Hi.
A table from 'TI Competitive Benchmark': slaa205b.pdf (July 2006).
Whetstone Code Size (in bytes) and Cycle Counts.
IAR C Compiler used (4.12A for AVR), full optimization.
MCU Code size Cycles
------------ --------- ------
MSP430FG4619 6544 107040
MSP430F149 6524 109837
PIC24FJ128GA 4605 108619
8051 8723 291836
H8/300H 4656 205910
MaxQ20 5392 158945
ARM7TDMI/Thumb 10532 60444
HCS12 7370 787635
ATmega8 4694 270991
avr-gcc 4.1.2 & avr-libc CVS HEAD (Jan 2008) gives:
MCU Code size Cycles Optimization
------------ --------- ------ ------------
ATmega8 4020 88563 -Os -morder1
ATmega8 4030 88565 -Os
ATmega8 5310 71743 -O3 -morder1
ATmega8 5332 71751 -O3
The dummy functions are added to prevent GCC optimization.
I am not sure in TI's accuracy. Results for both MSP430-es
are practicaly equal, despite the fact thar FG4619 has
considerable enhanced core. Also run time for ATmega8 is
too big. Can everbody to make any measurements with
IAR compiler?
Source in attachment.
Regards,
Dmitry.
/***************************************************************************
*
* Name : Whetstone
* Purpose : Benchmark the Whetstone code. The code focuses on scientific
* functions such as sine, cosine, exponents and logarithm on
* fixed and floating point numbers.
*
***************************************************************************/
/* List of changes (since TI slaa205b.pdf):
- void type of PA(), P0(), P3()
- dummy functions to prevent optimization
- float (not double) variants of math funcions and constants (this
is no matter for AVR)
*/
#include <math.h>
#ifdef __AVR__
# define atanf(x) atan(x)
# define cosf(x) cos(x)
# define expf(x) exp(x)
# define logf(x) log(x)
# define sinf(x) sin(x)
# define sqrtf(x) sqrt(x)
#endif
/* Extern functions. To prevent compiler's optimization. */
void dummy1 (float);
void dummy2 (float, float);
void PA (float E[5]);
void P0 (void);
void P3 (float *X, float *Y, float *Z);
float T, T1, T2, E1[5];
int J, K, L;
float X1, X2, X3, X4;
long ptime, time0;
void main (void)
{
int LOOP, I, II, JJ, N1, N2, N3, N4, N5, N6, N7, N8, N9, N10, N11;
float X, Y, Z;
T = .499975;
T1 = 0.50025;
T2 = 2.0;
LOOP = 1;
II = 1;
for (JJ = 1; JJ <= II; JJ++) {
N1 = 0;
N2 = 2 * LOOP;
N3 = 2 * LOOP;
N4 = 2 * LOOP;
N5 = 0;
N6 = 2 * LOOP;
N7 = 2 * LOOP;
N8 = 2 * LOOP;
N9 = 2 * LOOP;
N10 = 0;
N11 = 2 * LOOP;
/* Module 1: Simple identifiers */
X1 = 1.0;
X2 = -1.0;
X3 = -1.0;
X4 = -1.0;
if (N1 != 0) {
for (I = 1; I <= N1; I++) {
X1 = (X1 + X2 + X3 - X4) * T;
X2 = (X1 + X2 - X3 + X4) * T;
X3 = (X1 - X2 + X3 + X4) * T;
X4 = (-X1 + X2 + X3 + X4) * T;
}
}
/* Module 2: Array elements */
E1[1] = 1.0;
E1[2] = -1.0;
E1[3] = -1.0;
E1[4] = -1.0;
if (N2 != 0) {
for (I = 1; I <= N2; I++) {
E1[1] = (E1[1] + E1[2] + E1[3] - E1[4]) * T;
E1[2] = (E1[1] + E1[2] - E1[3] + E1[4]) * T;
E1[3] = (E1[1] - E1[2] + E1[3] + E1[4]) * T;
E1[4] = (-E1[1] + E1[2] + E1[3] + E1[4]) * T;
}
}
/* Module 3: Array as parameter */
if (N3 != 0) {
for (I = 1; I <= N3; I++) {
PA (E1);
}
}
/* Module 4: Conditional jumps */
J = 1;
if (N4 != 0) {
for (I = 1; I <= N4; I++) {
if (J == 1)
goto L51;
J = 3;
goto L52;
L51:
J = 2;
L52:
if (J > 2)
goto L53;
J = 1;
goto L54;
L53:
J = 0;
L54:
if (J < 1)
goto L55;
J = 0;
goto L60;
L55:
J = 1;
L60: ;
}
}
/* Module 5: Integer arithmetic */
J = 1;
K = 2;
L = 3;
if (N6 != 0) { /* ??? Where is N5? */
for (I = 1; I <= N6; I++) {
J = J * (K - J) * (L - K);
K = L * K - (L - J) * K;
L = (L - K) * (K + J);
E1[L - 1] = J + K + L;
E1[K - 1] = J * K * L;
}
}
/* Module 6: Trigonometric functions */
X = 0.5;
Y = 0.5;
if (N7 != 0) {
for (I = 1; I <= N7; I++) {
X = T * atanf (T2 * sinf (X) * cosf (X) /
(cosf (X + Y) + cosf (X - Y) - 1.0f));
Y = T * atanf (T2 * sinf (Y) * cosf (Y) /
(cosf (X + Y) + cosf (X - Y) - 1.0f));
}
}
dummy2 (X, Y);
/* Module 7: Procedure calls */
X = 1.0;
Y = 1.0;
Z = 1.0;
if (N8 != 0) {
for (I = 1; I <= N8; I++) {
P3 (&X, &Y, &Z);
}
}
/* Module 8: Array references */
J = 1;
K = 2;
L = 3;
E1[1] = 1.0;
E1[2] = 2.0;
E1[3] = 3.0;
if (N9 != 0) {
for (I = 1; I <= N9; I++) {
P0 ();
}
}
/* Module 9: Integer arithmetic */
J = 2;
K = 3;
if (N10 != 0) { /* TI skips this strange test. */
for (I = 1; I <= N10; I++) {
J = J + K;
K = J + K;
J = K - J;
K = K - J - J;
}
}
/* Module 10: Standard functions */
X = 0.75;
if (N11 != 0) {
for (I = 1; I <= N11; I++) {
X = sqrtf (expf (logf (X) / T1));
}
}
dummy1 (X);
}
}
void PA (float E[5])
{
int J1;
J1 = 0;
L10:
E[1] = (E[1] + E[2] + E[3] - E[4]) * T;
E[2] = (E[1] + E[2] - E[3] + E[4]) * T;
E[3] = (E[1] - E[2] + E[3] + E[4]) * T;
E[4] = (-E[1] + E[2] + E[3] + E[4]) / T2;
J1 = J1 + 1;
if ((J1 - 6) < 0)
goto L10;
return;
}
void P0 ()
{
E1[J] = E1[K];
E1[K] = E1[L];
E1[L] = E1[J];
return;
}
void P3 (float *X, float *Y, float *Z)
{
float Y1;
X1 = *X;
Y1 = *Y;
X1 = T * (X1 + Y1);
Y1 = T * (X1 + Y1);
*Z = (X1 + Y1) / T2;
return;
}
_______________________________________________
AVR-GCC-list mailing list
AVR-GCC-list@nongnu.org
http://lists.nongnu.org/mailman/listinfo/avr-gcc-list
