[B! CPU][AI] nminoruã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

nminoru id:nminoru

CPUã¨AIã«é–¢ã™ã‚‹nminoruã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (4)

${{author_name}}$

{{{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}}

æ·±å±¤å¦ç¿’ã«ç‰¹åŒ–ã—ãŸ ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ãƒ»ãƒ—ãƒã‚»ãƒƒã‚µãƒ¼ MN-Coreã‚’ç™ºè¡¨ã€‚2020å¹´æ˜¥ã€MN-Coreã«ã‚ˆã‚‹å¤§è¦æ¨¡ã‚¯ãƒ©ã‚¹ã‚¿ãƒ¼MN-3ã‚’ç¨¼åƒäºˆå®š â€“ Preferred Networks
æ·±å±¤å¦ç¿’ã«ç‰¹åŒ–ã—ãŸ ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ãƒ»ãƒ—ãƒã‚»ãƒƒã‚µãƒ¼ MN-Coreã‚’ç™ºè¡¨ã€‚2020å¹´æ˜¥ã€MN-Coreã«ã‚ˆã‚‹å¤§è¦æ¨¡ã‚¯ãƒ©ã‚¹ã‚¿ãƒ¼MN-3ã‚’ç¨¼åƒäºˆå®š æ ªå¼ä¼šç¤¾Preferred Networksï¼ˆæœ¬ç¤¾ï¼šæ±äº¬éƒ½åƒä»£ç”°åŒºã€ä»£è¡¨å–ç· å½¹ç¤¾é•·ï¼šè¥¿å·å¾¹ã€ãƒ—ãƒªãƒ•ã‚¡ãƒ¼ãƒ‰ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚¹ã€ä»¥ä¸‹ã€PFNï¼‰ã¯ã€æ·±å±¤å¦ç¿’ã«ç‰¹åŒ–ã—ãŸãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ãƒ»ãƒ—ãƒã‚»ãƒƒã‚µãƒ¼ MN-Core (TM)Â ï¼ˆã‚¨ãƒ ã‚¨ãƒŒãƒ»ã‚³ã‚¢ï¼‰ã‚’ç™ºè¡¨ã—ã¾ã™ã€‚æ±äº¬ãƒ“ãƒƒã‚°ã‚µã‚¤ãƒˆã§é–‹å‚¬ä¸ã® SEMICON Japan 2018ã«ãŠã„ã¦ã€MN-Coreãƒãƒƒãƒ—ã€ãƒœãƒ¼ãƒ‰ã€ã‚µãƒ¼ãƒãƒ¼ãªã©ã€PFNãŒç‹¬è‡ªé–‹ç™ºã—ãŸæ·±å±¤å¦ç¿’å‘ã‘ã®ãƒãƒ¼ãƒ‰ã‚¦ã‚§ã‚¢ã‚’å±•ç¤ºã—ã¾ã™ã€‚ PFNã¯ã€æ·±å±¤å¦ç¿’ã®å®Ÿä¸–ç•Œã¸ã®å¿œç”¨ã‚’ç›®æŒ‡ã—ã€ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹ã®æ·±å±¤å¦ç¿’ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯Chainer (TM)ï¼ˆãƒã‚§ã‚¤ãƒŠãƒ¼ï¼‰ã®é–‹ç™ºã‚„ã€ç ”ç©¶é–‹ç™ºã‚’æ”¯ãˆã‚‹å¤§è¦æ¨¡GPUã‚¯ãƒ©ã‚¹ã‚¿ãƒ¼MN-1ï¼ˆã‚¨ãƒ ã‚¨ãƒŒãƒ»ãƒ¯ãƒ³ï¼‰ã€MN-1bï¼ˆã‚¨ãƒ ã‚¨ãƒŒãƒ»ãƒ¯ãƒ³ãƒ»ãƒ“
nminoru 2018/12/12
AI

CPU
ãƒªãƒ³ã‚¯
NECã€ãƒ™ã‚¯ãƒˆãƒ«åž‹ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿ã«é©ã—ãŸæ©Ÿæ¢°å¦ç¿’å‘ã‘ãƒ‡ãƒ¼ã‚¿å‡¦ç†æŠ€è¡“ã‚’é–‹ç™º
NECã¯ã€ãƒ™ã‚¯ãƒˆãƒ«åž‹ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿(æ³¨1)ã«é©ã—ãŸã€æ©Ÿæ¢°å¦ç¿’ã‚’é«˜é€Ÿå®Ÿè¡Œã™ã‚‹ãƒ‡ãƒ¼ã‚¿å‡¦ç†æŠ€è¡“ã‚’é–‹ç™ºã—ã¾ã—ãŸã€‚æœ¬æŠ€è¡“ã«ã‚ˆã‚Šã€è¤‡æ•°ã‚µãƒ¼ãƒã‚’æŽ¥ç¶šã—ãŸã‚¯ãƒ©ã‚¹ã‚¿ä¸Šã®Spark(æ³¨2)ã§ã®åˆ©ç”¨ã¨æ¯”è¼ƒã—ã€50å€ä»¥ä¸Šé«˜é€Ÿã«æ©Ÿæ¢°å¦ç¿’ã‚’å®Ÿè¡Œã§ãã¾ã™ã€‚ ä»Šå›žã€å¤§è¦æ¨¡ãªæ©Ÿæ¢°å¦ç¿’ã§ã‚ˆãç”¨ã„ã‚‰ã‚Œã‚‹ã€Œç–Žè¡Œåˆ—(æ³¨3)ã€ã¨ã„ã†ãƒ‡ãƒ¼ã‚¿æ§‹é€ ã«ç€ç›®ã—ãŸæ¼”ç®—æŠ€è¡“ã¨é€šä¿¡æŠ€è¡“ã‚’é–‹ç™ºã™ã‚‹ã“ã¨ã§ã€æ©Ÿæ¢°å¦ç¿’ã«ãŠã„ã¦ã‚‚ãƒ™ã‚¯ãƒˆãƒ«åž‹ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿ãŒæŒã¤æ€§èƒ½ã‚’æœ€å¤§é™ã«ç™ºæ®ã—ãŸé«˜é€Ÿå®Ÿè¡Œã‚’å®Ÿç¾ã—ã¾ã—ãŸã€‚ ã¾ãŸæœ¬æŠ€è¡“ã‚’å–ã‚Šè¾¼ã‚“ã ãƒŸãƒ‰ãƒ«ã‚¦ã‚§ã‚¢ã‚’é–‹ç™ºã—ã€æœ¬æŠ€è¡“ã®åˆ©ç”¨ã‚’å®¹æ˜“ã«ã—ã¾ã—ãŸã€‚æœ¬ãƒŸãƒ‰ãƒ«ã‚¦ã‚§ã‚¢ã¯Sparkã‚„Python(æ³¨4)ã‹ã‚‰å‘¼ã³å‡ºã™ã“ã¨ãŒã§ãã€ã“ã‚Œã‚‰ã®ãƒ¦ãƒ¼ã‚¶ã¯ç‰¹åˆ¥ãªãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ç„¡ã—ã§ä½¿ç”¨ã§ãã¾ã™ã€‚ ã“ã‚Œã«ã‚ˆã‚Šã€ä¾‹ãˆã°ã€Webåºƒå‘Šå‡ºç¨¿ã®æœ€é©åŒ–ã‚„ãƒ¬ã‚³ãƒ¡ãƒ³ãƒ‰ã€æ–‡æ›¸åˆ†æžã®åˆ†é‡Žã§åˆ†æžãŒçŸæ™‚é–“ã§è¡Œãˆã‚‹ã‚ˆã†ã«ãªã‚‹ãŸã‚ã€ã‚¿ã‚¤ãƒ ãƒªãƒ¼ã«çµæžœã‚’åˆ©ç”¨ã§ãã¾ã™ã€‚ã¾ãŸã€å°‘æ•°ã®ã‚µãƒ¼ãƒ
nminoru 2017/07/03
ãªã‚“ã¨ã„ã†å”¯æˆ‘ç‹¬å°Šã€‚

NEC

AI

CPU
ãƒªãƒ³ã‚¯
Quantifying the performance of the TPU, our first machine learning chip | Google Cloud Blog
Quantifying the performance of the TPU, our first machine learning chip Editor's Note: Learn about our newly available Cloud TPUs. Weâ€™ve been using compute-intensive machine learning in our products for the past 15 years. We use it so much that we even designed an entirely new class of custom machine learning accelerator, the Tensor Processing Unit. Just how fast is the TPU, actually? Today, in co
nminoru 2017/04/06
Google

AI

CPU
ãƒªãƒ³ã‚¯
å¯Œå£«é€šã€ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°å°‚ç”¨AIãƒ—ãƒã‚»ãƒƒã‚µã€ŒDLUã€ã‚’é–‹ç™º
nminoru 2016/11/29
Fujitsu

AI

CPU
ãƒªãƒ³ã‚¯
1