System and method for assigning an engine measure metric to a computing system
The invention includes methods for deriving a computer system processing capacity metric for characterizing a computer system comprising at least one computer processor having a known processor clock speed, comprising the step of performing a computational function wherein at least one term of the function is determined by a value X derived at least in part by raising said clock speed to a non-integral exponential power. Processing capacity may be thus calculated by mathematical analysis alone, based on a limited number of inputs regarding a real or theoretical computer system combination, in a fashion that replicates accurately, and can replace, actual physical testing or benchmarkings of systems.
1. A method for deriving a computer system processing capacity metric for characterizing a computer system comprising at least one computer processor having a known processor clock speed, comprising the step of performing a computational function wherein at least one term of the function is determined by a value X derived at least in part by raising said clock speed to a non-integral exponential power.
2. The method of claim 1 , wherein said computational function is performed to yield the metric analytically without the need for actual physical testing of the system in operation.
3. The method of claim 2 , wherein said function is derived using a curve-fitting procedure based upon at least one set of empirically-derived data points from physical testing of at least one set of system configurations, whereby said function is operable for prediction of capacity metric data of additional system configurations without requirement of physical testing of the additional system configurations.
4. The method of claim 2 , wherein X is further derived at least in part from the number of central processing units in the system.
5. The method of claim 2 , wherein X is further derived at least in part from an architectural class value for the system.
6. The method of claim 2 , wherein the system capacity metric may be determined using only the system data of: (i) number of central processing units; (ii) clock speed of each central processing unit; and (iii) architecture class of the system.
7. The method of claim 2 , wherein the value of said function is derived using a computer spreadsheet program operable to calculate, display, and store system metric data on a plurality of system configurations.
8. The method of claim 2 , wherein the metric is derived according to the function:
f ( X )= aE−bX 3 −cE−dX 2 +gX−h,
wherein
X =[(number of system central processing units*(clock speed of each processor) n )+(number of system central processing units*(bit architecture number/ p ))],
wherein a, b, c, d, g, and h, and p are positive real numbers, and wherein n is a real number less than 1 and greater than 0.7.
9. The method of claim 8 , wherein:
a is between 1 and 10 inclusive;
b is between 10 and 12 inclusive;
c is between 1 and 4 inclusive;
d is between 5 and 7 inclusive;
g is between 0.35 and 0.5 inclusive;
h is between 0 and 20 inclusive; and
p is between 1 and 16 inclusive.
10. The method of claim 9 , wherein:
f ( X )=2 E− 11 X 3 −3 E −6 X 2 +0.4519 X −18.471,
and wherein
n=0.9979.
11. An apparatus for deriving a computer system processing capacity metric for characterizing a computer system comprising at least one computer processor having a known processor clock speed, the apparatus comprising an analyzer programmed to perform a computational function wherein at least one term of the function is determined by a value X derived at least in part by raising said clock speed to a non-integral exponential power.
12. The apparatus of claim 11 , wherein said computational function is performed to yield the metric analytically without the need for actual physical testing of the system in operation.
13. The apparatus of claim 12 , wherein said function is derived using a curve-fitting procedure based upon at least one set of empirically-derived data points from physical testing of at least one set of system configurations, whereby said function is operable for prediction of capacity metric data of additional system configurations without requirement of physical testing of the additional system configurations.
14. The apparatus of claim 12 , wherein X is further derived at least in part from the number of central processing units in the system.
15. The system of claim 12 , wherein X is further derived at least in part from an architectural class value for the system.
16. The system of claim 12 , wherein the system capacity metric may be determined using only the system data of: (i) number of central processing units; (ii) clock speed of each central processing unit; and (iii) architecture class of the system.
17. The system of claim 12 , wherein the value of said function is derived using a computer spreadsheet program for the analyzer operable to calculate, display, and store system metric data on a plurality of system configurations.
18. The system of claim 12 , wherein the metric is derived by the analyzer according to the function:
f ( X )= aE−bX 3 −cE−dX 2 +gX−h,
wherein
X =[(number of system central processing units*(clock speed of each processor) n )+(number of system central processing units*(bit architecture number/ p ))],
wherein a, b, c, d, g, and h, and p are positive real numbers, and wherein n is a real number less than 1 and greater than 0.7.
19. The system of claim 18 , wherein:
a is between 1 and 10 inclusive;
b is between 10 and 12 inclusive;
c is between 1 and 4 inclusive;
d is between 5 and 7 inclusive;
g is between 0.35 and 0.5 inclusive;
h is between 0 and 20 inclusive; and
p is between 1 and 16 inclusive.
20. The system of claim 19 , wherein:
f ( X )=2 E −11 X 3 −3 E −6 X 2 +0.4519 X −18.471,
and wherein
n=0.9979.
21. A method for estimating comparative performance characteristics of a plurality of computer systems having varying respective components and component operational traits without the need for physical testing of the systems or their respective components, the method comprising providing by mathematical analysis alone, by computation of a function, a system processing-operation-per-time capacity value for each of the plurality of systems, wherein said function value for each system derives at least in part from an exponential product of the clock speed of a processor for the system, and wherein the exponent for said product is greater than 0.9 but less than 1, and wherein performance may be estimated without knowledge of any programming application-specific parameters by comparison of said derived processing-operation-per-time capacity value of each respective system.
22. A computer program for deriving a computer system processing capacity metric for characterizing a computer system comprising at least one computer processor having a known processor clock speed, said program comprising programming instructions operable for performing a computational function wherein at least one term of the function is determined by a value X derived at least in part by raising said clock speed to a non-integral exponential power.