# US 2015/0149718 A1 · A lower energy consumption and high speed computer system and a marching main memory adapted for the computer system, without the memory bottleneck

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2015/0149718 A1
- Document ID: US20150149718A1
- Status: Published application
- Kind: A1
- Application number: 14/552,572
- Filed: 2014-11-25
- Published: 2015-05-28
- Earliest claimed: 2009-07-21
- Inventors: Tadao NAKAMURA; Michael J. Flynn
- Web page: https://patentgrep.com/patent/US20150149718A1

## CPC

- G11C 7/1072: {for memories with random access ports synchronised on clock signal pulse trains, e.g. synchronous memories, self timed memories}
- G06F 5/065: {Partitioned buffers, e.g. allowing multiple independent queues, bidirectional FIFO's}
- G06F 5/08: having a sequence of storage locations, the intermediate ones not being accessible for either enqueue or dequeue operations, e.g. using a shift register {(G06F5/065 takes precedence; shift registers per se G11C19/00)}
- G06F 9/30134: {Register stacks; shift registers}
- G06F 9/3824: {Operand accessing}
- G06F 9/3889: {controlled by multiple instructions, e.g. MIMD, decoupled access or execute}
- G06F 12/0875: with dedicated cache, e.g. instruction or stack
- G06F 15/7821: {Tightly coupled to memory, e.g. computational memory, smart memory, processor in memory}
- G11C 19/184: {with field-effect transistors, e.g. MOS-FET}
- G06F 1/08: Clock generators with changeable or programmable clock frequency
- G06F 7/57: Arithmetic logic units [ALU], i.e. arrangements or devices for performing two or more of the operations covered by groups G06F7/483 – G06F7/556 or for performing logical operations {(G06F7/49, G06F7/491 take precedence)}
- Y02D 10/00: Energy efficient computing, e.g. low power processors, power management or thermal management
- G11C: STATIC STORES (semiconductor memory devices H10B)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y02D: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE

## Abstract

A computer system encompasses a processor (11) including a control unit (111) and an ALU (112) configured to execute arithmetic and logic operations synchronized with the clock signal, and a marching main memory (31), which embraces an array of memory units, configured to store information in each of memory units and to transfer synchronously with the clock signal, providing the processor (11) with the stored information actively and sequentially so that the ALU (112) can execute the arithmetic and logic operations with the stored information. The results of the processing in the ALU (112) are sent out to the marching main memory (31), but there is only one way of instructions flow from the marching main memory (31) to the processor.

## Claims

### Claim at position 1

1-21. (canceled)

### Claim 22

A marching main memory adapted for a computer system having a processor and a marching main memory, the processor including a control unit having a clock generator generating a clock signal, and an arithmetic logic unit executing arithmetic and logic operations synchronized with the clock signal, the marching main memory comprising:
an array of memory units, in which a plurality of the memory units are aligned along a column direction, each of the memory units comprising an array of cells aligned along the column direction, each of the cells storing information, so that each of the memory units can store a unit of information aligned along the column direction, and a plurality of one-dimensional arrangements of the memory units aligned along the column direction are further deployed along a row direction;
an input-terminal column comprising a plurality of input terminals aligned along the column direction, each of the allocations of the input terminals are respectively assigned to the row positions of the cells in the memory units aligned along the column direction, through the input-terminal column, units of information are fed to the array of cells in each of the memory units aligned along the column direction; and
an output-terminal column comprising a plurality of output terminals aligned along the column direction, each of the allocations of the output terminals are respectively assigned to the row positions of the cells in the memory units aligned along the column direction, at least a part of the output terminals in the output-terminal column are connected to the arithmetic logic unit,
wherein the units of information stored in each of the memory units are transferred synchronously with the clock signal in an organized marching fashion in the array of the memory units, toward the output-terminal column, and the stored information is provided in parallel to the processor through the part of the output terminals, and the arithmetic logic unit can execute the arithmetic and logic operations synchronously with the clock signal, with the information being fed sequentially through the part of the output terminals.

### Claim 23

A computer system comprising:
a processor including:
a control unit having a clock generator configured to generate a clock signal, and
an arithmetic logic unit configured to execute arithmetic and logic operations synchronized with the clock signal; and
a marching main memory comprising an array of memory units, in which a plurality of the memory units are aligned along a column direction, each of the memory units comprising an array of cells aligned along the column direction, each of the cells storing information, so that each of the memory units can store a unit of information aligned along the column direction, and a plurality of one-dimensional arrangements of the memory units aligned along the column direction are further deployed along a row direction,
wherein the marching main memory further comprises:
an input-terminal column comprising a plurality of input terminals aligned along the column direction, each of the allocations of the input terminals are respectively assigned to the row positions of the cells in the memory units aligned along the column direction, through the input-terminal column, units of information are fed to the array of cells in each of the memory units aligned along the column direction; and
an output-terminal column comprising a plurality of output terminals aligned along the column direction, each of the allocations of the output terminals are respectively assigned to the row positions of the cells in the memory units aligned along the column direction, at least a part of the output terminals in the output-terminal column are connected to the arithmetic logic unit,
wherein the units of information stored in each of the memory units are transferred synchronously with the clock signal in an organized marching fashion in the array toward the output-terminal column, and the stored information is provided in parallel to the processor through the part of the output terminals, and the arithmetic logic unit can execute the arithmetic and logic operations synchronously with the clock signal, with the information being fed sequentially through the part of the output terminals, in addition the results of the processing in the arithmetic logic unit are sent out to the marching main memory through the part of the output terminals.

### Claim 24 (depends on 23)

The computer system of claim 23, wherein the memory units are arranged two dimensionally on a chip.

### Claim 25 (depends on 24)

The computer system of claim 24, wherein a plurality of chips are stacked vertically, sandwiching heat dissipating plates there between, so as to implement a three dimensional memory unit array.

### Claim 26 (depends on 25)

The computer system of claim 25, wherein the heat conducting plates are made of diamond.

### Claim 27 (depends on 23)

The computer system of claim 23, wherein the processor further includes a plurality of arithmetic pipelines configured to receive a flow of the unit of information through the part of the output terminals in the output-terminal column from the marching main memory, synchronously with the clock signal.

### Claim 28

A computer system comprising:
a processor including:
a control unit having a clock generator configured to generate a clock signal, and
an arithmetic logic unit configured to execute arithmetic and logic operations synchronized with the clock signal;
a marching main memory comprising an array of memory units, in which a plurality of the memory units are aligned along a column direction, each of the memory units comprising an array of cells aligned along the column direction, each of the cells storing information, so that each of the memory units can store a unit of information aligned along the column direction, and a plurality of one-dimensional arrangements of the memory units aligned along the column direction are further deployed along a row direction; and
a marching-cache memory having a first marching-cache array, in which a plurality of first cache memory units are aligned along a column direction, assigned to the location of at least part of the memory units aligned along the column direction, each of the first cache memory units store a unit of information aligned along the column direction, and a plurality of one-dimensional arrangements of the first cache memory units aligned along the column direction are further deployed along a row direction;
wherein the marching main memory further comprises an input-terminal column and an output-terminal column aligned along the column direction, the allocations of the input and output-terminal columns are respectively assigned to the row positions of the cells aligned along the column direction,
the marching-cache memory further comprises a cache input-terminal column of the first marching-cache array configured to receive the unit of stored information from the marching main memory, the cache input-terminal column being connected to the output-terminal column, and a cache output-terminal column, the cache output-terminal column is connected to the arithmetic logic unit through the input-terminal column, units of information are fed to the array of cells in each of the memory units aligned along the column direction in the marching main memory,
the units of information stored in each of the memory units are transferred synchronously with the clock signal in an organized marching fashion in the array, toward the output-terminal column, and the stored information is provided in parallel to the marching-cache memory through the output-terminal column and the cache input-terminal column,
the units of information are stored in each of the first cache memory units aligned along the column direction, and the stored units of information are transferred synchronously with the clock signal in an organized marching fashion, toward the cache output-terminal column, and the stored units of information are provided in parallel to the processor through the cache output-terminal column, and
the arithmetic logic unit can execute the arithmetic and logic operations synchronously with the clock signal, with the stored information being fed sequentially through the cache output-terminal column.

### Claim 29 (depends on 28)

The computer system of claim 28, wherein the arithmetic logic unit comprises a plurality of arithmetic pipelines, each of which includes a vector processing unit.

### Claim 30

A computer system comprising:
a processor including:
a control unit having a clock generator configured to generate a clock signal,
an arithmetic logic unit configured to execute arithmetic and logic operations synchronized with the clock signal, and
a plurality of arithmetic pipelines including either a plurality of vector processing units; and
a marching main memory comprising an array of memory units each having a unit of information, input terminals of the array and output terminals of the array, configured to store information in each of the memory units and to transfer synchronously with the clock signal, step by step, toward the output terminals;
a marching-cache memory having an array of cache memory units, located at locations corresponding to each for a unit of information, cache input terminals of the array configured to receive the stored information from the marching main memory, and cache output terminals of the array, configured to store information in each of cache memory units and to transfer, synchronously with the clock signal, step by step, the information each to an adjacent cache memory unit, toward the cache output terminals; and
a plurality of marching scalar/vector registers, each of which having an array of scalar/vector register units, located at locations corresponding to each for a unit of information, scalar/vector register input terminals of the array, configured to receive the stored vector information from the marching-cache memory, configured to store information in each of the scalar/vector register units and to transfer, synchronously with the clock signal, step by step, the stored vector information in each of scalar vector register units,
wherein, through the marching-cache memory and the marching scalar/vector registers, the marching main memory provides actively and sequentially vector data to the vector processing units so that the vector processing units can execute them, the results of the processing in the arithmetic logic unit are sent out to the marching main memory, through the marching-cache memory, except that in case of instructions movement, there is not the opposite direction to the information flow to be processed.

### Claim 31 (depends on 30)

The computer system of claim 30, further comprising a plurality of marching scalar/vector registers, each of which having an array of scalar/vector register units, located at locations corresponding to each for a unit of information, scalar/vector register input terminals of the array, configured to receive the stored scalar information from the marching-cache memory, configured to store scalar information in each of the scalar/vector register units and to transfer, synchronously with the clock signal, step by step, the stored scalar information in each of the vector register units, so as to provide actively and sequentially scalar data to the scalar processing units so that the scalar processing units can execute them, the results of the processing in the arithmetic logic unit are sent out to the marching-cache memory, except that in case of instructions movement, there is not the opposite direction to the information flow to be processed.

### Claim 32 (depends on 31)

The computer system of claim 31, wherein the control unit further has a field programmable gate array configured to switch-control operations of the marching main memory, the marching-cache memory, the marching-vector registers and marching-scalar registers.

### Claim 33 (depends on 32)

The computer system of claim 32, wherein the marching main memory and the marching-cache memory are merged on a same chip.

### Claim 34 (depends on 32)

The computer system of claim 32, wherein at least one of the vector processing units and at least one of the marching-vector registers are merged on a same chip.

### Claim 35 (depends on 24)

The computer system of claim 24, wherein a first chip merging the vector processing units and the marching-vector registers, a second chip merging the marching-cache memory and a third chip merging the marching main memory are stacked vertically, sandwiching heat dissipating plates there between.

### Claim 36 (depends on 23)

The computer system of claim 23, wherein the arithmetic logic unit implements a processor core.

### Claim 37 (depends on 36)

The computer system of claim 36, wherein the processor includes a plurality of processor cores.

### Claim 38 (depends on 37)

The computer system of claim 37, further comprising a marching-cache memory having an array of cache memory units, cache input terminals of the second array configured to receive the stored information from the marching main memory, and cache output terminals of the second array, configured to store information in each of cache memory units and to transfer successively, step by step, the stored information in each of cache memory units to an adjacent cache memory unit, synchronized with the clock signal from the cache memory units adjacent to the cache input terminals toward the cache memory units adjacent to the cache output terminals, so as to provide actively and sequentially the stored information to the processor through the cache output terminals so that the processor cores can operate with the stored information.

### Claim 39 (depends on 38)

The computer system of claim 38, further comprising a marching-register file having an array of register file cells, register file input terminals of the array configured to receive the stored information from the marching-cache memory, and register file output terminals of the array, configured to store information in each of register file cells and to transfer successively, step by step, the stored information in each of register file cells to an adjacent register file cell being synchronized with the clock signal from the register file cells adjacent to the register file input terminals toward the register file cells adjacent to the register file output terminals, so as to provide actively and sequentially the stored information to the processor cores through the register file output terminals.

### Claim 40 (depends on 39)

The computer system of claim 39, wherein the control unit further has a field programmable gate array configured to switch-control operations of the marching main memory, the marching-cache memory, and the marching-register file.

### Claim 41 (depends on 40)

The computer system of claim 40, wherein the marching main memory and the marching-cache memory are merged on a same chip.

### Claim 42 (depends on 40)

The computer system of claim 40, wherein the marching-register file and at least one of the processor cores are merged on a same chip.

### Claim 43 (depends on 23)

The computer system of claim 23, wherein each of the memory units aligned along the column direction comprises:
an array of first cells configured to transfer the units of information stored in each of the memory units, synchronously with the clock signal in the organized marching fashion, toward the output-terminal column, and
another array of second cells configured to transfer the units of information stored in each of the memory units, synchronously with a second clock signal in the organized marching fashion, toward the input-terminal column.

### Claim 44 (depends on 23)

The computer system of claim 23, wherein a part of the memory units aligned along the column direction comprises a first memory unit, which comprises an array of first cells configured to transfer the units of information synchronously with the clock signal toward the output-terminal column in the organized marching fashion, and
another part of the memory units aligned along the column direction comprises a second memory unit, which comprises an array of second cells configured to transfer the units of information synchronously with a second clock signal toward the input-terminal column in the organized marching fashion.

### Claim 45 (depends on 43)

The computer system of claim 43, further comprising a clock selector for selecting the clock signal or the second clock signal, and when the clock selector selects the second clock signal, being controlled by the control unit, results of the processing in the arithmetic logic unit are sent out sequentially to the marching main memory as a set of information, through the part of the output terminals in the output-terminal column, so that the set of information can be stored in the marching main memory, and the set of information is synchronously transferred toward the input-terminal column in the organized marching fashion in the array.

### Claim 46 (depends on 44)

The computer system of claim 44, further comprising a clock selector for selecting the clock signal or the second clock signal, and
when the clock selector selects the second clock signal, being controlled by the control unit, results of the processing in the arithmetic logic unit are sent out sequentially to the marching main memory as a set of information, through the part of the output terminals in the output-terminal column, so that the set of information can be stored in the marching main memory, and the set of information is synchronously transferred toward the input-terminal column in the organized marching fashion in the array.
