Á¤º¸°úÇÐȸ ³í¹®Áö C : ÄÄÇ»ÆÃÀÇ ½ÇÁ¦
Current Result Document : 2 / 2
ÇѱÛÁ¦¸ñ(Korean Title) |
CPU-GPU À̱âÁ¾ Ç÷§Æû¿¡¼ ÇÏµÓ ¸Ê¸®µà½ºÀÇ °¡¼Ó: CKY Æļ »ç·Ê ºÐ¼® |
¿µ¹®Á¦¸ñ(English Title) |
Accelerating Hadoop MapReduce on CPU-GPU Heterogeneous Platforms: A Case Study with CKY Parser |
ÀúÀÚ(Author) |
ÀÌ»õÇѽ½
ÀÌ¿µ¹Î
Sae-han-seul Yi
Youngmin Yi
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 20 NO. 06 PP. 0329 ~ 0338 (2014. 06) |
Çѱ۳»¿ë (Korean Abstract) |
ºòµ¥ÀÌÅÍ ½Ã´ë°¡ µµ·¡ÇÔ¿¡ µû¶ó ÇÏµÓ ¸Ê¸®µà½º¿Í °°Àº ÀÀ¿ëÀÌ ³Î¸® »ç¿ëµÇ°í ÀÖ´Ù. ÇÑÆí, ÃÖ±Ù GPGPU°¡ º¸ÆíȵǸé¼, ´Ù¾çÇÑ ºÐ¾ßÀÇ ÀÀ¿ëµéÀÌ GPU¸¦ ÀÌ¿ëÇÏ¿© °¡¼ÓµÇ°í ÀÖ´Ù. º» ³í¹®Àº ÇÏµÓ ¸Ê¸®µà½º¿¡¼ GPU¸¦ »ç¿ëÇÏ´Â ¹æ¹ýÀ» Á¦¾ÈÇÏ°í, CPU¿Í GPU°¡ ¸ðµÎ Æ÷ÇԵǴ À̱âÁ¾ ¼¹ö·Î ±¸¼ºµÈ ºÐ»êȯ°æ¿¡¼ ÃÖÀûÀÇ µ¥ÀÌÅÍ Ã³¸® ¼Óµµ¸¦ ¾ò±â À§ÇØ, CPU¿Í GPU¿¡ °¢°¢ ÇÒ´çµÇ´Â ¸Ê ŽºÅ©µé¿¡ ´ëÇÑ Á¤ÀûºÐÇÒ ¹× µ¿Àû ½ºÄÉÁÙ¸µ¿¡ ´ëÇÑ ±â¹ýÀ» Á¦¾ÈÇÏ¿´´Ù. ³ëµå¸¶´Ù 12°³ÀÇ CPU ÄÚ¾î¿Í 1°³ÀÇ GPU°¡ ÀåÂøµÈ 14-³ëµå Ŭ·¯½ºÅÍ È¯°æ¿¡¼ ÇÏµÓ ¸Ê¸®µà½º·Î CKY Æļ ÀÀ¿ëÀ» ¼öÇàÇÏ¿©, CPU ÄÚ¾î 1°³¸¸ »ç¿ëÇÑ ´ÜÀÏ ¼¹ö¿¡¼ÀÇ ¼öÇà½Ã°£ ´ëºñ 245¹è °¡¼ÓÀ» ÇÏ¿´°í, ³ëµåº°·Î GPU¸¦ »ç¿ëÇÏÁö ¾Ê°í CPU ÄÚ¾î 12°³¸¸ È°¿ëÇÏ´Â µ¿ÀÏ ÇÏµÓ Å¬·¯½ºÅÍ¿¡¼ÀÇ ¼öÇà½Ã°£ ´ëºñ 2.5¹è °¡¼ÓÀ» ÇÏ¿´´Ù. ¶ÇÇÑ Á¦¾ÈÇÏ´Â ±â¹ýÀ¸·Î CPU ÄÚ¾î12°³¿Í GPU¸¦ ¸ðµÎ »ç¿ëÇÏ´Â ÇÏµÓ Å¬·¯½ºÅÍ ¼öÇà½Ã°£ ´ëºñ ÃÑ 2.8¹è °¡¼ÓÀÌ µÇ¾ú´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
These days, big data computing is prevalent and Hadoop MapReduce framework is widely used for its simple programming model. On the other hand, General-Purpose Graphics Processing Unit (GPGPU) has become very popular and various domains of applications have been successfully accelerated using GPUs. In this paper, we propose a method to use GPU within Hadoop MapReduce framework. Then, we propose a static partitioning method that considers different capability of CPU mappers and GPU mappers, and a dynamic scheduling method that deals with a dynamic input size. Compared to a single CPU execution time, the CKY parser on a 14-node Hadoop cluster with 12 CPU cores and 1 GPU per node achieves 245 times speedup. Compared to the execution time on a 14-node Hadoop cluster with 12 CPU cores and no GPU per node, it also achieves 2.5 times speedup. Our proposed approach for both CPU and GPU mapper execution leads to an additional speedup, resulting in total of 2.8 times speedup.
|
Å°¿öµå(Keyword) |
ÇÏµÓ ¸Ê¸®µà½º
GPU ¸ÅÆÛ
CPU-GPU À̱âÁ¾ Ç÷§Æû
CKY Æļ
hadoop MapReduce
GPU mapper
CPU-GPU heterogeneous platform
CKY parser
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|