# US 2006/0155886 A1 · Methods and arrangements to manage on-chip memory to reduce memory latency

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

- Number: US 2006/0155886 A1
- Document ID: US20060155886A1
- Status: Published application
- Kind: A1
- Application number: 11/032,876
- Filed: 2005-01-11
- Published: 2006-07-13
- Earliest claimed: 2005-01-11
- Assignees: International Business Machines Corporation
- Inventors: Dilma da Silva; Elmootazbellah Elnozahy; Orran Krieger; Hazim Shafi; Xiaowei Shen; Balaram Sinharoy; Robert Tremaine
- Web page: https://patentgrep.com/patent/US20060155886A1

## CPC

- G06F 12/08: in hierarchically structured memory systems, e.g. virtual memory systems
- G06F 12/0893: Caches characterised by their organisation or structure
- G06F 2212/251: Local memory within processor subsystem
- G06F 2212/253: Centralized memory
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Methods, systems, and media for reducing memory latency seen by processors by providing a measure of control over on-chip memory (OCM) management to software applications, implicitly and/or explicitly, via an operating system are contemplated. Many embodiments allow part of the OCM to be managed by software applications via an application program interface (API), and part managed by hardware. Thus, the software applications can provide guidance regarding address ranges to maintain close to the processor to reduce unnecessary latencies typically encountered when dependent upon cache controller policies. Several embodiments utilize a memory internal to the processor or on a processor node so the memory block used for this technique is referred to as OCM.

## Claims

### Claim 1

A method for controlling data stored in an on-chip memory sharing a chip with a processor and a cache by a software application, comprising:
allocating within the on-chip memory of the chip a block of processor memory region for the data in response to an instruction initiated by the software application, the instruction being associated with the data;
transferring the data from a main memory to the block within the on-chip memory via a driver based upon an indication from the software application, the driver to transfer the data from the main memory and store the data in the block of the on-chip memory; and
accessing the block of the on-chip memory to obtain the data in response to a request for the data by the processor.

### Claim 2 (depends on 1)

The method of claim 1, further comprising updating a copy of the data in the main memory in response to modifying the data to maintain coherence.

### Claim 3 (depends on 1)

The method of claim 1, further comprising delaying access by another processor to a copy of the data that is maintained in the main memory to maintain coherence.

### Claim 4 (depends on 1)

The method of claim 1, further comprising issuing an instruction to delay access to the block of the on-chip memory while the data is being transferred to the block of the on-chip memory.

### Claim 5 (depends on 4)

The method of claim 4, wherein allocating the block of the on-chip memory comprises allocating a logical portion of the block of the on-chip memory, wherein the logical portion is to be associated with physical addresses in response to an access to the block of the on-chip memory.

### Claim 6 (depends on 1)

The method of claim 1, wherein transferring the data comprises transferring the data in response to the request for the data.

### Claim 7 (depends on 1)

The method of claim 1, wherein allocating the block of the on-chip memory comprises allocating the block of the on-chip memory in response to an explicit instruction from the software application, the explicit instruction being substantially determinative of allocation of the block of the on-chip memory.

### Claim 8 (depends on 7)

The method of claim 7, wherein allocating the block of the on-chip memory comprises allocating the block of the on-chip memory in response to the explicit instruction.

### Claim 9 (depends on 7)

The method of claim 7, wherein allocating the block of the on-chip memory comprises allocating the block of the on-chip memory after the explicit instruction and in response to an access of the allocated block of the on-chip memory.

### Claim 10 (depends on 1)

The method of claim 1, wherein allocating the block of the on-chip memory comprises allocating the block of the on-chip memory in response to an implicit instruction from the software application via an operating system, wherein the implicit instruction is a request for allocation of the block of the on-chip memory and the operating system determines whether to issue the instruction.

### Claim 11

An apparatus for controlling data stored on a chip by a software application, comprising:
an on-chip memory comprising a block for allocation to the software application, the on-chip memory sharing the chip with a processor and a cache;
a controller to allocate the block of the on-chip memory for the data in response to an instruction initiated by the software application, the instruction being associated with the data, and to access the block of the on-chip memory to obtain the data in response to a request for the data by the processor; and
a driver coupled with the on-chip memory and responsive to the controller to transfer the data from the main memory to store the data in the block of the on-chip memory based upon an indication from the software application.

### Claim 12 (depends on 11)

The apparatus of claim 11, wherein the controller is adapted to update a copy of the data in the main memory in response to modifying the data to maintain coherence.

### Claim 13 (depends on 11)

The apparatus of claim 11, wherein the controller is adapted to delay access to the block of the on-chip memory while the data is being transferred to the block of the on-chip memory.

### Claim 14 (depends on 11)

The apparatus of claim 11, wherein the controller is adapted to allocate the block of the on-chip memory in response to an explicit instruction from the software application.

### Claim 15 (depends on 11)

The apparatus of claim 11, wherein the controller is adapted to allocate the block of the on-chip memory in response to an implicit instruction from the software application via an operating system.

### Claim 16 (depends on 11)

The apparatus of claim 11, wherein the driver is adapted to access a library and to execute code based upon a content of the library to transfer data from the main memory to the block of the on-chip memory.

### Claim 17 (depends on 11)

The apparatus of claim 11, wherein the driver comprises a direct memory access engine responsive to the controller to page out a content of the block of the on-chip memory prior to copying the data into the block of the on-chip memory.

### Claim 18

A machine-accessible medium containing a program for controlling data stored in an on-chip memory sharing a chip with a processor and a cache by a software application, which, when executed, performs operations, the operations comprising:
allocating within the on-chip memory of the chip a block of processor memory region for the data in response to an instruction initiated by the software application, the instruction being associated with the data;
transferring the data from a main memory to the block within the on-chip memory via a driver based upon an indication from the software application, the driver to transfer the data from the main memory and store the data in the block of the on-chip memory; and
accessing the block of the on-chip memory to obtain the data in response to a request for the data by the processor.

### Claim 19 (depends on 18)

The machine-accessible medium of claim 18, wherein the operation further comprises updating a copy of the data in the main memory in response to modifying the data to maintain coherence.

### Claim 20 (depends on 18)

The machine-accessible medium of claim 18, wherein the operation further comprises issuing a wait instruction to delay access to the block of the on-chip memory while the data is being transferred to the block of the on-chip memory.
