GeForce RTX 5090
A 32 GB desktop card for local learning and supported small-model inference. The memory limit remains real even when the compute headline looks large.
GDDR7
Published peak, not measured application throughput
Laptop versions and desktop versions are different products. A card using four-bit weights still needs quantization metadata, cache and workspace; half a byte per parameter is only a floor.
When this is a sensible choice
Start here if…
Use a card you already own to learn serving, quantization and profiling. Pick a model that fits with its runtime and context; a short test is not evidence of stable long-context service.
Choose another configuration if…
For production, include power, cooling, physical slots, remote management and failure recovery in the comparison. A desktop card count is not equivalent to a managed multi-GPU server.
Specifications with their boundaries attached
- Architecture
- Blackwell
- Memory
- 32 GB GDDR7
- Memory bandwidth
- 1.79 TB/s per accelerator
- Peak compute
- Consumer AI TOPS are not normalized here against server BF16 peaks
- Scale-up interconnect
- No NVLink; motherboard PCIe topology matters
- Host attachment
- PCIe 5.0; board and lane allocation vary
- Power
- Board-specific; check exact vendor model
- Partitioning
- No server MIG guarantee
- Catalogue status
- Documented product
Compute figures are theoretical peaks at the stated precision. Dense and structured-sparse rates must not be mixed. Bandwidth labelled bidirectional combines both directions. See the source documents.
Three bandwidths, three different jobs
No NVLink; motherboard PCIe topology matters
InfiniBand, RoCE, EFA or provider-specific transport
Where it appears in provider documentation
| Provider / machine | Network scope | What changes the decision |
|---|---|---|
Runpod GPU type catalogue ↗ | Host-specific; require evidence for multi-node fabric | Server, workstation and Max-Q RTX PRO names differ. A marketplace GPU listing is not a topology guarantee. |
Model fit and software support
These publisher or serving-engine documents mention this hardware family. They have not been reproduced on our machines.
No model-specific recipe in our reviewed set certifies this exact hardware. The memory calculator can narrow candidates, but it cannot establish software support. Read the model register.
What is the memory floor?
Start with total parameters, then add the memory the workload needs. This arithmetic does not certify a serving configuration. All output sizes below are decimal GB.
Override device capacity with the memory exposed by your allocation, particularly for cloud B300 and partitioned devices. The starting 32 GB budget and 15% reserve are editable teaching assumptions. They are not measurements for the selected model. Mixed-precision tensors, quantization scales, vision encoders and draft models can increase the weight payload.
Raw weights only
320B × 8 bits ÷ 8
Budget per device
32 GB × (1 − 15%)
Arithmetic lower bound
round up ((weights + 32) ÷ budget)
This assumes perfectly balanced sharding. A result of three does not prove that a three-device parallel layout is supported. Check layer/expert divisibility, actual allocatable memory, precision kernels and the fabric before renting.
320B is the model card’s total; 18B active is not its storage size. The vLLM recipe reports about 306 GiB for the native FP8 checkpoint. Hopper requires BF16 KV for this model; the documented ROCm path targets gfx950, not every Instinct GPU. Read the model source ↗
For full training, also budget gradients, optimizer states, activations and communication buffers. The inference weight estimate above is insufficient.
Tokens per second, TTFT and TPOT are not certified for this hardware in our reference. Use the benchmark checklist to compare an exact model, software revision and workload.
Nearby memory capacities, different tradeoffs
These are comparison candidates selected by memory capacity, not performance rankings or drop-in replacements.
GeForce RTX 5080
A 16 GB desktop card for local learning and supported small-model inference. The memory limit remains real even when the compute headline looks large.
GeForce RTX 5070
A 12 GB desktop card for local learning and supported small-model inference. The memory limit remains real even when the compute headline looks large.
Sources and review scope
Manufacturer and cloud documentation checked 2026-09-12. We reviewed specifications and the stated product boundaries; we did not run training or serving benchmarks on this device. Cloud memory and availability can differ by configuration.
- NVIDIA GeForce specification comparison ↗ · checked 2026-09-12
Cite this reference
AI Infra Interviews. GeForce RTX 5090: specifications and workload fit. Reviewed . https://aiinfrainterviews.com/hardware/rtx-5090
Include your access date when citing a changing specification. Link to the specification section for hardware figures or the explanation for a sizing or workload decision.
Research method and limits
We compile manufacturer specifications, cloud documentation, model cards and serving recipes. The reference preserves source units and distinguishes individual devices from nodes and racks. Conflicting figures and unknown fields remain labelled.
Our contribution is the comparison, unit reconciliation, worked arithmetic and workload explanation. Published peaks are vendor specifications. Calculator results are estimates under the displayed assumptions. Neither is a measurement from our own accelerator lab.
For a manufacturer’s specification, consult the original source documents. Cite our page when using its analysis, and retain primary-source attribution for underlying figures. This curated reference does not establish market share, live availability or a universal performance ranking.
Found a discrepancy? Send a correction with the page URL, exact variant, disputed figure and supporting primary document.