|
|
|
|
|
|
|
Reliability. Data seems consistent with earlier work of Smith [259]. but since it is not based upon direct measurements, reliability should be interpreted as low (best available data). |
|
|
|
|
|
|
|
|
Stability. Variance may be high. |
|
|
|
|
|
|
|
|
Data Note 6: Effect of Architecture Family (Figure 5.31). |
|
|
|
|
|
|
|
|
Instruction cache data is taken from Mitchell [200]. Data caches vary little across different architecture families. |
|
|
|
|
|
|
|
|
Data Note 7: Not-in-TLB Rate (Figure 5.46). |
|
|
|
|
|
|
|
|
Amdahl V/6, IBM 3081 and VAX 11/780 data are published/reported measurement data [259, 54]. Data for 512-byte page size are simulation results taken from Clark and Emer [54]. Data for 4K-byte page size are projected from DTMR and are included for reference. |
|
|
|
|
|
|
|
|
Use of the Data. Data can be used directly. |
|
|
|
|
|
|
|
|
Reliability. Low. Data presented are based on limited reports. It appears consistent with manufacturers' projections. |
|
|
|
|
|
|
|
|
Stability. Data are expected to be reasonably stable over various environments. |
|
|
|
 |
|
|
|
|
Alan J. Smith. Cache memories. Computing Surveys, 14(3):473530, September 1982. |
|
|
|
|
|
|
|
|
M. D. Hill. A case for direct-mapped cache. IEEE Computer, pages 2540, December 1988. |
|
|
|
|
|
|
|
|
J. E. Smith and J. R. Goodman. A study of instruction cache organizations and replacement policies. IEEE Transactions on Software Engineering, SE-9(3):132137, 1983. |
|
|
|
|
|
|
|
|
D. W. Clark. Cache performance in the VAX-11/780. ACM Transactions on Computing Systems, 1(1):2437, February 1983. |
|
|
|
|
|
|
|
|
A. Agarwal, J. Hennessy, and M. Horowitz. Cache performance of operating system and multiprogramming workloads. ACM Transactions on Computer Systems, 6(4):393431, November 1988. |
|
|
|
|
|
|
|
|
D. W. Clark and J. S. Emer. Performance of the VAX-11/780 translation buffer: simulation and measurement. ACM Transactions on Computing Systems, 3(1):3162, February 1985. |
|
|
|
|
|