[B! GPU] nminoruã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

nminoru id:nminoru

GPUã«é–¢ã™ã‚‹nminoruã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (54)

${{author_name}}$

{{{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}}

NVLink Bridgeã§GPUã‚’ç¹‹ã„ã§ã‚‚1GPUã«ã¯ãªã‚Šã¾ã›ã‚“ | HPCã‚·ã‚¹ãƒ†ãƒ ã‚º Tech Blog
ã€ŽNVLink Bridgeã§è¤‡æ•°GPUã‚’ç¹‹ã„ã ã‚‰ã€ãã‚Œã‚‰ãŒ1GPUã¨ã—ã¦æ‰±ãˆã‚‹ã‚ˆã†ã«ãªã‚‹ã‚“ã§ã—ã‚‡ï¼Ÿã€ã¨ã„ã†èª¤è§£ã‚’ã•ã‚Œã¦ã„ã‚‹ãŠå®¢æ§˜ã‚’ã—ã°ã—ã°ãŠè¦‹å—ã‘ã„ãŸã—ã¾ã™ã€‚ã“ã¡ã‚‰ã®è¨˜äº‹ã§ã¯ã€ãã‚ŒãŒèª¤è§£ã§ã‚ã‚‹ã“ã¨ã€ã¾ãŸã€SLIã‚„Unified Memoryã¨ã„ã£ãŸé–¢é€£ã™ã‚‹æƒ…å ±ã«ã¤ã„ã¦ã‚‚æ•´ç†ã—ã¦è§£èª¬ã„ãŸã—ã¾ã™ã€‚é–“é•ã£ãŸæœŸå¾…ã‚’æŠ±ã„ã¦å¤±æ•—ã—ãªã„ã‚ˆã†ã«ã€æ£ã—ã„ç†è§£ã‚’æ·±ã‚ã¦ã„ãã¾ã—ã‚‡ã†ã€‚ GPUã®ãƒ¡ãƒ¢ãƒªç©ºé–“ã¯ä»–ãƒ‡ãƒã‚¤ã‚¹ã‹ã‚‰éš”çµ¶ã•ã‚Œã¦ã„ã¾ã™ GPUä¸Šã«ã¯æ¼”ç®—ã™ã‚‹ãŸã‚ã®ãƒ—ãƒã‚»ãƒƒã‚µã¨ã€ãƒ‡ãƒ¼ã‚¿ã‚’ä¸€æ™‚çš„ã«ç½®ã„ã¦ãŠããŸã‚ã®ãƒ¡ãƒ¢ãƒª(VRAM)ãŒæè¼‰ã•ã‚Œã¦ã„ã¾ã™ã€‚GPUã®ãƒ¡ãƒ¢ãƒªã‚’ã€CUDAã§æ›¸ã‹ã‚ŒãŸãƒ—ãƒã‚°ãƒ©ãƒ ã‹ã‚‰åˆ©ç”¨ã™ã‚‹ã«ã¯ã€cudaMallocã§ãƒ¡ãƒ¢ãƒªé ˜åŸŸã‚’ç¢ºä¿ã—ã€cudaMemcpyã‚’ä½¿ã£ã¦ãƒ›ã‚¹ãƒˆ(CPUå´)ã®ãƒ¡ãƒ¢ãƒªã¨ãƒ‡ãƒ¼ã‚¿ã®é€å—ä¿¡ã‚’è¡Œã„ã€GPUä¸Šã§æ¼”ç®—kernelã¨ã™ã‚‹é–¢æ•°ï¼ˆä»¥ä¸‹ã€GPU-Kernelï¼‰ã‚’å‘¼ã³å‡ºã—ã€æœ€å¾Œã«c
nminoru 2023/07/19
GPU
ãƒªãƒ³ã‚¯
ã€ã‚„ã˜ã†ã¾PC Watchã€‘ ã€Œä»®æƒ³é€šè²¨ã¯ç¤¾ä¼šã®ä½•ã®å½¹ã«ã‚‚ç«‹ãŸãªã„ã€NVIDIAã®CTOãŒç™ºè¨€
nminoru 2023/03/28
ä½•ã¨ã„ã†æ©çŸ¥ã‚‰ãšãªã‚“ã ã€‚

NVIDIA

GPU
ãƒªãƒ³ã‚¯
Intelã€ã‚³ãƒ³ã‚·ãƒ¥ãƒ¼ãƒžå‘ã‘é«˜æ€§èƒ½GPUãƒ–ãƒ©ãƒ³ãƒ‰ã€ŒArcã€ã€‚ç¬¬1å¼¾ã¯2022å¹´åˆ
nminoru 2021/08/17
IntelãŒæœ¬æ°—ã‚’å‡ºã—ã¦GPUã‚’ä½œã‚‹ã®ã‹ã€‚ã‚²ãƒ¼ãƒ æ¯Žã«ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã—ã¦ã‚‚ã‚‰ã‚ãªã„ã¨æ€§èƒ½ã§ãªã„ã ã‚ã†ã‘ã©PCå¸‚å ´ã ã¨NVIDIAã€AMDã€Apple Silicon M1Xã«ç¶šã4ç•ªæ‰‹ã ã‚ˆãªã€‚æµçŸ³ã«M1Xã¯ã™ãè¿½ã„è¶Šã›ã‚‹ã ã‚ã†ãŒã€‚

Itenl

GPU
ãƒªãƒ³ã‚¯
ã€ãƒ‹ãƒ¥ãƒ¼ã‚¹ãƒ»ãƒ•ãƒ©ãƒƒã‚·ãƒ¥ã€‘ELSAã€ç„¡ç™ºå…‰ä»•æ§˜ã®GeForce RTX 3060æè¼‰ãƒ“ãƒ‡ã‚ªã‚«ãƒ¼ãƒ‰
nminoru 2021/07/05
ã“ã‚Œã¾ã§ã®ãƒ“ãƒ‡ã‚ªã‚«ãƒ¼ãƒ‰ã¯ãƒ”ã‚«ãƒ”ã‚«ã¨å…‰ã‚‰ã–ã‚‹ãˆãªã‹ã£ãŸãŒæ–°ãƒ†ã‚¯ãƒŽãƒã‚¸ãƒ¼ã§ãã‚Œã‚’æŠ‘æ¢ã™ã‚‹ã“ã¨ãŒå¯èƒ½ã«ï¼ã¿ãŸã„ãªå°è±¡ã‚’å—ã‘ã‚‹ã€‚

GPU

NVIDIA
ãƒªãƒ³ã‚¯
ä¸å›½ã®æš—å·é€šè²¨è¦åˆ¶ã«ã‚ˆã‚Šã€GPUä¾¡æ ¼ãŒå…ƒã«æˆ»ã‚‹ã‹ã‚‚
ä¸å›½ã®æš—å·é€šè²¨è¦åˆ¶ã«ã‚ˆã‚Šã€GPUä¾¡æ ¼ãŒå…ƒã«æˆ»ã‚‹ã‹ã‚‚2021.06.23 14:0012,046 Sam Rutherford - Gizmodo US ï¼»åŽŸæ–‡ï¼½ ï¼ˆ å¡šæœ¬ç›´æ¨¹ ï¼‰ ãªã‚“ã¨ã‹ã—ã¦ï¼ˆåˆ‡æœ›ï¼‰ã€‚ æœ€è¿‘ã¯æ–°ã—ã„ã‚°ãƒ©ãƒ•ã‚£ãƒƒã‚¯ã‚«ãƒ¼ãƒ‰ã‚’æ‰‹é ƒãªä¾¡æ ¼ã§è¦‹ã¤ã‘ã‚‹ã®ã¯ã»ã¼ä¸å¯èƒ½ã ã£ãŸã®ã§ã™ãŒã€ä¸å›½ã®å››å·çœã€æ–°ç–†ã‚¦ã‚¤ã‚°ãƒ«è‡ªæ²»åŒºã€ãŠã‚ˆã³å†…ãƒ¢ãƒ³ã‚´ãƒ«è‡ªæ²»åŒºã§ã®æš—å·é€šè²¨ãƒžã‚¤ãƒ‹ãƒ³ã‚°ã®åœæ¢ãŒç›¸æ¬¡ã„ã ã“ã¨ã‹ã‚‰GPUä¾¡æ ¼ãŒ66%ã‚‚ä¸‹è½ã—ã€ã“ã†ã—ãŸçŠ¶æ³ãŒå¤‰ã‚ã‚‹ã‹ã‚‚ã—ã‚Œã¾ã›ã‚“ã€‚ South China Morning Postã®å ±é“ã«ã‚ˆã‚‹ã¨ã€ä¸å›½ã§ã¯æœ€è¿‘ã«ãªã£ã¦æš—å·é€šè²¨ã«å¯¾ã™ã‚‹å–ã‚Šç· ã¾ã‚ŠãŒè¡Œãªã‚ã‚Œã¦ãŠã‚Šã€åŒå›½ã®æš—å·é€šè²¨ãƒžã‚¤ãƒ‹ãƒ³ã‚°äº‹æ¥ã«å¤§æ‰“æ’ƒã‚’ä¸Žãˆã¦ã„ã¾ã™ã€‚ãã®çµæžœã€ã‚°ãƒ©ãƒ•ã‚£ãƒƒã‚¯ã‚«ãƒ¼ãƒ‰ã®éœ€è¦ãŒæ€¥æ¸›ã—ã€ä¸å›½ã®å¤šãã®å¤§æ‰‹ã‚ªãƒ³ãƒ©ã‚¤ãƒ³å°å£²åº—ã§GPUä¾¡æ ¼ãŒå¤§å¹…ã«ä¸‹è½ã—ã¦ã„ã¾ã™ã€‚ ç•°å¸¸ã ã£ãŸä¾¡æ ¼ãŒè½ã¡ç€ãå§‹ã‚ã¦ã„ã‚‹SCMPã«ã‚ˆã‚‹ã¨ã€
nminoru 2021/06/23
GPU

Bitcoin
ãƒªãƒ³ã‚¯
https://jp.techcrunch.com/2020/12/23/2020-12-21-horizon-robotics-700-million-funding/
nminoru 2020/12/23
GPU
ãƒªãƒ³ã‚¯
IntelãŒãƒŽãƒ¼ãƒˆPCå‘ã‘å¤–éƒ¨GPUã€ŒIris Xe MAX Graphicsã€ã‚’ç™ºå£²ã€€Tiger Lakeã¨ã®å”èª¿å‹•ä½œã«å¯¾å¿œ
Iris Xe MAX Graphicsã®ä¸»ãªä»•æ§˜ã¯ä»¥ä¸‹ã®é€šã‚Šã€‚ æ¼”ç®—ãƒ¦ãƒ‹ãƒƒãƒˆï¼ˆEUï¼‰æ•°ï¼š96åŸº ã‚°ãƒ©ãƒ•ã‚£ãƒƒã‚¯ã‚¹ãƒ¡ãƒ¢ãƒªï¼š4GBï¼ˆå¸¯åŸŸå¹…ã¯æ¯Žç§’68GBï¼‰ æŽ¥ç¶šã‚¤ãƒ³ã‚¿ãƒ•ã‚§ãƒ¼ã‚¹ï¼šPCI Express 4.0 x4 æ˜ åƒå‡ºåŠ›ï¼šDisplayPort 1.4ã€Embedded DisplayPortï¼ˆeDPï¼‰ 1.4ã€HDMI 2.0b æœ€å¤§è§£åƒåº¦ï¼š3840Ã—2160ãƒ”ã‚¯ã‚»ãƒ«ï¼ˆHDMIï¼eDPï¼‰ã€7680Ã—4320ãƒ”ã‚¯ã‚»ãƒ«ï¼ˆDisplayPortï¼‰ å¯¾å¿œAPIï¼šDirectX 12.1ã€Vulkanã€OpenGL ç¬¬11ä¸–ä»£Coreãƒ—ãƒã‚»ãƒƒã‚µï¼ˆé–‹ç™ºã‚³ãƒ¼ãƒ‰åï¼šTiger Lakeï¼‰ã¨çµ„ã¿åˆã‚ã›ãŸå ´åˆã€åŒCPUã¨å”èª¿å‹•ä½œã‚’ã™ã‚‹ã“ã¨ã§é›»åŠ›æ¶ˆè²»ã‚’æœ€é©åŒ–ã™ã‚‹ã€ŒIntel Deep Link techno logyã€ã«å¯¾å¿œã™ã‚‹ã€‚ã“ã‚Œã‚’åˆ©ç”¨ã™ã‚‹ã¨ã€CPUãƒªã‚½ãƒ¼ã‚¹ãŒå¿…è¦ãªå ´åˆã«Iris Xe MAX Graphic
nminoru 2020/11/02
Intel

GPU
ãƒªãƒ³ã‚¯
TechCrunch | Startup and Technology News
Bumble founder and executive chair Whitney Wolfe Herd raised eyebrows this week with her comments about how AI might change the dating experience. During an onstage interview, Bloombergâ€™s Emily Changâ€¦
nminoru 2020/05/16
NVIDIA

GPU
ãƒªãƒ³ã‚¯
BlazingSQLâ€Šâ€”â€ŠThe GPU SQL Engine now runs over 20X Faster than Apache Spark!
BlazingSQL â€” The GPU SQL Engine now runs over 20X Faster than Apache Spark! A few weeks ago we were very excited to show our first benchmark on an end to end workload. TL,DR you can check out the full article here. The key takeaways from the previous article were the following: We ran an end to end analytics workload: Data Lake â†’ ETL/Feature Engineering â†’ XGBoost TrainingWe built two price equival
nminoru 2019/02/24
Spark

GPU
ãƒªãƒ³ã‚¯
IBMã¨NVIDIAã€AIå‘ã‘ã‚³ãƒ³ãƒãƒ¼ã‚¸ãƒ‰ã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸ã‚·ã‚¹ãƒ†ãƒ ã§ææº - ZDNet Japan
å°åˆ·ã™ã‚‹ ãƒ¡ãƒ¼ãƒ«ã§é€ã‚‹ ãƒ†ã‚ã‚¹ãƒˆ HTML é›»åæ›¸ç± PDF ãƒ€ã‚¦ãƒ³ãƒãƒ¼ãƒ‰ ãƒ†ã‚ã‚¹ãƒˆ é›»åæ›¸ç± PDF ã‚¯ãƒªãƒƒãƒ—ã—ãŸè¨˜äº‹ã‚’Myãƒšãƒ¼ã‚¸ã‹ã‚‰èªã‚€ã“ã¨ãŒã§ãã¾ã™ IBMã¨NVIDIAã¯ã€äººå·¥çŸ¥èƒ½ãƒ¯ãƒ¼ã‚¯ãƒãƒ¼ãƒ‰ã‚„ã€ŒTensorFlowã€ã€ŒPyTorchã€ã€ŒSparkã€ãªã©ã®ãƒ‡ãƒ¼ã‚¿å‡¦ç†ãƒ„ãƒ¼ãƒ«å‘ã‘ã«è¨è¨ˆã•ã‚ŒãŸã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸ã‚·ã‚¹ãƒ†ãƒ ã‚’ç™ºå£²ã™ã‚‹ã€‚ ã“ã®ã€ŒIBM SpectrumAI with NVIDIA DGXã€ã¯ã€IBMã®ã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢å®šç¾©ãƒ•ã‚¡ã‚¤ãƒ«ã‚·ã‚¹ãƒ†ãƒ ã€ã‚ªãƒ¼ãƒ«ãƒ•ãƒ©ãƒƒã‚·ãƒ¥ã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸ã€NVIDIAã®GPUã‚·ã‚¹ãƒ†ãƒ ã€ŒDGX-1ã€ã‚’çµ„ã¿åˆã‚ã›ãŸã‚³ãƒ³ãƒãƒ¼ã‚¸ãƒ‰ã‚·ã‚¹ãƒ†ãƒ ã ã€‚ æœ€è¿‘ã§ã¯ã€ã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸ãƒ™ãƒ³ãƒ€ãƒ¼ãŒã€ã‚ªãƒ¼ãƒ«ãƒ•ãƒ©ãƒƒã‚·ãƒ¥ã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸ã‚’æŽ¡ç”¨ã—ãŸã€AIãƒ¯ãƒ¼ã‚¯ãƒãƒ¼ãƒ‰ã‚„æ©Ÿæ¢°å¦ç¿’ã«æœ€é©åŒ–ã•ã‚ŒãŸã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£ã‚’æŒã¤ã‚¹ãƒˆãƒ¬ãƒ¼ã‚¸è£½å“ã‚’æä¾›ã™ã‚‹ã‚±ãƒ¼ã‚¹ãŒå¢—ãˆã¦ãŠã‚Šã€IBM SpectrumAIã®ç™ºå£²ã‚‚ãã†ã—ãŸãƒˆãƒ¬ãƒ³ãƒ‰ã®ä¸€ç’°ã ã¨è¨€ãˆã‚‹ã€‚éŽåŽ»
nminoru 2018/12/13
IBM

NVIDIA

Storage

GPU
ãƒªãƒ³ã‚¯
æ©Ÿæ¢°å¦ç¿’ã‚¤ãƒ³ãƒ•ãƒ©ã®è¤‡é›‘ã•ã‚’è§£æ¶ˆã™ã‚‹ã‚¯ãƒ©ã‚¦ãƒ‰åž‹GPUãƒ„ãƒ¼ãƒ«ã‚¹ã‚¤ãƒ¼ãƒˆï¼šAWSãªã©å„ç¨®ãƒ™ãƒ³ãƒ€ãƒ¼ãŒæä¾› - TechTargetã‚¸ãƒ£ãƒ‘ãƒ³ ãƒ‡ãƒ¼ã‚¿åˆ†æž
é–¢é€£ã‚ãƒ¼ãƒ¯ãƒ¼ãƒ‰ PaaS | ãƒ‡ãƒ¼ã‚¿åˆ†æž | æ©Ÿæ¢°å¦ç¿’ æ·±å±¤å¦ç¿’ã‚„æ©Ÿæ¢°å¦ç¿’ã¯ã€ã“ã‚Œã¾ã§ã«ãªã„é©æ–°çš„ãªãƒãƒ¼ãƒ‰ã‚¦ã‚§ã‚¢ã®å‘¼ã³æ°´ã«ãªã‚‹ã€‚ã—ã‹ã—ãã®ã‚ˆã†ã«é«˜åº¦ãªã‚¤ãƒ³ãƒ•ãƒ©ã§ã‚¸ãƒ§ãƒ–ã‚’å®Ÿè¡Œã™ã‚‹ã®ã¯å®¹æ˜“ãªã“ã¨ã§ã¯ãªã„ã€‚æœŸå¾…ã®é«˜ã•ã«ã‚‚ã‹ã‹ã‚ã‚‰ãšã€ãƒãƒ¼ãƒ ã¯GPUã‚„TPUã‚’åˆ©ç”¨ã™ã‚‹æœ€æ–°ã®æ©Ÿæ¢°å¦ç¿’ã‚¤ãƒ³ãƒ•ãƒ©ã«ã‚¸ãƒ§ãƒ–ã‚’ç§»è¡Œã•ã›ã‚‹è©¦ã¿ã«é–‹ç™ºæ™‚é–“ã®å¤§åŠã‚’å¥ªã‚ã‚Œã¦ã„ã‚‹ã€‚ ã“ã†ã—ãŸå•é¡ŒãŒå¥‘æ©Ÿã¨ãªã‚Šã€GPUã‚’ã‚¯ãƒ©ã‚¦ãƒ‰å†…ã®æ©Ÿæ¢°å¦ç¿’ã‚¤ãƒ³ãƒ•ãƒ©ã§æ£ã—ãå‹•ä½œã•ã›ã‚‹ãŸã‚ã€ãƒ©ã‚¤ãƒ–ãƒ©ãƒªãªã©ã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢ã®è¦ç´ ã‚’ã‚³ãƒ³ãƒ†ãƒŠã«ã¾ã¨ã‚ã‚‹ä½œæ¥ã‚’è‡ªå‹•åŒ–ã™ã‚‹ã€ãƒ‡ãƒ¼ã‚¿ã‚µã‚¤ã‚¨ãƒ³ã‚¹ãƒ—ãƒ©ãƒƒãƒˆãƒ•ã‚©ãƒ¼ãƒ ã¸ã®é–¢å¿ƒãŒé«˜ã¾ã£ã¦ã„ã‚‹ã€‚ ä½µã›ã¦èªã¿ãŸã„ãŠè–¦ã‚ã®è¨˜äº‹ ã€‡é«˜ã¾ã‚‹GPUã®éœ€è¦ GPUãƒ‡ãƒ¼ã‚¿ãƒ™ãƒ¼ã‚¹ã¨ã¯ã€€CPUã¨ã®æ¯”è¼ƒã§åˆ†ã‹ã‚‹ãƒ“ãƒƒã‚°ãƒ‡ãƒ¼ã‚¿ä¸¦åˆ—å‡¦ç† GPUãŒã‘ã‚“å¼•ã™ã‚‹â€œãƒãƒ¼ãƒ‰é»„é‡‘æ™‚ä»£â€ã€ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚„AIå‡¦ç†ã§éœ€è¦é«˜ã¾ã‚‹ GPUã‚’æ´»ç”¨ã—ãŸã‚¤ãƒ³ãƒ•ãƒ©æ§‹ç¯‰ GPUãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹
nminoru 2018/10/03
AI

GPU
ãƒªãƒ³ã‚¯
GPUã‚’ç”¨ã„ãŸSSLãƒªãƒãƒ¼ã‚¹ãƒ—ãƒã‚ã‚·ã®å®Ÿè£…ã«ã¤ã„ã¦ - ã‚†ã†ã†ããƒ–ãƒã‚°
è¿‘å¹´ï¼Œæ±Žç”¨è¨ˆç®—ã®é«˜é€ŸåŒ–ã®ãŸã‚ã®ã‚¢ã‚¯ã‚»ãƒ©ãƒ¬ãƒ¼ã‚¿ã¨ã—ã¦æ³¨ç›®ã•ã‚Œã¦ã„ã‚‹GPUã‚’ï¼Œãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯å‡¦ç†ã«é©ç”¨ã™ã‚‹ä¸€ç’°ã¨ã—ã¦ï¼Œã‚µãƒ¼ãƒã‚µã‚¤ãƒ‰ã®SSLå‡¦ç†ã«æ³¨ç›®ã—ãŸè«–æ–‡ã‚’èªã‚“ã ã®ã§ï¼Œå†…å®¹ã‚’è»½ãç´¹ä»‹ã—ã¾ã™ï¼Ž SSLShader - GPU-accelerated SSL Proxy SSLShader SSLShader: Cheap SSL acceleration with commodity processors Proceedings of the 8th USENIX conference on Networked systems design and implementation 2011 ãªãŠï¼Œè©•ä¾¡ã«ä½¿ã‚ã‚ŒãŸå®Ÿè£…ã®ä¸€éƒ¨ã®ã‚½ãƒ¼ã‚¹ã‚³ãƒ¼ãƒ‰ãŒå…¬é–‹ã•ã‚Œã¦ã„ã¾ã™ï¼Ž http://shader.kaist.edu/sslshader/libgpu crypto/ ç´¹ä»‹ èƒŒæ™¯ SSL(Secure Socket
nminoru 2018/09/19
GPU

SSL
ãƒªãƒ³ã‚¯
Intelã®GPUéƒ¨é–€å‰¯ç¤¾é•·ãŒAMDã®Radeon Technologies Groupã¸ç§»ç± - 4Gamer.net
Intelã®GPUéƒ¨é–€å‰¯ç¤¾é•·ãŒAMDã®Radeon Techno logies Groupã¸ç§»ç± ç·¨é›†éƒ¨ï¼šä½ã€…å±±è–«éƒ Martin Ashtonæ°ï¼ˆâ€»å†™çœŸã¯æ°ã®LinkedInã‚¢ã‚«ã‚¦ãƒ³ãƒˆãƒšãƒ¼ã‚¸ã‚ˆã‚Šï¼‰ Imagination Techno logiesï¼ˆæ—§VideoLogicï¼‰ã«ãŠã„ã¦20å¹´ä»¥ä¸Šã®ã‚ãƒ£ãƒªã‚¢ã‚’æŒã¡ï¼Œ2016å¹´ã«Intelã¸ç§»ç±ã—ã¦ã„ãŸGPU ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã®Martin Ashtonï¼ˆãƒžãƒ¼ãƒ†ã‚£ãƒ³ãƒ»ã‚¢ã‚·ãƒ¥ãƒˆãƒ³ï¼‰æ°ãŒï¼ŒåŒ—ç±³æ™‚é–“2018å¹´7æœˆ9æ—¥ä»˜ã‘ã§AMDã¸ç§»ç±ã—ãŸã€‚ ã“ã‚Œã¯æ°ã®LinkedInã‚¢ã‚«ã‚¦ãƒ³ãƒˆãƒšãƒ¼ã‚¸ã§æ˜Žã‚‰ã‹ã«ãªã£ãŸã‚‚ã®ã ãŒï¼ŒIntelæ™‚ä»£ã®è‚©æ›¸ãã¯Vice President, Core and Visual Computing Groupå…¼Chief Engineer, VTTå…¼Director of Hardware and Co-Director of Architectu
nminoru 2018/07/11
Intel

AMD

GPU
ãƒªãƒ³ã‚¯
ãƒ˜ãƒ†ãƒDBç¤¾ã®GPUã‚’ç”¨ã„ãŸãƒ‡ãƒ¼ã‚¿ãƒ™ãƒ¼ã‚¹é«˜é€ŸåŒ–æŠ€è¡“ PG-Stromã®è²©å£²ã‚’é–‹å§‹ã„ãŸã—ã¾ã™ã€‚ - æ ªå¼ä¼šç¤¾ ã‚¨ãƒ«ã‚¶ ã‚¸ãƒ£ãƒ‘ãƒ³
ãƒ‹ãƒ¥ãƒ¼ã‚¹ãƒªãƒªãƒ¼ã‚¹ 2018/04/18 NEWS ãƒ˜ãƒ†ãƒDBç¤¾ã®GPUã‚’ç”¨ã„ãŸãƒ‡ãƒ¼ã‚¿ãƒ™ãƒ¼ã‚¹é«˜é€ŸåŒ–æŠ€è¡“ PG-Stromã®è²©å£²ã‚’é–‹å§‹ã„ãŸã—ã¾ã™ã€‚ ã“ã®åº¦ã€æ ªå¼ä¼šç¤¾ã‚¨ãƒ«ã‚¶ ã‚¸ãƒ£ãƒ‘ãƒ³ï¼ˆæ±äº¬éƒ½æ¸¯åŒºã€ä»£è¡¨å–ç· å½¹ç¤¾é•·ï¼šå†…è—¤ã€€ä¸‡ç¾©ï¼‰ã¯ã€ãƒ˜ãƒ†ãƒDBç¤¾ã®GPUã‚’ç”¨ã„ãŸãƒ‡ãƒ¼ã‚¿ãƒ™ãƒ¼ã‚¹é«˜é€ŸåŒ–æŠ€è¡“ã§ã‚ã‚‹PG-Stromã®ã‚µãƒ–ã‚¹ã‚¯ãƒªãƒ—ã‚·ãƒ§ãƒ³è£½å“ã€ãŠã‚ˆã³åŒã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢ã‚’æè¼‰ã—ãŸã‚¢ãƒ—ãƒ©ã‚¤ã‚¢ãƒ³ã‚¹è£½å“ã®è²©å£²ã‚’2018å¹´4æœˆ24æ—¥ã‚ˆã‚Šé–‹å§‹ã„ãŸã—ã¾ã™ã€‚ ç”»åƒã¯ã‚¤ãƒ¡ãƒ¼ã‚¸ã§ã™ã€‚ ãƒãƒ¼ãƒ‰ã‚¦ã‚§ã‚¢ã®èƒ½åŠ›ã‚’é™ç•Œã¾ã§å¼•ãå‡ºã—ã€ 1Uï¼ï¼’Uãƒ©ãƒƒã‚¯ã‚µãƒ¼ãƒã§DWHæ°´æº–ã®å‡¦ç†èƒ½åŠ›ã‚’å®Ÿç¾ã™ã‚‹ PG-Strom ãƒãƒ¼ã‚¸ãƒ§ãƒ³ 2.0 PG-Stromã¨ã¯ã€æœ€ã‚‚åºƒãä½¿ã‚ã‚Œã¦ã„ã‚‹ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹ã®ãƒªãƒ¬ãƒ¼ã‚·ãƒ§ãƒŠãƒ«ãƒ‡ãƒ¼ã‚¿ãƒ™ãƒ¼ã‚¹ã§ã‚ã‚‹PostgreSQLå‘ã‘ã«è¨è¨ˆã•ã‚ŒãŸæ‹¡å¼µãƒ¢ã‚¸ãƒ¥ãƒ¼ãƒ«ã§ã€SQLã‚³ãƒžãƒ³ãƒ‰ã‹ã‚‰è‡ªå‹•ç”Ÿæˆã—ãŸGPUãƒ—ãƒã‚°ãƒ©ãƒ ã‚’GPUã®æŒã¤æ•°åƒã‚³ã‚¢ã¨åºƒå¸¯åŸŸãƒ¡ãƒ¢ãƒª
nminoru 2018/04/18
PostgreSQL

GPU
ãƒªãƒ³ã‚¯
Exploring data with pandas and MapD using Apache Arrow
nminoru 2018/02/27
GPU

DWH
ãƒªãƒ³ã‚¯
SQL Server 2017 ã«ãŠã‘ã‚‹ GPU ã®åˆ©ç”¨ (ML Services ã® GPU åˆ©ç”¨) ã«ã¤ã„ã¦ at SE ã®é›‘è¨˜
SQL Server 2017 ã® GPU ã‚µãƒãƒ¼ãƒˆã«ã¤ã„ã¦ã€ã¾ã¨ã‚ã¦ãŠããŸã„ã¨æ€ã„ã¾ã™ã€‚ https://gist.github.com/MasayukiOzawa/7965ffce226a8f94ba457671dd844b92 ã«ä»Šå›žå‹‰å¼·ã‚’ã™ã‚‹ä¸ã§ä½¿ç”¨ã—ãŸã‚µãƒ³ãƒ—ãƒ«ã‚’ç½®ã„ã¦ã‚ã‚Šã¾ã™ã€‚ æƒ…å ±ã«ã¤ã„ã¦ã¯ Linux/Docker ç‰ˆã‚‚ç™»å ´! Microsoft SQL Server 2017 ã®æƒ…å ±ã¾ã¨ã‚ã€5/1 æ›´æ–°ã€‘ ã‚’ç¢ºèªã—ã¦ã„ãŸã ãã¨ã‚ˆã„ã‹ã¨è€ƒãˆã¦ã„ã¾ã™ãŒã€SQL Server 2017 ã® GPU ã®å¯¾å¿œã«ã¤ã„ã¦ã¯æ¬¡ã®ã‚ˆã†ãªè¨˜è¼‰ã¨ãªã£ã¦ã„ã¾ã™ã€‚ NVIDIA GPU ã‚¢ã‚¯ã‚»ãƒ©ãƒ¬ãƒ¼ã‚·ãƒ§ãƒ³ã®å®Ÿç¾: R/Python ã‚’åˆ©ç”¨ã—ã¦ GPU ã«ã‚ˆã‚‹é«˜é€Ÿè¨ˆç®—ã§ã®æ©Ÿæ¢°å¦ç¿’ã‚‚å®Ÿç¾ã€1 ç§’ã‚ãŸã‚Š100 ä¸‡ä»¶ã®äºˆæ¸¬å‡¦ç†ãŒå¯èƒ½ã§ã™ã€‚ SQL Server 2017 ãŒ GPU ã‚’ä½¿ç”¨å¯èƒ½ã«ãªã£ã¦ã„ã‚‹ç¯„
nminoru 2017/09/20
SQLServer

GPU
ãƒªãƒ³ã‚¯
OTOYã€ä½™ã£ã¦ã„ã‚‹GPUãƒªã‚½ãƒ¼ã‚¹ã‚’ãƒˆãƒ¼ã‚¯ãƒ³ã¨å¼•ãæ›ãˆã«å…±æœ‰ã™ã‚‹ãƒ–ãƒãƒƒã‚¯ãƒã‚§ãƒ¼ãƒ³ãƒ™ãƒ¼ã‚¹ã®åˆ†æ•£P2Pã€ŒGPUã€ãƒ¬ãƒ³ãƒ€ãƒªãƒ³ã‚°ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã€ŒRender Tokenï¼ˆRNDRï¼‰ã€ã‚’ç™ºè¡¨
OTOYã€ä½™ã£ã¦ã„ã‚‹GPUãƒªã‚½ãƒ¼ã‚¹ã‚’ãƒˆãƒ¼ã‚¯ãƒ³ã¨å¼•ãæ›ãˆã«å…±æœ‰ã™ã‚‹ãƒ–ãƒãƒƒã‚¯ãƒã‚§ãƒ¼ãƒ³ãƒ™ãƒ¼ã‚¹ã®åˆ†æ•£P2Pã€ŒGPUã€ãƒ¬ãƒ³ãƒ€ãƒªãƒ³ã‚°ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã€ŒRender Tokenï¼ˆRNDRï¼‰ã€ã‚’ç™ºè¡¨ 2017-09-02 OTOYã¯ã€GPUãƒªã‚½ãƒ¼ã‚¹ã‚’å…±æœ‰ã™ã‚‹ãƒ–ãƒãƒƒã‚¯ãƒã‚§ãƒ¼ãƒ³ãƒ™ãƒ¼ã‚¹ã®åˆ†æ•£P2Pã€€GPUãƒ¬ãƒ³ãƒ€ãƒªãƒ³ã‚°ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãƒ»ãƒ—ãƒã‚¸ã‚§ã‚¯ãƒˆã€ŒRender Tokenï¼ˆRNDRï¼‰ã€ã‚’ç™ºè¡¨ã—ã¾ã—ãŸã€‚ Render Tokenã¯ã€ä½™ã£ã¦ã„ã‚‹GPUãƒªã‚½ãƒ¼ã‚¹ã‚’ãƒˆãƒ¼ã‚¯ãƒ³ã¨å¼•ãæ›ãˆã«æä¾›ã™ã‚‹ã“ã¨ãŒã§ãã€è³¼å…¥ã—ãŸäººã¯ãã®GPUãƒ‘ãƒ¯ãƒ¼ã‚’ä½¿ã£ã¦Peer to Peerãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ä¸Šã§åˆ†æ•£ã—ã¦ãƒ¬ãƒ³ãƒ€ãƒªãƒ³ã‚°å‡¦ç†ã§ãã‚‹ã¨ã„ã†ã‚¨ã‚³ã‚·ã‚¹ãƒ ã§ã™ã€‚ ãƒ–ãƒãƒƒã‚¯ãƒã‚§ãƒ¼ãƒ³ãƒ»ãƒ—ãƒãƒˆã‚³ãƒ«ã«ã¯ã€ã€ŒEthereumã€ãŒä½¿ç”¨ã•ã‚Œã€å…¬å¹³æ€§ã€å®‰å…¨é¢ã‚’æ‹…ã„ã¾ã™ã€‚ ã“ã®ã“ã¨ã§ã€ä½¿ã£ã¦ã„ãªã„GPUãƒªã‚½ãƒ¼ã‚¹ã‚’æœ‰åŠ¹æ´»ç”¨ã™ã‚‹ã“ã¨ãŒã§ãã€ä¸€æ–¹ã§ãƒ¬ãƒ³ãƒ€ãƒªãƒ³ã‚°ã—ãŸã„äººã¯ä½Žã‚³ã‚¹ãƒˆã§å¤§é‡
nminoru 2017/09/03
GPU
ãƒªãƒ³ã‚¯
æ—¥æœ¬ä»®æƒ³åŒ–æŠ€è¡“ã€ã€ŒGPU on OpenStackã€ã‚’æŽ¨é€²--HPCãªã©ã«é©ç”¨
å°åˆ·ã™ã‚‹ ãƒ¡ãƒ¼ãƒ«ã§é€ã‚‹ ãƒ†ã‚ã‚¹ãƒˆ HTML é›»åæ›¸ç± PDF ãƒ€ã‚¦ãƒ³ãƒãƒ¼ãƒ‰ ãƒ†ã‚ã‚¹ãƒˆ é›»åæ›¸ç± PDF ã‚¯ãƒªãƒƒãƒ—ã—ãŸè¨˜äº‹ã‚’Myãƒšãƒ¼ã‚¸ã‹ã‚‰èªã‚€ã“ã¨ãŒã§ãã¾ã™ æ—¥æœ¬ä»®æƒ³åŒ–æŠ€è¡“ã¯7æœˆ18æ—¥ã€OpenStackç’°å¢ƒã§ã®GPU æŠ€è¡“ã€ŒGPU on OpenStackã€ã®æ´»ç”¨ã‚’æŽ¨é€²ã—ã¦ã„ãã¨ç™ºè¡¨ã—ãŸã€‚OpenStackç’°å¢ƒã®ä»®æƒ³ãƒžã‚·ãƒ³ã‚„ã‚³ãƒ³ãƒ†ãƒŠã«å¯¾ã—ã¦GPUã®è¨ˆç®—ãƒªã‚½ãƒ¼ã‚¹ã‚’å‰²ã‚Šå½“ã¦ã¦ã€ãƒã‚¤ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ãƒ†ã‚£ãƒ³ã‚°ã‚„ãƒ‡ã‚¹ã‚¯ãƒˆãƒƒãƒ—ä»®æƒ³åŒ–ãªã©ã®ã•ã¾ã–ã¾ãªç”¨é€”ã§åˆ©ç”¨ã§ãã‚‹ã¨ã„ã†ã€‚ åŒæŠ€è¡“ã‚’æ´»ç”¨ã™ã‚‹ã“ã¨ã§ã€GPUã‚’æ¥å‹™ã§æ´»ç”¨ã™ã‚‹ãƒ¦ãƒ¼ã‚¶ãƒ¼ã¯ã€åˆ©ä¾¿æ€§ã®å‘ä¸Šã‚„ä¸€å…ƒç®¡ç†ã«ã‚ˆã‚‹é‹ç”¨ã®åŠ¹çŽ‡åŒ–ã‚’å®Ÿç¾ã§ãã‚‹ãŒã€ç¾åœ¨ã€ç’°å¢ƒæ§‹ç¯‰ã‚„åˆ©ç”¨æ–¹æ³•ã«ã¤ã„ã¦ã®æƒ…å ±ã¯ååˆ†ã§ã¯ãªã„ã¨ã®èªè˜ã‚’ç¤ºã—ãŸã€‚ ãã“ã§ã€æ—¥æœ¬ä»®æƒ³åŒ–æŠ€è¡“ã¯ã€NVIDIAã¨ã‚¢ã‚¹ã‚¯ã¨å”åŠ›ã—ã¦ç’°å¢ƒæ§‹ç¯‰ã¨æŠ€è¡“è©•ä¾¡ã‚’è¡Œã„ã€åˆ©ç”¨è€…ã‚„ã‚¤ãƒ³ãƒ•ãƒ©ç®¡ç†è€…ã«ã¨ã£ã¦å½¹ç«‹ã¤æƒ…å ±ã‚’ç™ºä¿¡ã—ã¦ã„ãã“ã¨ã¨
nminoru 2017/07/21
OpenStack

GPU
ãƒªãƒ³ã‚¯
è£½å“
PG-StromGPUã¨NVMEã®èƒ½åŠ›ã‚’ãƒ•ãƒ«ã«å¼•ãå‡ºã—ã€ä¸¦åˆ—å‹•ä½œã™ã‚‹æ•°åƒã‚³ã‚¢ã®åŠ›ã§PostgreSQLã‚’çˆ†é€ŸåŒ–ã€ãƒ†ãƒ©ãƒã‚¤ãƒˆã‚’è¶…ãˆã‚‹å¤§é‡ãƒ‡ãƒ¼ã‚¿ã«æŒ‘ã‚€ãŸã‚ã®ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹æ‹¡å¼µãƒ¢ã‚¸ãƒ¥ãƒ¼ãƒ«ã€‚ PG-Stromã®æ¦‚è¦ PG-Stromã¯æœ€ã‚‚åºƒç¯„ã«åˆ©ç”¨ã•ã‚Œã¦ã„ã‚‹OSS-DBã‚·ã‚¹ãƒ†ãƒ ã®ä¸€ã¤ã€PostgreSQLå‘ã‘ã®æ‹¡å¼µãƒ¢ã‚¸ãƒ¥ãƒ¼ãƒ«ã§ã€æ•°åTBã€æ•°ç™¾å„„ä»¶ã¨ã„ã£ãŸè¦æ¨¡ã®å¤§é‡ãƒ‡ãƒ¼ã‚¿ã®æ¤œç´¢ã€é›†è¨ˆã€åŠ å·¥ã¨ã„ã£ãŸå‡¦ç†ã‚’ã€GPUã‚„NVME-SSDã¨ã„ã£ãŸæœ€æ–°ã®ãƒãƒ¼ãƒ‰ã‚¦ã‚§ã‚¢ã®èƒ½åŠ›ã‚’æœ€å¤§é™ã«å¼•ãå‡ºã—ã¦ã€ã‚·ãƒ³ãƒ—ãƒ«ãªã‚·ãƒ³ã‚°ãƒ«ãƒŽãƒ¼ãƒ‰æ§‹æˆã®ã‚·ã‚¹ãƒ†ãƒ ã§ã€é«˜é€Ÿã«å®Ÿè¡Œã™ã‚‹ãŸã‚ã®ã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢ã§ã™ã€‚ GPUã®ç‰¹å¾´ GPUã«ã¯æ•°åƒå€‹ã®ãƒ—ãƒã‚»ãƒƒã‚µã‚³ã‚¢ãŒæè¼‰ã•ã‚Œã€ä¸¦åˆ—è¨ˆç®—ã‚’æ¥µã‚ã¦åŠ¹çŽ‡ã‚ˆãå®Ÿè¡Œã™ã‚‹ã‚ˆã†ã«è¨è¨ˆã•ã‚Œã¦ã„ã¾ã™ã€‚ãã®ä¸€ã¤ãŒè¡Œåˆ—æ¼”ç®—ã§ã™ãŒã€ã“ã‚Œã¯å¤§é‡ã®å‡è³ªãªãƒ‡ãƒ¼ã‚¿ã«å¯¾ã—ã¦åŒä¸€ã®æ¼”ç®—ã‚’åŒæ™‚ã«å®Ÿè¡Œã™ã‚‹ã¨ã„ã†ç‰¹æ€§ãŒã‚ã‚Šã¾ã™ã€‚SQLã«ã‚‚é¡žä¼¼ã®å‡¦ç†ã¯
nminoru 2017/07/05
PostgreSQL

GPU
ãƒªãƒ³ã‚¯
AMDã€Vega10ã‚³ã‚¢æŽ¡ç”¨ã§å˜ç²¾åº¦12.3TFLOPSã®ã€ŒRadeon Instinct MI25ã€
nminoru 2017/06/24
AMD

GPU
ãƒªãƒ³ã‚¯
1 2 3 æ¬¡ã®ãƒšãƒ¼ã‚¸