Redefine What's Possible

AMD Instinct™ MI355X

Breakthrough AI performance with 288GB HBM3E and up to 2.8x training speedup. Purpose-built for frontier model development and massive-scale inference workloads.

AMD Instinct MI355X GPU

MI355X Specs

Our AMD Instinct™ MI355X features full-stack compatibility and zero lock-in for faster, frictionless deployment.

AMD Instinct MI355X GPU

MI355X Server Configuration

Our AMD Instinct MI355X clusters deliver consistent performance and scalability.

Whether you're developing frontier LLMs or deploying latency-sensitive inference applications, TensorWave's AMD MI355X cluster offers a seamless cost effective path from prototyping to production.

Media

Standard Node Config

Bare Metal

with optional managed Kubernetes & Slurm

Accelerators

8x AMD Instinct MI355X – Liquid Cooled

CPUs

2x AMD Turin 9575F

Memory

3 TiB DDR5 6000 MT/s

Local Storage

8x 3.84TB NVMe Drives+2x 893GB m.2 drives

Node Interconnects

3.2Tb/s

Peak Network Storage

50PB
Media

MI355X Server Configuration

Our AMD Instinct MI355X clusters deliver consistent performance and scalability.

Standard Node Config

Bare Metal

with optional managed Kubernetes & Slurm

Accelerators

8x AMD Instinct MI355X – Liquid Cooled

CPUs

2x AMD Turin 9575F

Memory

3 TiB DDR5 6000 MT/s

Local Storage

8x 3.84TB NVMe Drives+2x 893GB m.2 drives

Node Interconnects

3.2Tb/s

Peak Network Storage

50PB

Whether you're developing frontier LLMs or deploying latency-sensitive inference applications, TensorWave's AMD MI355X cluster offers a seamless cost effective path from prototyping to production.

AMD Instinct MI350 Series with Direct Liquid Cooling

The Cool Advantage

Sustained Efficiency and Performance Through Direct Liquid Cooling.

Sustain Peak Performance

Removing heat at the source, preventing throttling and keeping MI355X accelerators running at sustained peak performance.

Higher Density Efficiency

Liquid cooling enables denser racks with higher efficiency, smaller footprints, and lower total cost of ownership.

Improved System Reliability

Lower, more stable temperatures reduce thermal stress, improving hardware lifespan and overall system reliability.

Our purpose-built Direct Liquid Cooling infrastructure is a key engineering advantage that ensures the MI355X delivers its full potential.

AMD Instinct MI355X

Dual AMD Turin 9575F CPUs, 3TB DDR5 & 3Tb/s RoCEv2 (2x400Gb/s front-end + storage)

DIRECT LIQUID COOLING

Starting at
$2.95/GPU HR
Node Spec Summary
Memory: 288 GB HBM3E
Memory Bandwidth: 8Tb/s
Infinity Fabric™ Bandwidth Links: 7 x 153.6 GB/s
FP16: 2.52 PFLOPS | 5.03 PFLOPS with sparsity
INT8: 5.03 POPS | 10.1 POPS with sparsity

Related Blog Posts