NVIDIA RTX 6000 Ada Generation 48GB Specs, Benchmarks & Pricing
The NVIDIA RTX 6000 Ada Generation is the flagship of NVIDIA's Ada Lovelace professional workstation lineup, released on December 3, 2022. Built on the Ada Lovelace architecture using TSMC's 4nm process (4N), the GPU features 76.3 billion transistors on a 608.4mm² die. It combines 18,176 CUDA cores, 568 fourth-generation Tensor Cores, and 142 third-generation RT Cores with 48GB of GDDR6 ECC memory. The fourth-generation Tensor Cores deliver 728.5 TOPS of dense INT8 performance (1,457 TOPS with sparsity), providing up to 2X the AI performance over the previous RTX A6000 generation. With 91.1 TFLOPS of FP32 performance and 210.6 TFLOPS of RT core performance, the RTX 6000 Ada delivers 2-4X the performance of its predecessor while maintaining the same 300W TDP. Memory bandwidth of 960 GB/s is provided through a 384-bit memory interface running at 20 Gbps effective speed. The card features a dual-slot form factor with active blower cooling and supports PCIe 4.0 x16 interface with four DisplayPort 1.4a outputs.
- Release Date: December 3, 2022
- MSRP: $6,800 USD
- GPU Architecture: ada
- Hardware-Accelerated GEMM Operations:FP64 FP32 TF32 BF16 FP16 FP8 FP4 INT8 INT4
- CUDA Compute Capability : 8.9
Strengths
- Excellent FP32 compute performance (top 11% of GPUs)
- Excellent FP16 compute performance (top 15% of GPUs)
- Strong 3DMark Wild Life Extreme (4K) performance (top 15% of GPUs)
- Strong 3DMark Wild Life Extreme (1440p) performance (top 6% of GPUs)
Specifications for NVIDIA RTX 6000 Ada Generation
| Specification | Performance Ranking |
|---|---|
| FP32 TFLOPs | |
| FP16 TFLOPs | |
| Tensor Core Count | |
| Memory Capacity (GB) | |
| Memory Bandwidth (GB/s) | |
| Int8 TOPs | |
| FP8 TFLOPs | |
| FP4 TFLOPs | Data not available |
Gaming Benchmarks for NVIDIA RTX 6000 Ada Generation
Rankings based on 841 real-world benchmark results from OpenBenchmarking.org
| Benchmark | Performance Ranking |
|---|---|
| 3DMark Wild Life Extreme (4K) | |
| 3DMark Wild Life Extreme (1440p) |
Real-time NVIDIA RTX 6000 Ada Generation GPU Prices
Compare Price/Performance to other GPUs
Compare NVIDIA RTX 6000 Ada Generation to Another GPU
Price History
NVIDIA RTX 6000 Ada Generation Price History
This chart tracks NVIDIA RTX 6000 Ada Generation prices over the past 6 months based on eBay listings. The yellow line shows the average of the three lowest-priced listings each day—a useful indicator for finding deals.
Trend: Over this period, the lowest average price has been up 60.8%. The current lowest average is $9,299, compared to a typical $6,713 over the period. The lowest average price is calculated as the average of the three lowest-priced listings each day.
Product Identifiers
Manufacturer Part Numbers (4)
- NVIDIA Part Number
- 900-5G133-2250-000
- NVIDIA Part Number
- 900-5G133-2550-000
- NVIDIA Part Number
- 900-5G133-0150-001
- NVIDIA Part Number
- 900-5G133-2750-001
Available from 4 Partners (4 products)
- NVIDIA RTX 6000 Ada Graphics Card 48GB GDDR6
- VCNRTX6000ADA-PB(sku)
- Dell NVIDIA RTX 6000 Ada Generation, 48 GB GDDR6, full height, PCIe 4.0x16, 4 DP Graphics Card
- 490-BJMS(part number)
- HP NVIDIA RTX 6000 Ada Generation Graphics Card
- 79C23AA(part number)
- ThinkSystem NVIDIA RTX 6000 Ada 48GB PCIe Active GPU
- 4X67A89324(part number)
References
- https://www.nvidia.com/en-us/products/workstations/rtx-6000/
- https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/rtx-6000/proviz-print-rtx6000-datasheet-web-2504660.pdf
- https://images.nvidia.com/aem-dam/en-zz/Solutions/technologies/NVIDIA-ADA-GPU-PROVIZ-Architecture-Whitepaper_1.1.pdf
- https://www.leadtek.com/eng/products/workstation_graphics(2)/NVIDIA_RTX_6000_Ada_Generation(40949)/detail
- https://nvidianews.nvidia.com/news/nvidias-new-ada-lovelace-rtx-gpu-arrives-for-designers-and-creators
- https://lenovopress.lenovo.com/lp1940-thinksystem-nvidia-rtx-6000-ada-48gb-pcie-active-gpu
- https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/pdf/NVIDIA-RTX-Blackwell-PRO-GPU-Architecture-v1_1.pdf
Notes
- FP32 vector (CUDA core) throughput of 91.1 TFLOPS is the official peak single-precision performance published by NVIDIA.
- FP16 TFLOPS of 728.5 represents Peak Sparse Tensor Core performance (derived as half of FP8 Sparse performance). This metric is selected to align with the card's workstation AI capabilities; standard CUDA Core FP16 performance is 91.1 TFLOPS.
- int8TOPS of 728.5 represents non-sparse (dense) INT8 performance, derived from sparse INT8 of 1457 TOPS ÷ 2 per NVIDIA Ada Lovelace architecture. With 2:4 structured sparsity, INT8 performance is 1457 TOPS. Dense value used for consistent cross-vendor comparison.
- fp8TFLOPS of 728.5 represents non-sparse (dense) FP8 performance, derived from the peak FP8 tensor performance of 1,457 TFLOPS with sparsity (implied by this file's pre-existing fp16TFLOPS note, "FP16 TFLOPS represents ... half of FP8 Sparse performance") ÷ 2. NVIDIA's Ada Lovelace 4th-generation Tensor Cores process FP8 and INT8 at identical per-clock throughput (both are 8-bit formats), so dense FP8 (728.5) numerically equals dense INT8 (728.5). Dense value used for consistent cross-vendor comparison.
- Ada Lovelace has no FP4 hardware path (FP4 tensor core support was introduced in Blackwell) — fp4TFLOPS is left null and is a confirmed hardware negative, not a research gap, so it is not added to unverifiedSpecs.
- Memory bandwidth calculated as 384-bit × 20 Gbps ÷ 8 = 960 GB/s.
- CUDA Compute Capability 8.9 confirmed for Ada Lovelace architecture.
- The GPU uses the AD102 chip with 142 of 144 SMs enabled (18,176 of 18,432 CUDA cores).
- TDP is 300W, same as previous generation RTX A6000 despite 2-4X higher performance.
- MSRP of $6,800 USD from NVIDIA official page, confirmed by multiple sources at launch.
- Release date December 3, 2022 confirmed by multiple sources. Announced September 20, 2022.
- supportedHardwareOperations corrected during the 2026-09-02 Phase 5 audit to remove FP32 and FP64: per the canonical Ada Lovelace supportedHardwareOperations table (locked precision-filters plan definition, "hardware-accelerated matrix/tensor-core path support only"), FP32 is not a standalone Ada Tensor Core format — it only appears as plain shader/CUDA-core ALU throughput or as the TF32 reduced-precision tensor proxy — and no Ada SKU (including the RTX 6000 Ada) has a documented FP64 Tensor Core path (unlike datacenter Ampere/Blackwell); the NVIDIA Ada GPU Architecture whitepaper does not mention an FP64 Tensor Core row anywhere. Canonical set is FP8, FP16, BF16, TF32, INT8, INT4, uniform across all Ada SKUs (GeForce, professional, and datacenter).
- fp32TFLOPS of 182.1 is TF32 Tensor Core throughput without sparsity, per Table 4 of NVIDIA's RTX PRO Blackwell architecture whitepaper (https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/pdf/NVIDIA-RTX-Blackwell-PRO-GPU-Architecture-v1_1.pdf), which lists the RTX 6000 Ada at 182.1 TFLOPS dense TF32 and 91.1 TFLOPS FP32 vector.