NVIDIA · Desktop / server GPU card
NVIDIA H100 PCIe 80GB
Priced item: NVIDIA H100 PCIe 80GB Tensor Core GPU; seller PCIe variant
Seller SKU GW-NVIDIA-H100-PCIE-80GB
Supplied unit: One graphics card in the exact advertised priced variant.
Free India shipping. Dispatch: Within 14 calendar days after cleared bank-transfer payment. This is a dispatch timeframe; delivery transit time and the estimated delivery date will be stated with your invoice before payment.
Delivery, warranty & returns
- Payment
- Bank transfer against invoice
- Estimated transit
- 14 working days after dispatch within India
- Warranty
- New items. Manufacturer warranty as stated on the invoice. Review the manufacturer warranty details supplied with your invoice before payment.
- Returns & exchanges
- Returns or exchanges of unused and unopened items may be requested within 30 days of delivery. The customer pays return shipping. Contact sales@gpuworld.ai with your order reference for return or exchange instructions. Approved refunds are issued within 30 working days after receiving and approving the returned item.
Review the return and refund conditions above and the full order terms before ordering. Your invoice will show the estimated delivery date before payment.
Read online order termsProduct overview
NVIDIA H100 PCIe 80GB Tensor Core GPU; seller PCIe variant. Graphics or compute card in the Hopper data center family. Supplied unit: One graphics card in the exact advertised priced variant. NVIDIA model-level reference specifications — reference memory / capacity: 80 GB HBM2e with ECC; interface: PCIeGen5 x16; form factor: Full-height/full-length dual-slot passive PCIe card. Confirm the exact OEM configuration and its host, power, cooling and interconnect requirements before ordering.
Configuration overview
Review the GPU for a supported server deployment, matching the board edition, slot dimensions, auxiliary power and cooling design. Confirm operating-system, driver and application support for the exact GPU before ordering.
Specifications & compatibility
NVIDIA REFERENCE SPECIFICATIONS
Technical specifications
Specifications describe the identified NVIDIA model or reference configuration. Board-partner designs, OEM firmware, laptop power limits and complete-system configurations can differ.
- GPU architecture
- Hopper
- CUDA cores
- 14,592
- Tensor cores
- 456 fourth-generation Tensor cores
- GPU memory
- 80 GB HBM2e with ECC
- Memory bus width
- 5,120-bit
- FP32 theoretical peak
- 51 TFLOPS
- FP64 theoretical peak
- 26 TFLOPS
- FP64 matrix/Tensor theoretical peak
- 51 TFLOPS
- GPU base / boost clocks
- 1,125 MHz base / 1,755 MHz boost, product-brief reference
- System interface
- PCIe Gen5 x16; Gen5 x8 and Gen4 x16 supported
- Maximum board power
- 350 W; minimum configurable limit 200 W
- Auxiliary power
- One PCIe 16-pin connector
- MIG
- Up to 7 isolated instances
- Cooling
- Passive; qualified system airflow required
- Form factor
- 10.5 in, full-height/full-length, dual-slot
- Memory bandwidth, final architecture specification
- 2,039 GB/s
- TF32 Tensor peak
- 378 TFLOPS dense / 756 TFLOPS sparse
- FP16 / BF16 Tensor peak
- 756 TFLOPS dense / 1,513 TFLOPS sparse
- FP8 Tensor peak
- 1,513 TFLOPS dense / 3,026 TFLOPS sparse
- NVLink bridge bandwidth
- Up to 600 GB/s between two compatible PCIe H100 GPUs
80 GB HBM2e PCIe is distinct from H100 NVL 94 GB HBM3 and from H100 SXM. Confirm OEM board revision and NVLink bridge requirements.
- Capacity / memory
- 80 GB HBM2e with ECC
- Interface
- PCIeGen5 x16
- Form factor
- Full-height/full-length dual-slot passive PCIe card
Exact hardware variant from NVIDIA product brief/datasheet. SXM module and PCIe add-in card are different products; confirm OEM part and host.
Documented hardware reference; exact supply condition and software support by quote
Manufacturer reference ↗ (opens in a new tab)Browse categories & filters
Filters & sort
Ordering the right configurationCompatibility, equipment condition and delivery details
Match the part to its host.
NVIDIA model names identify GPUs and product editions; they do not establish one universal board-vendor part number. Laptop GPUs are supplied within complete laptops. SXM modules, HGX/DGX systems and integrated CPU–GPU platforms require their supported host and interconnect.
Memory type, rank, transfer rate and population rules must match the server and CPU. A storage capacity does not establish NVMe, SATA or SAS compatibility. SSD and mechanical HDD media, firmware, sector format, carriers and security variants need separate checks.
For a precise quotation, send:
- Complete manufacturer/OEM part number and label, including suffix.
- Quantity, acceptable condition and delivery city/country.
- Server, motherboard, CPU or storage-backplane model.
- Application, required configuration and compatibility requirements.
Where a price is not displayed, call or WhatsApp for the current GST-inclusive price. Confirm equipment condition, quantity, warranty, shipping and dispatch before ordering.