AMD Instinct MI350X vs NVIDIA L40 — Confronto GPU (Apr 2026)
AMD Instinct MI350X (288GB HBM3e, 1,800 TFLOPS FP16, CDNA 4) vs NVIDIA L40 (48GB GDDR6, 181 TFLOPS FP16, Ada Lovelace)
|
AMD Instinct MI350X
288GB HBM3e · CDNA 4
|
NVIDIA L40
48GB GDDR6 · Ada Lovelace
|
||
|---|---|---|---|
| Specifiche | |||
| Produttore | AMD | NVIDIA | |
| Architettura | CDNA 4 | Ada Lovelace | |
| VRAM | 288 GB HBM3e | 48 GB GDDR6 | |
| Larghezza di banda | 8,000 GB/s | 864 GB/s | |
| FP16 (Tensor) | 1800.0 TFLOPS | 181.0 TFLOPS | |
| FP32 | 72.0 TFLOPS | 90.5 TFLOPS | |
| TDP | 1000W | 300W | |
| Anno di rilascio | 2025 | 2023 | |
| Segmento | Data center | Data center | |
| Ideale per | Next-gen AI training inference at scale | Inference video processing rendering | |
| Prezzi Cloud | |||
| Più economico On-Demand | — | — | |
| Più economico Spot | — | — | |
| Provider | 1 | 0 | |
AMD Instinct MI350X
NVIDIA L40