专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明申请

US20190243780A1 SCALABLE APPLICATION-CUSTOMIZED MEMORY COMPRESSION 审中-公开
公开(公告)号：US20190243780A1
公开(公告)日：2019-08-08
申请号：US16380114
申请日：2019-04-10
申请人： Vinodh Gopal , Simon N. Peffers
发明人： Vinodh Gopal , Simon N. Peffers
IPC分类号： G06F12/1027 , G06F12/0815 , G06F12/1009 , G06F12/0811
CPC分类号： G06F12/1027 , G06F12/0811 , G06F12/0815 , G06F12/1009 , G06F2212/401
摘要： Methods and apparatus for scalable application-customized memory compression. Data is selectively stored in system memory using compressed formats or uncompressed format using a plurality of compression schemes. A compression ID is used to identify the compression scheme (or no compression) to be used and included with read and write requests submitted to a memory controller. For memory writes, the memory controller dynamically compresses data written to memory cache lines using compression algorithms (or no compression) identified by compression ID. For memory reads, the memory controller dynamically decompresses data stored memory cache lines in compressed formats using decompression algorithms identified by the compression ID. Page tables and TLB entries are augments to include a compression ID field. The format of memory cache lines includes a compression metabit indicating whether the data in the cache line is compressed. Support for DMA reads and writes from IO devices such as GPUs using selective memory compression is also provided.

2. 发明申请

US20190042496A1 DELAYED LINK COMPRESSION SCHEME 审中-公开
公开(公告)号：US20190042496A1
公开(公告)日：2019-02-07
申请号：US16140472
申请日：2018-09-24
申请人： Simon N. Peffers , Kirk S. Yap , Sean Gulley , Vinodh Gopal , Wajdi Feghali
发明人： Simon N. Peffers , Kirk S. Yap , Sean Gulley , Vinodh Gopal , Wajdi Feghali
IPC分类号： G06F13/16 , H04L29/06 , G06F13/28 , G06F13/40
摘要： Apparatus, systems and methods for implementing delayed decompression schemes. As a burst of packets comprising compressed packets and uncompressed packets are received over an interconnect link, they are buffered in a receive buffer without decompression. Subsequently, the packets are forwarded from the receive buffer to a consumer such as processor core, with the compressed packets being decompressed prior to reaching the processor core. Under a first delayed decompression approach, packets are decompressed when they are read from the receive buffer in conjunction with forwarding the uncompressed packet (or uncompressed data contained therein) to the consumer. Under a second delayed decompression scheme, the packets are read from the receive buffer and forwarded to a decompressor using a first datapath width matching the width of the packets, decompressed, and then forwarded to the consumer using a second datapath width matching the width of the uncompressed data.

3. 发明申请

US20220012563A1 METHODS AND APPARATUS FOR HIGH THROUGHPUT COMPRESSION OF NEURAL NETWORK WEIGHTS 有权
公开(公告)号：US20220012563A1
公开(公告)日：2022-01-13
申请号：US17484226
申请日：2021-09-24
申请人： Alejandro Castro Gonzalez , Praveen Nair , Somnath Paul , Sudheendra Kadri , Palanivel Guruvareddiar , Aaron Gubrud , Vinodh Gopal
发明人： Alejandro Castro Gonzalez , Praveen Nair , Somnath Paul , Sudheendra Kadri , Palanivel Guruvareddiar , Aaron Gubrud , Vinodh Gopal
IPC分类号： G06N3/04
摘要： Methods, apparatus, systems, and articles of manufacture are disclosed for high throughput compression of neural network weights. An example apparatus includes at least one memory, instructions in the apparatus and processor circuitry to execute the instructions to determine sizes of data lanes in a partition of neural network weights, determine a slice size based on a size difference between a first data lane and a second data lane of the data lanes in the partition, the first data lane including first data, the second data lane including second data, the second data of a smaller size than the first data, cut a portion of the first data from the first data lane based on the slice size, and append the portion of the first data to the second data lane.

4. 发明申请

US20180260218A1 INSTRUCTION SET ARCHITECTURES FOR FINE-GRAINED HETEROGENEOUS PROCESSING 审中-公开
公开(公告)号：US20180260218A1
公开(公告)日：2018-09-13
申请号：US15452150
申请日：2017-03-07
申请人： Vinodh Gopal
发明人： Vinodh Gopal
IPC分类号： G06F9/26 , G06F9/30 , G06F9/345
CPC分类号： G06F9/44547 , G06F1/3203 , G06F1/3287 , G06F8/447 , G06F8/47 , G06F9/268 , G06F9/30061 , G06F9/345 , G06F9/3557 , G06F9/3877 , G06F9/44521 , G06F9/5066 , G06F15/76
摘要： Instruction set architectures (ISA) for fine-grained heterogeneous processing and associated processors, methods, and compilers. The ISA includes instructions that are configured to be executed on processors having heterogeneous cores implementing different micro-architectures. Mechanisms are provided to enable respective code segments to be compiled/assembled for a target processor (or processor family) with heterogeneous cores and have appropriate code segments that has been compiled for specific types of processor core micro-architectures be dynamically called at run-time via execution of the ISA instructions. The ISA instructions include both unconditional and conditional branch and call instructions, in addition to instructions that support processors with three or more different types of cores. The instructions are configured to support dynamic migration of instruction threads across heterogeneous cores while adding substantially no overhead. A compiler is also provided to generate and assemble opcode segments configured to be executed on processors with heterogeneous cores.

5. 发明申请

US20180095760A1 INSTRUCTION SET FOR VARIABLE LENGTH INTEGER CODING 审中-公开
公开(公告)号：US20180095760A1
公开(公告)日：2018-04-05
申请号：US15281380
申请日：2016-09-30
申请人： James D. Guilford , Vinodh Gopal
发明人： James D. Guilford , Vinodh Gopal
IPC分类号： G06F9/30
CPC分类号： G06F9/3016 , G06F9/3001 , G06F9/30018 , G06F9/30025 , G06F9/30192
摘要： Instruction sets for variable length integer (varint) coding and associated methods and apparatus. The instructions sets include instructions for encoding and decoding varints, and may be included as a part of an instruction set architecture (ISA) for processors architectures such as x86 and Arm-based architectures, as well as other ISAs. In one aspect, the instructions include, a varint size encode instruction to encode a size of a varint, a varint encode instruction to encode a varint, a varint size decode instruction to decode a size of an encoded varint, and a varint decode instruction to decode an encoded varint. Varint encode size and encode instructions may be combined in a single instructions. Similarly, varint decode size and decode instructions may be combined in a single instruction. In one aspect, the instructions use a variable-length quantity (VLQ) encoding scheme under which varints are encoded into one or more VLQ octets.

6. 发明授权

US09632782B2 Method and apparatus to process SHA-2 secure hashing algorithm 有权
公开(公告)号：US09632782B2
公开(公告)日：2017-04-25
申请号：US13976274
申请日：2012-03-30
申请人： Kirk S. Yap , Gilbert M. Wolrich , James D. Guilford , Vinodh Gopal , Erdinc Ozturk , Sean M. Gulley , Wajdi K. Feghali , Martin G. Dixon
发明人： Kirk S. Yap , Gilbert M. Wolrich , James D. Guilford , Vinodh Gopal , Erdinc Ozturk , Sean M. Gulley , Wajdi K. Feghali , Martin G. Dixon
IPC分类号： G06F9/30 , G06F21/60 , H04L9/06 , H04L9/32
CPC分类号： G06F9/3016 , G06F9/30007 , G06F9/30036 , G06F9/30058 , G06F9/30098 , G06F9/30145 , G06F9/3802 , G06F9/384 , G06F12/0875 , G06F12/0897 , G06F12/1027 , G06F15/8007 , G06F21/602 , G06F2212/452 , G06F2212/68 , H04L9/0643 , H04L9/3239 , H04L2209/125
摘要： A processor includes an instruction decoder to receive a first instruction to process a secure hash algorithm 2 (SHA-2) hash algorithm, the first instruction having a first operand associated with a first storage location to store a SHA-2 state and a second operand associated with a second storage location to store a plurality of messages and round constants. The processor further includes an execution unit coupled to the instruction decoder to perform one or more iterations of the SHA-2 hash algorithm on the SHA-2 state specified by the first operand and the plurality of messages and round constants specified by the second operand, in response to the first instruction.

7. 发明授权

US09292548B2 Digest generation 有权
标题翻译：消化一代
公开(公告)号：US09292548B2
公开(公告)日：2016-03-22
申请号：US13995236
申请日：2011-11-01
申请人： Vinodh Gopal , James D. Guilford , Schuyler Eldridge , Gilbert M. Wolrich , Erdinc Ozturk , Wajdi K. Feghali
发明人： Vinodh Gopal , James D. Guilford , Schuyler Eldridge , Gilbert M. Wolrich , Erdinc Ozturk , Wajdi K. Feghali
IPC分类号： G06F17/30
CPC分类号： G06F17/30303 , G06F17/30306 , G06F17/3033
摘要： In one embodiment, circuitry may generate digests to be combined to produce a hash value. The digests may include at least one digest and at least one other digest generated based at least in part upon at least one CRC value and at least one other CRC value. The circuitry may include cyclical redundancy check (CRC) generator circuitry to generate the at least one CRC value based at least in part upon at least one input string. The CRC generator circuitry also may generate the at least one other CRC value based least in part upon at least one other input string. The at least one other input string resulting at least in part from at least one pseudorandom operation involving, at least in part, the at least one input string. Many modifications, variations, and alternatives are possible without departing from this embodiment.
摘要翻译：在一个实施例中，电路可以生成待组合的摘要以产生散列值。摘要可以至少部分地基于至少一个CRC值和至少一个其它CRC值来生成至少一个摘要和至少一个其他摘要。电路可以包括循环冗余校验（CRC）发生器电路，以至少部分地基于至少一个输入串来生成至少一个CRC值。 CRC发生器电路还可以至少部分地基于至少一个其他输入串来生成至少一个其它CRC值。所述至少一个其他输入字符串至少部分地由至少一个涉及至少一个输入字符串的伪随机操作产生。在不脱离本实施例的情况下，可以进行许多修改，变型和替换。

8. 发明授权

US09092645B2 Efficient multiplication, exponentiation and modular reduction implementations 有权
标题翻译：有效的乘法，乘法和模块化削减实现
公开(公告)号：US09092645B2
公开(公告)日：2015-07-28
申请号：US13994782
申请日：2011-12-05
申请人： Erdinc Ozturk , Vinodh Gopal , Gilbert M. Wolrich , Wajdi K. Feghali , James D. Guilford , Deniz Karakoyunlu , Martin G. Dixon , Kahraman D. Akdemir
发明人： Erdinc Ozturk , Vinodh Gopal , Gilbert M. Wolrich , Wajdi K. Feghali , James D. Guilford , Deniz Karakoyunlu , Martin G. Dixon , Kahraman D. Akdemir
IPC分类号： H04L29/00 , G06F21/71 , H04L9/30
CPC分类号： G06F21/71 , G06F7/723 , H04L9/3006 , H04L2209/12
摘要： In one embodiment, the present disclosure provides a method that includes segmenting an n-bit exponent e into a first segment et and a number t of k-bit segments ei in response to a request to determine a modular exponentiation result R, wherein R is a modular exponentiation of a generator base g for the exponent e and a q-bit modulus m, wherein the generator base g equals two and k is based at least in part on a processor configured to determine the result R; iteratively determining a respective intermediate modular exponentiation result for each segment ei, wherein the determining comprises multiplication, exponentiation and a modular reduction of at least one of a multiplication result and an exponentiation result; and generating the modular exponentiation result R=ge mod m based on, at least in part, at least one respective intermediate modular exponentiation result.
摘要翻译：在一个实施例中，本公开提供了一种方法，其包括响应于确定模幂运算结果R的请求，将n位指数e分割成第一段et和数目t的k比特段ei，其中R是指数e的发生器基数g和q位模数m的模幂运算，其中发生器基g等于2，并且k至少部分地基于被配置为确定结果R的处理器; 迭代地确定每个段ei的相应的中间模幂运算结果，其中所述确定包括相乘结果和求幂结果中的至少一个的乘法，乘法和模块化减少; 并且至少部分地基于至少一个相应的中间模幂运算结果来产生模幂运算结果R = ge mod m。

9. 发明申请

US20150082047A1 EFFICIENT MULTIPLICATION, EXPONENTIATION AND MODULAR REDUCTION IMPLEMENTATIONS 有权
标题翻译：有效的实施，授权和模块化的减少实施
公开(公告)号：US20150082047A1
公开(公告)日：2015-03-19
申请号：US13994782
申请日：2011-12-05
申请人： Vinodh Gopal , Gilbert M. Wolrich , Wajdi K. Feghali , James D. Guilford , Deniz Karakoyunlu , Martin G. Dixon , Kahraman D. Akdemir
发明人： Erdinc Ozturk , Vinodh Gopal , Gilbert M. Wolrich , Wajdi K. Feghali , James D. Guilford , Deniz Karakoyunlu , Martin G. Dixon , Kahraman D. Akdemir
IPC分类号： G06F21/71 , H04L9/30
CPC分类号： G06F21/71 , G06F7/723 , H04L9/3006 , H04L2209/12
摘要： In one embodiment, the present disclosure provides a method that includes segmenting an n-bit exponent e into a first segment et and a number t of k-bit segments ei in response to a request to determine a modular exponentiation result R, wherein R is a modular exponentiation of a generator base g for the exponent e and a q-bit modulus m, wherein the generator base g equals two and k is based at least in part on a processor configured to determine the result R; iteratively determining a respective intermediate modular exponentiation result for each segment ei, wherein the determining comprises multiplication, exponentiation and a modular reduction of at least one of a multiplication result and an exponentiation result; and generating the modular exponentiation result R=ge mod m based on, at least in part, at least one respective intermediate modular exponentiation result.
摘要翻译：在一个实施例中，本公开提供了一种方法，其包括响应于确定模幂运算结果R的请求，将n位指数e分割成第一段et和数目t的k比特段ei，其中R是指数e的发生器基数g和q位模数m的模幂运算，其中发生器基g等于2，并且k至少部分地基于被配置为确定结果R的处理器; 迭代地确定每个段ei的相应的中间模幂运算结果，其中所述确定包括相乘结果和求幂结果中的至少一个的乘法，乘法和模块化减少; 并且至少部分地基于至少一个相应的中间模幂运算结果来产生模幂运算结果R = ge mod m。

10. 发明授权

US08953785B2 Instruction set for SKEIN256 SHA3 algorithm on a 128-bit processor 有权
标题翻译： 128位处理器上的SKEIN256 SHA3算法指令集
公开(公告)号：US08953785B2
公开(公告)日：2015-02-10
申请号：US13631143
申请日：2012-09-28
申请人： Gilbert M. Wolrich , Kirk S. Yap , Vinodh Gopal
发明人： Gilbert M. Wolrich , Kirk S. Yap , Vinodh Gopal
IPC分类号： H04L9/28
CPC分类号： H04L9/0643 , G06F9/30007 , G06F9/30032 , G06F9/30036
摘要： According to one embodiment, a processor includes an instruction decoder to receive a first instruction to perform first SKEIN256 MIX-PERMUTE operations, the first instruction having a first operand associated with a first storage location to store a plurality of odd words, a second operand associated with a second storage location to store a plurality of even words, and a third operand. The processor further includes a first execution unit coupled to the instruction decoder, in response to the first instruction, to perform multiple rounds of the first SKEIN256 MIX-PERMUTE operations based on the odd words and even words using a first rotate value obtained from a third storage location indicated by the third operand, and to store new odd words in the first storage location indicated by the first operand.
摘要翻译：根据一个实施例，处理器包括指令解码器，用于接收执行第一SKEIN256 MIX-PERMUTE操作的第一指令，所述第一指令具有与第一存储位置相关联的第一操作数，以存储多个奇数字，第二操作数相关联具有存储多个偶数字的第二存储位置和第三操作数。处理器还包括响应于第一指令而耦合到指令解码器的第一执行单元，使用从第三指令获得的第一旋转值，基于奇数字和偶数字进行第一SKEIN256 MIX-PERMUTE操作的多轮由第三操作数指示的存储位置，并将新的奇数字存储在由第一操作数指示的第一存储位置中。

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式