# US 2003/0196004 A1 · Buffering and presenting chunks

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

- Number: US 2003/0196004 A1
- Document ID: US20030196004A1
- Status: Published application
- Kind: A1
- Application number: 10/444,816
- Filed: 2003-05-22
- Published: 2003-10-16
- Earliest claimed: 1999-12-29
- Assignees: Intel Corporation
- Inventors: David R. Jackson; Stephen W. Kiss; Miles F. Schwartz
- Web page: https://patentgrep.com/patent/US20030196004A1

## CPC

- G06F 13/1673: {using buffers}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A device and method for servicing data requests from a processor or other input/output interface in a multi-processor environment by accessing a full or partial cache line of data. A system data chip is used to access the cache line of data using a bit pattern supplied by a system address chip. This access and transmission of data to the processor or the input/output interface is controlled by a control/status unit in the system data chip based on the value of control valid signals which include a first valid (DxV) signal and a second valid (CxV) signal. Also, data may be stored and retrieved in a first data format (linear chunk order) or a second data format (critical chunk order). When control by the control/status unit is based on the DxV signal value, a read of a data chunk may occur immediately after a write to temporary storage if the data is in the same chunk order and a merge or combination operation is not taking place. When control by the control/status unit is based on the CxV signal, the entire cache line of data must be written to temporary storage prior to the reading of any data. In this way read operation are optimized to minimize latency and maximize throughput.

## Claims

### Claim 1

A device for servicing data reads and writes from at least one processor and an I/O interface connected to a plurality of I/O devices, comprising:
a system data chip to receive a read request for data from a processor of a plurality of processors or the I/O interface, wherein the system data chip further comprises:
a data buffer to store data in a first data format and a second data format being received by the system data chip as a result of the read request;
a control/status unit to control when writing the data to the data buffer occurs and when reading from data buffer occurs based on a first valid signal or a second valid signal.

### Claim 2 (depends on 1)

The device as recited in claim 1, wherein when the first valid signal is set data may be read from the data buffer immediately after it is written to the data buffer and immediately transmitted to the processor or the I/O interface that made the read request if the data is stored and transmitted in a same data format.

### Claim 3 (depends on 2)

The device as recited in claim 2, wherein data may only be immediately read from the data buffer after being written when the data is from a single source.

### Claim 4 (depends on 3)

The device as recited in claim 3, wherein when the data is from more than one source the system data chip performs a merge or combination function to write the data to the data buffer.

### Claim 5 (depends on 4)

The device as recited in claim 4, wherein when the system data chip performs the merge or combination function, the control/status unit will not cause a read from the data buffer immediately after a write to the data buffer and will set the second valid signal upon completion of the entire write of the data to the data buffer.

### Claim 6 (depends on 5)

The device as recited in claim 5, wherein the second valid signal is set the data written to the data buffer is read from the data buffer and transmitted to the processor or the I/O interface requesting the data.

### Claim 7 (depends on 3)

The device as recited in claim 3, wherein when the data written to the data buffer is in the second data format and the data is to be read from the data buffer is in the first data format then the data buffer enters a wait state.

### Claim 8 (depends on 7)

The device as recited in claim 7, wherein the data buffer will remain in the wait state until sufficient data is received to present the data in the second data format.

### Claim 9 (depends on 8)

The device as recited in claim 8, wherein the first data format is linear data order and the second data format is critical chunk order.

### Claim 10

A method of servicing data requests from a processor of a plurality of processors and an I/O interface having a plurality of I/O devices, comprising:
receiving a read request from a processor of the plurality of processors or the I/O interface to access data;
setting a first valid signal;
retrieving data from memory in a first data format or a second data format;
storing the data in a data buffer in a first data format or a second data format
transmitting the data to the processor or the I/O interface requesting the data immediately after it is written to the data buffer when the first valid signal is set and the data is retrieved from memory in the first data format and stored in the data buffer in the first data format;
setting a second valid signal when the storing of the data to the data buffer is completed; and
transmitting the data to the processor or the I/O interface requesting the data if the data has not been previously transmitted when the second valid signal is set.

### Claim 11 (depends on 10)

The method as recited in claim 10, wherein transmitting the data to the processor or the I/O interface requesting the data immediately after it is written to the data buffer when the first valid signal is set and the data is retrieved from memory in the first data format and stored in the data buffer in the first data format data may only be immediately read from the data buffer after being written when the data is from a single processor or the I/O interface.

### Claim 12 (depends on 11)

The method as recited in claim 11, further comprising:
merging the data when the data is from more than one source and not transmitting the data when the first valid signal is set.

### Claim 13 (depends on 12)

The method as recited in claim 12, setting the second valid signal when after merging the data.

### Claim 14 (depends on 13)

The method as recited in claim 13, transmitting the data when the second valid signal is set after merging the data.

### Claim 15 (depends on 14)

The method as recited in claim 14, wherein the first valid signal is a DxV signal and the second valid signal is a CxV signal.

### Claim 16 (depends on 11)

The method as recited in claim 11, further comprising:
entering a wait state in the data buffer when the data written to the data buffer is in the second data format and the data is to be read from the data buffer is in the first data format.

### Claim 17 (depends on 16)

The method as recited in claim 16, further comprising:
remaining in a wait state in the data buffer until sufficient data is received to present the data in the second data format.

### Claim 18 (depends on 17)

The method as recited in claim 17, wherein the first data format is linear data order and the second data format is critical chunk order.

### Claim 19

A method of servicing data requests from a processor of a plurality of processors and an I/O interface having a plurality of I/O devices, comprising:
receiving a read request from a processor of the plurality of processors or the I/O interface to access data;
setting a first valid signal;
retrieving data from memory in a linear chunk order or a critical chunk order;
merging or combining the data when the data is from more than one source;
storing the data in a data buffer in the linear chunk order or the critical chunk order;
transmitting the data to the processor or the I/O interface requesting the data immediately after it is written to the data buffer when the first valid signal is set and the data is retrieved from memory and stored in the data buffer in the same format, but not when the data is a result of a merge or combination from more than one source;
entering a wait state in the data buffer when the data written to the data buffer is in the critical chunk order and the data is to be read from the data buffer is in the linear chunk order;
setting a second valid signal when the storing of the data to the data buffer is completed; and
transmitting the data to the processor or the I/O interface requesting the data if the data has not been previously transmitted and the second valid signal is set.
