Files
calculet-npu-research-archive/history/llama.cpp/topics/calculet-authored/01-calrt-backend.tsv
T

11 KiB

1ebce9e1a548c2329aa97f53b8b2ad50d968088dce850ff4de25ad1221abc9bf3bb30df252ccd9c362026-05-27T09:38:00+08:00Yunzhe Jiaorigin/runtime_replaceRefactor cal-llm integration by removing runtime capacity loading and adjusting request handling
2e850ff4de25ad1221abc9bf3bb30df252ccd9c369985688f7fa96f755882f96a2cd8a18114fdc42d2026-05-26T10:59:45+08:00Yunzhe JiaEnhance cal-llm integration by adding runtime checks for CALRT_LIBRARY and updating link settings for Linux and Windows
39985688f7fa96f755882f96a2cd8a18114fdc42d38ca6895815671a87ceec86b4aa4eb976580ce762026-05-25T09:05:49+08:00Yunzhe JiaRefactor CalrtEngineConfig initialization by removing redundant parameters
438ca6895815671a87ceec86b4aa4eb976580ce76ddd60c928224b52eb4e1e60e01e8a1ecbb80ee452026-05-20T16:37:13+08:00Yunzhe JiaAdd support for cal-llm profile dumping to JSONL file
5ddd60c928224b52eb4e1e60e01e8a1ecbb80ee452716130010c58f1cdd9d94f0e2f8e047bc92dc662026-05-20T09:22:40+08:00Yunzhe JiaAdd cal-llm backend profiling support with command-line option
6c931ef3163940227c9b6291e04216245cce21429a43741244a646d3f8fa3ff01bb9b7d059223d0a52026-05-18T17:34:37+08:00wangbomengcalrt: fix prefill_by_ids.fix create_buf
72716130010c58f1cdd9d94f0e2f8e047bc92dc66b35bda124ccd4ed581dd6088d3ffa4931c2809b62026-05-18T08:40:54+08:00Yunzhe Jiafix n_ctx_slot error by adding runtime capacity loading for cal-llm
8a43741244a646d3f8fa3ff01bb9b7d059223d0a5b2003d6c90ef0fa26ab9e930ba3e22b355eb3d392026-05-15T14:39:53+08:00wangbomengrun llama-passkey successfully, fix llama-bench, copy 3 times input to infer qwen2.5vl
9b35bda124ccd4ed581dd6088d3ffa4931c2809b681a2cbd4cef5d28334ff513d6b69e593d9d1cb492026-05-15T09:19:31+08:00Yunzhe JiaImplement vocab-only model initialization and enhance cal-llm backend error handling
1081a2cbd4cef5d28334ff513d6b69e593d9d1cb493b14ed556c15529fe1f94b649f49db7156dfaf672026-05-14T16:06:37+08:00Yunzhe JiaEnhance CMake configuration for cal-llm integration by updating paths and adding library properties
113b14ed556c15529fe1f94b649f49db7156dfaf67dde8b8228081769c8d8f1e0751d47fa7875f84232026-05-14T14:34:25+08:00Yunzhe Jia- Removed references to CALRT from clip.h, mtmd-helper.cpp, mtmd-helper.h, mtmd.cpp, and mtmd.h. - Updated server.cpp to conditionally include CAL-LLM support. - Introduced option LLAMA_SERVER_USE_CAL_LLM in CMakeLists.txt to enable CAL-LLM integration. - Refactored server context to handle CAL-LLM requests and responses. - Updated image processing logic to remove CALRT-specific implementations. - Ensured compatibility with existing llama functionality while integrating new CAL-LLM features.
12fabcc7dac4b34d83fe430980bb364b81087c7ee91e9787962f20a7e82c71589ac1dd0585454e4bfe2026-04-30T16:17:17+08:00wangbomengcalrt: fix encode dump
13f584311c716dd078c4b737eebbfda97fe12b7274a339954cc2a145a80c5ea349e7896211916cbed92026-04-30T16:05:15+08:00wangbomengcalrt: fix encode dump
141bec0db5a675ddc60fb793be2a746e8f3c8855dc465df20c3e1f3f58d0f5531c796e0941a49c32b02026-04-30T16:02:59+08:00wangbomengorigin/dev-benbench: support 2 calbins
15465df20c3e1f3f58d0f5531c796e0941a49c32b0a339954cc2a145a80c5ea349e7896211916cbed92026-04-30T09:08:12+08:00wangbomengsupport one calbin
16431ad6f1c787f1b47e9e7be14b2e1a7d8565485cefa1876bb8b2a449a55054a8c3745892f2c0f2622026-04-27T16:52:15+08:00wangbomengupdate calrt_infer
174a2a3181f1cea170105c771e6ba69d851b82b937e3ea5541259f35871c3e995331770aaca297e9ec2026-04-26T10:56:02+08:00wangbomengadd calrt-version
18e3ea5541259f35871c3e995331770aaca297e9ec0b8e43943846a1cd05c26c9186b9ef778ebae6e22026-04-26T10:55:48+08:00wangbomengadd calrt-version
195eb46fffbd2e260c540fb112314ad67bbb8f56b6005112c51e00d998b457337ca3dad53fc46e2d0b2026-04-24T16:24:51+08:00wangbomengclip: add image_to_patches; calrt: vision copy data from bf16 to fp32
20005112c51e00d998b457337ca3dad53fc46e2d0b509670642e8090a1713f5d73590b549be827aaef2026-04-24T16:24:31+08:00wangbomengclip: add image_to_patches; calrt: vision copy data from bf16 to fp32
215d219bfcef239c1e3ddd1baecc0514a9462ff6a86787a53ab9b4897c5a7b51a4062d7b90e6697d352026-04-24T09:31:32+08:00wangbomengcalrt: recover base = batch_index * dict_len
226787a53ab9b4897c5a7b51a4062d7b90e6697d35e1a76abf66178ea85656479a5e19aea7785c09f42026-04-24T09:22:02+08:00wangbomengcalrt:input of multimodel from fp32 to bf16
23e1a76abf66178ea85656479a5e19aea7785c09f499dd308adb8488fa869bb669e3335e646a938eff2026-04-22T16:29:22+08:00wangbomengcalrt:add CalcoreRT version print
2499962021f2dd1994dbbf90d5712f0ca2633f20f89af6682ef101c6d006c743c703cf150ac25f466f2026-04-17T14:48:33+08:00wangbomengorigin/dev-otcalrt: add onetoken slice and untile_prefill
25e24f775ee26ea5249d5ff6cd23b8a0a17fe1438da206460e610aef995e36e50b33e8bd41fe4775412026-04-15T10:01:09+08:00wangbomengcalrt: add copy_data_to_ibuf for vision model
26a206460e610aef995e36e50b33e8bd41fe477541d8b2aaa00049978cebac15041f960d75b552f97b2026-04-14T10:23:43+08:00wangbomengcalrt: support mixture of prefill_by_ids and prefill_by_embeds
270cd44929b442860dc0e5f88976108be82350f9dcb8153b6b8f244b223c9f4c2940ae1f212634519e2026-04-09T15:08:50+08:00Bomeng Wangrm calculet0.txt
289626239f5a9e568c9975d67dd6745fef90279a87da724f3e2a8af2c2537e3edcff41e9415ac66fef2026-04-08T17:47:04+08:00wangbomengcalrt: fix MapBuf
29da724f3e2a8af2c2537e3edcff41e9415ac66fef18814f9bf1a423ebe8f32d6043c9cbc5b21b02182026-04-08T16:58:32+08:00Bomeng Wangcalrt: add MapBuf
30bbee06d6d851e3f84f75b578047967d7e0e4a8dce126c21ed7b793b648d63533ea7ee30885f5a82a2026-04-08T11:29:36+08:00wangbomengcalrt: add test time print
315603e050af07173589f3b8850121263d86da01fd1e8a16cd0e3ad3c9a0f8746852a0a38c6e52322e2026-04-01T05:57:55+08:00wangbomengcalrt: add t_incopy and t_outcopy
321e8a16cd0e3ad3c9a0f8746852a0a38c6e52322e893e52bda0fee99bbda1521ea733a760bc4201f12026-04-01T10:14:22+08:00wangbomengcalrt: fix slice and add infer/copy time
336d5e83902bdd3142a8ff0d3db19f02a8619e95a0e4eaab700333490e80ebefc8d023f6c1adfcc3122026-03-27T00:57:11+08:00wangbomengserver: printf calrt commit ID
34e4eaab700333490e80ebefc8d023f6c1adfcc312aeacc23a883400db0ebffb50e23355ada054da662026-03-25T09:03:55+08:00wangbomengcalrt: prefil to prefill
35aeacc23a883400db0ebffb50e23355ada054da66cbaa7492663630132adeddccd9fec5e44ffa33362026-03-25T08:49:22+08:00wangbomengcalrt: add slice only on decode
363c08ebf0e442cd730ddffd959ec676f9caf31c82613bc2b9a9d7af92efbe42d876ffcdb244f7a00e2026-03-24T02:43:03+08:00wangbomengcalrt: comment LOG_ERROR in untile_decode_cpy
37613bc2b9a9d7af92efbe42d876ffcdb244f7a00e4e194ecc6c4b8ebe4c72d93f6055e3605dbfcefa2026-03-24T02:30:12+08:00wangbomengcalrt: fix max_seq_len judgment
384e194ecc6c4b8ebe4c72d93f6055e3605dbfcefa16cbf81a731f5839a2f6b0eb9d8539826353748b2026-03-24T01:59:03+08:00wangbomengcalrt: fix ubatch.embd cpy
3916cbf81a731f5839a2f6b0eb9d8539826353748b8b4e17d990c23025148c4abadefe1010a0dd07342026-03-23T09:38:21+08:00wangbomengcalrt: fix model_name of untile_cpy
406dacdd2cf9574214ca272a86674e7bd7ba21c4f03e6ad64467771224f072e5bad90e270951ba4cdf2026-03-23T14:42:28+08:00wangbomengcalrt: add multimodal
413e6ad64467771224f072e5bad90e270951ba4cdfb37b7d9756feb3dac2db526eeab38b572a095d472026-03-23T14:36:41+08:00wangbomengcalrt: add multimodal
42d0a1b2c758440720d42fa108b3444f54593daa73b37b7d9756feb3dac2db526eeab38b572a095d472026-03-12T16:10:28+08:00Bomeng Wangcalrt: comment out dump
437ddafbdcb9426ae3af016925b8b596f11e8742d0059009eeaf20a569abfbac5eb37dad3bb93764282026-03-10T07:13:04+08:00wangbomengcalrt: commented out cal_ctx->encode(batch)
44059009eeaf20a569abfbac5eb37dad3bb93764281b073661d1311b5300173993b9f31ee7241d86062026-03-10T07:04:22+08:00wangbomengcalrt: add multimodel
456295273c5866b9249bacdedbfc3a31877e6e6a85c6ea3f0d2468e3b8eab5de80bacee7fe4c9dd3d72026-03-10T06:54:41+08:00wangbomengarg: update hf_repo url to https://download.calculet.tech:9443
46f8fba66b657c51a1df1efc7267bfea9f6140cebfdb4a61d2234c2f457b42ecc3f7219a1e1a35508a2026-03-06T15:06:36+08:00Yunzhe Jiafix calrt_install include path
47e672e02ca1a582f9e3a9d011d9e1b4fea4942a2527b82f64252779ad160a3fa76bd378c39421de36 7a3a9a0e76bb5f530beafcfe52e87e388e93117a2026-03-06T11:47:09+08:00Yunzhe JiaMerge branch 'dev' of http://192.168.10.26:1000/yunzhe/llama.cpp into dev
4827b82f64252779ad160a3fa76bd378c39421de360bd95719fed2f0247c8d0199dd897d24157dfae62026-03-06T11:42:45+08:00Yunzhe Jiaadapt to calrt_objs target
497a3a9a0e76bb5f530beafcfe52e87e388e93117a583c387efaf7bc030574e554088de79d09a27fbd2026-03-05T07:24:59+08:00wangbomengcalrt_infer: fix past_kv_len
50aa0a17d719326a63e0d6fea60aa0ced44e7abc33fa332e773da681f3e77d6ffe83a87edb557f758b2026-02-27T01:40:11+08:00wangbomengcalrt: add bf16_to_fp32
5149fea70a20f550bfdb59d0ffe041b54919c2244774a118df31b6a4acc8876b2a2db90d2395cc6ec32026-01-28T09:11:16+08:00Yunzhe Jiaadapt to new calbin test_case
5274a118df31b6a4acc8876b2a2db90d2395cc6ec398e8f9acfe87cd93646482cb2d942b8a132f08522026-01-13T14:32:47+08:00Yunzhe Jiasync with calrt update
538c8152a04036bb0d4756a4e08e54f514dae4e64d5d2387931528fbe04aa8d66de935147efb65ca4d2025-12-09T14:40:04+08:00Yunzhe Jiaadapt to calrt csr
54e6285a748657add8d974b78ca920c5b41753ee126b1ecdbbe3cc72ed8888be38f9e934cfd79b9fbb2025-11-27T16:58:35+08:00Yunzhe Jiaadapt to calrt
55240d9bb75241c91896afe125f2fc74698a7395f37eeb241aad429a68fb8cfa3ba85d000fb2348cd22025-11-24T12:03:00+08:00Yunzhe Jiaadapt to calrt modification
5698c8269abefdfa8fde47dfb2769ceebc8b704e862790ecc304e384d9075c55d0628d8bf440025dc42025-09-30T17:42:57+08:00Yunzhe Jiaserver: adapt to calrt
572790ecc304e384d9075c55d0628d8bf440025dc4f528e97d894b4d93934a0711a00253eea980e7992025-09-30T17:37:58+08:00Yunzhe Jiaadapt to calrt
58f528e97d894b4d93934a0711a00253eea980e799dcbe998a6d8f13b6c9cb36bbe7b355ced2ac38422025-09-11T14:13:35+08:00Yunzhe Jiaadd calrt mtmd support
59dcbe998a6d8f13b6c9cb36bbe7b355ced2ac384287932ec016f0ec81e98a13ce5212851f8c12458a2025-09-05T14:34:07+08:00Yunzhe Jiaadd calrt API used in server side
603a617cb18121a85eed17ac5585788810c5d9c1e0c42293848dd9f03eba9a6c1b85a600b398f065412025-08-13T10:58:59+08:00Yunzhe Jiaadd calrt_decode
61c42293848dd9f03eba9a6c1b85a600b398f065411cbf442f50a154aac0bf35d7b8b3436a2cdfd6f52025-08-06T15:46:06+08:00Yunzhe Jiasync with calrt API
621cbf442f50a154aac0bf35d7b8b3436a2cdfd6f5ba67d6c03fed3193987a4ef13050e6e2a4451df42025-07-24T15:50:19+08:00Yunzhe Jia1.sync with calrt prefill,decode,kv_update op 2.add hw pattern untile function
63ba67d6c03fed3193987a4ef13050e6e2a4451df48f52c6f8abb2ed9ad967cc0bfb90acb0730111932025-06-26T17:42:32+08:00Yunzhe Jiallama-calrt: infer-op: copy data from output buffer to dst-tensor.data
648f52c6f8abb2ed9ad967cc0bfb90acb073011193d2f325130c4ed77a3dcff2c23a926587fb08e2cc2025-06-26T15:10:45+08:00Yunzhe JiaSquashed commit of the following:
65d2f325130c4ed77a3dcff2c23a926587fb08e2cc6391ee7ffd97e723dfbdcc32f001e1b95a63cba62025-06-17T11:53:33+08:00Yunzhe Jiacalrt: add input_token in graph; input_token will be copy to calrt_in_buffer
66b9850050a5b5147cb2e6720b98ed8c1e8e02eff9933b205c32551848589b8339437d778cb4f2a46d2025-06-06T17:25:52+08:00Yunzhe Jiacalrt: add call chain for using calrt::infer
6733fea9a8592fd338abe279b89350a3af2e0ff1f7806f05a353d322e97d97e02f910de2b408a8b4ab2025-06-06T17:25:52+08:00Yunzhe Jiacalrt: add load calbin
68806f05a353d322e97d97e02f910de2b408a8b4abec9e0301fef6476df83e94842c3b625501c955662025-05-30T09:45:37+08:00Yunzhe Jiaadd calrt demo