NVIDIA
NVIDIA Tesla M40 24GB graphics card P8Y46A
Passive-cooled 24 GB GDDR5 compute GPU with 3,072 CUDA cores and 288 GB/s bandwidth for machine-learning workloads
- Memory InterfacePCI Express 3.0 x16
- 24 GB GDDR5 memory handles large datasets
- 3072 CUDA cores accelerate parallel compute workloads
- 288 GB/s bandwidth feeds data to the GPU fast
- 250 W passive cooling suits dense server racks
- PCIe 3.0 x16 interface connects to modern platforms
24GB Tesla M40 for compute
The NVIDIA Tesla M40 is a 24GB GDDR5 graphics card built on the GM200 GPU with 3072 CUDA cores and a 384-bit memory bus. It delivers 7 TFLOPS of single-precision performance and 0.21 TFLOPS of double-precision performance over a PCI Express 3.0 x16 interface. The card draws 250W TDP and uses an 8-pin CPU power connector.
Suits deep learning inference, not display
This accelerator targets machine learning inference and high-performance compute workloads that benefit from 24GB VRAM and 288 GB/s bandwidth. It has no video outputs and uses passive cooling, so it requires server chassis airflow and is unsuitable for workstations needing display connectivity or quiet desktop operation.
Fits dense GPU servers
The dual-slot passive heatsink relies on strong front-to-back system airflow, freeing internal space otherwise occupied by active coolers. The PCIe 3.0 x16 link and single 8-pin power connector simplify cabling in multi-GPU nodes, while the 250W TDP sets a clear thermal budget for rack power planning.
จุดเด่น
- 24 GB GDDR5 memory handles large datasets
- 3072 CUDA cores accelerate parallel compute workloads
- 288 GB/s bandwidth feeds data to the GPU fast
- 250 W passive cooling suits dense server racks
- PCIe 3.0 x16 interface connects to modern platforms
ข้อมูลจำเพาะ
| Brand | Nvidia |
|---|---|
| Model | Nvidia Tesla M40 |
| Memory Interface | PCI Express 3.0 x16 |
คำถามเกี่ยวกับสินค้านี้
Does the passive cooler require a server chassis with forced airflow?
Yes. The card has no fan and relies on high-volume front-to-back airflow from the chassis; typical workstation cases will not provide enough cooling.
What power connector and PSU capacity are needed?
The board draws 250 W and uses a single 8-pin CPU power connector; a quality PSU rated at least 250 W with that connector is required.
Can this card drive a display for local debugging?
No display outputs are present; the GPU is designed for headless compute workloads and cannot connect a monitor directly.
อื่น ๆ ในหมวดนี้
ลูกค้ามักเปรียบเทียบสินค้าเหล่านี้
หมวดเดียวกัน การชำระเงินแบบเดียวกัน — แปดสกุลเงินคริปโตและอัตราแลกเปลี่ยนที่ล็อกไว้ 30 นาที



