NVIDIA
NVIDIA Tesla P4 8GB GDDR5 Low-Profile graphics card
Low-profile GPU with 8 GB GDDR5 memory, 2560 CUDA cores and 192 GB/s bandwidth for data-centre inference
- Memory TypeGDDR5
- 8 GB GDDR5 memory handles large inference datasets
- 5 TFLOPS single-precision performance accelerates compute workloads
- 22 TOPS INT8 throughput speeds up deep-learning inference
- 75 W max power with passive heatsink fits dense servers
- Low-profile PCIe form factor enables space-efficient deployment
NVIDIA Tesla P4 8GB GDDR5 Low-Profile
The NVIDIA Tesla P4 is an 8GB GDDR5 accelerator built on the Pascal architecture for inference workloads. It delivers 5.5 teraFLOPS single-precision performance and 22 TOPS for INT8 operations. A passive heatsink and 75W maximum power draw suit dense server deployments.
Data-centre inference servers needing low-profile
This card targets rack-mounted inference serving where space and airflow are constrained. The low-profile PCIe form factor fits 1U chassis, while ECC memory and page migration engine support sustained operation. Workloads requiring active cooling, display outputs, or higher compute density should evaluate alternative accelerators.
75W passive low-profile PCIe form factor
The 75W thermal envelope enables operation from slot power alone without auxiliary connectors. Passive cooling eliminates fan failure points and reduces acoustic load in dense racks. Verify that the host chassis provides sufficient airflow across the heatsink for continuous 75W dissipation.
จุดเด่น
- 8 GB GDDR5 memory handles large inference datasets
- 5 TFLOPS single-precision performance accelerates compute workloads
- 22 TOPS INT8 throughput speeds up deep-learning inference
- 75 W max power with passive heatsink fits dense servers
- Low-profile PCIe form factor enables space-efficient deployment
ข้อมูลจำเพาะ
| Memory Type | GDDR5 |
|---|
คำถามเกี่ยวกับสินค้านี้
Will this card fit in a low-profile server chassis?
Yes, the card uses a low-profile PCI Express form factor and a passive heatsink, so it fits in standard low-profile server slots without needing extra clearance for fans.
What external power cables are required?
None. The card draws up to 75 W entirely from the PCIe slot, so no auxiliary power connectors are needed.
Can it run inference workloads with 8-bit precision?
Yes. The Pascal architecture delivers 22 TOPS of INT8 performance, making it suitable for 8-bit inference tasks.
อื่น ๆ ในหมวดนี้
ลูกค้ามักเปรียบเทียบสินค้าเหล่านี้
หมวดเดียวกัน การชำระเงินแบบเดียวกัน — แปดสกุลเงินคริปโตและอัตราแลกเปลี่ยนที่ล็อกไว้ 30 นาที



