Intel 210200-002 manual MC6809

APPENDIX

Table 3. Execution Times with "Equal" Memory Access Times (5 MHz 88/10 vs 2 MHz 6809)
		Absolute Time		Normalized Time
Benchmark Program	iAPX 88110		MC6809*	iAPX88/10	MC6809
Computer Graphics	2.32 sec.		57.1 sec.		24.61
16-Bit Multiply	40.8	us	91.9 us		2.25
Vector Add	295.0	us	369.0 us		1.25
Block Move	328.0	us	763.0 us		2.33
Block Translate	1507.0	us	3016.0 us		2.00
Character Search	136.0	us	324.0 us		2.38
Word Shift	14.4	us	49.1 us		3.78
Reentrant Call	87.6	us	84.1 us		0.96
Single-Vectored Interrupt	102.6	us	30.1 us		0.29
Multi-Vectored Interrupt	24.6	us	55.3 us		2.25
Average Normalized Execution Time·"					4.21
Adjusted Average Normalized Execution Time··					2.15

*Times for the MC6809 include one wait state on memory accesses.

··See note, Table 2, for description of average calculations.

of the ten programs. The MC6809's Average Normal- ized Time of 4.17 greatly reflects (as it did in Table 2) the fact that the iAPX 88/10 outperformed the MC6809 by a large margin (more than 24 to 1) in the Computer Graphics benchmark. The Adjusted Average Normal- ized Time of 2.10 indicates that, after eliminating the Computer Graphics and Single-Vectored Interrupt, the iAPX 88/10 was more than twice as fast as the MC6809.

Table 4 compares the performance of the iAPX 88 and the MC6809 in terms of memory use, or coding efficien- cy. The results in this table show that the iAPX 88 used less code for nine of the ten programs. The two pro-

grams in which the largest performance differences oc- curred were the interrupt response benchmarks. The MC6809 won on the Single-Vectored Interrupt, largely due to the use of its IRQ interrupt which automatically stacks all the MC6809's registers. The iAPX 88/10 per- formed better for the Multi-Vectored Interrupt because its interrupt response requires no extra code to accom- modate multiple interrupt vectors. For the other pro- grams, the iAPX 88 provides significant advantages due to its string instructions and its efficient handling of 16-bit quantities. Tne Adjusted Average Normalized Number of Bytes shows the iAPX 88 with better than a 2 to 1 advantage over the MC6809 in coding efficiency.

	Table 4. Memory Utilization (Bytes)
	Bytes of Code		Normalized Bytes
Benchmark Program	iAPX 88/10	MC6809	iAPX 88/10	MC6809
Computer Graphics	40	180		4.50
16-Bit Multiply	14	56		4.00
Vector Add	18	21		1.17
Block Move	15	26		1.73
Block Translate	24	37		1.54
Character Search	18	19		1.06
Word Shift	6	18		3.00
Reentrant Call	48	49		1.02
Single-Vectored Interrupt	15	1		0.D7
Multi- Vectored Interrupt		15		15.00
Average Normalized Number of Bytes of Code"				3.31
Adjusted Average Normalized Number of Bytes of Code·				2.25

"'See note, Table 2, for description of average calculations.

25 AFN 01532A