TY - GEN
T1 - A high speed DMA transaction method for PCI express devices
AU - Peng, Yu
AU - Li, Bo
AU - Liu, Datong
AU - Peng, Xiyuan
PY - 2009
Y1 - 2009
N2 - PCI Express is being more and more widely deployed for its attractive bandwidth in virtual instrument design. However, very few PCI Express transaction methods can be found in literature. So, this paper presents a novel PCI Express DMA transaction method based on bridge chip PEX 8311. Furthermore, we propose a new method to optimize PCI Express DMA Transaction through improving both bus-efficiency and DMA-efficiency. A novel FSM design to respond data and address cycles on PCI Express bus is introduced, and a continuous data burst is realized, which greatly promote busefficiency. As the foundation of the whole design, a successful 2.5G PCI Express interface design is presented. Then, to make a whole solution, in software design, WDM driver framework and three successful DMA optimizing options for PCI Express devices are presented to improve DMA-efficiency. At last, a FSM-based test to data transaction speed is proposed. Experiments show that method discussed here can reach a maximum DMA WRITE speed up to 166MBytes/s and DMA READ speed up to 136MBytes/s. Both of them exceed PCI theoretical maximum speed (133MBytes/s). In fact, this paper provides not noly a PCI Express example, but also PCI Express interface solution and DMA transaction method which can be directly extended into high speed PXI-Express applications and so forth. That is quite meaningful to next generation instrumentation.
AB - PCI Express is being more and more widely deployed for its attractive bandwidth in virtual instrument design. However, very few PCI Express transaction methods can be found in literature. So, this paper presents a novel PCI Express DMA transaction method based on bridge chip PEX 8311. Furthermore, we propose a new method to optimize PCI Express DMA Transaction through improving both bus-efficiency and DMA-efficiency. A novel FSM design to respond data and address cycles on PCI Express bus is introduced, and a continuous data burst is realized, which greatly promote busefficiency. As the foundation of the whole design, a successful 2.5G PCI Express interface design is presented. Then, to make a whole solution, in software design, WDM driver framework and three successful DMA optimizing options for PCI Express devices are presented to improve DMA-efficiency. At last, a FSM-based test to data transaction speed is proposed. Experiments show that method discussed here can reach a maximum DMA WRITE speed up to 166MBytes/s and DMA READ speed up to 136MBytes/s. Both of them exceed PCI theoretical maximum speed (133MBytes/s). In fact, this paper provides not noly a PCI Express example, but also PCI Express interface solution and DMA transaction method which can be directly extended into high speed PXI-Express applications and so forth. That is quite meaningful to next generation instrumentation.
KW - Bus-efficiency
KW - DMA
KW - DMA-efficiency
KW - PCI Express
UR - https://www.scopus.com/pages/publications/70149119201
U2 - 10.1109/CAS-ICTD.2009.4960747
DO - 10.1109/CAS-ICTD.2009.4960747
M3 - 会议稿件
AN - SCOPUS:70149119201
SN - 9781424425877
T3 - 2009 IEEE Circuits and Systems International Conference on Testing and Diagnosis, ICTD'09
BT - 2009 IEEE Circuits and Systems International Conference on Testing and Diagnosis, ICTD'09
T2 - 2009 IEEE Circuits and Systems International Conference on Testing and Diagnosis, ICTD'09
Y2 - 28 April 2009 through 29 April 2009
ER -