Chip comparison
NVIDIA L40S vs Tenstorrent Blackhole
Side-by-side specs
Straight from each vendor's datasheet. Winner in bold: higher throughput, memory, or transistor count; lower power draw or price. Sparse (2:4) throughput roughly doubles the FP8/INT8 numbers shown.
| Field | NVIDIA L40S | Tenstorrent Blackhole |
|---|---|---|
| Process node | TSMC 4N | TSMC N6 |
| Transistors (B) | 76.3 | — |
| TDP | 350 W | 300 W |
| Memory | 48 GB GDDR6 | 32 GB GDDR6 |
| Memory bandwidth | 864 GB/s | 512 GB/s |
| FP16 (dense) | 362 TFLOPS | 745 TFLOPS |
| BF16 (dense) | 362 TFLOPS | — |
| FP8 (dense) | 733 TFLOPS | — |
| INT8 (dense) | 733 TOPS | 2,980 TOPS |
| Form factor | PCIe | PCIe |
| Announced | 2023-08-08 | 2024-07-01 |
| Released | — | 2025-01-01 |
Robots running NVIDIA L40S
No robots publicly running it yet.
Robots running Tenstorrent Blackhole
No robots publicly running it yet.
Data centers with NVIDIA L40S
No data centers publicly running it yet.
Data centers with Tenstorrent Blackhole
No data centers publicly running it yet.
Common questions
NVIDIA L40S vs Tenstorrent Blackhole: which is faster for training?
Tenstorrent Blackhole has 1.02x the dense FP16/BF16 throughput of the other (NVIDIA L40S: 733 TFLOPS; Tenstorrent Blackhole: 745 TFLOPS). Real-world training speed also depends on memory bandwidth, interconnect, and software stack: see the spec diff for those inputs.
NVIDIA L40S vs Tenstorrent Blackhole: which has more memory?
NVIDIA L40S carries more HBM (NVIDIA L40S: 48 GB GDDR6; Tenstorrent Blackhole: 32 GB GDDR6). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.
NVIDIA L40S vs Tenstorrent Blackhole: which is more power-efficient?
Dense FP16/BF16 TFLOPS per watt: NVIDIA L40S 2.09 (733 ÷ 350 W); Tenstorrent Blackhole 2.48 (745 ÷ 300 W). Tenstorrent Blackhole wins at the die level; system-level efficiency also depends on cooling and interconnect.
See also: every chip comparison · NVIDIA L40S full page · Tenstorrent Blackhole full page.