Outer product-based matrix-vector multiplication operation apparatus for accelerating vector operation and method using the same
US-2024362297-A1 · Oct 31, 2024 · US
US9747073B2 · US · B2
| Field | Value |
|---|---|
| Publication number | US-9747073-B2 |
| Application number | US-201414198746-A |
| Country | US |
| Kind code | B2 |
| Filing date | Mar 6, 2014 |
| Priority date | Mar 6, 2014 |
| Publication date | Aug 29, 2017 |
| Grant date | Aug 29, 2017 |
A practical reading order for non-experts. Skip the full description unless you need deep technical detail.
What the patent document calls the invention.
A short plain-language summary of the technical disclosure.
Who owns or filed the patent and who is credited as inventor.
Filing, priority, publication, and grant dates set the timeline.
The legal scope of protection — read this for what is actually claimed.
Technology tags used to group this patent with similar filings.
Prior art links and similar publications in this corpus.
Official abstract text for this publication.
Embodiments of a processor are disclosed for performing arithmetic operations on a machine independent number format. The processor may include a floating point unit, and a number unit. The number format may include a sign/exponent block, a length block, and multiple mantissa digits. The number unit may be configured to perform an operation on two operands by converting the digit format of each mantissa digit of each operand, to perform the operation using the converted mantissa digits, and then to convert each mantissa digit of the result of the operation back into the original digit format.
Opening claim text (preview).
What is claimed is: 1. A processor, comprising: a floating point unit including a plurality of execution pipelines each of which is configured to perform respective arithmetic or graphics processing operations; and a number unit coupled to the floating point unit, wherein the number unit includes one or more logic circuits configured to: receive information specifying an operation, a first operand and a second operand; wherein each of the first operand and the second operand includes one or more mantissa digits, wherein each mantissa digit is formatted with an initial format; convert the one or more mantissa digits of the first operand from the initial format to a first format to generate a first converted operand; convert the one or more mantissa digits of the second operand from the initial format to the first format to generate a second converted operand; in response to a determination that the operation is an addition operation or a subtraction operation: perform the operation in the first format using the first converted operand and the second converted operand to generate a first result, wherein the first result includes one or more mantissa digits; convert the one or more mantissa digits of the first result from the first format to the initial format; otherwise: convert the one or more mantissa digits of the first converted operand from the first format to a second format to generate a first reformatted operand; convert the one or more mantissa digits of the second converted operand from the first format to the second format to generate a second reformatted operand; perform the operation in the second format using the first reformatted operand and the second reformatted operand to generate a second result, wherein the second result includes one or more mantissa digits formatted with the second format; and convert the one or more mantissa digits of the second result from the second format to the first format; wherein a particular mantissa digit formatted with the initial format includes one of a set of positive integers between 1 and 100, wherein each one of the set of positive integers corresponds to a respective one of a set of base 100 numbers between 0 and 99; wherein a particular mantissa digit formatted with the first format includes one of a set of base 100 numbers between 0 and 99; and wherein the second format includes binary coded decimal. 2. The processor of claim 1 , wherein the number unit is further configured to: clear a particular mantissa digit of the first converted operand in response to a determination that the particular mantissa digit of the first converted operand is unused; and clear a particular mantissa digit of the second converted operand in response to a determination that the particular mantissa digit of the second converted operand is unused. 3. The processor of claim 1 , wherein the number unit is further configured to clear a mantissa digit of the first result in response to a determination that the mantissa digit of the result is unused.
Methods or arrangements for data conversion without changing the order or content of the data handled · CPC title
Computations with numbers represented by a non-linear combination of denominational numbers, e.g. rational numbers, logarithmic number system or floating-point numbers {(G06F7/4806, G06F7/4824, G06F7/49, G06F7/491, G06F7/544 take precedence)} · CPC title
Computations with decimal numbers {radix 12 or 20. (G06F7/4824 takes precedence)} · CPC title
Non-specified BCD representation · CPC title
Accepting numbers of variable word length · CPC title
Related publications grouped by family.
Answers are generated from the same data shown on this page.