Claims Analysis
Total Claims: 3
Claim 1
Independent Claim
440 words
In a data processing system, having a system bus; and a main memory, a cache memory, a central processing unit (CPU) and a communications controller all connected in parallel to said system bus, said controller operating to supervise interconnection between said units via siad bus for transferring data therebetween, said CPU including memory request means for generating data requests, and said cache memory including private interface means connecting said cache memory directly to said CPU for permitting direct transmission of data requests from said CPU to said cache memory and direct transmission of requested data from said cache memory to said CPU; cache directory and data buffer means for evaluating said data requests and generating a first predetermined output when the requested data is not present in said cache memory; and system bus interface means connecting said cache memory to said system bus for obtaining CPU requested data not found in said cache memory from said main memory via said system bus in response to said first predetermined output from said cache directory and data buffer means; the improvement comprising: selection means in said main memory responsive to a system operator for controlling the configuration of said main memory, said main memory being partitioned in modules, said modules having data stored therein in sequentially numbered addresses to constitute a banked storage mode, or said modules being arranged in pairs of modules, the first modules in each said pairs having even numbered addresses and the second modules in each of said pairs having odd numbered addresses, said pairs of modules having successive elements of said data stored sequentially therein in said even and odd numbered addresses, alternating between said modules to constitute an interleaved storage mode; read request means included in said system bus interface means, responsive to said first predetermined signal and said selection means, for obtaining a word containing the data requested by said CPU either from a specific address in one of said modules in said banked storage mode, or for obtaining two words containing data requested by said CPU from two successive adjacent addresses in two of said interleaved mode modules, one word being obtained from the first module of one of said pairs and the second of said words being obtained from the second module in said one of said pairs; said cache memory generating a second predetermined output; and said read request means also including banked double pull means responsive to said second predetermined output from said cache memory, and said selection means, for obtaining two words containing CPU requested data from two successive addresses in one of said modules in said banked storage mode.
Claim 2
Dependent Claim
440 words
In a data processing system, having a system bus; and a main memory, a cache memory, a central processing unit (CPU) and a communications controller all connected in parallel to said system bus, said controller operating to supervise interconnection between said units via siad bus for transferring data therebetween, said CPU including memory request means for generating data requests, and said cache memory including private interface means connecting said cache memory directly to said CPU for permitting direct transmission of data requests from said CPU to said cache memory and direct transmission of requested data from said cache memory to said CPU; cache directory and data buffer means for evaluating said data requests and generating a first predetermined output when the requested data is not present in said cache memory; and system bus interface means connecting said cache memory to said system bus for obtaining CPU requested data not found in said cache memory from said main memory via said system bus in response to said first predetermined output from said cache directory and data buffer means; the improvement comprising: selection means in said main memory responsive to a system operator for controlling the configuration of said main memory, said main memory being partitioned in modules, said modules having data stored therein in sequentially numbered addresses to constitute a banked storage mode, or said modules being arranged in pairs of modules, the first modules in each said pairs having even numbered addresses and the second modules in each of said pairs having odd numbered addresses, said pairs of modules having successive elements of said data stored sequentially therein in said even and odd numbered addresses, alternating between said modules to constitute an interleaved storage mode; read request means included in said system bus interface means, responsive to said first predetermined signal and said selection means, for obtaining a word containing the data requested by said CPU either from a specific address in one of said modules in said banked storage mode, or for obtaining two words containing data requested by said CPU from two successive adjacent addresses in two of said interleaved mode modules, one word being obtained from the first module of one of said pairs and the second of said words being obtained from the second module in said one of said pairs; said cache memory generating a second predetermined output; and said read request means also including banked double pull means responsive to said second predetermined output from said cache memory, and said selection means, for obtaining two words containing CPU requested data from two successive addresses in one of said modules in said banked storage mode.
Claim 3
Dependent Claim
89 words
The data processing system of claim 1 wherein said read request means also includes interleaved double pull means responsive to said second predetermined output from said cache memory and said selection means for obtaining four words containing CPU requested data from two adjacent addresses in two of said interleaved mode modules, two words being obtained from two successive addresses in the first of one of said pairs of modules and two words being obtained from two successive addresses in the second of said one of said pairs of modules.