Loading the catalog…
ULTRASTACK 30 - AMD MI350P (Liqid)
Blurred values are subscriber-only. Sign in to start a free trial — 2 full pages a day — or subscribe for unlimited access.
| Max Gpu Count | 8 |
|---|---|
| Description | |
| Key Feature | |
| Onboard Network | |
| Product Name |
| SKU | ULTRASTACK 30 - AMD MI350P |
|---|
| Tagline | |
|---|---|
| GPU count (headline) | |
| Aggregate HBM (headline) | |
| FP8 performance (headline) | |
| Scale-up description (marketing) |
| GPU limits of traditional servers (marketing) |
|---|
| CPU platform (marketing) |
|---|
| Tokenomics - tokens per second |
|---|
| Tokenomics - tokens per dollar |
|---|
| Tokenomics - tokens per watt |
|---|
| Deployment cost reduction (marketing) |
|---|
| Power reduction (marketing) |
|---|
| Why Liqid |
|---|
| Key Advantage - Breakthrough Scale-Up Density |
|---|
| Key Advantage - Superior Tokenomics |
|---|
| Key Advantage - Lower Cost & Power |
|---|
| Key Advantage - CXL memory-pooling ready |
|---|
| Built For |
|---|
| Ideal for |
|---|
| Node type (headline) |
|---|
| Manufacturer |
|---|
| Manufacturer address |
|---|
| Contact - office phone |
|---|
| Contact - email |
|---|
| Patent No. |
|---|
| Datasheet revision date |
|---|
| Host Server |
|---|
| Management Appliance |
|---|
| GPUs |
|---|
| AI Performance |
|---|
| GPU HBM Memory |
|---|
| GPU Pooling Fabric |
|---|
| Expansion Chassis |
|---|
| NVMe Storage |
|---|
| Orchestration |
|---|
| Memory Expansion |
|---|
| Total System Power |
|---|
| Performance per kW |
|---|
| HBM per kW |
|---|
| HBM per PFLOP |
|---|
| Efficiency Proof Points footnote |
|---|
| System composition diagram |
|---|
| Key Feature |
|---|
| Key Feature |
|---|
| Key Feature |
|---|
| Key Feature |
|---|
| Key Feature |
|---|
| Key Feature |
|---|
