[B! gpu] masterqã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

masterq id:masterq

gpuã«é–¢ã™ã‚‹masterqã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (119)

${{author_name}}$

{{author_name}} {{created}}

{{{comment_expanded}}}

{{label}}

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

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

{{#following_bookmarks}}

${{author_name}}$

{{author_name}} {{created}}

{{{comment_expanded}}}

{{label}}

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

{{/following_bookmarks}}

{{/is_wiped}}

FireAlpaca SE 3.0 ãƒªãƒªãƒ¼ã‚¹ï¼
å…ˆã»ã©ç„¡äº‹ãƒªãƒªãƒ¼ã‚¹ã•ã‚Œã¾ã—ãŸã€‚ ãƒšã‚¤ãƒ³ãƒˆãƒ„ãƒ¼ãƒ« FireAlpaca SE ã“ã‚Œã«åˆã‚ã›ã¦å¹¾ã¤ã‹ãƒ•ã‚£ãƒ«ã‚¿ã‚’MFGã‚¹ãƒˆã‚¢ã«è¿½åŠ ã™ã‚‹ä½œæ¥ã‚’ã—ã¦ã„ã¾ã™ã€‚ ã“ã®ä½œæ¥ãŒçµ‚ã‚ã‚Œã°ç¾æ™‚ç‚¹ã§ã¯10å€‹ãã‚‰ã„ã®ãƒ•ã‚£ãƒ«ã‚¿ã«ãªã‚Šã¾ã™ã€‚ ã¾ã ã¾ã å¢—ã‚„ã—ã¦ã„ããŸã„ã€‚ è‡ªåˆ†ãŒã“ã“æ•°å¹´é–‹ç™ºã—ã¦ã„ãŸã‚‚ã®ã®é›†å¤§æˆã§ã™ã€‚ãªã‹ãªã‹æ„Ÿæ…¨æ·±ã„ã§ã™ãã€‚åéŸ¿ãŒã‚ã‚‹ã¨ã„ã„ãªãã€‚ ç§ã¨@nattou_orgã§åˆ†æ‹…ã—ã¦ä½œã£ã¦ã„ãŸæ„Ÿã˜ã§ã™ãŒã€ã“ã“ã§ã¯è‡ªåˆ†ãŒæ‹…å½“ã—ãŸéƒ¨åˆ†ã®è©±ã‚’å¹¾ã¤ã‹ã€‚ ã§ã‹ã„ã‚ãƒ£ãƒ³ãƒã‚¹ã¨å¤§é‡ã®ãƒ¬ã‚¤ãƒ¤ãƒ¼ã§ã‚‚çˆ†é€Ÿã§å‹•ã„ã¦ã€æœ€å¼·ã®ãƒ–ãƒ©ã‚·ã¨æœ€å¼·ã®ãƒ•ã‚£ãƒ«ã‚¿ã‚’å‚™ãˆã‚‹ãƒšã‚¤ãƒ³ãƒˆã‚½ãƒ•ãƒˆ ä½œã‚ã†ã¨ã—ãŸã®ã¯ã€ã ã„ãŸã„ã“ã®ä¸‰ã¤ã‚’å‚™ãˆãŸãƒšã‚¤ãƒ³ãƒˆã‚½ãƒ•ãƒˆã€ã¨ãªã‚Šã¾ã™ã€‚ ãƒ—ãƒãƒ•ã‚§ãƒƒã‚·ãƒ§ãƒŠãƒ«ãƒ¦ãƒ¼ã‚¹ã¨ã—ã¦ã€æœ¬æ ¼çš„ãªPCã§ä½¿ãˆã°ãã®æ€§èƒ½ã‚’éºæ†¾ç„¡ãç™ºæ®ã™ã‚‹ã€ã¨ã„ã†ã®ã¯å¼·ãæ„è˜ã—ãŸã¨ã“ã‚ã§ã™ã€‚ æ™®é€šã®ä½¿ç”¨ã§ã¯ãŠç›®ã«ã‹ã‹ã‚Œãªã„ã‚ˆã†ãªå¤§ããªãƒ‡ãƒ¼ã‚¿ã§ã‚‚ååˆ†ã«ä½¿ã„ç‰©ã«ãªã‚‹ã‚ˆã†ãªã€
masterq 2025/11/21
ã„ã„è©±

paint

graphics

gpu

language

future

gui

ui

mfg
ãƒªãƒ³ã‚¯
Intel CPUã¨NVIDIA GPUã‚’çµ±åˆã—ãŸå¤¢ã®x86 SoCå®Ÿç¾ã¸ã€‚NVIDIAãŒIntelã«7,400å„„å††æŠ•è³‡
masterq 2025/09/19
intel

nvidia

gpu

cpu

x86
ãƒªãƒ³ã‚¯
è¥¿å·å–„å¸ã®3DGEï¼šAMDã®GPUã€ŒRDNA 4ã€å®Œå…¨è§£èª¬ã€‚æ—¢å˜ã®ã‚²ãƒ¼ãƒ ã‚’ãã®ã¾ã¾é«˜é€ŸåŒ–ã™ã‚‹æ–°æŠ€è¡“ã®æ•°ã€…ãŒæ˜Žã‚‰ã‹ã«
è¥¿å·å–„å¸ã®3DGEï¼šAMDã®GPUã€ŒRDNA 4ã€å®Œå…¨è§£èª¬ã€‚æ—¢å˜ã®ã‚²ãƒ¼ãƒ ã‚’ãã®ã¾ã¾é«˜é€ŸåŒ–ã™ã‚‹æ–°æŠ€è¡“ã®æ•°ã€…ãŒæ˜Žã‚‰ã‹ã« ãƒ©ã‚¤ã‚¿ãƒ¼ï¼šè¥¿å·å–„å¸ AMDã®æœ€æ–°GPUã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£ã§ã‚ã‚‹ã€ŒRDNA 4ã€ã¯ï¼Œ2025å¹´3æœˆã«ï¼Œæœ€åˆã®æŽ¡ç”¨è£½å“ã€ŒRadeon RX 9070 XTã€ã¨ã€ŒRadeon RX 9070ã€ãŒç™»å ´ã€‚2025å¹´6æœˆã«ã¯ï¼ŒãƒŸãƒ‰ãƒ«ã‚¯ãƒ©ã‚¹å¸‚å ´å‘ã‘ã®ã€ŒRadeon RX 9060 XTã€ãŒï¼Œ2025å¹´8æœˆã«ã¯ï¼Œã€ŒRadeon RX 9060ã€ãŒç™»å ´ã—ãŸã€‚ Radeon RX 9070 XTæè¼‰ã‚«ãƒ¼ãƒ‰ã®Sapphire Techno logyè£½ã€ŒPURE Radeon RX 9070 XT GPUã€ XFXè£½ã®Radeon AI PRO R9700æè¼‰ã‚«ãƒ¼ãƒ‰ã‚’æŽ²ã’ã‚‹Scott Olschewskyæ°ï¼ˆRadeon Graphicsï¼ŒProduct Management Directorï¼ŒAMDï¼‰
masterq 2025/09/05
gpu

amd

doc

ã‚ã¨ã§èªã‚€
ãƒªãƒ³ã‚¯
GitHub - lean-dojo/LeanCopilot: LLMs as Copilots for Theorem Proving in Lean
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert
masterq 2025/05/08
suggest_tacticsã‚’ä½¿ã†ã¨Copilotã®ã‚ˆã†ã«LLMã¨å¯¾è©±ã—ãªãŒã‚‰è¨¼æ˜Žã§ãã‚‹

llm

ai

lean

proof

verify

gpu
ãƒªãƒ³ã‚¯
Runpod | The cloud built for AI
Instant AI workloadsâ€”no setup, scaling, or idle costs.
masterq 2025/05/04
cloud

gpu

ai

ã‚ã¨ã§èªã‚€

nvidia
ãƒªãƒ³ã‚¯
NVIDIA Jetson Edge AI Solutions - Jetson Nano, Jetson Xavier NX, Jetson AGX Orin - Seeed Studio
masterq 2025/02/07
seeed

nvidia

gpgpu

gpu

hardware
ãƒªãƒ³ã‚¯
DeepSeek R1ã‚’ã»ã¼æº–å‚™ãªã—ã‹ã‚‰ãƒãƒ¼ã‚«ãƒ«GPUã§å‹•ã‹ã™ï½œã—ã´ã¡ã‚ƒã‚“
ã—ã´ã´ã´ï¼ Vtuberã®ã—ã´ã¡ã‚ƒã‚“ (https://www.youtube.com/@CP-chan)ã§ã™ã€‚æ™®æ®µã¯ã‚²ãƒ¼ãƒ é…ä¿¡ã—ã‹ã—ã¦ã¾ã›ã‚“ãŒã€ãŸã¾ã«AIã«ã¤ã„ã¦ç™ºä¿¡ã™ã‚‹ã“ã¨ãŒã‚ã‚Šã¾ã™ã€‚ä»Šæ—¥ã¯AIã®è¨˜äº‹ã®æ–¹ã€‚ ç¾åœ¨ã¯ãƒãƒ¼ã‚«ãƒ«AIãƒ¢ãƒ‡ãƒ«ã«é–¢ã™ã‚‹é€£è¼‰ã‚’ã—ã¦ã„ã¾ã™ã€‚ ç¬¬ä¸€å¼¾ã€€æœ¬è¨˜äº‹ï¼ˆDeepSeek R1ã‚’ã»ã¼æº–å‚™ãªã—ã‹ã‚‰ãƒãƒ¼ã‚«ãƒ«GPUã§å‹•ã‹ã™ï¼‰ ç¬¬äºŒå¼¾ã€€Clineï¼‹ãƒãƒ¼ã‚«ãƒ«ç‰ˆDeepSeek R1ã§AIã‚³ãƒ¼ãƒ‡ã‚£ãƒ³ã‚°ã‚’ä½¿ã„æ”¾é¡Œã«ã™ã‚‹ï¼ˆé«˜ã‚¹ãƒšãƒƒã‚¯ãƒžã‚·ãƒ³å‘ã‘ï¼‰ æœ€è¿‘è©±é¡Œã®ãƒãƒ¼ã‚«ãƒ«ã§å‹•ãLLMã€DeepSeek R1 ã‚’ãƒãƒ¼ã‚«ãƒ«GPUç’°å¢ƒï¼ˆNVIDIAï¼‰ã§å‹•ã‹ã—ã¦ã¿ã¾ã—ã‚‡ã†ã€‚ å¤šå°‘ã®ã‚³ãƒžãƒ³ãƒ‰ãƒ©ã‚¤ãƒ³ã®æ“ä½œãŒã§ãã‚Œã°ã€äº‹å‰ã«ãƒãƒ¼ã‚«ãƒ«LLMã‚’è§¦ã£ãŸã“ã¨ãŒãªãã¦ã‚‚ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ã§ãã‚‹ã‚ˆã†ã«æ›¸ãã¤ã‚‚ã‚Šã§ã™ã€‚ ã‚µãƒ ãƒã¯DeepSeek R1ãã‚“ã«ç”»åƒã‚’ç”Ÿæˆã—ã¦ã‚‚ã‚‰ãŠã†ã¨ã—ãŸã¨ã“ã‚ã§ã™ï¼ˆãã‚“ãªæ©Ÿèƒ½ã¯ã‚ã‚Šã¾ã›ã‚“ï¼‰ã€‚
masterq 2025/01/28
"æ—¥æœ¬èªžæ€§èƒ½ã«ã‹ãªã‚Šå·®ãŒã‚ã‚‹ã®ã§ã€æœ€ä½Žã§ã‚‚æ—¥æœ¬äººã¯14Bä»¥ä¸Šã®ãƒ¢ãƒ‡ãƒ«ã‚’ä½¿ã£ãŸæ–¹ãŒè‰¯ã„"/"GPU NVIDIA RTX 3080 VRAM 10GBï¼ˆã‚‚ã£ã¨ä½Žã‚¹ãƒšãƒƒã‚¯ã§ã‚‚å‹•ã"

gpu

gpgpu

ai

llm

deepseek

japanese

language

cuda
ãƒªãƒ³ã‚¯
ä¸–ç•Œä¸€å®‰ã„GPUã‚’æ±‚ã‚ã¦2024å†¬ - Qiita
çµè«– ã‚¹ãƒãƒƒãƒˆã§ä½¿ã†ãªã‚‰Colabã€RunpodãŒæœ€å¼· ã‚²ãƒ¼ãƒŸãƒ³ã‚°PCã§ã„ã„ãªã‚‰ã©ã“ã‹ã§è²·ã†ã‹BREAJONã§ã‚µãƒ–ã‚¹ã‚¯ã—ã‚ˆã† ã ã„ãŸã„Runpodã•ã‚“ãŒãªã‚“ã¨ã‹ã—ã¦ãã‚Œã‚‹ ä»Šå›žã®è¨˜äº‹ã«ã¤ã„ã¦ ã“ã‚“ã«ã¡ã¯ï¼é€†ç€¬å· (Â https://x.com/gyakuse ) ã§ã™ï¼ä»Šæ—¥ã¯æœ€å¼·ã®GPUãƒ—ãƒãƒã‚¤ãƒ€ãƒ¼æ±ºå®šæˆ¦ã‚’ã—ã¾ã™ã€‚ä¸–ã¯å¤§GPUæ™‚ä»£ã¨ãªã‚Šã¾ã—ãŸã€‚Valorantã‚’ã‚„ã‚‹ã«ã‚‚ã€APEXã‚’ã‚„ã‚‹ã«ã‚‚ã€ã‚ã‚‹ç¨‹åº¦é«˜æ€§èƒ½ãªGPUãŒå¿…è¦ã§ã™ã€‚League of Legends ã¯ãƒŽãƒ¼ãƒˆãƒ‘ã‚½ã‚³ãƒ³ã§ã‚‚éŠã¹ã‚‹ã®ã§ãŠã™ã™ã‚ã§ã™ã€‚ ãã®ä»–ã®åˆ©ç”¨ã¨ã—ã¦æ©Ÿæ¢°å¦ç¿’ã®ãƒ¢ãƒ‡ãƒ«ã‚’ãƒˆãƒ¬ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ã—ãŸã„ã¨ã„ã†éœ€è¦ãŒã‚ã‚‹ã¨æ€ã„ã¾ã™ã€‚æŽ¨è«–ã‚‚ã•ã›ãŸã„ã§ã™ã‚ˆãã€‚è¨€èªžãƒ¢ãƒ‡ãƒ«ã‚„éŸ³å£°èªè˜ãƒ¢ãƒ‡ãƒ«ã€éŸ³å£°åˆæˆãƒ¢ãƒ‡ãƒ«â€¦ã„ã‚ã‚“ãªã‚‚ã®ã‚’å‹•ã‹ãã†ã¨ã™ã‚‹ã¨ã€ã¤ã‚‰ã„ã€‚ã•ã‚‰ã«ã€ãƒˆãƒ¬ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ã—ã‚ˆã†ã¨ã™ã‚‹ã¨ã‚‚ã£ã¨ã¤ã‚‰ã„ã€‚ã§ã™ãŒã€æœ€è¿‘ã¯éžå¸¸ã«å®‰ä¾¡ãªã‚¯ãƒ©ã‚¦ãƒ‰GPUãŒå°é ã—
masterq 2024/12/23
"ã ã„ãŸã„Runpodã•ã‚“ãŒãªã‚“ã¨ã‹ã—ã¦ãã‚Œã‚‹"

gpu

gpgpu

cost

ai

cloud

ã‚ã¨ã§èªã‚€
ãƒªãƒ³ã‚¯
CUDAç”¨ã®emacsè¨å®šã¨lspã®å°Žå…¥æ–¹æ³• - pyopyopyo - Linuxã¨ã‹ãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ã®è¦šãˆæ›¸ã -
CUDAã®ãƒ—ãƒã‚°ãƒ©ãƒ ã‚’æ›¸ããŸã‚ã® emacs ã®è¨å®šã§ã™ cuda-mode ã¾ãšcuda-modeã‚’ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ã—ã¾ã™ è¨å®šã¯ã“ã‚Œã ã‘ (use-package cuda-mode :mode ("\\.cu\\'" "\\.cuh\\'") ) ã“ã‚Œã§CUDAé–¢é€£ã®äºˆç´„èªžãªã©ãŒhighlightè¡¨ç¤ºã•ã‚Œã¾ã™ lsp é–¢é€£ clangd çµŒç”±ã§CUDAã®ã‚³ãƒ¼ãƒ‰ã§ã‚‚lspãŒä½¿ãˆã¾ã™ è¨å®šã‚‚ç°¡å˜ã§ã™ è¨å®šãƒ•ã‚¡ã‚¤ãƒ« .clangd ã‚’ç”¨æ„ C/C++ã¨åŒæ§˜ã« lspã‚’ä½¿ã† ã“ã‚Œã ã‘ã§ã™ .clangdã¯ã‚½ãƒ¼ã‚¹ã‚³ãƒ¼ãƒ‰ã¨åŒã˜ãƒ‡ã‚£ãƒ¬ã‚¯ãƒˆãƒªã«é…ç½®ã—ã¾ã™ .clangd ã®å†…å®¹ã¯ä»¥ä¸‹ã®ã¨ãŠã‚Š CompileFlags: Add: - --cuda-path=/usr/lib/cuda - -L/usr/lib/cuda/lib64 - -I/usr/lib/cuda/include ã“ã‚Œã§ clangd
masterq 2024/12/01
cuda

emacs

lsp

gpu

gpgpu
ãƒªãƒ³ã‚¯
GitHub - XiongjieDai/GPU-Benchmarks-on-LLM-Inference: Multiple NVIDIA GPUs or Apple Silicon for Large Language Model Inference?
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert
masterq 2024/11/22
model

nvidia

apple

llm

gpu

gpgpu

benchmark

ai
ãƒªãƒ³ã‚¯
GitHub - tmcdonell/cuda: Haskell FFI bindings to CUDA
masterq 2024/11/22
cuda

gpu

gpgpu

nvidia

ffi

haskell
ãƒªãƒ³ã‚¯
nvidia/cuda - Docker Image
masterq 2024/11/12
nvidia

cuda

gpu

gpgpu

linux

docker
ãƒªãƒ³ã‚¯
CUDA-X
NVIDIA CUDA-X Libraries NVIDIA CUDA-Xâ„¢ Libraries, built on CUDAÂ®, is a collection of libraries that deliver dramatically higher performanceâ€”compared to CPU-only alternativesâ€”across application domains, including AI and high-performance computing. NVIDIA libraries run everywhere from resource-constrained IoT devices to self-driving cars to the largest supercomputers on the planet. As a result, user
masterq 2024/11/12
åˆ¥ã€…ã®ãƒ—ãƒã‚¸ã‚§ã‚¯ãƒˆã‚’ã²ã¨ã¾ã¨ã‚ã«ã—ã¦ã¿ãŸã‚ˆæ„Ÿã¯ã‚ã‚Šã¾ã™ã€‚ã–ã£ã¨è¦‹ã‚‹ã‹ãŽã‚Škernelã‚’è‡ªåˆ†ã§æ›¸ã‹ãªãã¦ã‚‚æ¸ˆã‚€ãƒ©ã‚¤ãƒ–ãƒ©ãƒªé›†

nvidia

cuda

lib

math

gpu

gpgpu
ãƒªãƒ³ã‚¯
GitHub - lukstafi/ocaml-cudajit: Bindings to the `cuda` and `nvrtc` libraries with a unified interface
masterq 2024/09/30
ocaml

cuda

gpu

ã‚ã¨ã§èªã‚€
ãƒªãƒ³ã‚¯
GitHub - mathiasbourgoin/SPOC: Stream Processing with OCaml
SPOC is a set of tools for GPGPU programming with OCaml. The SPOC library enables the detection and use of GPGPU devices with OCaml using Cuda and OpenCL. There is also a camlp4 syntax extension to handle external Cuda or OpenCL kernels, as well as a DSL to express GPGPU kernels from the OCaml code. This work was part of my PhD thesis (UPMC-LIP6 laboratory, Paris, France) and was partially funded
masterq 2024/09/30
gpgpu

gpu

ocaml

ã‚ã¨ã§èªã‚€
ãƒªãƒ³ã‚¯
SCALE GPGPU Programming Language
SCALE is a GPGPU programming toolkit that allows CUDA applications to be natively compiled for AMD GPUs.
masterq 2024/07/16
"CUDAã®ãƒ—ãƒã‚°ãƒ©ãƒ ã‚’å…¨ãå¤‰æ›´ã™ã‚‹ã“ã¨ãªãAMDè£½GPUã§å®Ÿè¡Œã§ãã‚‹ã‚ˆã†ã«ã‚³ãƒ³ãƒ‘ã‚¤ãƒ«ã™ã‚‹ãƒ„ãƒ¼ãƒ«ã‚ãƒƒãƒˆ"

cuda

gpu

gpgpu

linux
ãƒªãƒ³ã‚¯
AMD ROCmâ„¢ Software
masterq 2024/07/07
amd

gpu

ã‚ã¨ã§èªã‚€

gpgpu
ãƒªãƒ³ã‚¯
GitHub - AccelerateHS/accelerate-llvm: LLVM backend for Accelerate
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert
masterq 2024/07/06
haskell

nvidia

gpu

gpgpu

llvm
ãƒªãƒ³ã‚¯
CUDA 10.2 or CUDA 11.0 for RTX3060
masterq 2024/07/05
"RTX 3060 uses the Ampere architecture, which requires CUDA 11.x."

cuda

gpu

gpgpu

nvidia
ãƒªãƒ³ã‚¯
RTX3060x2ã®GPUã§æ¿€å®‰ãƒãƒ¼ã‚«ãƒ«LLMãƒžã‚·ãƒ³ã‚’æ§‹ç¯‰ã€‚ | blueqat
ã“ã‚“ã«ã¡ã¯ã€å®‰ãLLMã‚’æ§‹ç¯‰ã—ãŸã„ã§ã™ãã€‚ãŠã™ã™ã‚æ§‹æˆã‚’ãŠçŸ¥ã‚‰ã›ã—ã¾ã™ã€‚ LLMã¯å¤§è¦æ¨¡è¨€èªžãƒ¢ãƒ‡ãƒ«ã®ã“ã¨ã§ã€AIãŒChatGPTã¿ãŸã„ã«æ–‡ç« ã‚’ä½œã£ã¦ãã‚Œã¾ã™ã€‚ç„¡æ–™ã§ä½¿ãˆã¾ã™ãŒã€æ¥å‹™ã§ä½¿ãŠã†ã¨ã™ã‚‹ã¨çµæ§‹é›£ç‚¹ãŒã‚ã‚‹ã—ã€ãƒ‡ãƒ¼ã‚¿ã®æ¼æ´©ã¨ã‹å«ã‚ã¦å›°ã‚Šã¾ã™ãã€‚æ‰‹å…ƒã®ãƒãƒ¼ã‚«ãƒ«ã§LLMã‚’ä½œã‚‹éš›ã®ãŠã™ã™ã‚æ§‹æˆã‚’ãŠçŸ¥ã‚‰ã›ã—ã¾ã™ã€‚ ã¾ãšã€LLMã‚’ä½œã‚‹ã«ã¯ã€ãƒžã‚¶ãƒ¼ãƒœãƒ¼ãƒ‰ã€CPUã€ãƒ¡ãƒ¢ãƒªã€SSDã€é›»æºã€ã‚±ãƒ¼ã‚¹ã€ãã—ã¦GPUãŒå¿…è¦ã§ã™ã€‚ã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢ã‚„OSã¯ã™ã¹ã¦ç„¡æ–™ã§æ‰‹ã«å…¥ã‚Šã¾ã™ã®ã§ã€å¿ƒé…å…¥ã‚Šã¾ã›ã‚“ã€‚ ã¾ãšã€ã‚±ãƒ¼ã‚¹ã«ãƒžã‚¶ãƒ¼ãƒœãƒ¼ãƒ‰ã€CPUã€ãƒ¡ãƒ¢ãƒªã€SSDã€é›»æºã‚’ã‚»ãƒƒãƒˆã—ã¾ã™ã€‚ã“ã®è¾ºã‚Šã¯æ¨™æº–çš„ãªæ§‹æˆã‚’ãŠå‹§ã‚ã—ã¾ã™ã€‚ è‚å¿ƒãªã®ã¯GPUã§ã€VRAMã®ã‚µã‚¤ã‚ºãŒæè¼‰ã§ãã‚‹LLMã®æ€§èƒ½ã‚’å·¦å³ã—ã¾ã™ã€‚ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿æ•°7Bã®ãƒ¢ãƒ‡ãƒ«ãŒæ¨™æº–çš„ã«ä½¿ã‚ã‚Œã¦ã„ã¾ã™ã®ã§ã€ 14GBç¨‹åº¦ã®VRAMãŒå¿…è¦ã§ã™ã€‚ ãã“ã§ã€ã‚ˆãä½¿ã‚ã‚Œã‚‹ã®ãŒãŠãã‚‰ãå®‰ä¾¡ã§VR
masterq 2024/07/05
"ã‚ˆãä½¿ã‚ã‚Œã‚‹ã®ãŒãŠãã‚‰ãå®‰ä¾¡ã§VRAMãŒè±Šå¯ŒãªNVIDIA RTX3060 12Gã§ã™ã€‚æ—¥æœ¬å††ã§40000-50000å††ç¨‹åº¦ã§12Gä½¿ãˆã¾ã™ã®ã§ã€2æžšè²·ã†ã¨8-10ä¸‡å††ç¨‹åº¦ã§åˆè¨ˆ24GBç¢ºä¿ã§ãã¾ã™ã€‚LLMã§ã¯ã€VRAMã®åˆè¨ˆå€¤ã§ã†ã¾ãè¨ˆç®—ã—ã¦ãã‚Œã¾ã™"

gpu

gpgpu

llm

nvidia

hardware

ai
ãƒªãƒ³ã‚¯
1 2 3 4 5 6 æ¬¡ã®ãƒšãƒ¼ã‚¸

ãŠçŸ¥ã‚‰ã›

ã‚‚ã£ã¨èªã‚€

å…¬å¼Twitter

@HatenaBookmark
ãƒªãƒªãƒ¼ã‚¹ã€éšœå®³æƒ…å ±ãªã©ã®ã‚µãƒ¼ãƒ“ã‚¹ã®ãŠçŸ¥ã‚‰ã›
@hatebu
æœ€æ–°ã®äººæ°—ã‚¨ãƒ³ãƒˆãƒªãƒ¼ã®é…ä¿¡

ã‚ãƒ¼ãƒœãƒ¼ãƒ‰ã‚·ãƒ§ãƒ¼ãƒˆã‚«ãƒƒãƒˆä¸€è¦§

jæ¬¡ã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

kå‰ã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

lã‚ã¨ã§èªã‚€

eã‚³ãƒ¡ãƒ³ãƒˆä¸€è¦§ã‚’é–‹ã

oãƒšãƒ¼ã‚¸ã‚’é–‹ã

è¨å®šã‚’å¤‰æ›´ã—ã¾ã—ãŸx