RTX 4090 Specifications
Best raw FLOPS per dollar of any NVIDIA part, but 24 GB, no NVLink, no ECC, and licence terms restrict datacenter use.
Buy it for
Local experimentation and fine-tuning on a desktop. Best raw FLOPS per dollar of any NVIDIA part.
Look elsewhere if
You are deploying to a datacenter — NVIDIA's driver licence restricts GeForce in datacenters, and there is no ECC or NVLink.
Will it fit your system?
- Form factor
- PCIe triple-slot
- Cooling
- active
- Power connectors
- 1x 16-pin CEM5 (12VHPWR) or 3x 8-pin
- Board power
- 450 W
- Suggested PSU
- 1000 W or better
- Interface
- PCIe Gen 4
PSU figure assumes one card plus a mainstream CPU and drives, with headroom for transient spikes — modern cards can draw well above their rated TDP for milliseconds, which is what trips an undersized supply.
Supported formats & features
What the RTX 4090 does and does not support in hardware.
Blackwell-generation feature — B200 has it
E4M3 and E5M2 — the standard format for modern inference
Wider exponent than FP16 — fewer overflow issues when training
Drop-in FP32 speedup with no code changes
Quantised inference supported
Not yet confirmed against vendor documentation.
Precision must be managed manually
Multi-GPU traffic goes over PCIe — a real limit for distributed training
Hardware ray tracing for rendering workloads
No ECC, so silent corruption is possible on multi-day runs
What fits in memory
Largest model that fits on one RTX 4090, assuming ~20% of the 24 GB is reserved for KV cache and activations.
| Model | FP16 / BF16 | FP8 |
|---|---|---|
| Llama 8B | Fits | Fits |
| Llama 70B | 8× GPUs | 4× GPUs |
| Mixtral 8x22B | 15× GPUs | 8× GPUs |
| Llama 405B | 43× GPUs | 22× GPUs |
Weights only. Long contexts, large batch sizes and optimiser state during training all need substantially more — full fine-tuning typically needs 3–4× the weight memory.
Workload character
Plenty of bandwidth per unit of compute. Memory-bound work — single-stream token decode above all — runs close to its ceiling, and compute is usually what runs out first.
Cost per unit of compute
Breakeven counts hardware cost against rental only. Power, cooling, hosting and depreciation push the real crossover substantially later — see colocation pricing for what hosting actually costs.
Cost to run
Electricity for one card at 100% duty, including cooling and PSU overhead (5,519 kWh/year).
At the US average rate, three years of continuous operation costs $1,987 — roughly 124% of the card's $1,599 launch price, before any cooling or hosting fees.
Where it sits in the line
A newer part existing does not make this one a bad buy — it makes it cheaper. What matters is price per unit of the throughput you actually need.
Availability & market
Retail availability is normal. Read the licence terms before deploying commercially — this is the usual trap.
Memory
- Capacity
- 24 GB
- Type
- GDDR6X
- Bandwidth
- 1,008 GB/s
- Bus width
- 384-bit
- GB per TFLOPMemory relative to compute — higher favours large models
- 0.145
Compute
- FP64
- —
- FP32
- 82.6 TFLOPS
- TF32 dense
- —
- FP16 dense
- 165.2 TFLOPS
- FP16 sparseRequires structured sparsity; rarely reached in practice
- 330.3 TFLOPS
- FP8 dense
- 660 TFLOPS
- INT8 dense
- —
Silicon
- Codename
- AD102
- Process node
- TSMC 4N
- Transistors
- 76.3B
- Die size
- 609 mm²
- CUDA cores
- 16,384
- Tensor cores
- 512
- SM count
- 128
- Boost clock
- 2,520 MHz
Platform & power
- TDP
- 450 W
- TFLOPS per wattDense FP16 ÷ TDP — drives your power bill at scale
- 0.37
- Form factor
- PCIe triple-slot
- Cooling
- active
- Interconnect
- None
- Interconnect speed
- —
- PCIe generation
- Gen 4
- MIG
- No
- Launch
- 2022-10
- MSRP at launch
- $1,599
RTX 4090 — common questions
Answered from the specifications above. Questions this card's data cannot answer with confidence are left out rather than guessed.
- How much VRAM does the RTX 4090 have?
- The RTX 4090 has 24 GB of GDDR6X memory with 1,008 GB/s of bandwidth. In practice plan on roughly 19 GB for weights: KV cache, activations and fragmentation take the rest, and long contexts or large batches take considerably more.
- What size model can the RTX 4090 run?
- On a single RTX 4090, about 9B parameters at FP16, or roughly 19B quantised to FP8. Anything larger has to be split across several GPUs, which is slow on this card because it has no NVLink. These are weight-only figures — a long context window can consume tens of gigabytes on its own.
- Can you put the RTX 4090 in a desktop PC?
- Yes. The RTX 4090 is a self-cooled PCIe triple-slot card and works in a normal desktop, provided the case has clearance and the power supply is rated around 1000 W or better.
- How much power does the RTX 4090 use?
- The RTX 4090 is rated at 450 W board power. It is fed through 1x 16-pin CEM5 (12VHPWR) or 3x 8-pin. For a single-card system, a 1000 W supply or better leaves headroom for the CPU, drives and the transient spikes that trip undersized units.
- How much does it cost to run the RTX 4090?
- At the US commercial average of $0.12/kWh, running one RTX 4090 continuously costs about $662 a year in electricity. That figure includes cooling and power-supply overhead at a factor of 1.4 over board power, and over three years comes to roughly 124% of what the card cost at launch.
- Does the RTX 4090 support FP8?
- Yes. The RTX 4090 has hardware FP8 tensor support, so FP8 weights and activations run at full tensor-core rate — 660 dense TFLOPS. This is the format most current inference servers default to for large models.
- Does the RTX 4090 support FP4?
- No. The RTX 4090 predates hardware FP4. You can run 4-bit quantised weights on it and the memory saving is real, but they are dequantised before compute, so there is no arithmetic speedup.
- Does the RTX 4090 support INT8?
- Yes. The RTX 4090 supports INT8 acceleration. INT8 remains the most widely supported quantised inference format across frameworks.
- Does the RTX 4090 support BF16?
- Yes. The RTX 4090 supports bfloat16. Its wider exponent range makes training far less prone to the overflow that plagues FP16, which is why most current training recipes assume it.
- Does the RTX 4090 have NVLink?
- No. The RTX 4090 has no NVLink. Multi-GPU work goes over PCIe Gen 4, which is an order of magnitude slower and makes tensor parallelism across cards expensive.
- Does the RTX 4090 have RT cores?
- Yes. The RTX 4090 includes dedicated ray-tracing cores, so it is usable for rendering and visualisation as well as compute.
- Does the RTX 4090 have ECC memory?
- No. The RTX 4090 has no ECC. For short interactive work this rarely matters; for multi-week training runs it is a real risk, and it is one reason datacentre parts cost more than consumer ones.
- What replaced the RTX 4090?
- The RTX 5090 (2025) is its successor, with 32 GB of GDDR7 at 1,792 GB/s. It is about 1.3x faster in dense FP16. On the used market that gap is what sets a fair price for the RTX 4090.
- Is the RTX 4090 worth buying?
- Buy it for: Local experimentation and fine-tuning on a desktop. Best raw FLOPS per dollar of any NVIDIA part. Look elsewhere if: You are deploying to a datacenter — NVIDIA's driver licence restricts GeForce in datacenters, and there is no ECC or NVLink.