TI C6000 C代码优化.pptVIP

  1. 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. 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. 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. 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. 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. 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. 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
查看更多
TI C6000 C代码优化

Texas Instruments C6000 DSP C code optimization Optimization Fields Optimization is a system project, there are many fields should be considered: Framework optimization Algorithm optimization Code optimization Memory optimization Framework and Algorithm optimization are application dependent. They are not covered here. Agenda Optimization preparation Basic Optimization methods Software Pipeline Memory/Cache Optimization Other optimization tips and tricks Optimization Procedure After decide the optimization goal, current runtime characteristics should be determined to decide whether the implementation can meet the goal. After getting the benchmark data, the DSP capability need to be analyzed. Then the optimization direction can be determined. For each optimization field, relative optimization methods can be applied to do optimization. This is a multi-loop procedure between profile and optimization. If the goal is met, current procedure can terminate. If not, other optimization methods should be used or the goal should be adjusted accordingly. C66x DSP VLIW Architecture Two (almost independent) sides, A and B 8 functional units, M, L, S, D Up to 8 instructions sustained dispatch rate Basic capabilities of C64x+ core Operation Precision Operations per cycle Function Units Notes MAC Real 8 x 8 2 x 4 = 8 M1, M2 1 Real 16 x 16 2 x 2 = 4 Real 32 x 32 2 x 1 = 2 Complex (16,16)x(16,16) 2 x 1 = 2 Complex (32,32)x(32,32) N/A Memory Access 8 bit 16 bit 32 bit 64 bit 2 x 1 = 2 D1, D2 2, 3 Arithmetic Logical 8 bit 4 x 4 = 16 L1, L2, S1, S2 16 bit 4 x 2 = 8 32 bit 4 x 1 = 4 1, if no multiplication, M1, M2 may be used for some other operations. 2, if no memory read/write, D1, D2 may be used for other operations. 3, try to used 64 bit read/write as much as possible. 4, division and module is not supported with single instruction, try to avoid it in your algorithm Notes Basic capabilities of C66x core Operation Precision Operations per cycle Function Units Notes MAC Real 8 x 8

文档评论(0)

xxj1658888 + 关注
实名认证
文档贡献者

教师资格证持证人

该用户很懒,什么也没介绍

领域认证该用户于2024年04月12日上传了教师资格证

1亿VIP精品文档

相关文档