Overview
The d-Matrix Corsair is a full-height full-length PCIe inference accelerator built around a two-tier memory system rather than HBM. Each card carries 2 GB of on-die performance memory running at 150 TB/s, backed by up to 256 GB of off-die capacity memory, and delivers 2,400 TFLOPS of MXINT8 and 9,600 TFLOPS of MXINT4. d-Matrix labels those figures "dense" itself, so the basis is stated rather than inferred, and it publishes no sparse figure. Cards are joined by a 512 GB/s card-to-card DMX Bridge. Note that d-Matrix publishes card figures only for what it calls the "Corsair Dual Card", a bridged pair; the per-card values here are that pair halved, which is corroborated nine separate ways by d-Matrix's own eight-card server and sixty-four-card rack figures, every one of which divides cleanly onto a single card. The 150 TB/s belongs to the on-die SRAM and is deliberately not recorded as memory bandwidth, which on this site means DRAM bandwidth. d-Matrix publishes no TDP, no process node and no PCIe generation for this card, so those fields are left blank rather than estimated.
Performance Metrics
Peak theoretical throughput by precision type
| Precision | Bits | Peak TFLOPS | |
|---|---|---|---|
| INT8 | 8 | 2400.0 | |
| INT4 | 4 | 9600.0 |
Power Specifications
TDP
--
Max Power
--
Power Connector
PCIe Slot
Cooling
Air
Memory Specifications
Capacity
256 GB
Type
--
Bandwidth
--
Interface
--
Hardware & Design
Form Factor
PCIe
Architecture
Corsair
Process Node
--
Launch Year
--
Variant
Standard
Market Segment
Professional
Full Specifications
| Memory | |
|---|---|
| VRAM | 256 GB |
| On-Die SRAM | 2.0 GB |
| Interconnect & I/O | |
| GPU-to-GPU | DMX Bridge |
| Interconnect Bandwidth | 512 GB/s |
| Power & Thermal | |
| General | |
| Form Factor | PCIe |
| Architecture | Corsair |
Documentation & Resources
Common Use Cases
The d-Matrix Corsair is optimized for high-performance computing tasks with Corsair architecture delivering high TFLOPS of compute power.
Where to Rent
Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.
Browse GPU Cloud ProvidersStay Updated on GPU Releases
Get notified when new GPUs are added or specifications are updated.
No spam, unsubscribe anytime.