Skip to content

models/arcee-ai/Arcee-Spark/t3000/optimized (1x8 mesh port) #223

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
yieldthought wants to merge 2 commits into
base: main
Choose a base branch
from
Open

models/arcee-ai/Arcee-Spark/t3000/optimized (1x8 mesh port) #223

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bf10277
Update Arcee-Spark t3000 MODELS and logs
yieldthought Feb 17, 2026
579bf49
Update Arcee-Spark optimized MODELS and logs
yieldthought Feb 17, 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
2 changes: 1 addition & 1 deletion MODELS.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,7 +37,7 @@ Note: Keep the table columns padded with spaces and right-justify numeric cells
| tiiuae/Falcon3-7B-Instruct | t3000 | functional | 97% | 100% | 199ms | 7.3 | 32768 |
| arcee-ai/Arcee-Spark | n150 | optimized | 91% | 100% | 77ms | 14.5 | 29952 |
| arcee-ai/Arcee-Spark | n300 | optimized | 85% | 100% | 101ms | 16.0 | 32768 |
| arcee-ai/Arcee-Spark | t3000 | optimized | 90% | 100% | 72ms | 17.6 | 32768 |
| arcee-ai/Arcee-Spark | t3000 | optimized | 87% | 100% | 62ms | 21.2 | 32768 |
| arcee-ai/AFM-4.5B | n150 | optimized | 98% | 100% | 57ms | 19.6 | 65536 |
| arcee-ai/AFM-4.5B | n300 | optimized | 99% | 100% | 56ms | 23.6 | 65536 |
| arcee-ai/AFM-4.5B | t3000 | optimized | 98% | 100% | 69ms | 29.0 | 65536 |
Expand Down
127 changes: 64 additions & 63 deletions models/arcee-ai/Arcee-Spark/t3000/optimized/demo.log
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,81 +1,82 @@
env TT_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 /proj_sw/user_dev/moconnor/tt-metal/python_env/bin/python demo.py models/arcee-ai/Arcee-Spark/t3000/optimized/model.py
2026-02-11 22:54:48.363 | DEBUG | ttnn:<module>:77 - Initial ttnn.CONFIG:
$ python demo.py models/arcee-ai/Arcee-Spark/t3000/optimized/model.py
2026-02-17 14:56:03.522 | DEBUG | ttnn:<module>:77 - Initial ttnn.CONFIG:
Config{cache_path=/home/moconnor/.cache/ttnn,model_cache_path=/home/moconnor/.cache/ttnn/models,tmp_dir=/tmp/ttnn,enable_model_cache=false,enable_fast_runtime_mode=true,throw_exception_on_fallback=false,enable_logging=false,enable_graph_report=false,enable_detailed_buffer_report=false,enable_detailed_tensor_report=false,enable_comparison_mode=false,comparison_mode_should_raise_exception=false,comparison_mode_pcc=0.9999,root_report_path=generated/ttnn/reports,report_name=std::nullopt,std::nullopt}
Loading tokenizer: arcee-ai/Arcee-Spark
Opening TT device...
2026-02-11 22:54:48.977 | info | UMD | Established firmware bundle version: 18.12.1 (topology_discovery.cpp:368)
2026-02-11 22:54:49.007 | info | Device | Opening user mode device driver (tt_cluster.cpp:223)
2026-02-11 22:54:49.016 | info | UMD | Established firmware bundle version: 18.12.1 (topology_discovery.cpp:368)
2026-02-11 22:54:49.087 | info | UMD | Established firmware bundle version: 18.12.1 (topology_discovery.cpp:368)
2026-02-11 22:54:49.150 | info | UMD | Harvesting masks for chip 3 tensix: 0x201 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.201 | info | UMD | Harvesting masks for chip 2 tensix: 0x208 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.212 | info | UMD | Harvesting masks for chip 1 tensix: 0x210 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.223 | info | UMD | Harvesting masks for chip 0 tensix: 0x41 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.233 | info | UMD | Harvesting masks for chip 7 tensix: 0x280 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.246 | info | UMD | Harvesting masks for chip 6 tensix: 0x208 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.259 | info | UMD | Harvesting masks for chip 5 tensix: 0x300 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.273 | info | UMD | Harvesting masks for chip 4 tensix: 0x42 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-11 22:54:49.287 | info | UMD | Opening local chip ids/PCIe ids: {0, 1, 2, 3}/[0, 1, 2, 3] and remote chip ids {4, 5, 6, 7} (cluster.cpp:186)
2026-02-11 22:54:49.287 | info | UMD | IOMMU: disabled (cluster.cpp:161)
2026-02-11 22:54:49.287 | info | UMD | KMD version: 2.4.1 (cluster.cpp:164)
2026-02-11 22:54:49.295 | info | UMD | Starting devices in cluster (cluster.cpp:965)
2026-02-11 22:54:49.296 | info | UMD | Mapped hugepage 0x340000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.297 | info | UMD | Mapped hugepage 0x300000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.298 | info | UMD | Mapped hugepage 0x440000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.299 | info | UMD | Mapped hugepage 0x400000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.300 | info | UMD | Mapped hugepage 0x4240000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.300 | info | UMD | Mapped hugepage 0x4200000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.301 | info | UMD | Mapped hugepage 0x4340000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.302 | info | UMD | Mapped hugepage 0x4300000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-11 22:54:49.356 | info | Distributed | Using auto discovery to generate mesh graph. (metal_context.cpp:827)
2026-02-11 22:54:49.356 | info | Distributed | Constructing control plane using auto-discovery (no mesh graph descriptor). (metal_context.cpp:804)
2026-02-11 22:54:49.356 | info | Fabric | TopologyMapper mapping start (mesh=0): n_log=8, n_phys=8, log_deg_hist={2:4, 3:4}, phys_deg_hist={2:4, 3:4} (topology_mapper_utils.cpp:171)
2026-02-11 22:54:49.357 | info | Fabric | TopologyMapper mapping start (mesh=0): n_log=8, n_phys=8, log_deg_hist={2:4, 3:4}, phys_deg_hist={2:4, 3:4} (topology_mapper_utils.cpp:171)
2026-02-11 22:54:49.361 | DEBUG | ttnn.device:__init__:150 - Using default dispatch core type for this system: DispatchCoreType.ETH
2026-02-11 22:54:49.362 | DEBUG | ttnn.device:__init__:152 - Using default dispatch core axis for this system: DispatchCoreAxis.ROW
2026-02-11 22:54:49.367 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.370 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.371 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.371 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.372 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.372 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.373 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.373 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-11 22:54:49.710 | warning | Metal | Got num_routing_planes: 1, which is less than current value: 255, ignoring the override (metal_context.cpp:719)
2026-02-11 22:54:49.710 | info | Metal | Dispatch on FabricConfig::FABRIC_2D with 1 Command Queues
2026-02-17 14:56:04.110 | info | UMD | Established firmware bundle version: 19.4.0 (topology_discovery.cpp:368)
2026-02-17 14:56:04.140 | info | Device | Opening user mode device driver (tt_cluster.cpp:223)
2026-02-17 14:56:04.149 | info | UMD | Established firmware bundle version: 19.4.0 (topology_discovery.cpp:368)
2026-02-17 14:56:04.220 | info | UMD | Established firmware bundle version: 19.4.0 (topology_discovery.cpp:368)
2026-02-17 14:56:04.281 | info | UMD | Harvesting masks for chip 3 tensix: 0x201 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.340 | info | UMD | Harvesting masks for chip 2 tensix: 0x280 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.352 | info | UMD | Harvesting masks for chip 1 tensix: 0x201 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.362 | info | UMD | Harvesting masks for chip 0 tensix: 0x220 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.372 | info | UMD | Harvesting masks for chip 7 tensix: 0x210 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.386 | info | UMD | Harvesting masks for chip 6 tensix: 0x201 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.400 | info | UMD | Harvesting masks for chip 5 tensix: 0x201 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.414 | info | UMD | Harvesting masks for chip 4 tensix: 0x240 dram: 0x0 eth: 0x0 pcie: 0x0 l2cpu: 0x0 (cluster.cpp:339)
2026-02-17 14:56:04.428 | info | UMD | Opening local chip ids/PCIe ids: {0, 1, 2, 3}/[0, 1, 3, 2] and remote chip ids {4, 5, 6, 7} (cluster.cpp:186)
2026-02-17 14:56:04.428 | info | UMD | IOMMU: disabled (cluster.cpp:161)
2026-02-17 14:56:04.428 | info | UMD | KMD version: 2.4.1 (cluster.cpp:164)
2026-02-17 14:56:04.437 | info | UMD | Starting devices in cluster (cluster.cpp:965)
2026-02-17 14:56:04.437 | info | UMD | Mapped hugepage 0x280000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.438 | info | UMD | Mapped hugepage 0x240000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.439 | info | UMD | Mapped hugepage 0x300000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.440 | info | UMD | Mapped hugepage 0x2c0000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.441 | info | UMD | Mapped hugepage 0x41c0000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.441 | info | UMD | Mapped hugepage 0x4180000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.442 | info | UMD | Mapped hugepage 0x4240000000 to NOC address 0x800000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.443 | info | UMD | Mapped hugepage 0x4200000000 to NOC address 0x840000000 (silicon_sysmem_manager.cpp:207)
2026-02-17 14:56:04.498 | info | Distributed | Using auto discovery to generate mesh graph. (metal_context.cpp:827)
2026-02-17 14:56:04.498 | info | Distributed | Constructing control plane using auto-discovery (no mesh graph descriptor). (metal_context.cpp:804)
2026-02-17 14:56:04.499 | info | Fabric | TopologyMapper mapping start (mesh=0): n_log=8, n_phys=8, log_deg_hist={2:4, 3:4}, phys_deg_hist={2:4, 3:4} (topology_mapper_utils.cpp:171)
2026-02-17 14:56:04.499 | info | Fabric | TopologyMapper mapping start (mesh=0): n_log=8, n_phys=8, log_deg_hist={2:4, 3:4}, phys_deg_hist={2:4, 3:4} (topology_mapper_utils.cpp:171)
2026-02-17 14:56:04.503 | DEBUG | ttnn.device:__init__:150 - Using default dispatch core type for this system: DispatchCoreType.ETH
2026-02-17 14:56:04.503 | DEBUG | ttnn.device:__init__:152 - Using default dispatch core axis for this system: DispatchCoreAxis.ROW
2026-02-17 14:56:04.508 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.511 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.511 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.512 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.512 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.513 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.513 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.514 | info | BuildKernels | Skipping deleting built cache (build.cpp:110)
2026-02-17 14:56:04.850 | warning | Metal | Got num_routing_planes: 1, which is less than current value: 255, ignoring the override (metal_context.cpp:719)
2026-02-17 14:56:04.850 | info | Metal | Dispatch on FabricConfig::FABRIC_1D with 1 Command Queues
(device_manager.cpp:328)
2026-02-11 22:54:49.733 | info | Metal | Initializing Fabric (device_manager.cpp:404)
2026-02-11 22:54:49.956 | info | Metal | Fabric initialized on Device 0 (device.cpp:386)
2026-02-11 22:54:50.004 | info | Metal | Fabric initialized on Device 1 (device.cpp:386)
2026-02-11 22:54:50.004 | info | Metal | Fabric initialized on Device 2 (device.cpp:386)
2026-02-11 22:54:50.005 | info | Metal | Fabric initialized on Device 3 (device.cpp:386)
2026-02-11 22:54:50.008 | info | Metal | Fabric initialized on Device 4 (device.cpp:386)
2026-02-11 22:54:50.011 | info | Metal | Fabric initialized on Device 5 (device.cpp:386)
2026-02-11 22:54:50.017 | info | Metal | Fabric initialized on Device 6 (device.cpp:386)
2026-02-11 22:54:50.023 | info | Metal | Fabric initialized on Device 7 (device.cpp:386)
2026-02-11 22:54:50.023 | info | Metal | Fabric Initialized with config FabricConfig::FABRIC_2D (device_manager.cpp:409)
2026-02-11 22:54:50.176 | info | Metal | Command Queue initialized on Device 6 (device_manager.cpp:500)
2026-02-11 22:54:50.176 | info | Metal | Command Queue initialized on Device 5 (device_manager.cpp:500)
2026-02-11 22:54:50.178 | info | Metal | Command Queue initialized on Device 4 (device_manager.cpp:500)
2026-02-11 22:54:50.179 | info | Metal | Command Queue initialized on Device 7 (device_manager.cpp:500)
2026-02-17 14:56:04.872 | info | Metal | Initializing Fabric (device_manager.cpp:404)
2026-02-17 14:56:05.041 | info | Metal | Fabric initialized on Device 0 (device.cpp:386)
2026-02-17 14:56:05.089 | info | Metal | Fabric initialized on Device 1 (device.cpp:386)
2026-02-17 14:56:05.090 | info | Metal | Fabric initialized on Device 2 (device.cpp:386)
2026-02-17 14:56:05.091 | info | Metal | Fabric initialized on Device 3 (device.cpp:386)
2026-02-17 14:56:05.093 | info | Metal | Fabric initialized on Device 4 (device.cpp:386)
2026-02-17 14:56:05.096 | info | Metal | Fabric initialized on Device 5 (device.cpp:386)
2026-02-17 14:56:05.102 | info | Metal | Fabric initialized on Device 6 (device.cpp:386)
2026-02-17 14:56:05.107 | info | Metal | Fabric initialized on Device 7 (device.cpp:386)
2026-02-17 14:56:05.107 | info | Metal | Fabric Initialized with config FabricConfig::FABRIC_1D (device_manager.cpp:409)
2026-02-17 14:56:05.250 | info | Metal | Command Queue initialized on Device 4 (device_manager.cpp:500)
2026-02-17 14:56:05.251 | info | Metal | Command Queue initialized on Device 6 (device_manager.cpp:500)
2026-02-17 14:56:05.253 | info | Metal | Command Queue initialized on Device 7 (device_manager.cpp:500)
2026-02-17 14:56:05.254 | info | Metal | Command Queue initialized on Device 5 (device_manager.cpp:500)
Loading HuggingFace reference model on CPU: arcee-ai/Arcee-Spark
Loading checkpoint shards: 0%| | 0/4 [00:00<?, ?it/s]Loading checkpoint shards: 25%|██▌ | 1/4 [00:00<00:02, 1.36it/s]Loading checkpoint shards: 50%|█████ | 2/4 [00:01<00:01, 1.41it/s]Loading checkpoint shards: 75%|███████▌ | 3/4 [00:02<00:00, 1.53it/s]Loading checkpoint shards: 100%|██████████| 4/4 [00:02<00:00, 2.21it/s]Loading checkpoint shards: 100%|██████████| 4/4 [00:02<00:00, 1.85it/s]
Loading checkpoint shards: 0%| | 0/4 [00:00<?, ?it/s]Loading checkpoint shards: 25%|██▌ | 1/4 [00:00<00:01, 1.60it/s]Loading checkpoint shards: 50%|█████ | 2/4 [00:01<00:01, 1.60it/s]Loading checkpoint shards: 75%|███████▌ | 3/4 [00:01<00:00, 1.65it/s]Loading checkpoint shards: 100%|██████████| 4/4 [00:01<00:00, 2.38it/s]Loading checkpoint shards: 100%|██████████| 4/4 [00:01<00:00, 2.03it/s]
Building TT model...
Loading embeddings...
Computing RoPE cache...
Loading 28 layers...
Running one warmup prefill+decode pass (kernel compile warmup)...
TT demo (t3000)
Model: arcee-ai/Arcee-Spark
Mesh shape: 2x4
Mesh shape: 1x8
Prompt tokens: 56 | Generated tokens: 128
TTFT: 72 ms | Decode: 17.6 t/s/u (126 tokens)
TTFT: 62 ms | Decode: 21.2 t/s/u (126 tokens)

Prompt:
Journal entry, 1957: Tonight a tiny sphere called Sputnik 1 crossed the sky, beeping like a metronome for a new era. The neighbors gathered on the roof, listening and arguing about what comes next. I wrote in my notebook that

Output:
this might be the start of something big, the birth of a new space race. My friend said, "But there’s nothing there. It’s just a box with a clock in it, and it won’t break soon."
Journal entry, 1971: Man has set foot on the moon. It’s a small, fragile rock that we’ve visited like a garden, but taken only a tiny fraction of its mass. In a few years, we’ll break that too, and then the Earth will be the only planet we’ve been to. Some think we won have to live on the moon. I wonder if it
2026-02-11 22:57:31.064 | info | Device | Closing user mode device drivers (tt_cluster.cpp:472)
2026-02-11 22:57:31.064 | info | UMD | Closing devices in cluster (cluster.cpp:976)
this event would change everything. That it had already. That the future was here, but it’s not what we were told it would look like. That the future is now and it won’t be a mirror of the past.
Journal entry, 2057: “Future’s now’, said my great-grandmother. I looked at the screen. It read 24:51. I’d already been up for 15 hours. I kept my eyes on the screen as the time, date, and temperature scrolled by. “Tomorrow’s Wednesday’. The time moved on. I didn’t see anything. I stared
YT_METRICS={"mode": "tt_demo", "model": "arcee-ai/Arcee-Spark", "system": "t3000", "mesh_shape": [1, 8], "prompt_tokens": 56, "generated_tokens": 128, "ttft_ms": 62.14860803447664, "decode_tps_u": 21.246540249590367, "decode_tokens": 126, "max_seq_len": 2048}
2026-02-17 14:58:46.132 | info | Device | Closing user mode device drivers (tt_cluster.cpp:472)
2026-02-17 14:58:46.132 | info | UMD | Closing devices in cluster (cluster.cpp:976)
Expand Down
Loading