Vector based matrix multiplication
A method is provided that includes performing, by a processor in response to a vector matrix multiply instruction, multiplying an m×n matrix (A matrix) and a n×p matrix (B matrix) to generate elements of an m×p matrix (R matrix), and storing the elements of the R matrix in a storage location specified by the vector matrix multiply instruction.
1. A system comprising:
a first permute component including a first plurality of permute networks;
a first multiplexer including an output and inputs coupled to the first plurality of permute networks;
a second permute component including a second plurality of permute networks;
a second multiplexer including an output and inputs coupled to the second plurality of permute networks;
a third multiplexer including:
a first input coupled to the output of the first multiplexer;
a second input coupled to the output of the second multiplexer; and
an output; and
a multiplication component coupled to the output of the third multiplexer.
2. The system of claim 1 , wherein the first plurality of permute networks includes a first permute network configured to permute data for performing a finite impulse response operation.
3. The system of claim 1 , wherein the first plurality of permute networks includes a second permute network configured to permute data for performing matrix multiplication.
4. The system of claim 1 , further comprising a logical OR circuit coupled to the third multiplexer and the multiplication component.
5. The system of claim 4 , further comprising:
a third permute component; and
a fourth multiplexer coupled to the third permute component, wherein the logical OR circuit is coupled to the fourth multiplexer.
6. The system of claim 5 , wherein the logical OR circuit is configured to concatenate an output of the third multiplexer and an output of the fourth multiplexer.
7. The system of claim 1 , further comprising:
a first buffer coupled to the first permute component; and
a second buffer coupled to the second permute component.
8. The system of claim 7 , wherein input packets of a first set of data elements and a second set of data elements are alternated in a ping pong fashion between the first buffer and the second buffer.
9. The system of claim 7 ,
wherein the first and second buffers are configured to receive a set of data elements, and
wherein the third multiplexer is configured to select between the first permute component and the second permute component based on which of the first buffer and the second buffer stores a next data element of the set of data elements.
10. The system of claim 9 , further comprising an instruction decoder circuit configured to:
generate a select signal based on which buffer of the first buffer and the second buffer stores the next data element; and
send the select signal to the third multiplexer.
11. The system of claim 1 , wherein the multiplication component is configured to perform vector multiplication.
12. A method comprising:
selecting, by a first multiplexer, among a first plurality of permute networks;
selecting, by a second multiplexer, among a second plurality of permute networks;
selecting, by a third multiplexer, among a third plurality of permute networks;
selecting, by a fourth multiplexer, among a fourth plurality of permute networks;
selecting, by a fifth multiplexer, between an output of the first multiplexer and an output of the second multiplexer;
selecting, by a sixth multiplexer, between an output of the third multiplexer and an output of the fourth multiplexer;
performing a logical OR operation on an output of the fifth multiplexer and an output of the sixth multiplexer; and
performing a multiplication operation on an output of the logical OR operation.
13. The method of claim 12 , further comprising:
by a first permute network in the first plurality of permute networks, permuting data for performing a finite impulse response operation; and
by a second permute network in the first plurality of permute networks, permuting data for performing matrix multiplication.
14. The method of claim 12 , wherein performing the logical OR operation comprises concatenating the output of the fifth multiplexer and the output of the sixth multiplexer.
15. A system comprising:
a first permute component including a first plurality of permute networks;
a first multiplexer coupled to the first plurality of permute networks;
a second permute component including a second plurality of permute networks;
a second multiplexer coupled to the second plurality of permute networks;
a third permute component including a third plurality of permute networks;
a third multiplexer coupled to the third plurality of permute networks;
a fourth permute component including a fourth plurality of permute networks;
a fourth multiplexer coupled to the fourth plurality of permute networks;
a fifth multiplexer coupled to the first multiplexer and the second multiplexer;
a sixth multiplexer coupled to the third multiplexer and the fourth multiplexer;
a logical OR circuit coupled to the fifth multiplexer and the sixth multiplexer; and
a multiplication component coupled to the logical OR circuit.
16. The system of claim 15 , wherein the first plurality of permute networks includes:
a first permute network configured to permute data for performing a finite impulse response operation; and
a second permute network configured to permute data for performing matrix multiplication.
17. The system of claim 15 , wherein the logical OR circuit is configured to concatenate an output of the fifth multiplexer and an output of the sixth multiplexer.
18. The system of claim 15 , further comprising:
a first buffer coupled to the first permute component; and
a second buffer coupled to the second permute component,
wherein the third multiplexer is configured to select between the first permute component and the second permute component based on which of the first buffer and the second buffer stores a next data element of a set of data elements, and
wherein input packets of a first set of data elements and a second set of data elements are alternated in a ping pong fashion between the first buffer and the second buffer.
19. The system of claim 18 , further comprising an instruction decoder circuit configured to:
generate a select signal based on which buffer of the first buffer and the second buffer stores the next data element; and
send the select signal to the third multiplexer.
20. The system of claim 15 , wherein the multiplication component is configured to perform vector multiplication.