VOBLA: A Vehicle for Optimized Basic Linear Algebra

被引：0

作者：

Beaugnon, Ulysse ^{[1
,2
,3
]}

Kravets, Alexey ^{[1
]}

van Haastregt, Sven ^{[1
]}

Baghdadi, Riyadh ^{[2
,3
]}

Tweed, David ^{[1
]}

Absar, Javed ^{[1
]}

Lokhmotov, Anton ^{[1
]}

机构：

[1] ARM, Berlin, Germany

[2] INRIA, Paris, France

[3] Ecole Normale Super, Paris, France

来源：

ACM SIGPLAN NOTICES | 2014年 / 49卷 / 05期

关键词：

Algorithms; languages; performance; linear algebra; GPU; domain-specific language; parallel; sparse matrix; BLAS;

D O I：

10.1145/2666357.2597818

中图分类号：

TP31 [计算机软件];

学科分类号：

081202 ; 0835 ;

摘要：

We present VOBLA, a domain-specific language designed for programming linear algebra libraries. VOBLA is compiled to PENCIL, a domain independent intermediate language designed for efficient mapping to accelerator architectures such as GPGPUs. PENCIL is compiled to efficient, platform-specific OpenCL code using techniques based on the polyhedral model. This approach addresses both the programmer productivity and performance portability concerns associated with accelerator programming. We demonstrate our approach by using VOBLA to implement a BLAS library. We have evaluated the performance of OpenCL code generated using our compilation flow on ARM Mali, AMD Radeon, and AMD Opteron platforms. The generated code is currently on average 1 : 9 x slower than highly hand-optimized OpenCL code, but on average 8 : 1 x faster than straightforward OpenCL code. Given that the VOBLA coding takes significantly less effort compared to hand-optimizing OpenCL code, we believe our approach leads to improved productivity and performance portability.

引用

页码：115 / 124

页数：10

共 17 条

[1] [Anonymous], 2013, GPR, DOI [DOI 10.1145/2400682.2400713, DOI 10.19476/J.YSXB.1004.0609.2013.09.014]
[2] Antoine C.W., 2000, Parallel Computing, V27, P2001
[3] Baghdadi R., 2012, WORKSH DOM SPEC LANG
[4] Faucher A., 2013, C AMP BLAS
[5] Fowler M., 2011, DOMAIN SPECIFIC LANG
[6] Goto K. ., 2013, GOTOBLAS TEXAS ADV C
[7] Worst Practices for Domain-Specific Modeling
Kelly, Steven
Pohjonen, Risto
[J]. IEEE SOFTWARE, 2009, 26 (04) : 22 - 29
[8] Kravets A., 2014, VOBLA PENSIL TOOLS
[9] Lawson C. L., 1979, ACM Transactions on Mathematical Software, V5, P308, DOI 10.1145/355841.355848
[10] IMPLEMENTING DOMAIN-SPECIFIC LANGUAGES FOR HETEROGENEOUS PARALLEL COMPUTING
Lee, HyoukJoong
Brown, Kevin J.
Sujeeth, Arvind K.
Chafi, Hassan
Olukotun, Kunle
Rompf, Tiark
Odersky, Martin
[J]. IEEE MICRO, 2011, 31 (05) : 42 - 52

← 1 2 →