enne2/qwentts.cpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

logs

Browse Source
This commit is contained in:
Pascal
2026-07-20 20:36:48 +02:00
parent 521503f50b
commit ba4c7f7838
12 changed files with 327 additions and 259 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/clone-CPU-BF16.log
+11 -11
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 6 ms
[Pipeline] Speaker encoder lazy loaded in 3 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 53 ms
[Pipeline] Codec encoder lazy loaded in 43 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.112500 13.760122 -0.329032 2.993861
[Debug] enc-transformer-out: [430, 512] first4: 0.526661 3.208130 -0.164395 0.605249
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.054070 -34.409252 -7.069129 28.624571
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.240086 -1.222942 -0.223565 0.888025
[Debug] talker-logits-prefill: [3072] first4: 1.339535 -7.642556 -6.988994 -10.529062
[Batch] Admit slot=0 T_ctx=225 prefill=391.8 ms build=2.3 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=384.7 ms build=2.5 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.073552 -0.034240 -0.008814 0.042007
@@ -93,14 +93,14 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000529 0.000353 -0.000865 -0.000453
[Perf] PromptBuild 2.3 ms
[Perf] Prefill 391.8 ms (T_ctx prefill)
[Perf] TTFA 1549.7 ms (first frame codes)
[Perf] TalkerDecode 2864.7 ms (64 frames, 44.76 ms/frame)
[Perf] CodePredictor 1446.2 ms (22.60 ms/frame)
[Perf] PromptBuild 2.5 ms
[Perf] Prefill 384.7 ms (T_ctx prefill)
[Perf] TTFA 1506.9 ms (first frame codes)
[Perf] TalkerDecode 2886.2 ms (64 frames, 45.10 ms/frame)
[Perf] CodePredictor 1518.3 ms (23.72 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 1112.5 ms
[Perf] Total 6949.6 ms (64 frames, 67.36 ms/frame AR, audio 5.12 s, RTF 1.357)
[Perf] CodecDecode 968.0 ms
[Perf] Total 6854.8 ms (64 frames, 68.82 ms/frame AR, audio 5.12 s, RTF 1.339)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-CPU-F32.log
+10 -10
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 46 ms
[Pipeline] Codec encoder lazy loaded in 33 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.120292 13.742489 -0.348104 3.009387
[Debug] enc-transformer-out: [430, 512] first4: 0.526076 3.125370 -0.189241 0.630764
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.133089 -34.474190 -7.108908 28.674923
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.242892 -1.225840 -0.224931 0.890014
[Debug] talker-logits-prefill: [3072] first4: 1.330122 -7.768507 -6.935926 -10.338600
[Batch] Admit slot=0 T_ctx=225 prefill=701.2 ms build=4.3 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=638.7 ms build=4.9 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.073561 -0.034233 -0.008814 0.042004
@@ -93,14 +93,14 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000450 0.000327 -0.000890 -0.000492
[Perf] PromptBuild 4.3 ms
[Perf] Prefill 701.2 ms (T_ctx prefill)
[Perf] TTFA 1963.7 ms (first frame codes)
[Perf] TalkerDecode 5589.4 ms (64 frames, 87.33 ms/frame)
[Perf] CodePredictor 3961.5 ms (61.90 ms/frame)
[Perf] PromptBuild 4.9 ms
[Perf] Prefill 638.7 ms (T_ctx prefill)
[Perf] TTFA 1849.1 ms (first frame codes)
[Perf] TalkerDecode 5501.4 ms (64 frames, 85.96 ms/frame)
[Perf] CodePredictor 3904.9 ms (61.01 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 1026.8 ms
[Perf] Total 12474.5 ms (64 frames, 149.23 ms/frame AR, audio 5.12 s, RTF 2.436)
[Perf] CodecDecode 992.6 ms
[Perf] Total 12188.6 ms (64 frames, 146.98 ms/frame AR, audio 5.12 s, RTF 2.381)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-CPU-Q4_K_M.log
+10 -10
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 3 ms
[Pipeline] Speaker encoder lazy loaded in 4 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 4.580647 -33.465454 -6.676281 23.526917
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.155397 -1.185542 -0.210456 0.727513
[Debug] talker-logits-prefill: [3072] first4: -0.397744 -4.641821 -7.453447 -10.679588
[Batch] Admit slot=0 T_ctx=225 prefill=429.9 ms build=2.9 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=414.6 ms build=3.4 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 417.000000
[Debug] next-emb-step0: [2048] first4: 0.035528 0.008631 -0.031989 0.034729
@@ -93,14 +93,14 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 417.000000
[Debug] output-audio: [122880] first4: -0.000160 -0.000284 -0.001051 -0.000894
[Perf] PromptBuild 2.9 ms
[Perf] Prefill 429.9 ms (T_ctx prefill)
[Perf] TTFA 1546.8 ms (first frame codes)
[Perf] TalkerDecode 1013.1 ms (64 frames, 15.83 ms/frame)
[Perf] CodePredictor 404.5 ms (6.32 ms/frame)
[Perf] PromptBuild 3.4 ms
[Perf] Prefill 414.6 ms (T_ctx prefill)
[Perf] TTFA 1527.6 ms (first frame codes)
[Perf] TalkerDecode 950.0 ms (64 frames, 14.84 ms/frame)
[Perf] CodePredictor 362.7 ms (5.67 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 980.9 ms
[Perf] Total 3938.8 ms (64 frames, 22.15 ms/frame AR, audio 5.12 s, RTF 0.769)
[Perf] CodecDecode 972.2 ms
[Perf] Total 3806.5 ms (64 frames, 20.51 ms/frame AR, audio 5.12 s, RTF 0.743)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-CPU-Q8_0.log
+9 -9
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 6.696979 -34.327644 -7.302014 28.627213
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.227865 -1.219680 -0.230862 0.887843
[Debug] talker-logits-prefill: [3072] first4: 1.309391 -7.886597 -6.855894 -10.555799
[Batch] Admit slot=0 T_ctx=225 prefill=545.3 ms build=2.6 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=541.1 ms build=2.7 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.010612 -0.008921 0.074940 0.036301
@@ -93,14 +93,14 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000527 0.000324 -0.000983 -0.000676
[Perf] PromptBuild 2.6 ms
[Perf] Prefill 545.3 ms (T_ctx prefill)
[Perf] TTFA 1680.2 ms (first frame codes)
[Perf] TalkerDecode 1713.7 ms (64 frames, 26.78 ms/frame)
[Perf] CodePredictor 707.1 ms (11.05 ms/frame)
[Perf] PromptBuild 2.7 ms
[Perf] Prefill 541.1 ms (T_ctx prefill)
[Perf] TTFA 1609.8 ms (first frame codes)
[Perf] TalkerDecode 1637.8 ms (64 frames, 25.59 ms/frame)
[Perf] CodePredictor 624.7 ms (9.76 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 1025.4 ms
[Perf] Total 5114.5 ms (64 frames, 37.83 ms/frame AR, audio 5.12 s, RTF 0.999)
[Perf] CodecDecode 1003.3 ms
[Perf] Total 4865.2 ms (64 frames, 35.35 ms/frame AR, audio 5.12 s, RTF 0.950)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-CUDA0-BF16.log
+10 -10
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 18 ms
[Pipeline] Speaker encoder lazy loaded in 4 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 116 ms
[Pipeline] Codec encoder lazy loaded in 44 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.111631 13.770183 -0.302467 2.984188
[Debug] enc-transformer-out: [430, 512] first4: 0.536580 3.214924 -0.152684 0.613806
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.045298 -34.398609 -7.069448 28.624956
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.239789 -1.222574 -0.223577 0.888043
[Debug] talker-logits-prefill: [3072] first4: 0.944606 -7.870981 -6.684478 -10.343767
[Batch] Admit slot=0 T_ctx=225 prefill=22.4 ms build=1.4 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=12.4 ms build=1.4 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.013383 -0.012534 0.088655 0.033408
@@ -95,13 +95,13 @@ ggml_backend_cuda_graph_compute: CUDA graph warmup complete
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000498 0.000316 -0.001016 -0.000733
[Perf] PromptBuild 1.4 ms
[Perf] Prefill 22.4 ms (T_ctx prefill)
[Perf] TTFA 913.7 ms (first frame codes)
[Perf] TalkerDecode 187.0 ms (64 frames, 2.92 ms/frame)
[Perf] CodePredictor 259.9 ms (4.06 ms/frame)
[Perf] Prefill 12.4 ms (T_ctx prefill)
[Perf] TTFA 778.8 ms (first frame codes)
[Perf] TalkerDecode 166.9 ms (64 frames, 2.61 ms/frame)
[Perf] CodePredictor 257.6 ms (4.03 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 14.9 ms
[Perf] Total 1368.8 ms (64 frames, 6.98 ms/frame AR, audio 5.12 s, RTF 0.267)
[Perf] CodecDecode 14.7 ms
[Perf] Total 1211.3 ms (64 frames, 6.64 ms/frame AR, audio 5.12 s, RTF 0.237)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-CUDA0-F32.log
+11 -11
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 45.8 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 17 ms
[Pipeline] Speaker encoder lazy loaded in 4 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 90 ms
[Pipeline] Codec encoder lazy loaded in 32 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.134336 13.731120 -0.351051 3.013485
[Debug] enc-transformer-out: [430, 512] first4: 0.525800 3.116711 -0.187038 0.631811
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.131224 -34.471714 -7.108000 28.674042
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.242833 -1.225771 -0.224906 0.890001
[Debug] talker-logits-prefill: [3072] first4: 1.171040 -7.796190 -6.822366 -10.286754
[Batch] Admit slot=0 T_ctx=225 prefill=45.0 ms build=47.6 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=27.3 ms build=1.4 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.003672 0.000908 0.039506 0.094594
@@ -94,14 +94,14 @@ ggml_backend_cuda_graph_compute: CUDA graph warmup complete
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000739 0.000593 -0.000747 -0.000435
[Perf] PromptBuild 47.6 ms
[Perf] Prefill 45.0 ms (T_ctx prefill)
[Perf] TTFA 1027.4 ms (first frame codes)
[Perf] TalkerDecode 608.3 ms (64 frames, 9.50 ms/frame)
[Perf] CodePredictor 1654.1 ms (25.84 ms/frame)
[Perf] PromptBuild 1.4 ms
[Perf] Prefill 27.3 ms (T_ctx prefill)
[Perf] TTFA 767.0 ms (first frame codes)
[Perf] TalkerDecode 275.7 ms (64 frames, 4.31 ms/frame)
[Perf] CodePredictor 347.0 ms (5.42 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 24.8 ms
[Perf] Total 3261.0 ms (64 frames, 35.35 ms/frame AR, audio 5.12 s, RTF 0.637)
[Perf] CodecDecode 14.8 ms
[Perf] Total 1397.9 ms (64 frames, 9.73 ms/frame AR, audio 5.12 s, RTF 0.273)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-CUDA0-Q4_K_M.log
+11 -11
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 27 ms
[Pipeline] Speaker encoder lazy loaded in 3 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 43 ms
[Pipeline] Codec encoder lazy loaded in 8 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.134336 13.731120 -0.351051 3.013485
[Debug] enc-transformer-out: [430, 512] first4: 1.140990 3.713432 0.131578 0.616251
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 4.146309 -31.912390 -6.918371 22.418171
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.140643 -1.130366 -0.218058 0.693131
[Debug] talker-logits-prefill: [3072] first4: -0.461190 -5.760653 -7.184410 -10.423358
[Batch] Admit slot=0 T_ctx=225 prefill=24.0 ms build=6.9 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=15.2 ms build=1.5 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1042.000000 390.000000 881.000000
[Debug] next-emb-step0: [2048] first4: -0.033968 -0.020822 -0.037756 -0.013151
@@ -94,14 +94,14 @@ ggml_backend_cuda_graph_compute: CUDA graph warmup complete
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1042.000000 390.000000 881.000000
[Debug] output-audio: [122880] first4: 0.000087 0.000212 -0.000383 -0.000346
[Perf] PromptBuild 6.9 ms
[Perf] Prefill 24.0 ms (T_ctx prefill)
[Perf] TTFA 844.4 ms (first frame codes)
[Perf] TalkerDecode 191.5 ms (64 frames, 2.99 ms/frame)
[Perf] CodePredictor 735.8 ms (11.50 ms/frame)
[Perf] PromptBuild 1.5 ms
[Perf] Prefill 15.2 ms (T_ctx prefill)
[Perf] TTFA 738.7 ms (first frame codes)
[Perf] TalkerDecode 101.3 ms (64 frames, 1.58 ms/frame)
[Perf] CodePredictor 193.3 ms (3.02 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 19.7 ms
[Perf] Total 1766.6 ms (64 frames, 14.49 ms/frame AR, audio 5.12 s, RTF 0.345)
[Perf] CodecDecode 16.2 ms
[Perf] Total 1044.4 ms (64 frames, 4.61 ms/frame AR, audio 5.12 s, RTF 0.204)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-CUDA0-Q8_0.log
+11 -11
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 26 ms
[Pipeline] Speaker encoder lazy loaded in 4 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 46 ms
[Pipeline] Codec encoder lazy loaded in 9 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.134336 13.731120 -0.351051 3.013485
[Debug] enc-transformer-out: [430, 512] first4: 0.628775 2.909739 -0.209228 0.726215
@@ -76,7 +76,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 6.677233 -34.349281 -7.298450 28.605591
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.227255 -1.220783 -0.230812 0.887415
[Debug] talker-logits-prefill: [3072] first4: 1.706040 -8.254342 -7.318320 -10.816376
[Batch] Admit slot=0 T_ctx=225 prefill=15.5 ms build=1.6 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=15.1 ms build=1.5 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: -0.034936 -0.024575 0.029866 0.045185
@@ -94,14 +94,14 @@ ggml_backend_cuda_graph_compute: CUDA graph warmup complete
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000596 0.000259 -0.001224 -0.000817
[Perf] PromptBuild 1.6 ms
[Perf] Prefill 15.5 ms (T_ctx prefill)
[Perf] TTFA 784.9 ms (first frame codes)
[Perf] TalkerDecode 126.9 ms (64 frames, 1.98 ms/frame)
[Perf] CodePredictor 196.0 ms (3.06 ms/frame)
[Perf] PromptBuild 1.5 ms
[Perf] Prefill 15.1 ms (T_ctx prefill)
[Perf] TTFA 752.7 ms (first frame codes)
[Perf] TalkerDecode 136.9 ms (64 frames, 2.14 ms/frame)
[Perf] CodePredictor 194.9 ms (3.04 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 15.2 ms
[Perf] Total 1117.9 ms (64 frames, 5.05 ms/frame AR, audio 5.12 s, RTF 0.218)
[Perf] CodecDecode 15.1 ms
[Perf] Total 1094.5 ms (64 frames, 5.19 ms/frame AR, audio 5.12 s, RTF 0.214)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
tests/clone-Vulkan0-BF16.log
+61 -44
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 46 ms
[Pipeline] Codec encoder lazy loaded in 45 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.110593 13.765666 -0.317055 2.981106
[Debug] enc-transformer-out: [430, 512] first4: 0.536260 3.195749 -0.156796 0.610111
@@ -76,51 +76,33 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.051195 -34.399265 -7.070650 28.631254
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.239991 -1.222603 -0.223616 0.888243
[Debug] talker-logits-prefill: [3072] first4: 0.882721 -7.881639 -6.796915 -10.414125
[Batch] Admit slot=0 T_ctx=225 prefill=32.2 ms build=2.1 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=31.9 ms build=2.0 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: -0.008599 -0.015697 0.093591 0.086640
/mnt/workspace/git/qwentts.cpp/ggml/src/ggml-vulkan/ggml-vulkan.cpp:11497: GGML_ASSERT(dst->op != GGML_OP_GET_ROWS || (a_offset == 0 && b_offset == 0 && d_offset == 0)) failed
[New LWP 121432]
[New LWP 121431]
[New LWP 121430]
[New LWP 121429]
[New LWP 121428]
[New LWP 121427]
[New LWP 121426]
[New LWP 121425]
[New LWP 121424]
[New LWP 121423]
[New LWP 121422]
[New LWP 121421]
[New LWP 121420]
[New LWP 121419]
[New LWP 121418]
[New LWP 121417]
[New LWP 121416]
[New LWP 121415]
[New LWP 121414]
[New LWP 121413]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
0x00007f5e576799ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#0 0x00007f5e576799ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#1 0x00007f5e5766e668 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#2 0x00007f5e5766e6ad in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#3 0x00007f5e576d97c7 in wait4 () from /lib/x86_64-linux-gnu/libc.so.6
#4 0x00007f5e57b89a93 in ggml_print_backtrace () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#5 0x00007f5e57b89bfe in ggml_abort () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#6 0x00007f5c0e2d4bf1 in void init_pushconst_tensor_offsets<vk_op_binary_push_constants>(ggml_backend_vk_context*, vk_op_binary_push_constants&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#7 0x00007f5c0e449201 in void ggml_vk_op_f32<vk_op_binary_push_constants>(ggml_backend_vk_context*, std::shared_ptr<vk_context_struct>&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*, ggml_op, vk_op_binary_push_constants&&) [clone .constprop.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#8 0x00007f5c0e464f48 in ggml_vk_build_graph(ggml_backend_vk_context*, ggml_cgraph*, int, ggml_tensor*, int, bool, bool, bool) [clone .isra.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#9 0x00007f5c0e4679bc in ggml_backend_vk_graph_compute(ggml_backend*, ggml_cgraph*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#10 0x00007f5e57ba196e in ggml_backend_graph_compute () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#11 0x000055bbb3e94365 in tts_engine_step(TtsEngine*, std::vector<TtsJob*, std::allocator<TtsJob*> >*) ()
#12 0x000055bbb3ea166a in pipeline_tts_synthesize(PipelineTTS*, BPETokenizer*, qt_tts_params const*, long, qt_audio*) ()
#13 0x000055bbb3e7db80 in qt_synthesize ()
#14 0x000055bbb3e74cdd in run(Args const&) ()
#15 0x000055bbb3e72a83 in main ()
[Inferior 1 (process 121412) detached]
[Debug] talker-hidden-step1: [2048] first4: 0.466706 2.304709 -0.154099 0.228768
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Pipeline] Generated 8 frames (slot 0)
[Pipeline] Generated 16 frames (slot 0)
[Pipeline] Generated 24 frames (slot 0)
[Pipeline] Generated 32 frames (slot 0)
[Pipeline] Generated 40 frames (slot 0)
[Pipeline] Generated 48 frames (slot 0)
[Pipeline] Generated 56 frames (slot 0)
[Pipeline] Generated 64 frames (slot 0)
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000515 0.000357 -0.000915 -0.000575
[Perf] PromptBuild 2.0 ms
[Perf] Prefill 31.9 ms (T_ctx prefill)
[Perf] TTFA 812.5 ms (first frame codes)
[Perf] TalkerDecode 212.2 ms (64 frames, 3.32 ms/frame)
[Perf] CodePredictor 374.8 ms (5.86 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 32.1 ms
[Perf] Total 1425.9 ms (64 frames, 9.18 ms/frame AR, audio 5.12 s, RTF 0.278)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
@@ -136,3 +118,38 @@ Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] BF16 -> ../models/qwen-talker-1.7b-base-BF16.gguf + ../models/qwen-tokenizer-12hz-BF16.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-BF16.gguf --codec ../models/qwen-tokenizer-12hz-BF16.gguf --seed 42 --ref-wav ../examples/freeman.wav --ref-text ../examples/freeman.txt --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 76.83% (3440 values)
[Cossim] Embed cos: 0.962709 max: 1.9437e-01 mean: 1.0689e-02
[Cossim] TrailingText cos: 1.000000 max: 2.9802e-08 mean: 6.7932e-10
[Cossim] TTSPadEmbed cos: 1.000000 max: 2.9802e-08 mean: 6.7932e-10
[Cossim] L0 cos: 0.983221 max: 7.7030e-01 mean: 1.3992e-02
[Cossim] L7 cos: 0.999989 max: 1.2884e+01 mean: 3.3235e-02
[Cossim] L14 cos: 0.999964 max: 1.2886e+01 mean: 7.1843e-02
[Cossim] L21 cos: 0.999531 max: 1.6955e+01 mean: 2.8087e-01
[Cossim] L27 cos: 0.998209 max: 4.0700e+01 mean: 6.4207e-01
[Cossim] Final cos: 0.990658 max: 1.5839e+01 mean: 2.0917e-01
[Cossim] Logits cos: 0.999301 max: 1.1222e+00 mean: 1.3438e-01
[Cossim] NextEmbStep0 cos: 0.818255 max: 1.4225e-01 mean: 3.0455e-02
[Cossim] TalkerHiddenStep1 cos: 0.993880 max: 1.7297e+00 mean: 2.3463e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 9.1669e-04 mean: 1.6221e-05
[Cossim] SeanetResnet0 cos: 0.999961 max: 2.1108e-03 mean: 1.6910e-04
[Cossim] SeanetStage0 cos: 0.999983 max: 4.4128e-02 mean: 7.5091e-04
[Cossim] SeanetStage1 cos: 0.999862 max: 7.2962e-01 mean: 5.2996e-03
[Cossim] SeanetStage3 cos: 0.999994 max: 2.1304e+00 mean: 4.7641e-02
[Cossim] SeanetOut cos: 0.999953 max: 3.4988e-01 mean: 2.0273e-02
[Cossim] EncTransformer cos: 0.999963 max: 8.5211e-01 mean: 1.0424e-02
[Cossim] CodecPreFSQ cos: 0.999962 max: 4.6668e-01 mean: 2.2088e-02
[Cossim] SpkFrontend cos: 0.999997 max: 2.5180e-02 mean: 9.4975e-04
[Cossim] SpkBlock3 cos: 0.999991 max: 1.0191e-01 mean: 4.1519e-03
[Cossim] SpkMFA cos: 0.999979 max: 7.0797e-02 mean: 1.0369e-03
[Cossim] SpkASP cos: 1.000000 max: 2.3533e-03 mean: 1.4947e-04
[Cossim] SpeakerEmb cos: 1.000000 max: 6.7258e-04 mean: 1.2809e-04
[Cossim] CodesFull exact: 2.28% (1008 values)
[Cossim] Audio cos: 0.003079
[Cossim] WAV stft_cos: 0.208123 samples: 120960
tests/clone-Vulkan0-F32.log
+62 -45
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 45.8 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 6 ms
[Pipeline] Speaker encoder lazy loaded in 5 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -51,7 +51,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[EncDownsample] Loaded: k=4 stride=2, 512 -> 512 channels, weights 2.0 MB
[WeightCtx] Loaded 20 tensors, 34.0 MB into backend
[EncQuantizer] Loaded: 16 codebooks (1 semantic + 15 acoustic), 2048 entries x 256 dim, hidden 512, weights 34.0 MB
[Pipeline] Codec encoder lazy loaded in 36 ms
[Pipeline] Codec encoder lazy loaded in 35 ms
[Debug] audio-input: [412800] first4: -0.000546 -0.001162 -0.000871 0.000340
[Debug] seanet-out: [430, 512] first4: -0.138303 13.699992 -0.351357 3.025051
[Debug] enc-transformer-out: [430, 512] first4: 0.526569 3.116264 -0.195330 0.625080
@@ -76,51 +76,33 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 7.115197 -34.442551 -7.104288 28.617508
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.242297 -1.224786 -0.224798 0.888284
[Debug] talker-logits-prefill: [3072] first4: 1.013672 -7.734375 -6.857910 -10.250000
[Batch] Admit slot=0 T_ctx=225 prefill=1426.7 ms build=21.4 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=20.0 ms build=2.4 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.034006 -0.013787 0.039109 0.091488
/mnt/workspace/git/qwentts.cpp/ggml/src/ggml-vulkan/ggml-vulkan.cpp:11497: GGML_ASSERT(dst->op != GGML_OP_GET_ROWS || (a_offset == 0 && b_offset == 0 && d_offset == 0)) failed
[New LWP 121280]
[New LWP 121279]
[New LWP 121278]
[New LWP 121277]
[New LWP 121276]
[New LWP 121275]
[New LWP 121274]
[New LWP 121273]
[New LWP 121272]
[New LWP 121271]
[New LWP 121270]
[New LWP 121269]
[New LWP 121268]
[New LWP 121267]
[New LWP 121266]
[New LWP 121265]
[New LWP 121264]
[New LWP 121263]
[New LWP 121262]
[New LWP 121261]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
0x00007f5ef94a69ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#0 0x00007f5ef94a69ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#1 0x00007f5ef949b668 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#2 0x00007f5ef949b6ad in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#3 0x00007f5ef95067c7 in wait4 () from /lib/x86_64-linux-gnu/libc.so.6
#4 0x00007f5ef9a60a93 in ggml_print_backtrace () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#5 0x00007f5ef9a60bfe in ggml_abort () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#6 0x00007f5cb08d4bf1 in void init_pushconst_tensor_offsets<vk_op_binary_push_constants>(ggml_backend_vk_context*, vk_op_binary_push_constants&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#7 0x00007f5cb0a49201 in void ggml_vk_op_f32<vk_op_binary_push_constants>(ggml_backend_vk_context*, std::shared_ptr<vk_context_struct>&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*, ggml_op, vk_op_binary_push_constants&&) [clone .constprop.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#8 0x00007f5cb0a64f48 in ggml_vk_build_graph(ggml_backend_vk_context*, ggml_cgraph*, int, ggml_tensor*, int, bool, bool, bool) [clone .isra.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#9 0x00007f5cb0a679bc in ggml_backend_vk_graph_compute(ggml_backend*, ggml_cgraph*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#10 0x00007f5ef9a7896e in ggml_backend_graph_compute () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#11 0x000055af16f11365 in tts_engine_step(TtsEngine*, std::vector<TtsJob*, std::allocator<TtsJob*> >*) ()
#12 0x000055af16f1e66a in pipeline_tts_synthesize(PipelineTTS*, BPETokenizer*, qt_tts_params const*, long, qt_audio*) ()
#13 0x000055af16efab80 in qt_synthesize ()
#14 0x000055af16ef1cdd in run(Args const&) ()
#15 0x000055af16eefa83 in main ()
[Inferior 1 (process 121260) detached]
[Debug] talker-hidden-step1: [2048] first4: 0.742126 2.656314 -0.143770 -0.182433
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Pipeline] Generated 8 frames (slot 0)
[Pipeline] Generated 16 frames (slot 0)
[Pipeline] Generated 24 frames (slot 0)
[Pipeline] Generated 32 frames (slot 0)
[Pipeline] Generated 40 frames (slot 0)
[Pipeline] Generated 48 frames (slot 0)
[Pipeline] Generated 56 frames (slot 0)
[Pipeline] Generated 64 frames (slot 0)
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000673 0.000478 -0.000828 -0.000482
[Perf] PromptBuild 2.4 ms
[Perf] Prefill 20.0 ms (T_ctx prefill)
[Perf] TTFA 796.6 ms (first frame codes)
[Perf] TalkerDecode 328.0 ms (64 frames, 5.13 ms/frame)
[Perf] CodePredictor 457.1 ms (7.14 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 29.3 ms
[Perf] Total 1604.0 ms (64 frames, 12.27 ms/frame AR, audio 5.12 s, RTF 0.313)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
@@ -136,3 +118,38 @@ Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] F32 -> ../models/qwen-talker-1.7b-base-F32.gguf + ../models/qwen-tokenizer-12hz-F32.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-F32.gguf --codec ../models/qwen-tokenizer-12hz-F32.gguf --seed 42 --ref-wav ../examples/freeman.wav --ref-text ../examples/freeman.txt --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 82.41% (3440 values)
[Cossim] Embed cos: 0.972129 max: 1.9482e-01 mean: 8.1111e-03
[Cossim] TrailingText cos: 1.000000 max: 2.9802e-08 mean: 6.7932e-10
[Cossim] TTSPadEmbed cos: 1.000000 max: 2.9802e-08 mean: 6.7932e-10
[Cossim] L0 cos: 0.987066 max: 6.8534e-01 mean: 1.0819e-02
[Cossim] L7 cos: 0.999990 max: 1.4637e+01 mean: 2.7206e-02
[Cossim] L14 cos: 0.999971 max: 1.4633e+01 mean: 5.9840e-02
[Cossim] L21 cos: 0.999707 max: 1.4642e+01 mean: 2.2449e-01
[Cossim] L27 cos: 0.998906 max: 2.5744e+01 mean: 5.2221e-01
[Cossim] Final cos: 0.994366 max: 7.7967e+00 mean: 1.6928e-01
[Cossim] Logits cos: 0.999568 max: 8.3677e-01 mean: 1.1911e-01
[Cossim] NextEmbStep0 cos: 0.822712 max: 2.7881e-01 mean: 3.0097e-02
[Cossim] TalkerHiddenStep1 cos: 0.990742 max: 1.7658e+00 mean: 2.9620e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 4.1014e-04 mean: 3.4146e-06
[Cossim] SeanetResnet0 cos: 0.999987 max: 1.9526e-03 mean: 9.4296e-05
[Cossim] SeanetStage0 cos: 0.999985 max: 4.2238e-02 mean: 6.5089e-04
[Cossim] SeanetStage1 cos: 0.999866 max: 7.7454e-01 mean: 4.9543e-03
[Cossim] SeanetStage3 cos: 0.999995 max: 1.4657e+00 mean: 3.8722e-02
[Cossim] SeanetOut cos: 0.999963 max: 4.0791e-01 mean: 1.6984e-02
[Cossim] EncTransformer cos: 0.999974 max: 3.3284e-01 mean: 7.6420e-03
[Cossim] CodecPreFSQ cos: 0.999972 max: 4.5137e-01 mean: 1.8013e-02
[Cossim] SpkFrontend cos: 1.000000 max: 2.6600e-03 mean: 1.1964e-04
[Cossim] SpkBlock3 cos: 1.000000 max: 1.2138e-02 mean: 5.2076e-04
[Cossim] SpkMFA cos: 1.000000 max: 8.9911e-03 mean: 1.3077e-04
[Cossim] SpkASP cos: 1.000000 max: 2.2851e-04 mean: 1.7482e-05
[Cossim] SpeakerEmb cos: 1.000000 max: 7.8417e-05 mean: 1.3521e-05
[Cossim] CodesFull exact: 3.17% (1008 values)
[Cossim] Audio cos: 0.023761
[Cossim] WAV stft_cos: 0.225096 samples: 120960
tests/clone-Vulkan0-Q4_K_M.log
+61 -44
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -33,7 +33,7 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Audio-Resample] Done: 380672 -> 414337 samples
[WeightCtx] Loaded 76 tensors, 34.9 MB into backend
[SpeakerEncoder] Loaded: enc_dim=2048 sr=24000 mel_dim=128 hidden=512 mfa=1536 asp_attn=128 se=128 scale=8
[Pipeline] Speaker encoder lazy loaded in 5 ms
[Pipeline] Speaker encoder lazy loaded in 4 ms
[Debug] mel-spk: [1618, 128] first4: -6.330339 -6.170074 -6.011195 -6.243161
[Debug] mel-hann: [1024] first4: 0.000000 0.000009 0.000038 0.000085
[Debug] mel-basis: [128, 513] first4: 0.000000 0.033550 0.008569 0.000000
@@ -76,51 +76,33 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 4.921398 -31.754539 -6.813190 21.707518
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.166135 -1.119393 -0.213715 0.667947
[Debug] talker-logits-prefill: [3072] first4: -0.151123 -5.516602 -7.045898 -10.537109
[Batch] Admit slot=0 T_ctx=225 prefill=1887.5 ms build=1227.7 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=18.7 ms build=2.8 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 417.000000
[Debug] next-emb-step0: [2048] first4: 0.005714 0.034105 -0.000204 0.073642
/mnt/workspace/git/qwentts.cpp/ggml/src/ggml-vulkan/ggml-vulkan.cpp:11497: GGML_ASSERT(dst->op != GGML_OP_GET_ROWS || (a_offset == 0 && b_offset == 0 && d_offset == 0)) failed
[New LWP 121736]
[New LWP 121735]
[New LWP 121734]
[New LWP 121733]
[New LWP 121732]
[New LWP 121731]
[New LWP 121730]
[New LWP 121729]
[New LWP 121728]
[New LWP 121727]
[New LWP 121726]
[New LWP 121725]
[New LWP 121724]
[New LWP 121723]
[New LWP 121722]
[New LWP 121721]
[New LWP 121720]
[New LWP 121719]
[New LWP 121718]
[New LWP 121717]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
0x00007f9b896a69ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#0 0x00007f9b896a69ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#1 0x00007f9b8969b668 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#2 0x00007f9b8969b6ad in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#3 0x00007f9b897067c7 in wait4 () from /lib/x86_64-linux-gnu/libc.so.6
#4 0x00007f9b89c75a93 in ggml_print_backtrace () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#5 0x00007f9b89c75bfe in ggml_abort () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#6 0x00007f99408d4bf1 in void init_pushconst_tensor_offsets<vk_op_binary_push_constants>(ggml_backend_vk_context*, vk_op_binary_push_constants&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#7 0x00007f9940a49201 in void ggml_vk_op_f32<vk_op_binary_push_constants>(ggml_backend_vk_context*, std::shared_ptr<vk_context_struct>&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*, ggml_op, vk_op_binary_push_constants&&) [clone .constprop.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#8 0x00007f9940a64f48 in ggml_vk_build_graph(ggml_backend_vk_context*, ggml_cgraph*, int, ggml_tensor*, int, bool, bool, bool) [clone .isra.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#9 0x00007f9940a679bc in ggml_backend_vk_graph_compute(ggml_backend*, ggml_cgraph*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#10 0x00007f9b89c8d96e in ggml_backend_graph_compute () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#11 0x000055f31eca3365 in tts_engine_step(TtsEngine*, std::vector<TtsJob*, std::allocator<TtsJob*> >*) ()
#12 0x000055f31ecb066a in pipeline_tts_synthesize(PipelineTTS*, BPETokenizer*, qt_tts_params const*, long, qt_audio*) ()
#13 0x000055f31ec8cb80 in qt_synthesize ()
#14 0x000055f31ec83cdd in run(Args const&) ()
#15 0x000055f31ec81a83 in main ()
[Inferior 1 (process 121716) detached]
[Debug] talker-hidden-step1: [2048] first4: 0.590231 2.653425 0.827661 0.831199
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Pipeline] Generated 8 frames (slot 0)
[Pipeline] Generated 16 frames (slot 0)
[Pipeline] Generated 24 frames (slot 0)
[Pipeline] Generated 32 frames (slot 0)
[Pipeline] Generated 40 frames (slot 0)
[Pipeline] Generated 48 frames (slot 0)
[Pipeline] Generated 56 frames (slot 0)
[Pipeline] Generated 64 frames (slot 0)
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 417.000000
[Debug] output-audio: [122880] first4: -0.000101 0.000003 -0.000779 -0.000457
[Perf] PromptBuild 2.8 ms
[Perf] Prefill 18.7 ms (T_ctx prefill)
[Perf] TTFA 748.1 ms (first frame codes)
[Perf] TalkerDecode 140.4 ms (64 frames, 2.19 ms/frame)
[Perf] CodePredictor 286.6 ms (4.48 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 30.2 ms
[Perf] Total 1200.8 ms (64 frames, 6.67 ms/frame AR, audio 5.12 s, RTF 0.235)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
@@ -136,3 +118,38 @@ Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] Q4_K_M -> ../models/qwen-talker-1.7b-base-Q4_K_M.gguf + ../models/qwen-tokenizer-12hz-Q4_K_M.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-Q4_K_M.gguf --codec ../models/qwen-tokenizer-12hz-Q4_K_M.gguf --seed 42 --ref-wav ../examples/freeman.wav --ref-text ../examples/freeman.txt --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 49.83% (3440 values)
[Cossim] Embed cos: 0.917128 max: 2.2245e-01 mean: 2.1853e-02
[Cossim] TrailingText cos: 0.999967 max: 1.5749e-03 mean: 2.3645e-04
[Cossim] TTSPadEmbed cos: 0.999967 max: 1.5749e-03 mean: 2.3645e-04
[Cossim] L0 cos: 0.961486 max: 8.1847e-01 mean: 2.8364e-02
[Cossim] L7 cos: 0.997433 max: 3.9307e+01 mean: 1.6337e-01
[Cossim] L14 cos: 0.997359 max: 3.8675e+01 mean: 2.4737e-01
[Cossim] L21 cos: 0.995877 max: 3.8921e+01 mean: 7.7256e-01
[Cossim] L27 cos: 0.956688 max: 1.7483e+03 mean: 1.9751e+00
[Cossim] Final cos: 0.955767 max: 3.3655e+01 mean: 5.4944e-01
[Cossim] Logits cos: 0.992621 max: 2.9744e+00 mean: 4.5055e-01
[Cossim] NextEmbStep0 cos: 0.819095 max: 1.4988e-01 mean: 3.0364e-02
[Cossim] TalkerHiddenStep1 cos: 0.957568 max: 3.9459e+00 mean: 6.2838e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 4.1014e-04 mean: 3.4146e-06
[Cossim] SeanetResnet0 cos: 0.999987 max: 1.9526e-03 mean: 9.4296e-05
[Cossim] SeanetStage0 cos: 0.999985 max: 4.2238e-02 mean: 6.5089e-04
[Cossim] SeanetStage1 cos: 0.999866 max: 7.7454e-01 mean: 4.9543e-03
[Cossim] SeanetStage3 cos: 0.999995 max: 1.4657e+00 mean: 3.8722e-02
[Cossim] SeanetOut cos: 0.999963 max: 4.0791e-01 mean: 1.6984e-02
[Cossim] EncTransformer cos: 0.992979 max: 3.5659e+00 mean: 1.6375e-01
[Cossim] CodecPreFSQ cos: 0.996263 max: 2.2477e+00 mean: 2.4022e-01
[Cossim] SpkFrontend cos: 1.000000 max: 2.6600e-03 mean: 1.1964e-04
[Cossim] SpkBlock3 cos: 1.000000 max: 1.2137e-02 mean: 5.2085e-04
[Cossim] SpkMFA cos: 1.000000 max: 8.9911e-03 mean: 1.3079e-04
[Cossim] SpkASP cos: 1.000000 max: 2.2919e-04 mean: 1.7381e-05
[Cossim] SpeakerEmb cos: 1.000000 max: 8.0518e-05 mean: 1.3621e-05
[Cossim] CodesFull exact: 2.38% (1008 values)
[Cossim] Audio cos: -0.014421
[Cossim] WAV stft_cos: 0.205816 samples: 120960
tests/clone-Vulkan0-Q8_0.log
+60 -43
View File
@@ -1,4 +1,4 @@
[Qwen] qwentts.cpp 53c84fe (2026-07-20)
[Qwen] qwentts.cpp 521503f (2026-07-20)
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 97247 MiB):
Device 0: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, compute capability 12.0, VMM: yes, VRAM: 97247 MiB
load_backend: loaded CUDA backend from /mnt/workspace/git/qwentts.cpp/build/libggml-cuda.so
@@ -76,51 +76,33 @@ load_backend: loaded CPU backend from /mnt/workspace/git/qwentts.cpp/build/libgg
[Debug] talker-hidden-prefill-l27: [225, 2048] first4: 6.677526 -34.405407 -7.383842 28.511698
[Debug] talker-hidden-prefill-final: [225, 2048] first4: 0.227110 -1.221941 -0.233353 0.883898
[Debug] talker-logits-prefill: [3072] first4: 1.264160 -7.826172 -7.041016 -10.638672
[Batch] Admit slot=0 T_ctx=225 prefill=2227.7 ms build=1539.1 ms (stall for 0 active slots)
[Batch] Admit slot=0 T_ctx=225 prefill=17.0 ms build=2.9 ms (stall for 0 active slots)
[Sample] step=0 c0=27 u=-1.0000000000 subseq=0
[Debug] codes-step0: [16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] next-emb-step0: [2048] first4: 0.021546 -0.036059 0.003111 0.073181
/mnt/workspace/git/qwentts.cpp/ggml/src/ggml-vulkan/ggml-vulkan.cpp:11497: GGML_ASSERT(dst->op != GGML_OP_GET_ROWS || (a_offset == 0 && b_offset == 0 && d_offset == 0)) failed
[New LWP 121584]
[New LWP 121583]
[New LWP 121582]
[New LWP 121581]
[New LWP 121580]
[New LWP 121579]
[New LWP 121578]
[New LWP 121577]
[New LWP 121576]
[New LWP 121575]
[New LWP 121574]
[New LWP 121573]
[New LWP 121572]
[New LWP 121571]
[New LWP 121570]
[New LWP 121569]
[New LWP 121568]
[New LWP 121567]
[New LWP 121566]
[New LWP 121565]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
0x00007fd1dd7b69ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#0 0x00007fd1dd7b69ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#1 0x00007fd1dd7ab668 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#2 0x00007fd1dd7ab6ad in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#3 0x00007fd1dd8167c7 in wait4 () from /lib/x86_64-linux-gnu/libc.so.6
#4 0x00007fd1ddd2ca93 in ggml_print_backtrace () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#5 0x00007fd1ddd2cbfe in ggml_abort () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#6 0x00007fcf948d4bf1 in void init_pushconst_tensor_offsets<vk_op_binary_push_constants>(ggml_backend_vk_context*, vk_op_binary_push_constants&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#7 0x00007fcf94a49201 in void ggml_vk_op_f32<vk_op_binary_push_constants>(ggml_backend_vk_context*, std::shared_ptr<vk_context_struct>&, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor const*, ggml_tensor*, ggml_op, vk_op_binary_push_constants&&) [clone .constprop.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#8 0x00007fcf94a64f48 in ggml_vk_build_graph(ggml_backend_vk_context*, ggml_cgraph*, int, ggml_tensor*, int, bool, bool, bool) [clone .isra.0] () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#9 0x00007fcf94a679bc in ggml_backend_vk_graph_compute(ggml_backend*, ggml_cgraph*) () from /mnt/workspace/git/qwentts.cpp/build/libggml-vulkan.so
#10 0x00007fd1ddd4496e in ggml_backend_graph_compute () from /mnt/workspace/git/qwentts.cpp/build/libggml-base.so.0
#11 0x000055b420b50365 in tts_engine_step(TtsEngine*, std::vector<TtsJob*, std::allocator<TtsJob*> >*) ()
#12 0x000055b420b5d66a in pipeline_tts_synthesize(PipelineTTS*, BPETokenizer*, qt_tts_params const*, long, qt_audio*) ()
#13 0x000055b420b39b80 in qt_synthesize ()
#14 0x000055b420b30cdd in run(Args const&) ()
#15 0x000055b420b2ea83 in main ()
[Inferior 1 (process 121564) detached]
[Debug] talker-hidden-step1: [2048] first4: 0.372459 2.555521 -0.076819 0.171399
[Sample] step=1 c0=1902 u=-1.0000000000 subseq=16
[Pipeline] Generated 8 frames (slot 0)
[Pipeline] Generated 16 frames (slot 0)
[Pipeline] Generated 24 frames (slot 0)
[Pipeline] Generated 32 frames (slot 0)
[Pipeline] Generated 40 frames (slot 0)
[Pipeline] Generated 48 frames (slot 0)
[Pipeline] Generated 56 frames (slot 0)
[Pipeline] Generated 64 frames (slot 0)
[Pipeline] Generation done : 64 frames
[Debug] codes-full: [64, 16] first4: 27.000000 1408.000000 1824.000000 400.000000
[Debug] output-audio: [122880] first4: 0.000658 0.000394 -0.000943 -0.000566
[Perf] PromptBuild 2.9 ms
[Perf] Prefill 17.0 ms (T_ctx prefill)
[Perf] TTFA 771.5 ms (first frame codes)
[Perf] TalkerDecode 158.9 ms (64 frames, 2.48 ms/frame)
[Perf] CodePredictor 290.3 ms (4.54 ms/frame)
[Perf] HostCompose 0.1 ms (c0 sample + next emb)
[Perf] CodecDecode 31.1 ms
[Perf] Total 1247.3 ms (64 frames, 7.02 ms/frame AR, audio 5.12 s, RTF 0.244)
[WAV] Wrote cpp/clone/clone-cpp.wav: 122880 samples, 24000 Hz, mono S16
[Pipeline] Wrote 122880 samples (5.12 s) -> cpp/clone/clone-cpp.wav
[Input] Prompt: 100 chars: qwentts.cpp is a minimal C++17 and GGML port of Qwen3-TTS fo...
[Input] RefAudio: ../examples/freeman.wav
[Input] RefText: 213 chars: If you go into different cultures, they have different conce...
@@ -136,3 +118,38 @@ Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[Python] Audio: 120960 samples 24000 Hz 5.04s -> python/clone/clone-python.wav
[Quant] Q8_0 -> ../models/qwen-talker-1.7b-base-Q8_0.gguf + ../models/qwen-tokenizer-12hz-Q8_0.gguf
[GGML] Cmd: ../build/qwen-tts --model ../models/qwen-talker-1.7b-base-Q8_0.gguf --codec ../models/qwen-tokenizer-12hz-Q8_0.gguf --seed 42 --ref-wav ../examples/freeman.wav --ref-text ../examples/freeman.txt --lang english --max-new 64 --dump cpp/clone -o cpp/clone/clone-cpp.wav --greedy
[GGML] Audio: 122880 samples 24000 Hz 5.12s -> cpp/clone/clone-cpp.wav
[Cossim] PromptIDs exact: 100.00% (38 values)
[Cossim] RefCodes exact: 81.16% (3440 values)
[Cossim] Embed cos: 0.970174 max: 1.9001e-01 mean: 8.9738e-03
[Cossim] TrailingText cos: 0.999999 max: 1.9968e-04 mean: 4.5182e-05
[Cossim] TTSPadEmbed cos: 0.999999 max: 1.9968e-04 mean: 4.5182e-05
[Cossim] L0 cos: 0.985826 max: 7.7038e-01 mean: 1.1813e-02
[Cossim] L7 cos: 0.999979 max: 5.4001e+00 mean: 3.6051e-02
[Cossim] L14 cos: 0.999957 max: 5.4233e+00 mean: 7.0612e-02
[Cossim] L21 cos: 0.999605 max: 1.8733e+01 mean: 2.5057e-01
[Cossim] L27 cos: 0.998543 max: 5.3202e+01 mean: 5.6948e-01
[Cossim] Final cos: 0.992844 max: 1.9032e+01 mean: 1.8018e-01
[Cossim] Logits cos: 0.999587 max: 7.9291e-01 mean: 1.0517e-01
[Cossim] NextEmbStep0 cos: 0.848337 max: 2.7267e-01 mean: 2.8037e-02
[Cossim] TalkerHiddenStep1 cos: 0.993831 max: 1.5021e+00 mean: 2.3902e-01
[Cossim] MelHann cos: 1.000000 max: 1.7881e-07 mean: 3.2043e-08
[Cossim] MelBasis cos: 1.000000 max: 1.7881e-07 mean: 3.9379e-10
[Cossim] MelMag cos: 1.000000 max: 3.1307e-05 mean: 5.6396e-07
[Cossim] MelSpk cos: 1.000000 max: 2.5492e-03 mean: 9.3754e-06
[Cossim] SeanetInit cos: 1.000000 max: 4.1014e-04 mean: 3.4146e-06
[Cossim] SeanetResnet0 cos: 0.999987 max: 1.9526e-03 mean: 9.4296e-05
[Cossim] SeanetStage0 cos: 0.999985 max: 4.2238e-02 mean: 6.5089e-04
[Cossim] SeanetStage1 cos: 0.999866 max: 7.7454e-01 mean: 4.9543e-03
[Cossim] SeanetStage3 cos: 0.999995 max: 1.4657e+00 mean: 3.8722e-02
[Cossim] SeanetOut cos: 0.999963 max: 4.0791e-01 mean: 1.6984e-02
[Cossim] EncTransformer cos: 0.999932 max: 3.3955e-01 mean: 1.5945e-02
[Cossim] CodecPreFSQ cos: 0.999951 max: 4.5186e-01 mean: 2.6992e-02
[Cossim] SpkFrontend cos: 1.000000 max: 2.6600e-03 mean: 1.1964e-04
[Cossim] SpkBlock3 cos: 1.000000 max: 1.2137e-02 mean: 5.2085e-04
[Cossim] SpkMFA cos: 1.000000 max: 8.9911e-03 mean: 1.3079e-04
[Cossim] SpkASP cos: 1.000000 max: 2.2919e-04 mean: 1.7381e-05
[Cossim] SpeakerEmb cos: 1.000000 max: 8.0518e-05 mean: 1.3621e-05
[Cossim] CodesFull exact: 2.38% (1008 values)
[Cossim] Audio cos: 0.013854
[Cossim] WAV stft_cos: 0.183103 samples: 120960