selfhost.directory on Nostr: 🚀 New release · llama.cpp b9948 ggml : process data in smaller chunks in CUDA ...
🚀 New release · llama.cpp b9948
ggml : process data in smaller chunks in CUDA ggml_top_k() and ggml_argsort() to reduce temporary buffers memory usage (#24776)
• ggml : process data in smaller chunks in CUDA ggml_top_k() implementation to reduce temporary buffers memory usage
• ggml : allocate tmp_dst only only once before the loop
• chore : whitespaces
Co-authored-by: Georgi Gerganov ggerganov@gmail.com
Check and self-host it →
https://selfhost.directory/project/llamacpp#update-3749202 
Published at
2026-07-10 07:55:00 UTCEvent JSON
{
"id": "ae95315319a39f87a5a76f9d77e95565c7abe1720690b8ef47e4bf51612ebd42",
"pubkey": "80e8f63915c8d718c20c5a281cf7847ffef267e926c7b4fdd9e66961748b681b",
"created_at": 1783670100,
"kind": 1,
"tags": [],
"content": "🚀 New release · llama.cpp b9948\n\nggml : process data in smaller chunks in CUDA ggml_top_k() and ggml_argsort() to reduce temporary buffers memory usage (#24776)\n• ggml : process data in smaller chunks in CUDA ggml_top_k() implementation to reduce temporary buffers memory usage\n• ggml : allocate tmp_dst only only once before the loop\n• chore : whitespaces\nCo-authored-by: Georgi Gerganov ggerganov@gmail.com\n\nCheck and self-host it → https://selfhost.directory/project/llamacpp#update-3749202\n\nhttps://selfhost.directory/_og/d/c_SelfhostDirectory,title_llama.cpp,description_~bGxhbWEuY3BwIGlzIGEgaGlnaC1wZXJmb3JtYW5jZSwgc2VsZi1ob3N0YWJsZSBMYXJnZSBMYW5ndWFnZSBNb2RlbCAoTExNKSBpbmZlcmVuY2UgZW5naW5lIHdyaXR0ZW4gaW4gQy9DKysuIEl0IGVuYWJsZXMgdXNlcnMgdG8gcnVuIHN0YXRlLW9mLXRoZS1hcnQgTExNcyBlbuKApg,p_Ii9wcm9qZWN0L2xsYW1hY3BwIg.png",
"sig": "975d0cb58fe8212b5dccd835beb290afd0aebd31a56535dc620cbdffe122aaad12ee3464d661bd903b793272f82a95d51d54c9a8788b8c794359ded2d3caa143"
}
