CARDINALH200 SXM 141GB
Both versions have 141 GB per GPU; their power and integration requirements differ
SPECIFICATIONS
| Memory per accelerator | 141 GB HBM3e |
|---|---|
| Memory bandwidth | 4800 GB/s |
| Form factor | SXM module |
| Maximum accelerator power | 700 W |
Manufacturer specifications describe the product, not the condition of an offered unit. Checked 2026-09-08
COMPUTE SPECIFICATIONS
Manufacturer peak arithmetic per accelerator. These values do not predict application throughput or latency.
| Precision | Execution | Peak rate | Sparsity | Reference |
|---|---|---|---|---|
| FP64 | Vector | 34 TFLOPS | Dense | Source ↗ConditionsManufacturer peak per GPU; not sustained workload performance |
| FP64 | Tensor | 67 TFLOPS | Dense | Source ↗ConditionsManufacturer peak per GPU; not sustained workload performance |
| FP32 | Vector | 67 TFLOPS | Dense | Source ↗ConditionsManufacturer peak per GPU; not sustained workload performance |
| TF32 | Tensor | 989 TFLOPS | Structured sparse | Source ↗ConditionsManufacturer specification marked with sparsity; full GPU, not an NVL pair or MIG slice |
| BF16 | Tensor | 1979 TFLOPS | Structured sparse | Source ↗ConditionsManufacturer specification marked with sparsity; full GPU, not an NVL pair or MIG slice |
| FP16 | Tensor | 1979 TFLOPS | Structured sparse | Source ↗ConditionsManufacturer specification marked with sparsity; full GPU, not an NVL pair or MIG slice |
| FP8 | Tensor | 3958 TFLOPS | Structured sparse | Source ↗ConditionsManufacturer specification marked with sparsity; full GPU, not an NVL pair or MIG slice |
- Peak arithmetic rates are not token throughput, latency or a guarantee of application performance
- Structured-sparse rates require eligible sparsity and supported kernels; do not compare them directly with dense rates
- Dense low-precision tensor rates are not derived by halving sparse ratings
- Source labels H200 specifications preliminary
- INT8 omitted because both checked webpage and PDF incorrectly label integer rates TFLOPS; awaiting unambiguous unit provenance
PLATFORM AND SOFTWARE
- Requires a qualified SXM server platform
- Manufacturer page labels these specifications preliminary
- Validate the intended inference engine and precision on the quoted platform
- Confirm software entitlement separately from physical hardware
Manufacturer maximum accelerator rating; actual draw and supported limits depend on system
INTERCONNECT
NVLink
900 GB/s. Manufacturer aggregate bidirectional bandwidth; topology depends on platform
PCIe Gen5 x16
128 GB/s. Theoretical aggregate bidirectional interface bandwidth
MANUFACTURER SOURCES
NVIDIA H200 specifications ↗NVIDIA · Checked 2026-09-08NVIDIA H200 datasheet ↗NVIDIA · Checked 2026-09-08HOW CARDINAL HELPS
Share your quantity, destination and server model, if known. We help turn those requirements into a clearly specified request.
- Help confirm the exact configuration and platform requirements
- Request supplier records for condition, unit identifiers and available testing evidence
- Clarify included accessories, delivery and proposed warranty terms before you decide
Any additional testing or guarantees must be agreed in the written offer.