[B! gpgpu] yassã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

yass id:yass

gpgpuã«é–¢ã™ã‚‹yassã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (28)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

SSD-to-GPU Peer-to-Peer DMAã¨ãƒãƒƒãƒ•ã‚¡ç®¡ç†ï¼ˆãã®ï¼‘ï¼‰ - KaiGaiã®ä¿ºãƒ¡ãƒ¢
æ˜¨å¹´ã®æš®ã‚Œã€JPUGã‚«ãƒ³ãƒ•ã‚¡ãƒ¬ãƒ³ã‚¹ã®LTã§ã€ŽSQL+GPU+SSD=âˆžã€ã¨é¡Œã—ãŸã‚¹ãƒ”ãƒ¼ãƒã‚’è¡Œã£ãŸã€‚ SQL+GPU+SSD=âˆž (Japanese) from Kohei KaiGai www.slideshare.net ã“ã‚Œã¯ã‹ã„ã¤ã¾ã‚“ã§è¨€ãˆã°ã€ã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸ã‹ã‚‰ãƒ‡ãƒ¼ã‚¿ã‚’CPU+RAMã¸ã¨ãƒãƒ¼ãƒ‰ã™ã‚‹ã‚ˆã‚Šå‰ã«ä¸€æ—¦GPUã¸ã¨ãƒ‡ãƒ¼ã‚¿ã‚’è»¢é€ã—ã€ãã“ã§ä¸è¦ãªãƒ‡ãƒ¼ã‚¿ã‚’å‰ŠãŽè½ã—ã¦ã‹ã‚‰CPU+RAMã¸ã¨æ¸¡ã—ã¦ã‚„ã‚‹äº‹ã§ã€CPUè² è·ã®è»½æ¸›ã¨RAMã®æœ‰åŠ¹æ´»ç”¨ãŒè¨ˆã‚Œã‚‹ã¨ã„ã†ã‚¢ã‚¤ãƒ‡ã‚¢ã§ã‚ã‚‹ã€‚ å®Ÿè£…ã¨ã—ã¦ã¯ã€PCI-Eãƒ‡ãƒã‚¤ã‚¹é–“ã§ã®P2P DMAæ©Ÿèƒ½ã‚’åˆ©ç”¨ã™ã‚‹äº‹ã«ã‚ˆã£ã¦NVMe SSDã®ç‰¹å®šãƒ–ãƒãƒƒã‚¯ã‹ã‚‰GPU RAMä¸Šã®ç‰¹å®šã®é ˜åŸŸã¸DMAã‚’å®Ÿè¡Œã™ã‚‹ã¨ã„ã†ã‚‚ã®ãªã®ã§ã€ã“ã“ã¯åˆ¥ã«æ–°ã—ãã‚‚ä½•ã¨ã‚‚ãªã„ã€‚ ä»¥ä¸‹ã®å›³ã¯ã€å¾“æ¥ã®ä»•çµ„ã¿ã«ãŠã‘ã‚‹ãƒ‡ãƒ¼ã‚¿ã®æµã‚Œã‚’ç¤ºã—ãŸã‚‚ã®ã€‚ SSDã‹ã‚‰èªã¿å‡ºã•ã‚ŒãŸãƒ‡ãƒ¼ã‚¿ã¯å…ˆãšCPU+RAMã«ãƒãƒƒãƒ•ã‚¡ã•ã‚Œ
yass 2016/02/14
SSD

GPU

gpgpu

dma
ãƒªãƒ³ã‚¯
OpenCLã‚„ã‚‹å‰ã«SIMDä½¿ã„åˆ‡ã‚Œã£ã¦ã„ã†å¹»æƒ³ - aokomoriuta's blog
ãŠè©«ã³ ã‚¢ãƒ©ã‚¤ãƒ³å¿˜ã‚Œã¦ã¾ã—ãŸã”ã‚ã‚“ãªã•ã„ã€ã§ã‚‚æ™‚é–“ã‚ã‚“ã¾ã‚Šå¤‰ã‚ã£ã¦ãªã„ã‹ã‚‰è¨±ã—ã¦ãƒ»ãƒ»ãƒ» https://t.co/JNtq2U2kMqâ€” é’åå®ˆæŒ (@aokomoriuta) April 29, 2015 ã§ã¯æœ¬ç·¨ã©ã†ãžâ†“ æœ¬ç·¨ è‹¥å¹²è©±é¡Œã«ä¹—ã‚Šé…ã‚ŒãŸæ„Ÿã‚ã‚Šã¾ã™ãŒã€‚ d.hatena.ne.jp ã‘ã©ã€SSEã‚‚çŸ¥ã‚‰ããƒ¼ã€SIMDã‚‚çŸ¥ã‚‰ããƒ¼ã€ãªã‚“ã‹ä¿ºãŒæ›¸ã„ãŸã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ é…ã„ã‘ã©ã¨ã‚Šã‚ãˆãšOpenCLã¨ã‹ã§é«˜é€ŸåŒ–ã—ã‚ˆã£ã‹ãªãƒ¼ã¨ã‹ãã€ç”˜ã£ãŸã‚Œã¦ã‚“ã˜ã‚ƒããˆã‚ˆã€‚CPUã‚’ã‚‚ã£ã¨ä½¿ã„ãã‚Œã‚ˆã€‚ãŠå‰ã®ã‚¢ã‚ªã„ã‚³ãƒ¼ãƒ‰ã®ã›ã„ã§CPUãŒæ³£ã„ã¦ã‚‹ã‚ˆã€‚ã£ã¦ã„ã†è©±ã§ã™ã‚ˆã€‚ GPGPUãªã‚“ã¦ã®ã¯SIMDã‚’ä½¿ã„åˆ‡ã£ãŸå¾Œã®è©±ã§ã—ã‚‡ã€‚ GPGPUã™ã‚‹ã®ã«GPUã®ãƒ‘ãƒ¯ãƒ¼ã¨ãƒ¡ãƒ¢ãƒªãŒè¶³ã‚Šã¾ã›ã‚“ã¨ã‹è¨€ã†å‰ã«ã¾ãšSIMDã‹ã‚‰ã ã‚ã€‚ ã¨ã‹è¨€ã‚ã‚Œã¦ãŸã®ã§ã€æ¤œè¨¼ã™ã‚‹ã“ã¨ã«ã—ãŸï¼ˆã‚„ã£ã¤ã‘ï¼‰ã€‚ ç’°å¢ƒ OS: Windows 7 Profession
yass 2015/04/29
AVX

opencl

GPGPU

simd
ãƒªãƒ³ã‚¯
NVIDIA cuDNN | NVIDIA Developer
Accelerated Learning cuDNN provides kernels, targeting Tensor Cores, to deliver best available performance on compute-bound operations. It offers heuristics for choosing the right kernel for a given probl em size. Expressive Op Graph API The user defines computations as a graph of operations on tensors. The cuDNN library has both a direct C API and an open-source C++ frontend for convenience. Most
yass 2014/09/08
" NVIDIA cuDNN is a GPU-accelerated library of primitives for deep neural networks.Â "

machinelearning

gpu

gpgpu
ãƒªãƒ³ã‚¯
ã€å¾Œè—¤å¼˜èŒ‚ã®Weeklyæµ·å¤–ãƒ‹ãƒ¥ãƒ¼ã‚¹ã€‘ AMDã®å®Œæˆå½¢APUã€ŒKaveriã€ã®ã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£
yass 2014/01/14
" ä»Šå›žã‚µãƒãƒ¼ãƒˆã•ã‚ŒãŸãƒ¡ãƒ¢ãƒªãƒ¢ãƒ‡ãƒ«ã€ŒhUMAã€ã¨ã‚ãƒ¥ãƒ¼ã‚¤ãƒ³ã‚°ãƒ¢ãƒ‡ãƒ«ã€ŒhQã€ã¯ã€ä¸ã§ã‚‚æœ€é‡è¦ã®æ©Ÿèƒ½ã ã€‚ãªãœãªã‚‰ã€ã“ã®2ã¤ã®æ©Ÿèƒ½ã«ã‚ˆã£ã¦ã€GPUã§èµ°ã‚‹ã‚°ãƒ©ãƒ•ã‚£ãƒƒã‚¯ã‚¹ä»¥å¤–ã®ãƒ—ãƒã‚°ãƒ©ãƒ ã‚’æ›¸ãã“ã¨ãŒé£›èºçš„ã«ç°¡å˜ã«ãªã‚‹ã‹ã‚‰ã "

AMD

cpu

gpu

gpgpu

hsa
ãƒªãƒ³ã‚¯
The Vector Programming Language
Vector was our teamâ€™s final project for COMS 4115: Programming Languages and Translators. Vector is a high-level language for general-purpose GPU computing. Over the course of the semester, our team developed a compiler that compiled Vector into CUDA, which could then be compiled to run on Nvidia GPUs. My teammates were Zachary Newman, Sidharth Shanker, Jonathan Yu, and Harry Lee. Rationale Genera
yass 2013/12/22
gpgpu

cuda

programming
ãƒªãƒ³ã‚¯
GPUã§ãƒ¬ã‚³ãƒ¡ãƒ³ãƒ‡ãƒ¼ã‚·ãƒ§ãƒ³ã®è¨ˆç®—ã®é«˜é€ŸåŒ– - Thinking out loud
å®Ÿéš›ã®ã¨ã“ã‚ã€ãƒªã‚¢ãƒ«ã‚¿ã‚¤ãƒ ã§ãƒ¬ã‚³ãƒ¡ãƒ³ãƒ‡ãƒ¼ã‚·ãƒ§ãƒ³ã‚’èµ°ã‚‰ã›ã‚‹ãŸã‚ã«ã¯ã€é€Ÿã„è¨ˆç®—æ©ŸãŒå¿…è¦ã§ã‚ã‚‹ã€‚ã€€è¨ˆç®—é‡ã‚’é£Ÿã‚ãªã„ç°¡ç•¥åŒ–ã—ãŸãƒã‚¸ãƒƒã‚¯ã‚’å‹•ã‹ã™ã¨ã‹ã€è¨ˆç®—æ©Ÿã‚’ã‚¯ãƒ©ã‚¹ã‚¿åŒ–ã—ã¦ç’°å¢ƒã‚’æ§‹ç¯‰ã™ã‚‹ã€ãªã©ã®ã‚¢ãƒ—ãƒãƒ¼ãƒã‚‚ã‚ã‚‹ãŒã€GPUã‚’ä½¿ã£ã¦è¨ˆç®—ã‚’é«˜é€ŸåŒ–ã™ã‚‹ã¨ã„ã†ã‚¢ãƒ—ãƒãƒ¼ãƒã¯ãªã‹ãªã‹è¦‹ã‹ã‘ãªã„ã®ã§é¢ç™½ã„ã€‚ ï¼ˆç¤¾å†…ã§ã‚„ã£ã¦ã„ã‚‹ã¨ã“ã‚ã¯ã‚ã‚‹ã‹ã‚‚ã—ã‚Œãªã„ã‘ã‚Œã©ã€å…¬é–‹ã‚’ã—ã¦ã„ãªã„ã ã‘ã€ã ã£ãŸã‚Šã—ã¦ã€‚ï¼‰ "You Might Also Like: A Multi-GPU Recommendation System" You Might Also Like: A Multi-GPU Recommendation System from NVIDIA GPUæ¼”ç®—ã§ã€(SimpleãªãŒã‚‰ã‚‚)ãƒ¬ã‚³ãƒ¡ãƒ³ãƒ‡ãƒ¼ã‚·ãƒ§ãƒ³(recommendation)ã®è¨ˆç®—ãŒã€CPUã¨æ¯”ã¹ã¦20å€ã€œ300å€ã®æ—©ã•ã«ãªã£ãŸã‚ˆã€ã¨ã„ã†ã‚¹ãƒ©ã‚¤ãƒ‰ã€‚ è¦ã¯ã€Util
yass 2013/12/07
gpu

gpgpu

recommend
ãƒªãƒ³ã‚¯
CUDA 6é€Ÿå ±(1): Unified Memory - ãƒˆãƒ¼ã‚¿ãƒ«ãƒ»ãƒ‡ã‚£ã‚¹ã‚¯ãƒãƒ¼ã‚¸ãƒ£ãƒ»ã‚µã‚¤ãƒˆï¼ˆäº‹å®Ÿã‚’ã‚ã‚Šã®ã¾ã¾ã«ï¼‰
å…ˆæ—¥ç™ºè¡¨ã•ã‚ŒãŸCUDA 6ã«ã¤ã„ã¦ã€NVIDIAãŒç´¹ä»‹ã—ã¦ã„ã‚‹æ–°æ©Ÿèƒ½ã‚’ã„ãã¤ã‹ç´¹ä»‹ã™ã‚‹ã€‚ãªãŠã€Toolkitã¯ã¾ã è©•ä¾¡ã§ãã‚‹çŠ¶æ…‹ã§ã¯ãªã„ãŸã‚ã€NVIDIAã®ç´¹ä»‹ã‹ã‚‰ã®æŽ¨æ¸¬ã«ã‚ˆã‚‹è¨˜äº‹ã§ã‚ã‚‹ã€‚ãã®ãŸã‚ã€æ£ç¢ºæ€§ã«æ¬ ã‘ã‚‹å¯èƒ½æ€§ã«ã¤ã„ã¦æ°—ã‚’ä»˜ã‘ãŸã„ã€‚ Unified Memoryã¨ã¯ CPUã¨GPUã¨ã®ä¸¡æ–¹ã‹ã‚‰ã‚¢ã‚¯ã‚»ã‚¹ã•ã‚Œã‚‹ã“ã¨ã‚’ç›®çš„ã¨ã—ãŸãƒ¡ãƒ¢ãƒªé ˜åŸŸã§ã‚ã‚‹ã€‚CPUã¨GPUã¨ã®ä¸¡æ–¹ã«ãƒ¡ãƒ¢ãƒªã‚’ç¢ºä¿ã—ã€ãƒ‡ãƒ¼ã‚¿ã®ã‚³ãƒ’ãƒ¼ãƒ¬ãƒ³ã‚·ã‚’ä¿ã¤ã‚ˆã†ã«é©å®œåŒæœŸãŒè¡Œã‚ã‚Œã‚‹ã‚ˆã†ãªå®Ÿè£…ã§ã‚ã‚‹ã¨èª¬æ˜Žã•ã‚Œã¦ã„ã‚‹ã€‚ ã—ã°ã—ã°ã€ŒCPUã¨GPUã®ä¸¡æ–¹ã®ãƒ¡ãƒ¢ãƒªã«ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ãŒç›´æŽ¥ã‚¢ã‚¯ã‚»ã‚¹ã§ãã‚‹ã‚ˆã†ã«ãªã‚‹ã€ã¨ã„ã†ã‚ˆã†ãªè¡¨ç¾ã§ç´¹ä»‹ã—ã¦ã„ã‚‹ãƒ‹ãƒ¥ãƒ¼ã‚¹ã‚µã‚¤ãƒˆãŒã‚ã‚‹ãŒã€CPUä¸Šã®ãƒ¡ãƒ¢ãƒªã‚’GPUãŒç›´æŽ¥ã‚¢ã‚¯ã‚»ã‚¹ã™ã‚‹æ©Ÿèƒ½ã¯ã™ã§ã«CUDA 2.2ã®ã¨ãã«ã€ŒMapped Memoryã€ã¨ã„ã†æ©Ÿèƒ½ã§å®Ÿç¾ã•ã‚Œã¦ã„ã‚‹ã€‚ã•ã‚‰ã«ã€CUDA 4.0ã®ã¨ãã«å®Ÿè£…ã•ã‚ŒãŸã€Œ
yass 2013/11/23
" Unified Memoryã¨ã¯ CPUã¨GPUã¨ã®ä¸¡æ–¹ã‹ã‚‰ã‚¢ã‚¯ã‚»ã‚¹ã•ã‚Œã‚‹ã“ã¨ã‚’ç›®çš„ã¨ã—ãŸãƒ¡ãƒ¢ãƒªé ˜åŸŸã§ã‚ã‚‹ã€‚CPUã¨GPUã¨ã®ä¸¡æ–¹ã«ãƒ¡ãƒ¢ãƒªã‚’ç¢ºä¿ã—ã€ãƒ‡ãƒ¼ã‚¿ã®ã‚³ãƒ’ãƒ¼ãƒ¬ãƒ³ã‚·ã‚’ä¿ã¤ã‚ˆã†ã«é©å®œåŒæœŸãŒè¡Œã‚ã‚Œã‚‹ã‚ˆã†ãªå®Ÿè£…ã§ã‚ã‚‹ã¨èª¬æ˜Žã•ã‚Œã¦ã„ã‚‹ã€‚"

cuda

gpgpu
ãƒªãƒ³ã‚¯
ã‚µãƒ¼ãƒ“ã‚¹çµ‚äº†ã®ãŠçŸ¥ã‚‰ã›
ã‚µãƒ¼ãƒ“ã‚¹çµ‚äº†ã®ãŠçŸ¥ã‚‰ã› ã„ã¤ã‚‚Yahoo! JAPANã®ã‚µãƒ¼ãƒ“ã‚¹ã‚’ã”åˆ©ç”¨ã„ãŸã ãèª ã«ã‚ã‚ŠãŒã¨ã†ã”ã–ã„ã¾ã™ã€‚ ãŠå®¢æ§˜ãŒã‚¢ã‚¯ã‚»ã‚¹ã•ã‚ŒãŸã‚µãƒ¼ãƒ“ã‚¹ã¯æœ¬æ—¥ã¾ã§ã«ã‚µãƒ¼ãƒ“ã‚¹ã‚’çµ‚äº†ã„ãŸã—ã¾ã—ãŸã€‚ ä»Šå¾Œã¨ã‚‚Yahoo! JAPANã®ã‚µãƒ¼ãƒ“ã‚¹ã‚’ã”æ„›é¡§ãã ã•ã„ã¾ã™ã‚ˆã†ã€ã‚ˆã‚ã—ããŠé¡˜ã„ã„ãŸã—ã¾ã™ã€‚
yass 2013/11/23
" Amazonã®ã‚¯ãƒ©ã‚¦ãƒ‰ã§NVIDIAã®GRID GPU / GRID K2 GPUã¯GK104ãƒãƒƒãƒ—ã‚’ä½¿ã£ã¦ãŠã‚Šï¼Œå€ç²¾åº¦ã®æµ®å‹•å°æ•°ç‚¹æ¼”ç®—æ€§èƒ½ã¯ï¼Œ2å€‹ã§190GFlopsã¨ã‚ã¾ã‚Šé«˜ãã¯æœ‰ã‚Šã¾ã›ã‚“ãŒï¼Œå˜ç²¾åº¦ã®æ€§èƒ½ã¯2å€‹ã§4577GFlopsã¨é«˜ã ï¼Œ3Dã‚°ãƒ©ãƒ•ã‚£ãƒƒã‚¯ã‚¹ã§ã¯é«˜ã„æ€§èƒ½ã‚’ç™ºæ®"

ec2

gpu

gpgpu

nvidia
ãƒªãƒ³ã‚¯
About NVIDIA
çµ±åˆãƒ¡ãƒ¢ãƒªã€ãƒ‰ãƒãƒƒãƒ—ã‚¤ãƒ³ãƒ»ãƒ©ã‚¤ãƒ–ãƒ©ãƒªãªã©ã®ãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°æ©Ÿèƒ½ã«ã‚ˆã‚Šã€ æ¬¡ä¸–ä»£ã®GPUé–‹ç™ºè€…ã‚’æ”¯æ´ 2013å¹´11æœˆ14æ—¥ ï¼ ã‚«ãƒªãƒ•ã‚©ãƒ«ãƒ‹ã‚¢å·žã‚µãƒ³ã‚¿ã‚¯ãƒ©ãƒ© ï¼NVIDIAï¼ˆæœ¬ç¤¾ï¼šç±³å›½ã‚«ãƒªãƒ•ã‚©ãƒ«ãƒ‹ã‚¢å·žã‚µãƒ³ã‚¿ã‚¯ãƒ©ãƒ©ã€ç¤¾é•·å…¼CEOï¼š ã‚¸ã‚§ãƒ³ã‚¹ãƒ³ãƒ»ãƒ•ã‚¢ãƒ³(Jen-Hsun Huang)ã€Nasdaqï¼šNVDAï¼‰ã¯æœ¬æ—¥ã€ä¸–ç•Œã§æœ€ã‚‚æ™®åŠã—ã¦ã„ã‚‹ä¸¦åˆ—ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ãƒ†ã‚£ãƒ³ã‚°ã®ãƒ—ãƒ©ãƒƒãƒˆãƒ•ã‚©ãƒ¼ãƒ ãŠã‚ˆã³ãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ãƒ¢ãƒ‡ãƒ«ã®æœ€æ–°ãƒãƒ¼ã‚¸ãƒ§ãƒ³ã€NVIDIAÂ® CUDAÂ® 6ã‚’ç™ºè¡¨ã—ã¾ã—ãŸã€‚ CUDA 6ãƒ—ãƒ©ãƒƒãƒˆãƒ•ã‚©ãƒ¼ãƒ ã§ã¯ä¸¦åˆ—ãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ãŒã„ã¾ã¾ã§ä»¥ä¸Šã«ã‚„ã‚Šã‚„ã™ããªã‚Šã€ç§‘å¦ã‚„å·¥å¦ã€ã‚¨ãƒ³ã‚¿ãƒ¼ãƒ—ãƒ©ã‚¤ã‚ºã®ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã‚’GPUã§é«˜é€ŸåŒ–ã™ã‚‹ãŸã‚ã«å¿…è¦ãªæ™‚é–“ã¨åŠ´åŠ›ãŒå¤§å¹…ã«å‰Šæ¸›ã•ã‚Œã¾ã™ã€‚ ã„ã¾ã¾ã§ä½¿ã£ã¦ã„ãŸCPUãƒ™ãƒ¼ã‚¹ã®ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã‚’ç½®ãæ›ãˆã‚‹ã ã‘ã§å®Ÿè¡Œé€Ÿåº¦ã‚’æœ€å¤§8å€ã«é«˜ã‚ã‚‰ã‚Œã‚‹ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ã®æ‹¡å¼µæ©Ÿèƒ½ã‚‚ç”¨æ„ã•ã‚Œã¦ã„ã¾ã™ã€‚CUD
yass 2013/11/23
" çµ±åˆãƒ¡ãƒ¢ãƒªï¼CPUã¨GPUã®ãƒ¡ãƒ¢ãƒªã¸ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ãŒè‡ªç”±ã«ã‚¢ã‚¯ã‚»ã‚¹ã§ãã‚‹ãŸã‚ã€ãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ãŒã‚·ãƒ³ãƒ—ãƒ«ã«ãªã‚Šã¾ã™ã€‚ä¸¡è€…ã®é–“ã§ãƒ‡ãƒ¼ã‚¿ã‚’ã„ã¡ã„ã¡ã‚³ãƒ”ãƒ¼ã™ã‚‹å¿…è¦ãŒãªããªã‚‹ã®ã§ã™ã€‚"

CUDA

gpgpu

gpu
ãƒªãƒ³ã‚¯
ã€é€£è¼‰ã€‘Kepler GPUã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£ã¨ãƒ—ãƒã‚°ãƒ©ãƒ æœ€é©åŒ–
NVIDIAã®ã€ŒGK110(Kepler)ã€GPUã‚’ä½¿ã†ã€ŒTesla K20/K20xã€ã¨ã€ŒGeForce Titanã€ã¯ã€ç¾çŠ¶ã§ã¯æœ€å¼·ã®ç§‘å¦æŠ€è¡“è¨ˆç®—èƒ½åŠ›ã‚’èª‡ã£ã¦ãŠã‚Šã€K20xã¯2012å¹´11æœˆã®Top500ã§1ä½ã‚’ç²å¾—ã—ãŸã‚ªãƒ¼ã‚¯ãƒªãƒƒã‚¸å›½ç«‹ç ”ç©¶æ‰€ã®Titanã‚¹ãƒ‘ã‚³ãƒ³ã«ã‚‚æŽ¡ç”¨ã•ã‚Œã¦ã„ã‚‹ã€‚GPUã¯é«˜ã„æµ®å‹•å°æ•°ç‚¹æ¼”ç®—èƒ½åŠ›ã‚’æŒã£ã¦ã„ã‚‹ãŒã€å˜ã«ã€æµ®å‹•å°æ•°ç‚¹æ¼”ç®—å™¨ã®æ•°ã‚’å¢—ã‚„ã—ã¦å¾“æ¥ã®CPUã‚’ç›¸ä¼¼å½¢ã§æ‹¡å¤§ã—ãŸã‚‚ã®ã§ã¯ãªãã€ã‹ãªã‚Šç•°ãªã£ãŸä½œã‚Š(ã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£)ã¨ãªã£ã¦ã„ã‚‹ã€‚ãã®ãŸã‚ã€GPUã®é«˜ã„æ¼”ç®—èƒ½åŠ›ã‚’å¼•ãå‡ºã™ãŸã‚ã«ã¯ã€CPUã®ãƒ—ãƒã‚°ãƒ©ãƒ ã¨ã¯ç•°ãªã‚‹æ³¨æ„ãŒå¿…è¦ã§ã‚ã‚‹ã€‚ãã“ã§æœ¬é€£è¼‰ã§ã¯ã€Kepler GPUã‚’ä¾‹ã«ã¨ã‚Šã€ãã®ã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£ã¨CUDAãƒ—ãƒã‚°ãƒ©ãƒ ã®æœ€é©åŒ–ã®ãƒã‚¤ãƒ³ãƒˆã‚’è¦‹ã¦ã„ããŸã„ã¨æ€ã†ã€‚
yass 2013/11/23
gpu

gpgpu
ãƒªãƒ³ã‚¯
AWS News Blog
AWS Week in Review â€“Â AWS Documentation Updates, Amazon EventBridge is Faster, and More â€“ May 22, 2023 Here are your AWS updates from the previous 7 days. Last week I was in Turin, Italy for CloudConf, a conference Iâ€™ve had the pleasure to participate in for the last 10 years. AWS Hero Anahit Pogosova was also there sharing a few serverless tips in front of a full house. Hereâ€™s a picture I [â€¦] Amaz
yass 2013/11/23
" cg1.4xlarge / 2ã¤ã®NVIDIA Tesla M2050 "Fermi" GPU / 2ã¤ã®quad-core Intel "Nehalem" X5570ãƒ—ãƒã‚»ãƒƒã‚µã€åˆè¨ˆ33.5 ECUs / 22 GBã®RAM / å€‹ã€…ã®Tesla M2050ã¯448ã‚³ã‚¢ã€3GBã®ECC RAMã‚’å«ã¿ã€å€ç²¾åº¦æµ®å‹•å°æ•°ç‚¹æ¼”ç®—ã§515gigaflopsã¾ã§å¯èƒ½ "

ec2

aws

gpu

gpgpu
ãƒªãƒ³ã‚¯
é€²åŒ–ã™ã‚‹GPUä»®æƒ³åŒ–æŠ€è¡“ã§ãƒ¯ãƒ¼ã‚¯ã‚¹ãƒ†ãƒ¼ã‚·ãƒ§ãƒ³ã‚’ãƒ‡ãƒ¼ã‚¿ã‚»ãƒ³ã‚¿ãƒ¼ã« - NVIDIA GRID VGXã‚½ãƒªãƒ¥ãƒ¼ã‚·ãƒ§ãƒ³
TechCenterã‹ã‚‰ç§»è¡Œã•ã‚ŒãŸãƒ†ã‚¯ãƒ‹ã‚«ãƒ« ãƒªã‚½ãƒ¼ã‚¹ 2018å¹´8æœˆæ™‚ç‚¹ã§ã€ã‚¢ã‚¯ãƒ†ã‚£ãƒ–ãªTechCenterã®ã‚³ãƒ³ãƒ†ãƒ³ãƒ„ãŒç§»è¡Œã•ã‚Œã€Dell.com ã®Dellã‚µãƒãƒ¼ãƒˆã®ä¸€éƒ¨ã«ãªã‚Šã€ãƒ•ã‚©ãƒ¼ãƒ©ãƒ ãŒDellã‚³ãƒŸãƒ¥ãƒ‹ãƒ†ã‚£ã«ç§»è¡Œã•ã‚Œã¾ã—ãŸã€‚ æ¦‚è¦: 2018å¹´8æœˆæ™‚ç‚¹ã§ã€ã‚¢ã‚¯ãƒ†ã‚£ãƒ–ãªTechCenterã®ã‚³ãƒ³ãƒ†ãƒ³ãƒ„ãŒç§»è¡Œã•ã‚Œã€Dell.com ã®Dellã‚µãƒãƒ¼ãƒˆã®ä¸€éƒ¨ã«ãªã‚Šã€ãƒ•ã‚©ãƒ¼ãƒ©ãƒ ãŒDellã‚³ãƒŸãƒ¥ãƒ‹ãƒ†ã‚£ã«ç§»è¡Œã•ã‚Œã¾ã—ãŸã€‚
yass 2013/11/23
NVIDIA

GPU

gpgpu

NVIDIA GRID
ãƒªãƒ³ã‚¯
ã€ä»®æƒ³åŒ–é“å ´ã€‘ ã€GTC 2013ãƒ¬ãƒãƒ¼ãƒˆã€‘NVIDIAãŒæä¾›ã™ã‚‹ã‚¯ãƒ©ã‚¦ãƒ‰/ä»®æƒ³ç’°å¢ƒå‘ã‘ã®GPUã€ŒNVIDIA GRIDã€
yass 2013/11/23
NVIDIA

GPU

gpgpu

NVIDIA GRID
ãƒªãƒ³ã‚¯
AWS News Blog
Announcing Amazon Managed Service for Apache Flink Renamed from Amazon Kinesis Data Analytics Today we are announcing the rename of Amazon Kinesis Data Analytics to Amazon Managed Service for Apache Flink, a fully managed and serverless service for you to build and run real-time streaming applications using Apache Flink. We continue to deliver the same experience in your Flink applications without
yass 2013/11/23
" NVIDIA GRIDâ„¢ (GK104 "Kepler") GPU, 1,536 CUDA ã‚³ã‚¢ ã¨ 4 GB ãƒ“ãƒ‡ã‚ªRAM / cg1ã® GPGPU ã¯ã€å€ç²¾åº¦æµ®å‹•å°æ•°ç‚¹æ¼”ç®—ã¨ã‚¨ãƒ©ãƒ¼è¨‚æ£ãƒ¡ãƒ¢ãƒª / ãã‚Œã«æ¯”ã¹ g2.2xlarge ã®GPUã¯ã€å˜ç²¾åº¦æµ®å‹•å°æ•°ç‚¹æ¼”ç®—ã‚’ã‚µãƒãƒ¼ãƒˆ / ã‚¨ãƒ©ãƒ¼è¨‚æ£ãƒ¡ãƒ¢ãƒªã¯ã‚µãƒãƒ¼ãƒˆã—ã¾ã›ã‚“"

ec2

GPU

NVIDIA

AWS

gpgpu

NVIDIA GRID
ãƒªãƒ³ã‚¯
GTC Japan 2013 - CUDAã®ç™ºæ˜Žè€…ã®Ian Buckæ°ã«èžã„ãŸGPUã‚½ãƒ•ãƒˆã®å°†æ¥
NVIDIAãŒ7æœˆ30æ—¥ã«é–‹å‚¬ã—ãŸã€ŒGTC Japan 2013ã€ã«ãŠã„ã¦GPU Computing Softwareéƒ¨é–€ã®ã‚¸ã‚§ãƒãƒ©ãƒ«ãƒžãƒãƒ¼ã‚¸ãƒ£ã®Ian Buckæ°ã‹ã‚‰ã€GPUã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢ã®å°†æ¥åƒã«ã¤ã„ã¦è©±ã‚’èžã„ãŸã€‚Buckæ°ã¯CUDAã®ç”Ÿã¿ã®è¦ªã§ã‚ã‚Šã€ç¾åœ¨ã‚‚NVIDIAã®GPUã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢é–‹ç™ºã®ä¸å¿ƒã¨ãªã£ã¦ã„ã‚‹äººç‰©ã§ã‚ã‚‹ã€‚ ä»Šå›žã®GTCã«ãŠã„ã¦ã€NVIDIAãŒã‚³ãƒ³ãƒ‘ã‚¤ãƒ©ã®ãƒ¡ãƒ¼ã‚«ãƒ¼ã§ã‚ã‚‹PGI(Portland Group International)ã‚’è²·åŽã—ãŸã“ã¨ãŒç™ºè¡¨ã•ã‚ŒãŸã€‚ã“ã‚Œã«é–¢ã—ã¦ã€NVIDIAã¨PGIã®åˆ†æ‹…ã¯ã©ã†ãªã‚‹ã®ã‹ã‚’è³ªå•ã—ãŸã¨ã“ã‚ã€PGIã¯ã€ã“ã‚Œã¾ã§ã®ã‚³ãƒ³ãƒ‘ã‚¤ãƒ©ã®ãƒ©ã‚¤ã‚»ãƒ³ã‚¹ãƒ»è²©å£²ã‚’ç¶šã‘ã‚‹ã€‚ã—ã‹ã—ã€NVIDIAã®ä¸€å“¡ã¨ãªã£ãŸã®ã§ã€Maxwellã‚„Voltaã¨ã„ã£ãŸå°†æ¥ã®GPUã‚„CUDA6/7ã®æŠ€è¡“æƒ…å ±ã«ã‚¢ã‚¯ã‚»ã‚¹ã§ãã‚‹ã‚ˆã†ã«ãªã‚Šã€ã‚ˆã‚Šç·Šå¯†ãªå”åŠ›ãŒè¡Œãˆã‚‹ã‚ˆã†ã«ãªã‚‹ã€‚ã¾ãŸã€PGI
yass 2013/10/17
"OpenCLã¯ GPUãƒ‡ãƒã‚¤ã‚¹ã‚’åˆ¶å¾¡ã™ã‚‹ã¨ã„ã†æ€æƒ³ã§ä½œã‚‰ã‚ŒãŸè¨€èªžã§ã‚ã‚‹ã®ã«å¯¾ã—ã¦ã€CUDAã¯ã€Cè¨€èªžã‚’ãƒ™ãƒ¼ã‚¹ã«ä¸¦åˆ—è¨ˆç®—ã®ãŸã‚ã®æœ€å°é™ã®æ‹¡å¼µã‚’ä»˜ã‘åŠ ãˆãŸã‚‚ã®ã§ã€ãƒ‡ãƒã‚¤ã‚¹åˆ¶å¾¡ã¨ã¯é€†æ–¹å‘ã®æ‹¡å¼µã‚’ç›®æŒ‡ã™ã‚‚ã®ã§ã‚ã‚Šã€æ–¹å‘æ€§ãŒé•ã†"

cuda

gpu

opencl

gpgpu
ãƒªãƒ³ã‚¯
The 13 application areas where OpenCL and CUDA can be used - StreamHPC
Our Services and ProductsWe are a software house that writes extremely fast code for the world's most demanding companies. Software Development â€“ From architecture to delivery, making fast softwareDevelop Fast SoftwareMaking the release version of prototype codeMaking CUDA-code work on AMD hardware (HIP)Porting CUDA to OpenCLTraining â€“ From crash-course to full in-house trainings1-day Crash Course
yass 2013/10/12
" what kind of algorithms are faster when using accelerators and OpenCL? "

opencl

gpgpu

gpu
ãƒªãƒ³ã‚¯
GPUã¸ã®MapReduceã®é©ç”¨ã«é–¢ã™ã‚‹èª¿æŸ» - nishio-dens's diary
æ¦‚è¦ è¿‘å¹´ã€GPUã®æ€§èƒ½ã¯é£›èºçš„ã«å‘ä¸Šã—ã¦ãŠã‚Šã€ã‚°ãƒ©ãƒ•ã‚£ãƒƒã‚¯å°‚ç”¨ã®å‡¦ç†è£…ç½®ã¨ã—ã¦ã§ã¯ãªãã€æ•°å€¤è¨ˆç®—ç‰ã®æ±Žç”¨å‘ã‘ã®å‡¦ç†ã«åˆ©ç”¨ã™ã‚‹ã€GPGPUã«é–¢ã™ã‚‹ç ”ç©¶ãŒç››ã‚“ã«è¡Œã‚ã‚Œã¦ã„ã‚‹ã€‚GPUã¯å†…éƒ¨ã«å¤šãã®ã‚³ã‚¢ã‚’å‚™ãˆã¦ãŠã‚Šã€NVIDIAç¤¾ã®GeForce GTX 580ã§ã¯ã€512å€‹ã‚‚ã®ã‚³ã‚¢ã‚’ä¿æŒã—ã¦ã„ã‚‹ã€‚ã“ã‚Œã‚‰ã®ã‚³ã‚¢ã™ã¹ã¦ã‚’åŠ¹çŽ‡ã‚ˆãåˆ©ç”¨ã™ã‚‹ã“ã¨ã§ã€GPUã®æŒã£ã¦ã„ã‚‹é«˜ã„ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ã‚’å¼•ãå‡ºã™ã“ã¨ãŒå‡ºæ¥ã‚‹ãŒã€GPUã®é«˜ã„ä¸¦åˆ—æ€§ã‚’åˆ©ç”¨ã™ã‚‹ãŸã‚ã«ã¯GPUç‰¹æœ‰ã®å‡¦ç†ã‚’å®Ÿè£…ã™ã‚‹å¿…è¦ãŒã‚ã‚Šã€GPUãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ã«ãªã˜ã¿ã®ç„¡ã„åˆ©ç”¨è€…ã«ã¨ã£ã¦ã¯å‡¦ç†ã®è¨˜è¿°ãŒå›°é›£ã§ã‚ã‚‹ã€‚ ãã“ã§ã€GPUã‚’å¤§é‡ã®ã‚³ã‚¢ã‚’æŒã£ãŸãƒ—ãƒã‚»ãƒƒã‚µã ã¨è€ƒãˆã€ã“ã‚Œã‚‰ã«å¯¾ã—ã¦MapReduceã‚’é©ç”¨ã™ã‚‹ã“ã¨ã‚’æ¤œè¨Žã™ã‚‹ã€‚è¦ªã—ã¿ã‚„ã™ã„MapReduceã‚¤ãƒ³ã‚¿ãƒ¼ãƒ•ã‚§ãƒ¼ã‚¹ã«ã‚ã¦ã¯ã‚ã¦å‡¦ç†ã‚’è¨˜è¿°ã™ã‚‹ã ã‘ã§ã€GPUã®é«˜ã„ä¸¦åˆ—æ€§ã‚’ç”Ÿã‹ã—ãŸãƒ—ãƒã‚°ãƒ©ãƒ ãŒå®Ÿè£…å¯èƒ½ã¨ãªã‚‹ã€‚åŠ
yass 2013/09/29
" GPUãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ã«ãªã˜ã¿ã®ç„¡ã„åˆ©ç”¨è€…ã«ã¨ã£ã¦ã¯å‡¦ç†ã®è¨˜è¿°ãŒå›°é›£ã§ã‚ã‚‹ã€‚ãã“ã§ã€GPUã‚’å¤§é‡ã®ã‚³ã‚¢ã‚’æŒã£ãŸãƒ—ãƒã‚»ãƒƒã‚µã ã¨è€ƒãˆã€ã“ã‚Œã‚‰ã«å¯¾ã—ã¦MapReduceã‚’é©ç”¨ã™ã‚‹ã“ã¨ã‚’æ¤œè¨Žã™ã‚‹ã€‚"

gpgpu

MapReduce

gpu
ãƒªãƒ³ã‚¯
2013å¹´åº¦ | ï¼§ï¼°ï¼µã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ãƒ†ã‚£ãƒ³ã‚° - TOKYO TECH OCW
è¬›ç¾©è³‡æ–™ã‚’å…¨ä¸–ç•Œã«å‘ã‘ã¦ç„¡å„Ÿã§å…¬é–‹ã—ã€æœ€é«˜æ°´æº–ã®ç†å·¥ç³»æ•™è‚²ã‚’å…¨ä¸–ç•Œã®å…±æœ‰è²¡ç”£ã¨ã™ã¹ãæä¾›ã™ã‚‹ãƒ—ãƒ©ãƒƒãƒˆãƒ•ã‚©ãƒ¼ãƒ ã§ã™
yass 2013/09/26
GPGPU

CUDA
ãƒªãƒ³ã‚¯
2009/12/10 GPUã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ãƒ†ã‚£ãƒ³ã‚°ã®ç¾çŠ¶ã¨ã‚¹ãƒ¼ãƒ‘ãƒ¼ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ãƒ†ã‚£ãƒ³ã‚°ã®æœªæ¥
PFIç¤¾å†…ã‚»ãƒŸãƒŠãƒ¼ 2009å¹´12æœˆ10æ—¥ 20:00-21:00(äºˆå®š) GPUã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ãƒ†ã‚£ãƒ³ã‚°ã®ç¾çŠ¶ã¨ã‚¹ãƒ¼ãƒ‘ãƒ¼ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ãƒ†ã‚£ãƒ³ã‚°ã®æœªæ¥ ç™ºè¡¨è€…: æ‘ä¸» å´‡è¡Œï¼ˆãƒ—ãƒªãƒ•ã‚¡ãƒ¼ãƒ‰ã‚¤ãƒ³ãƒ•ãƒ©ã‚¹ãƒˆãƒ©ã‚¯ãƒãƒ£ãƒ¼ ç ”ç©¶é–‹ç™ºéƒ¨é–€ãƒ»äº¬éƒ½å¤§å¦å¤§å¦é™¢ã€€ç‰©ç†å¦ç¬¬äºŒæ•™å®¤ï¼‰ ã‚»ãƒŸãƒŠãƒ¼éŒ²ç”»URL: http://www.ustream.tv/recorded/2837689 ã“ã®ã‚¹ãƒ©ã‚¤ãƒ‰ã¯ã€ç™ºè¡¨å¾Œã«ã¿ãªã•ã¾ã‹ã‚‰ã„ãŸã ã„ãŸè²´é‡ãªæ„è¦‹ã‚’ã‚‚ã¨ã«æ”¹è¨‚ã—ãŸç‰ˆã§ã™ã€‚ç™ºè¡¨æ™‚ç‚¹ã§ã®ç‰ˆã¯ã“ã¡ã‚‰: http://www.slideshare.net/pfi/20091210-gpu-2735685Read less
yass 2013/09/26
gpgpu

gpu

HPC
ãƒªãƒ³ã‚¯
ã€ã‚¤ãƒ™ãƒ³ãƒˆãƒ¬ãƒãƒ¼ãƒˆã€‘ ã€GTCãƒ¬ãƒãƒ¼ãƒˆã€‘æ—¥æœ¬ãƒ¦ãƒ‹ã‚·ã‚¹ã€CUDAã§ãƒªã‚³ãƒ¡ãƒ³ãƒ‰ã‚·ã‚¹ãƒ†ãƒ ã‚’å®Ÿè£…
yass 2013/09/26
"ãƒžãƒ«ãƒGPUã§å‹•ä½œã™ã‚‹ãƒªã‚³ãƒ¡ãƒ³ãƒ‰ã‚·ã‚¹ãƒ†ãƒ ã‚’ç´¹ä»‹ã™ã‚‹ã‚»ãƒƒã‚·ãƒ§ãƒ³"

CUDA

SVD

recommend

k-NN

gpgpu
ãƒªãƒ³ã‚¯
1 2 æ¬¡ã®ãƒšãƒ¼ã‚¸