机构选择它的理由
NVIDIA HGX B300 with eight GPUs and 1.8 TB/s NVLink.
Air cooling for data halls without liquid-cooling infrastructure.
可选型号
此系列共有 2 款型号。
E3-G894-SD3
- Form factor
- 8U air-cooled GPU server
- Processor
- 2× Intel Xeon 6 or 2× AMD EPYC 9005 (alternative platforms)
- Accelerator
- NVIDIA HGX B300, 8 GPUs, 1.8 TB/s NVLink
- Storage
- Gen5 NVMe, designed to workload
E3-G894-ZD3
- Form factor
- 8U air-cooled GPU server
- Processor
- 2× Intel Xeon 6 or 2× AMD EPYC 9005 (alternative platforms)
- Accelerator
- NVIDIA HGX B300, 8 GPUs, 1.8 TB/s NVLink
- Storage
- Gen5 NVMe, designed to workload
| Part number | Form factor | Processor | Accelerator | Storage |
|---|---|---|---|---|
| E3-G894-SD3 | 8U air-cooled GPU server | 2× Intel Xeon 6 or 2× AMD EPYC 9005 (alternative platforms) | NVIDIA HGX B300, 8 GPUs, 1.8 TB/s NVLink | Gen5 NVMe, designed to workload |
| E3-G894-ZD3 | 8U air-cooled GPU server | 2× Intel Xeon 6 or 2× AMD EPYC 9005 (alternative platforms) | NVIDIA HGX B300, 8 GPUs, 1.8 TB/s NVLink | Gen5 NVMe, designed to workload |
技术规格
Platform
- Form factor
- 8U air-cooled GPU server
- Processor
- 2× Intel Xeon 6 or 2× AMD EPYC 9005 (alternative platforms)
- GPU / accelerator
- NVIDIA HGX B300, 8 GPUs, 1.8 TB/s NVLink
Storage & cooling
- Storage
- Gen5 NVMe, designed to workload
- Cooling
- Air cooled
规格说明
Processor options are alternative platforms, not a mixed configuration; each part number corresponds to one platform, confirmed on quotation. Memory, storage capacity, networking and power are configured to the workload.
咨询此配置Servers 中的其他产品
Wasp 1U
1U rack server · General compute
Sentry 1U
1U dual-socket rack server · General compute
Sentry 2U
2U dual-socket rack server · General compute
Globemaster 60
4U storage server · Storage
Harrier E2
Short-depth 2U edge server · Edge
Gripen-4
2U NVIDIA MGX GPU server · GPU / AI
Gripen-8
4U NVIDIA RTX PRO Server · GPU / AI
Lancer B300 LC
4U liquid-cooled GPU server · GPU / AI
Lancer MI355X
8U air-cooled GPU server · GPU / AI
Lancer MI355X LC
4U liquid-cooled GPU server · GPU / AI
Squadron NVL72
Complete liquid-cooled rack, 18 compute trays · Rack-scale AI

