Panel Discussion High Level Design and ESL Who Cares.pptVIP

下载本文档

0
0
约2.46千字
约 6页
2016-10-07 发布于重庆
举报
版权申诉

Panel Discussion High Level Design and ESL Who Cares.ppt

1、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。。
2、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
3、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。
4、该文档为VIP文档，如果想要下载，成为VIP会员后，下载免费。
5、成为VIP后，下载本文档将扣除1次下载权益。下载后，不支持退款、换文档。如有疑问请联系我们。
6、成为VIP后，您将拥有八大权益，权益包括：VIP文档下载权益、阅读免打扰、文档格式转换、高级专利检索、专属身份标志、高级客服、多端互通、版权登记。
7、VIP文档为合作方或网友上传，每下载1次，网站将根据用户上传文档的质量评分、类型等，对文档贡献者给予高额补贴、流量扶持。如果你也想贡献VIP文档。上传文档

Panel Discussion High Level Design and ESL Who Cares

Panel DiscussionHigh Level Design and ESL: Who Cares? Masahiro Fujita VLSI Design and Education Center (VDEC) University of Tokyo Questions from the organizer (1) With HL design models verification becomes easier or harder ? HL models could be golden for RTL verification This is, I believe, why industrial people are writing HL models There are RTL specific properties to be verified, such as clock crossing issues Questions from the organizer (2) Starting with HL is good in general ? Good for more application specific ones, such as SoC Good for the designs where high level synthesis tools work pretty well, such as DSP type computation This is, I believe, why industrial people are writing HL models May not be good for microprocessor designs where performance are critical, and designs themselves are very similar to one another How about BlueSpec ? It may be useful Questions from the organizer (3) Rationale to start with HL (cost effective ?) Industrial designers (architects) like to have golden model in the beginning to make sure the overall behaviors Then, how can we use it ? Also they like to explore more design space with HL models Need reasonably good high/system-level synthesis tools For HW/SW co-design, yes Discuss in the next slides Equivalence checking may not be so powerful For speed up of scientific computing, must be Discuss in the following Algorithmic design description High level synthesizable description Register Transfer Level description Many steps of manual refinement Many steps of manual refinement Design optimization High level synthesis Design optimization Design optimization Static/ Model checking Sequential Equivalence Checking High level design flow and verification (Ideal case for HW/SW co-design) Tools have been actually developed under Japanese Space Agency (JAXA) JAXA, Toshiba, NEC, Fujitsu, UCI (Prof. Gajski) and U. of Tokyo Really used for space satellite designs in Japan My conclusions If you can have correct RTL by itself, you do no