How to execute SPEC95 benchmarks in Simplescalar?

Let us see how to run the SPEC95 benchmarks (little endian) namely compress95, anagram, go, cc1 and perl in Simplescalar simulator software. Each and every benchmark has a corresponding input file that specifies the input to the program and an output file which we can use to verify if our test was successful. Also we will be writing the output of our simulation to a file with a .out extension and we will be printing the execution trace with all the run statistics to a file with a .trace extension.

The little endian version of the benchmarks and the input and output files can be download from the below location. After extracting the archive,  place the folder BenchMarks_Little in the simplesim-3.0 folder of your simplescalar installation folder. All the benchmarks shall be executed from the simplesim-3.0 directory.

BenchMarks_Little.tar.gz

compress95:

The below command will execute the compress95 benchmark and print the output to compress95.out and log the execution trace to compress95.trace in the Results folder

./sim-outorder BenchMarks_Little/Programs/compress95.ss < BenchMarks_Little/Input/compress95.in  2> BenchMarks_Little/Results/compress95.trace > BenchMarks_Little/Results/compress95.out

go:

The below command will execute the go benchmark and print the output to go.out and log the execution trace to go.trace in the Results folder

./sim-outorder BenchMarks_Little/Programs/go.ss 50 9 BenchMarks_Little/Input/2stone9.in  2> BenchMarks_Little/Results/go.trace > BenchMarks_Little/Results/go.out

anagram:

Before executing the anagram program place the words file from the Input folder in the simplesim-3.0 directory. The below command will execute the anagram benchmark and print the output to anagram.out and log the execution trace to anagram.trace in the Results folder

./sim-outorder BenchMarks_Little/Programs/anagram.ss words < BenchMarks_Little/Input/anagram.in  2> BenchMarks_Little/Results/anagram.trace > BenchMarks_Little/Results/anagram.out

cc1:

The below command will execute the cc1 benchmark and print the output to 1stmt.s in the Programs folder and log the execution trace to cc1.trace in the Results folder

./sim-outorder BenchMarks_Little/Programs/cc1.ss -O BenchMarks_Little/Input/1stmt.i 2> BenchMarks_Little/Results/cc1.trace

perl:

Before executing the perl program place the perl-tests.pl file from the Input folder in the simplesim-3.0 directory. The below command will execute the perl benchmark and print the output to perl.out and log the execution trace to perl.trace in the Results folder

./sim-outorder BenchMarks_Little/Programs/perl.ss < perl-tests.pl 2> BenchMarks_Little/Results/perl.trace > BenchMarks_Little/Results/perl.out

References:

http://www.simplescalar.com/
http://www.igoy.in/simplescalar-installation-made-simple/
http://users.ece.gatech.edu/~hamblen/4100/course/simplescalar/Spec95%20Benchmark%
20Command%20Lines.htm
http://www.spec.org/osg/cpu95/

The MegaPixel (MP) Rating of a Digital Camera

The resolution of a digital camera is perceived to be proportional to its pixel count. A digital image is nothing but a continuous arrangement of pixels or picture elements. Resolution is expressed as the number of horizontal pixels multiplied by the number of vertical pixels (E.g. 1600x1200 => 1.92 MP). Digital cameras record visual image with the help of CCD or CMOS sensors and each pixel of the output image on screen will map to a specific pixel on the image sensor in the digital camera.

Digital Camera Resolution Formula MegapixelDigital Camera CCD CMOS PixelMegapixel Resolution Myth

In order to improve the resolution, one tends to simply increase the pixel count. The more the number of pixels the more details the image will provide and the more it will resemble to the real world image. But for a given image or sensor area, increasing the number of pixels tends to saturate at a certain point and further increase in pixels doesn't mean much to the human eye. A common man will not be able to accurately identify the difference in quality of a usual 10cm x 15cm photograph taken by digital cameras having resolution greater than 5 MP. But higher pixel count will be beneficial only when you are trying to generate a bigger picture from a smaller one or when you are trying to crop a particular portion of the picture and try to enlarge it.

There are some common misinterpretations of resolution which a camera manufacturer might try to exploit. Since resolution is more like a square function (length x height), gain in pixel count will not be directly related to resolution gain. Also some pixels on a digital camera sensor will be used for other compensation mechanisms and not really for capturing the image. Companies may include this overhead in order to bump up the pixel rating. In color displays such as LCD screens, subpixel rendering is used in which a single pixel will be composed of a three sub pixels (R, G & B) and so the pixel count should never be exaggerated. The size of the camera sensor also plays a part in deciding resolution, sharpness and image quality for digital cameras having the same pixel rating apart from other factors like lens quality, sensitivity of the image sensors, optical zoom capability, shutter speed, etc.