公开(公告)号：US20210150770A1

公开(公告)日：2021-05-20

申请号：US17095544

申请日：2020-11-11

Applicant: Intel Corporation

Inventor： ABHISHEK R. APPU ,  PRASOONKUMAR SURTI ,  JILL BOYCE ,  SUBRAMANIAM MAIYURAN ,  MICHAEL APODACA ,  ADAM T. LAKE ,  JAMES HOLLAND ,  VASANTH RANGANATHAN ,  ALTUG KOKER ,  LIDONG XU ,  NIKOS KABURLASOS

IPC: G06T9/00 ,  G06N3/04 ,  G06T15/00

Abstract: Embodiments described herein provided for an instruction and associated logic to enable a processing resource including a tensor accelerator to perform optimized computation of sparse submatrix operations. One embodiment provides hardware logic to apply a numerical transform to matrix data to increase the sparsity of the data. Increasing the sparsity may result in a higher compression ratio when the matrix data is compressed.

公开(公告)号：US20230377209A1

公开(公告)日：2023-11-23

申请号：US18322194

申请日：2023-05-23

Applicant: Intel Corporation

Inventor： ABHISHEK R. APPU ,  PRASOONKUMAR SURTI ,  JILL BOYCE ,  SUBRAMANIAM MAIYURAN ,  MICHAEL APODACA ,  ADAM T. LAKE ,  JAMES HOLLAND ,  VASANTH RANGANATHAN ,  ALTUG KOKER ,  LIDONG XU ,  NIKOS KABURLASOS

IPC: G06T9/00 ,  G06T15/00 ,  G06N3/045

CPC classification number: G06T9/002 ,  G06T9/007 ,  G06T15/005 ,  G06T9/008 ,  G06N3/045

Abstract: Embodiments described herein provided for an instruction and associated logic to enable a processing resource including a tensor accelerator to perform optimized computation of sparse submatrix operations. One embodiment provides a parallel processor comprising a processing cluster coupled with the cache memory. The processing cluster includes a plurality of multiprocessors coupled with a data interconnect, where a multiprocessor of the plurality of multiprocessors includes a tensor core configured to load tensor data and metadata associated with the tensor data from the cache memory, wherein the metadata indicates a first numerical transform applied to the tensor data, perform an inverse transform of the first numerical transform, perform a tensor operation on the tensor data after the inverse transform is performed, and write output of the tensor operation to a memory coupled with the processing cluster.