Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
43 commits
Select commit Hold shift + click to select a range
3d52219
feat(tensor): checkpoint llama proving campaign
hero78119 Aug 27, 2026
9d24605
feat(tensor): add compact Llama precompile path
hero78119 Aug 28, 2026
dcfca5e
perf(setup): profile symbolic registry construction
hero78119 Aug 28, 2026
08cd4d4
perf(setup): fingerprint fixed trace setup
hero78119 Aug 28, 2026
a65eec3
feat: add proof-bound tensor bus resident segments
hero78119 Aug 28, 2026
8b9fd7b
feat: validate k11008 tensor syscall with cuda provider
hero78119 Aug 28, 2026
98cfeef
verify K11008 CUDA syscall provider
hero78119 Aug 28, 2026
4f3bbd4
feat(tensor): prove default-width resident segment
hero78119 Aug 28, 2026
83fa561
feat(tensor): prove four resident default-width segments
hero78119 Aug 28, 2026
7780712
feat(tensor): prove outer-only resident block
hero78119 Aug 29, 2026
2119b8b
feat(tensor): sweep production resident guest blocks
hero78119 Aug 29, 2026
1f500a5
bench(tensor): add synthetic 7b shape raw-hint probe
hero78119 Aug 29, 2026
6cbc1b7
feat(tensor): verify 32-layer resident topology and atomic AOT split
hero78119 Aug 29, 2026
ffdfbba
perf(tensor): report resident pipeline phase metrics
hero78119 Aug 29, 2026
fb5ef17
docs(tensor): define zkllm-compatible 7b workload
hero78119 Aug 30, 2026
43731b4
docs(tensor): bind zkllm workload to shard proofs
hero78119 Aug 30, 2026
46c81af
feat(tensor): prove batched matmul sections
hero78119 Aug 30, 2026
eb7c909
feat(tensor): fuse matmul claims into batched main
hero78119 Aug 30, 2026
ee5b5c0
feat(recursion): replay fused matmul claims
hero78119 Aug 30, 2026
5676948
docs(tensor): record fused matmul architecture
hero78119 Aug 30, 2026
9350d3f
feat(tensor): prove complete tiny llama layer
hero78119 Aug 30, 2026
84a65e1
refactor(tensor): use row-oriented llama stages
hero78119 Aug 30, 2026
ba75891
docs(tensor): define multi-layer segment topology
hero78119 Aug 31, 2026
c07e260
prove chained llama tiny layers on GPU
hero78119 Aug 31, 2026
7152fed
prove four chained llama tiny layers on GPU
hero78119 Aug 31, 2026
197c688
feat: checkpoint exact production Llama layer
hero78119 Aug 31, 2026
a54a76d
feat(tensor): add two-segment tiny GPU gate
hero78119 Aug 31, 2026
e5bba7d
feat(tensor-vm): fuse head projection with attention
hero78119 Sep 1, 2026
40f80de
perf(tensor-vm): reduce PV dynamic range tower
hero78119 Sep 1, 2026
bd02f12
perf(tensor-vm): reduce PV peak proving memory
hero78119 Sep 1, 2026
faef682
perf(tensor-vm): spill PCS digest around main proving
hero78119 Sep 2, 2026
9e54852
docs(campaign): checkpoint PCS digest spill evidence
hero78119 Sep 2, 2026
1cda84f
fix(campaign): gate unsupported head1 matrix verifier
hero78119 Sep 2, 2026
c6a1c18
docs(campaign): record TensorVM assignment blocker
hero78119 Sep 2, 2026
6a13a7a
docs(campaign): start TensorVM record ownership repair
hero78119 Sep 2, 2026
94d7453
docs(campaign): record PV ownership repair result
hero78119 Sep 2, 2026
351abc3
docs(campaign): authorize provider ownership repair
hero78119 Sep 2, 2026
ef20a4c
docs(campaign): record provider ownership architecture blocker
hero78119 Sep 2, 2026
cf71083
skip agent workspace file
hero78119 Sep 2, 2026
2c59407
feat(tensor): prove generic production attention layer
hero78119 Sep 2, 2026
1158130
perf(prover): keep digest resident and profile GPU phases
hero78119 Sep 3, 2026
6fced8b
fix(sumcheck): frontload matrix correction openings
hero78119 Sep 3, 2026
b7096fe
feat(zkvm): fuse production qk shift softmax
hero78119 Sep 3, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
126 changes: 112 additions & 14 deletions Cargo.lock

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

26 changes: 13 additions & 13 deletions Cargo.toml
Original file line number Diff line number Diff line change
Expand Up @@ -128,20 +128,20 @@ lto = "thin"
#ceno_crypto_primitives = { path = "../ceno-patch/crypto-primitives", package = "ceno_crypto_primitives" }
#ceno_syscall = { path = "../ceno-patch/syscall", package = "ceno_syscall" }

#[patch."https://github.com/scroll-tech/ceno-gpu-mock.git"]
#ceno_gpu = { path = "../ceno-gpu/cuda_hal", package = "cuda_hal" }
[patch."https://github.com/scroll-tech/ceno-gpu-mock.git"]
ceno_gpu = { path = "../ceno-gpu/cuda_hal", package = "cuda_hal" }
#
#[patch."https://github.com/scroll-tech/gkr-backend.git"]
#ff_ext = { path = "../gkr-backend/crates/ff_ext", package = "ff_ext" }
#mpcs = { path = "../gkr-backend/crates/mpcs", package = "mpcs" }
#multilinear_extensions = { path = "../gkr-backend/crates/multilinear_extensions", package = "multilinear_extensions" }
#p3 = { path = "../gkr-backend/crates/p3", package = "p3" }
#poseidon = { path = "../gkr-backend/crates/poseidon", package = "poseidon" }
#sp1-curves = { path = "../gkr-backend/crates/curves", package = "sp1-curves" }
#sumcheck = { path = "../gkr-backend/crates/sumcheck", package = "sumcheck" }
#transcript = { path = "../gkr-backend/crates/transcript", package = "transcript" }
#whir = { path = "../gkr-backend/crates/whir", package = "whir" }
#witness = { path = "../gkr-backend/crates/witness", package = "witness" }
[patch."https://github.com/scroll-tech/gkr-backend.git"]
ff_ext = { path = "../gkr-backend/crates/ff_ext", package = "ff_ext" }
mpcs = { path = "../gkr-backend/crates/mpcs", package = "mpcs" }
multilinear_extensions = { path = "../gkr-backend/crates/multilinear_extensions", package = "multilinear_extensions" }
p3 = { path = "../gkr-backend/crates/p3", package = "p3" }
poseidon = { path = "../gkr-backend/crates/poseidon", package = "poseidon" }
sp1-curves = { path = "../gkr-backend/crates/curves", package = "sp1-curves" }
sumcheck = { path = "../gkr-backend/crates/sumcheck", package = "sumcheck" }
transcript = { path = "../gkr-backend/crates/transcript", package = "transcript" }
whir = { path = "../gkr-backend/crates/whir", package = "whir" }
witness = { path = "../gkr-backend/crates/witness", package = "witness" }

# [patch."https://github.com/scroll-tech/openvm.git"]
# openvm = { path = "../openvm-scroll-tech/crates/toolchain/openvm", default-features = false }
Expand Down
6 changes: 6 additions & 0 deletions ceno_emul/Cargo.toml
Original file line number Diff line number Diff line change
Expand Up @@ -13,12 +13,14 @@ version.workspace = true
anyhow.workspace = true
ceno_rt = { path = "../ceno_rt" }
ceno_syscall.workspace = true
cudarc = { workspace = true, optional = true, features = ["nvrtc"] }
elf = "0.7"
ff_ext.workspace = true
halo2curves-axiom = { version = "=0.7.2", features = ["asm"] }
itertools.workspace = true
libloading = { version = "0.8", optional = true }
multilinear_extensions.workspace = true
p3.workspace = true
num.workspace = true
num-derive.workspace = true
num-traits.workspace = true
Expand Down Expand Up @@ -46,4 +48,8 @@ typenum.workspace = true
aot-x86_64 = ["dep:libloading", "dep:tempfile"]
default = ["forbid_overflow"]
forbid_overflow = []
llama-tiny = ["ceno_rt/llama-tiny"]
production-heads-1 = ["ceno_rt/production-heads-1"]
production-heads-2 = ["ceno_rt/production-heads-2"]
tensor-cuda = ["dep:cudarc"]
u16limb_circuit = []
Loading
Loading