1 / 12

Hash Sorting for Triplet Matching

Hash Sorting for Triplet Matching. Jinyuan Wu, M. Wang For BTeV collaboration Dec. 2002. Hash Sorter. 7. 6. 5. 4. 3. 2. 1. 0. Hash Sorting and Then?. DATA. DATA. Directly send to C algorithm. Send to hardware algorithm and then C algorithm. Hash Sorter. Hash Sorter. DSP.

cianna
Télécharger la présentation

Hash Sorting for Triplet Matching

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Hash Sorting for Triplet Matching Jinyuan Wu, M. Wang For BTeV collaboration Dec. 2002

  2. Hash Sorter 7 6 5 4 3 2 1 0

  3. Hash Sorting and Then? DATA DATA • Directly send to C algorithm. • Send to hardware algorithm and then C algorithm. Hash Sorter Hash Sorter DSP Hardware Triplet Matcher DSP Buffer Manager Buffer Manager

  4. However Hash sorting alone can not do the whole triplet matching job.

  5. Hash Sorting and Feed to C Code • Data are just sorted, no cutting. • Sorted data let C code finds possible match in small number of combinations. DATA Hash Sorter DSP Buffer Manager

  6. Hash Sorting and Feed to Hardware Triplet Matcher • Data are just sorted, no cutting. • Sorted data reduce the functional block array size, so that it will fit in FPGA. DATA Hash Sorter Hardware Triplet Matcher DSP Buffer Manager

  7. Hash Sorting (For Triplet Matching) Triplets sorted by (dx/dz)

  8. Hash Sorter (How) Int. Triplets Ext. Triplets dx

  9. Hash Sorter (Implementation) DOUT DIN A_write DATA RAM B B(DIN) (B,C) Count for DOUT (C+1) C • Large number of bins. • 2 CLK/write. • 1 CLK/read. • Ping-pong structure. COUNT RAM

  10. Results See the “hashT2Tr.pdf” file

  11. Summary • Int. and Ext. triplets are sorted in 800um bins. • < 0.5% matching triplets are beyond 2 neighboring bins. • Feeding to C code: • Factor of 12 combination reduction. (both 132ns and 396ns cases) • Feeding to Hardware Triplet Matcher: • Max. query list size: 16 for 132ns case. (>80 without hash binning) • Max. query list size: 32 for 396ns case. (>200 without hash binning)

  12. End

More Related