000000 # ./bzip2/bzip200.peak.ev6 ./bzip2/input.graphic 58 > ./bzip2/bzip2.2.out
bzip2NS.3.arg:45800000000 # ./bzip2/bzip200.peak.ev6 ./bzip2/input.program 58 > ./bzip2/bzip2.3.out
Full Run 1: Successful (108878091794 insts)
Reference Output 1
Full Run 2: Successful (143565170088 insts)
Reference Output 2
Full Run 3: Successful (124927058156 insts)
Reference Output 3

CRAFTY

craftyNS.1.arg:77400000000 # ./crafty/crafty00.peak.ev6 < ./crafty/crafty.in > ./crafty/crafty.1.out
Full Run: Successful (191882992596 insts)
Reference Output

EON

eonNS.1.arg:40300000000 # ./eon/eon00.peak.ev6 ./chair.control.cook ./chair.camera ./chair.surfaces ./chair.cook.ppm ./eon/ppm ./pixels_out.cook > ./eon/eon.1.out
eonNS.2.arg:40300000000 # ./eon/eon00.peak.ev6 ./chair.control.rushmeier ./chair.camera ./chair.surfaces ./chair.rushmeier.ppm ./eon/ppm ./pixels_out.rushmeier > ./eon/eon.2.out
eonNS.3.arg:40300000000 # ./eon/eon00.peak.ev6 ./chair.control.kajiya ./chair.camera ./chair.surfaces ./chair.kajiya.ppm ./eon/ppm ./pixels_out.kajiya > ./eon/eon.3.out
pixels_out not generated but .ppm file is valid for all three methods
Full Run 1: Successful (80592509966 insts)
Reference Output 1
Full Run 2: Successful (57849044927 insts)
Reference Output 2
Full Run 3: Successful (101260087364 insts)
Reference Output 3

EQUAKE

equakeNS.1.arg:81200000000 # ./equake/equake00.peak.ev6 < ./equake/inp.in > ./equake/equake.1.out
Full Run: Successful (131518588096 insts)
Reference Output

FACEREC

facerecNS.1.arg:37500000000 # ./facerec/facerec00.peak.ev6 < ./facerec/ref.in > ./facerec/facerec.1.out
Full Run: No, output varies. (211030287982 insts)
Reference Output
Note: The binary may be a problem (old size: 1228800). We recompiled with on Tru64 (new size: 2188672):

3x times:
f90 -O4 -non_shared *.f90 encode decode_256 -c
f90 -O4 -non_shared *.o -o facerec

Removing "-fast" appears to fix this issue (perhaps because of "-assume noaccuracy_sensitive"?)

FMA3D

fma3dNS.1.arg:254100000000 # ./fma3d/fma3d00.peak.ev6 < ./fma3d/fma3d.in > ./fma3d/fma3d.1.out
The reference data doesn't include the "fmaelo" output.
Standard Run: Successful (268370754732 insts)
Reference Output

GALGEL

galgelNS.1.arg:249100000000 # ./galgel/galgel00.peak.ev6 < ./galgel/galgel.in > ./galgel/galgel.1.out
Diff:

162c162
<             Newton iteration #  1    Maximal derivative = 0.2453E-04
---
>             Newton iteration #  1    Maximal derivative = 0.2454E-04
186c186
<             Newton iteration #  0    Maximal derivative = 0.1194E-06
---
>             Newton iteration #  0    Maximal derivative = 0.1193E-06
190c190
<           Eigenvalue:       Re(Lambda)= 0.4988E-09     Im(Lambda)=  92.75
---
>           Eigenvalue:       Re(Lambda)= 0.6332E-09     Im(Lambda)=  92.75

Full Run: Successful (409354922233 insts)
Reference Output

GAP

gapNS.1.arg:67400000000 # ./gap/gap00.peak.ev6 -l ./gap/ -q -m 192M < ./gap/ref.in > ./gap/gap.1.out
Full Run: Successful (269035884280 insts)
Reference Output

GCC

gccNS.1.arg:400000000 # ./gcc/gcc00.peak.ev6 ./gcc/166.i -o ./gcc/166.S > ./gcc/gcc.1.out
gccNS.2.arg:400000000 # ./gcc/gcc00.peak.ev6 ./gcc/200.i -o ./gcc/200.S > ./gcc/gcc.2.out
gccNS.3.arg:400000000 # ./gcc/gcc00.peak.ev6 ./gcc/expr.i -o ./gcc/expr.S > ./gcc/gcc.3.out
gccNS.4.arg:400000000 # ./gcc/gcc00.peak.ev6 ./gcc/integrate.i -o ./gcc/integrate.S > ./gcc/gcc.out
gccNS.5.arg:400000000 # ./gcc/gcc00.peak.ev6 ./gcc/scilab.i -o ./gcc/scilab.S > ./gcc/gcc.5.out
Full Run 1: Successful (46917925304 insts)
Reference Output 1
Full Run 2: Successful (108628654285 insts)
Reference Output 2
Full Run 3: Successful (12086091942 insts)
Reference Output 3
Full Run 4: Successful (13163437164 insts)
Reference Output 4
Full Run 5: Successful (62037081002 insts)
Reference Output 5

GZIP

gzipNS.1.arg:26500000000 # ./gzip/gzip00.peak.ev6 ./gzip/input.source 60 > ./gzip/gzip.source.out
gzipNS.2.arg:26500000000 # ./gzip/gzip00.peak.ev6 ./gzip/input.log 60 > ./gzip/gzip.log.out
gzipNS.3.arg:26500000000 # ./gzip/gzip00.peak.ev6 ./gzip/input.graphic 60 > ./gzip/gzip.graphic.out
gzipNS.4.arg:26500000000 # ./gzip/gzip00.peak.ev6 ./gzip/input.random 60 > ./gzip/gzip.random.out
gzipNS.5.arg:26500000000 # ./gzip/gzip00.peak.ev6 ./gzip/input.program 60 > ./gzip/gzip.program.out
input.graphic is processed in smaller chunks but with the same result
Full Run 1: Successful (84367396325 insts)
Reference Output 1
Full Run 2: Successful (39526723706 insts)
Reference Output 2
Full Run 3: Successful (103706490377 insts)
Reference Output 3
Full Run 4: Successful (82167474921 insts)
Reference Output 4
Full Run 5: Successful (168868067396 insts)
Reference Output 5

LUCAS

lucasNS.1.arg:54500000000 # ./lucas/lucas00.peak.ev6 < ./lucas/lucas2.in > ./lucas/lucas.1.out
Full Run: Successful (142398811927 insts)
Reference Output

MCF

mcfNS.1.arg:54300000000 # ./mcf/mcf00.peak.ev6 ./mcf/inp.in > ./mcf/mcf.1.out
Full Run: Successful (61867331832 insts)
Reference Output

MESA

mesaNS.1.arg:113500000000 # ./mesa/mesa00.peak.ev6 -frames 1000 -meshfile ./mesa/mesa.in -ppmfile ./mesa/mesa.ppm > ./mesa/mesa.1.out
Full Run: Successful (281694636357 insts)
Reference Output

MGRID

mgridNS.1.arg:329200000000 # ./mgrid/mgrid00.peak.ev6 < ./mgrid/mgrid.in > ./mgrid/mgrid.1.out
Full Run: Successful (419156005761 insts)
Reference Output

PARSER

parserNS.1.arg:114600000000 # ./parser/parser00.peak.ev6 ./parser/2.1.dict -batch < ./parser/ref.in > ./parser/parser.1.out
Full Run: Successful (546749970243 insts)
Reference Output

PERLBMK

perlbmkNS.1.arg:500000000 # ./perlbmk/perlbmk00.peak.ev6 -I./perlbmk/lib ./perlbmk/diffmail.pl 2 550 15 24 23 100 > ./perlbmk/perlbmk.1.out
perlbmkNS.2.arg:500000000 # ./perlbmk/perlbmk00.peak.ev6 -I./perlbmk/lib ./perlbmk/makerand.pl > ./perlbmk/perlbmk.2.out
perlbmkNS.3.arg:500000000 # ./perlbmk/perlbmk00.peak.ev6 -I./perlbmk/lib ./perlbmk/perfect.pl b 3 m 4 > ./perlbmk/perlbmk.3.out
perlbmkNS.4.arg:500000000 # ./perlbmk/perlbmk00.peak.ev6 -I./perlbmk/lib ./perlbmk/splitmail.pl 850 5 19 18 1500 > ./perlbmk/perlbmk.4.out
perlbmkNS.5.arg:500000000 # ./perlbmk/perlbmk00.peak.ev6 -I./perlbmk/lib ./perlbmk/splitmail.pl 704 12 26 16 836 > ./perlbmk/perlbmk.5.out
perlbmkNS.6.arg:500000000 # ./perlbmk/perlbmk00.peak.ev6 -I./perlbmk/lib ./perlbmk/splitmail.pl 535 13 25 24 1091 > ./perlbmk/perlbmk.6.out
perlbmkNS.7.arg:500000000 # ./perlbmk/perlbmk00.peak.ev6 -I./perlbmk/lib ./perlbmk/splitmail.pl 957 12 23 26 1014 > ./perlbmk/perlbmk.7.out
Full Run 1: Successful (39921703305 insts)
Reference Output 1
Full Run 2: Successful (2009018368 insts)
Reference Output 2
Full Run 3: Successful (BigInt code needs correction (29467522470 insts)
Reference Output 3
Full Run 4: Successful (127492988438 insts)
Reference Output 4
Full Run 5: Successful (66746519637 insts)
Reference Output 5
Full Run 6: Successful (62697168490 insts)
Reference Output 6
Full Run 7: Successful (110833983783 insts)
Reference Output 7

SIXTRACK

sixtrackNS.1.arg:304300000000 # ./sixtrack/sixtrack00.peak.ev6 < ./sixtrack/inp.in > ./sixtrack/sixtrack.1.out
Differences appear to be floating point related but they propagate quite a bit
Full Run: Uncertain (470950198586 insts)
Reference Output

SWIM

swimNS.1.arg:207900000000 # ./swim/swim00.peak.ev6 < ./swim/swim.in > ./swim/swim.1.out
Full Run: Successful (225830996336 insts)
Reference Output

TWOLF

twolfNS.1.arg:106600000000 # ./twolf/twolf00.peak.ev6 ./twolf/ref > ./twolf/twolf.1.out
Rename was not supported but only affected intermediate results.
Full Run: Successful (346484470084 insts)
Reference Output

VORTEX

vortexNS.1.arg:27100000000 # ./vortex/vortex00.peak.ev6 ./vortex/lendian1.raw > ./vortex/vortex.1.out
vortexNS.2.arg:27100000000 # ./vortex/vortex00.peak.ev6 ./vortex/lendian2.raw > ./vortex/vortex.2.out
vortexNS.3.arg:27100000000 # ./vortex/vortex00.peak.ev6 ./vortex/lendian3.raw > ./vortex/vortex.3.out
Output is not sent to stdout, instead, directly to vortex1, vortex2, vortex3.
Full Run 1: Successful (118971693858 insts)
Reference Output 1
Full Run 2: Successful (138672082687 insts)
Reference Output 2
Full Run 3: Successful (133044304523 insts)
Reference Output 3

VPR

vprNS.1.arg:47600000000 # ./vpr/vpr00.peak.ev6 ./vpr/net.in ./vpr/arch.in ./vpr/place.out ./vpr/dum.out -nodisp -place_only -init_t 5 -exit_t 0.005 -alpha_t 0.9412 -inner_num 2 > ./vpr/vpr.1.out
vprNS.2.arg:47600000000 # ./vpr/vpr00.peak.ev6 ./vpr/net.in ./vpr/arch.in ./vpr/place.in ./vpr/route.out -nodisp -route_only -route_chan_width 15 -pres_fac_mult 2 -acc_fac 1 first_iter_pres_fac 4 -initial_pres_fac 8 > ./vpr/vpr.2.out
Route (2nd): Diff:

265,266c265,266
< Assuming no buffer sharing (pessimistic). Total: 1.67424e+07  Per clb: 1978.07
< Assuming buffer sharing (slightly optimistic). Total: 1.35633e+07  Per clb: 1602.47
---
> Assuming no buffer sharing (pessimistic). Total: 1.67447e+07  Per clb: 1978.34
> Assuming buffer sharing (slightly optimistic). Total: 1.35651e+07  Per clb: 1602.68

Full Run 1: Success (112263684960 insts)
Reference Output 1
Full Run 2: Success (84068615835 insts)
Reference Output 2

WUPWISE

wupwiseNS.1.arg:323700000000 # ./wupwise/wupwise00.peak.ev6 < ./wupwise/wupwise.in > ./wupwise/wupwise.1.out
Precision is different but within tolerance, uses "D" notation instead of "E"
Standard Run: Successful (349623906323 insts)
Reference Output

Personal tools