公开(公告)号：DE602004027543D1

公开(公告)日：2010-07-15

申请号：DE602004027543

申请日：2004-12-06

Applicant: IBM

Inventor： MAKHERVAKS VADIM ,  MACHULSKY ZORIK ,  BIRAN GIORA ,  SHALEV LEAH

IPC: G06F15/16 ,  H04L29/06

公开(公告)号：AT470187T

公开(公告)日：2010-06-15

申请号：AT04813127

申请日：2004-12-06

Applicant: IBM

Inventor： MAKHERVAKS VADIM ,  MACHULSKY ZORIK ,  BIRAN GIORA ,  SHALEV LEAH

IPC: G06F15/16 ,  H04L29/06

Abstract: An RNIC implementation that performs direct data placement to memory where all segments of a particular connection are aligned, or moves data through reassembly buffers where all segments of a particular connection are non-aligned. The type of connection that cuts-through without accessing the reassembly buffers is referred to as a "Fast" connection because it is highly likely to be aligned, while the other type is referred to as a "Slow" connection. When a consumer establishes a connection, it specifies a connection type. The connection type can change from Fast to Slow and back. The invention reduces memory bandwidth, latency, error recovery using TCP retransmit and provides for a "graceful recovery" from an empty receive queue. The implementation also may conduct CRC validation for a majority of inbound DDP segments in the Fast connection before sending a TCP acknowledgement (Ack) confirming segment reception.

公开(公告)号：CA2548085C

公开(公告)日：2014-04-29

申请号：CA2548085

申请日：2004-12-08

Applicant: IBM

Inventor： BIRAN GIORA ,  MACHULSKY ZORIK ,  MAKHERVAKS VADIM ,  SHALEV LEAH

IPC: H04L1/22 ,  H03M13/09 ,  H04L29/08

Abstract: A network interface controller implementation that performs direct data placement to memory where all segments of a particular connection are aligned, i.e. cuts-through without accessing reassembly buffers (i.e., a "Fast" connection type), or if all segments are non~aligned moves data through reassembly buffers (i.e., a "Slow" connection type). The type of connection can change from Fast to Slow and back. The implementation preferably conducts CRC validation for a majority of inbound DDP segments in the Fast connection before sending a TCP acknowledgement confirming segment reception. The method of computing a CRC value according to the present invention assumes that each TCP segment starts with an aligned DDP segment and that the first two bytes of a TCP payload is an MPA length field of an MPA frame.

公开(公告)号：CA2548085A1

公开(公告)日：2005-07-07

申请号：CA2548085

申请日：2004-12-08

Applicant: IBM

Inventor： SHALEV LEAH ,  MAKHERVAKS VADIM ,  MACHULSKY ZORIK ,  BIRAN GIORA

IPC: H04L1/22 ,  H04L29/08

Abstract: A network interface controller implementation that performs direct data placement to memory where all segments of a particular connection are aligne d, i.e. cuts-through without accessing reassembly buffers (i.e., a "Fast" connection type), or if all segments are non~aligned moves data through reassembly buffers (i.e., a "Slow" connection type). The type of connection can change from Fast to Slow and back. The implementation preferably conduct s CRC validation for a majority of inbound DDP segments in the Fast connection before sending a TCP acknowledgement confirming segment reception. The metho d of computing a CRC value according to the present invention assumes that eac h TCP segment starts with an aligned DDP segment and that the first two bytes of a TCP payload is an MPA length field of an MPA frame.