找到合适的加速器。

Legacy

VRAM

32 GB

内存

HBM2

带宽

1.2 TB/s

TDP

300 W

PCIe dual-slot FHFL · HPC / AI

Legacy

VRAM

64 GB

内存

HBM2E

带宽

1.6 TB/s

TDP

300 W

PCIe dual-slot FHFL · HPC / AI

Legacy

VRAM

128 GB

内存

HBM2E

带宽

3.2 TB/s

TDP

560 W

OAM Module · HPC / AI training

Previous

VRAM

192 GB

内存

HBM3

带宽

5.3 TB/s

TDP

750 W

OAM Module · AI training / inference / HPC

Previous

VRAM

256 GB

内存

HBM3E

带宽

6 TB/s

TDP

1000 W

OAM Module · AI training / inference / HPC

Current

VRAM

288 GB

内存

HBM3E

带宽

8 TB/s

TDP

1000 W

OAM Module · AI training / inference / HPC

Current

VRAM

288 GB

内存

HBM3E

带宽

8 TB/s

TDP

1400 W

OAM Module · AI training / inference / HPC

Current

VRAM

432 GB

内存

HBM4

带宽

23.3 TB/s

Accelerator module · HPC / scientific AI / sovereign AI

Current

VRAM

432 GB

内存

HBM4

带宽

23.3 TB/s

Enhanced Accelerator Module (EAM) · AI training / inference

Previous

VRAM

96 GB

内存

HBM2E

带宽

2.45 TB/s

TDP

600 W

Mezzanine accelerator · AI training / inference

Current

VRAM

128 GB

内存

HBM2E

带宽

3.7 TB/s

AI training / inference

Previous

VRAM

24 GB

内存

GDDR6

带宽

0.6 TB/s

TDP

150 W

PCIe single-slot FHFL · AI inference / graphics / VDI

Previous

VRAM

40 GB

内存

HBM2

带宽

1.555 TB/s

AI training / inference / HPC

Previous

VRAM

40 GB

内存

HBM2

带宽

1.555 TB/s

TDP

250 W

PCIe dual-slot · AI training / inference / HPC

Previous

VRAM

40 GB

内存

HBM2

带宽

1.555 TB/s

TDP

400 W

SXM · AI training / inference / HPC

Previous

VRAM

80 GB

内存

HBM2E

AI training / inference / HPC

Previous

VRAM

80 GB

内存

HBM2E

带宽

1.935 TB/s

TDP

300 W

PCIe dual-slot · AI training / inference / HPC

Previous

VRAM

80 GB

内存

HBM2E

带宽

2.039 TB/s

TDP

400 W

SXM · AI training / inference / HPC

Previous

VRAM

20 GB

AI training / inference

Previous

VRAM

80 GB

内存

HBM2E

AI training / inference / HPC

Previous

VRAM

64 GB

内存

GDDR6

带宽

0.8 TB/s

TDP

250 W

PCIe dual-slot FHFL · VDI / graphics virtualization / cloud workstations

Previous

VRAM

16 GB

内存

GDDR6

带宽

0.2 TB/s

TDP

60 W

PCIe low-profile single-slot · Edge AI / inference

Previous

VRAM

24 GB

内存

HBM2

带宽

0.933 TB/s

TDP

165 W

PCIe dual-slot FHFL · AI inference / training / HPC

Previous

VRAM

48 GB

内存

GDDR6

带宽

0.696 TB/s

TDP

300 W

PCIe dual-slot · Graphics / AI / virtual workstation

Previous

VRAM

40 GB

内存

HBM2

带宽

1.555 TB/s

TDP

240 W

PCIe dual-slot active · AI / data science / HPC workstation

Previous

VRAM

80 GB

内存

HBM2E

带宽

1.935 TB/s

TDP

300 W

PCIe dual-slot FHFL · AI training / inference / HPC

Previous

VRAM

80 GB

内存

HBM2E

带宽

2.039 TB/s

TDP

500 W

SXM4 / HGX · AI training / inference / HPC

Current

VRAM

180 GB

内存

HBM3E

带宽

8 TB/s

TDP

1000 W

SXM · AI training / inference / HPC

Current

VRAM

192 GB

内存

HBM3E

带宽

8 TB/s

TDP

1000 W

SXM · AI training / inference / HPC

Current

VRAM

192 GB

内存

HBM3E

带宽

8 TB/s

TDP

1000 W

SXM · AI training / inference / HPC

Current

VRAM

192 GB

内存

HBM3E

带宽

8 TB/s

TDP

1000 W

SXM · AI training / inference / HPC

Current

VRAM

288 GB

内存

HBM3E

带宽

8 TB/s

TDP

1400 W

SXM · AI training / inference / HPC

Current

VRAM

186 GB

内存

HBM3E

带宽

8 TB/s

TDP

1200 W

Grace Blackwell Superchip / NVL · Large-scale AI training / inference

Current

VRAM

288 GB

内存

HBM3E

带宽

8 TB/s

TDP

1400 W

Rack-scale NVL module · Frontier AI training / reasoning inference / HPC

VRAM

11 GB

内存

GDDR5X

VRAM

12 GB

内存

GDDR6

TDP

170 W

PCIe consumer

VRAM

8 GB

内存

GDDR6

带宽

0.448 TB/s

TDP

220 W

PCIe consumer

VRAM

8 GB

内存

GDDR6X

TDP

290 W

PCIe consumer

VRAM

10 GB

内存

GDDR6X

带宽

0.76 TB/s

TDP

320 W

PCIe consumer

VRAM

12 GB

内存

GDDR6X

TDP

350 W

PCIe consumer

Legacy

VRAM

24 GB

内存

GDDR6X

带宽

0.936 TB/s

TDP

350 W

PCIe consumer · AI / consumer compute

VRAM

24 GB

内存

GDDR6X

TDP

450 W

PCIe consumer

Current

VRAM

16 GB

内存

GDDR6

PCIe consumer · Cloud GPU / AI workloads

Current

VRAM

12 GB

PCIe consumer · Cloud GPU / AI workloads

VRAM

12 GB

内存

GDDR6X

TDP

285 W

PCIe consumer

Previous

VRAM

16 GB

内存

GDDR6X

带宽

0.672 TB/s

TDP

285 W

PCIe consumer · AI / generative AI / rendering / consumer compute

Previous

VRAM

16 GB

内存

GDDR6X

带宽

0.717 TB/s

TDP

320 W

PCIe consumer triple-slot · AI / generative AI / rendering / consumer compute

Previous

VRAM

16 GB

内存

GDDR6X

带宽

0.736 TB/s

TDP

320 W

PCIe consumer triple-slot · AI / generative AI / rendering / consumer compute

Previous

VRAM

24 GB

内存

GDDR6X

带宽

1.008 TB/s

TDP

450 W

PCIe consumer · AI / generative AI / consumer compute

Current

VRAM

8 GB

内存

GDDR7

PCIe consumer · Cloud GPU / AI workloads

Current

VRAM

16 GB

内存

GDDR7

PCIe consumer · Cloud GPU / AI workloads

Current

VRAM

12 GB

内存

GDDR7

带宽

0.672 TB/s

TDP

250 W

PCIe consumer · AI inference / generative AI / consumer compute

Current

VRAM

16 GB

内存

GDDR7

带宽

0.896 TB/s

TDP

300 W

PCIe consumer · AI / generative AI / rendering / consumer compute

Current

VRAM

16 GB

内存

GDDR7

带宽

0.96 TB/s

TDP

360 W

PCIe consumer dual-slot · AI / generative AI / rendering / consumer compute

Current

VRAM

32 GB

内存

GDDR7

带宽

1.792 TB/s

TDP

575 W

PCIe consumer · AI / generative AI / consumer compute

Current

VRAM

24 GB

内存

GDDR7

带宽

0.896 TB/s

Laptop GPU · Cloud GPU / AI workloads

Previous

VRAM

144 GB

内存

HBM3E

带宽

4.9 TB/s

TDP

900 W

Grace Hopper Superchip · AI training / inference / HPC / large-memory workloads

Previous

VRAM

96 GB

内存

HBM3

带宽

4 TB/s

TDP

900 W

Grace Hopper Superchip · AI training / inference / HPC

Previous

VRAM

80 GB

AI training / inference / HPC

Previous Generation

VRAM

94 GB

内存

HBM3

带宽

3.9 TB/s

TDP

400 W

PCIe · AI Training, AI Inference, HPC

Previous

VRAM

80 GB

AI training / inference / HPC

Previous

VRAM

80 GB

内存

HBM2E

带宽

2 TB/s

TDP

350 W

PCIe dual-slot · AI training / inference / HPC

Previous Generation

VRAM

80 GB

内存

HBM3

带宽

3.35 TB/s

TDP

700 W

SXM · AI Training, AI Inference, HPC

Current

VRAM

141 GB

内存

HBM3E

带宽

4 TB/s

TDP

400 W

HGX SXM5 / 8-GPU platform · Large-model inference / AI training

Current

VRAM

96 GB

内存

HBM3

带宽

4 TB/s

TDP

400 W

HGX SXM5 / 8-GPU platform · AI inference / training / China-market AI compute

Previous

VRAM

141 GB

内存

HBM3E

带宽

4.8 TB/s

AI training / inference / HPC

Previous

VRAM

141 GB

内存

HBM3E

带宽

4.8 TB/s

TDP

600 W

PCIe dual-slot · AI inference / HPC / enterprise

Previous

VRAM

141 GB

内存

HBM3E

带宽

4.8 TB/s

TDP

700 W

SXM · AI training / inference / HPC

Previous

VRAM

80 GB

内存

HBM2E

带宽

2 TB/s

TDP

350 W

PCIe dual-slot FHFL · AI training / inference / HPC

Previous

VRAM

80 GB

内存

HBM3

带宽

3.35 TB/s

TDP

700 W

SXM5 / HGX · AI training / inference / HPC

Previous

VRAM

94 GB

内存

HBM3

带宽

3.9 TB/s

TDP

400 W

PCIe dual-slot FHFL · AI training / inference / enterprise AI

Current

VRAM

48 GB

内存

GDDR6

带宽

0.864 TB/s

TDP

275 W

PCIe dual-slot FHFL · AI inference / graphics / China-market compute

Current

VRAM

24 GB

内存

GDDR6

带宽

0.3 TB/s

TDP

72 W

PCIe low-profile single-slot · AI inference / video / graphics

Previous

VRAM

48 GB

内存

GDDR6

带宽

0.864 TB/s

TDP

300 W

PCIe dual-slot FHFL · AI inference / graphics / virtual workstation

Current

VRAM

48 GB

内存

GDDR6

带宽

0.864 TB/s

TDP

350 W

PCIe dual-slot · AI inference / graphics / rendering

VRAM

8 GB

内存

GDDR5

带宽

0.192 TB/s

TDP

120 W

PCIe single-slot

VRAM

8 GB

内存

GDDR5

带宽

0.243 TB/s

TDP

105 W

PCIe single-slot

VRAM

16 GB

内存

GDDR5X

带宽

0.288 TB/s

TDP

180 W

PCIe dual-slot

VRAM

24 GB

内存

GDDR5X

带宽

0.432 TB/s

TDP

250 W

PCIe dual-slot

VRAM

8 GB

内存

GDDR6

带宽

0.416 TB/s

TDP

125 W

PCIe single-slot

VRAM

16 GB

内存

GDDR6

带宽

0.448 TB/s

TDP

230 W

PCIe dual-slot

Legacy

VRAM

24 GB

内存

GDDR6

带宽

0.672 TB/s

TDP

260 W

PCIe dual-slot · AI / rendering / professional visualization

Legacy

VRAM

48 GB

内存

GDDR6

带宽

0.672 TB/s

TDP

260 W

PCIe dual-slot · AI / rendering / professional visualization

Previous

VRAM

20 GB

内存

GDDR6

带宽

0.36 TB/s

TDP

130 W

PCIe single-slot · AI inference / rendering / workstation

Previous

VRAM

24 GB

内存

GDDR6

带宽

0.432 TB/s

TDP

210 W

PCIe dual-slot · AI / rendering / workstation

Previous

VRAM

32 GB

内存

GDDR6

带宽

0.576 TB/s

TDP

250 W

PCIe dual-slot · AI / rendering / workstation / inference

Previous

VRAM

48 GB

内存

GDDR6

带宽

0.96 TB/s

TDP

300 W

PCIe dual-slot · AI / rendering / workstation

Previous

VRAM

12 GB

内存

GDDR6

带宽

0.288 TB/s

TDP

70 W

PCIe low-profile dual-slot · AI inference / rendering / workstation

Previous

VRAM

16 GB

内存

GDDR6

带宽

0.448 TB/s

TDP

140 W

PCIe single-slot · AI / rendering / workstation

Previous

VRAM

20 GB

内存

GDDR6

带宽

0.64 TB/s

TDP

200 W

PCIe dual-slot · AI / rendering / workstation

Previous

VRAM

24 GB

内存

GDDR6

带宽

0.768 TB/s

TDP

230 W

PCIe dual-slot · AI / rendering / workstation

Previous

VRAM

24 GB

内存

GDDR6

带宽

0.768 TB/s

TDP

230 W

PCIe dual-slot · AI / rendering / workstation

Previous

VRAM

48 GB

内存

GDDR6

带宽

0.768 TB/s

TDP

300 W

PCIe dual-slot · AI / rendering / workstation

Current

VRAM

16 GB

内存

GDDR7

带宽

0.288 TB/s

TDP

70 W

PCIe low-profile dual-slot · Edge AI / professional graphics / inference

Current

VRAM

24 GB

内存

GDDR7

带宽

0.672 TB/s

TDP

145 W

PCIe single-slot · AI / rendering / workstation / inference

Current

VRAM

32 GB

内存

GDDR7

带宽

0.8 TB/s

TDP

165 W

PCIe Gen5 single-slot FHFL · AI inference / data science / enterprise compute / video

Current

VRAM

32 GB

内存

GDDR7

带宽

0.896 TB/s

TDP

200 W

PCIe Gen5 dual-slot · AI / rendering / professional visualization / data science

Current

VRAM

48 GB

内存

GDDR7

带宽

1.344 TB/s

TDP

300 W

PCIe Gen5 dual-slot · AI development / inference / data science / rendering

Current

VRAM

72 GB

内存

GDDR7

带宽

1.344 TB/s

TDP

300 W

PCIe Gen5 dual-slot · AI development / LLM inference / data science / rendering

VRAM

96 GB

内存

GDDR7

Current

VRAM

96 GB

内存

GDDR7

带宽

1.792 TB/s

TDP

300 W

PCIe Gen5 dual-slot · AI development / inference / data science / rendering

Current

VRAM

96 GB

内存

GDDR7

带宽

1.597 TB/s

TDP

600 W

PCIe dual-slot / single-slot liquid · AI inference / graphics / enterprise compute

Current

VRAM

288 GB

内存

HBM4

带宽

22 TB/s

Rack-scale / liquid-cooled module · AI training / inference / HPC

Legacy

VRAM

16 GB

内存

GDDR6

带宽

0.32 TB/s

TDP

70 W

PCIe low-profile single-slot · AI inference / video

Legacy

VRAM

16 GB

内存

GDDR5

带宽

0.32 TB/s

TDP

300 W

PCIe dual-slot · VDI / graphics virtualization / legacy compute

Legacy

VRAM

16 GB

内存

HBM2

带宽

0.732 TB/s

TDP

250 W

PCIe dual-slot FHFL · HPC / legacy AI training

Legacy

VRAM

16 GB

内存

HBM2

带宽

0.732 TB/s

TDP

300 W

SXM2 · HPC / legacy AI training

Legacy

VRAM

8 GB

内存

GDDR5

带宽

0.192 TB/s

TDP

75 W

PCIe low-profile single-slot · AI inference / video transcoding

Legacy

VRAM

24 GB

内存

GDDR5

带宽

0.346 TB/s

TDP

250 W

PCIe dual-slot FHFL · Legacy AI inference

Legacy

VRAM

24 GB

内存

GDDR6

带宽

0.672 TB/s

TDP

280 W

PCIe dual-slot · AI research / data science / rendering

Legacy

VRAM

16 GB

内存

HBM2

带宽

0.9 TB/s

TDP

250 W

PCIe dual-slot FHHL · HPC / AI training

Legacy

VRAM

32 GB

内存

HBM2

带宽

0.9 TB/s

TDP

250 W

PCIe dual-slot FHHL · HPC / AI training

Legacy

VRAM

32 GB

内存

HBM2

带宽

0.9 TB/s

TDP

300 W

SXM2 · AI training / HPC

Legacy

VRAM

16 GB

内存

HBM2

带宽

0.9 TB/s

TDP

300 W

SXM2 · HPC / AI training

Legacy

VRAM

32 GB

内存

HBM2

SXM3 · HPC / AI training

Legacy

VRAM

32 GB

内存

HBM2

带宽

1.134 TB/s

TDP

250 W

PCIe dual-slot FHHL · HPC / AI training