NVIDIA RTX A6000 (48GB)

Buying advice

The RTX A6000 is a premier professional accelerator featuring a massive 48GB of memory, specifically engineered for enterprise-scale AI training, large language model inference, and high-end 3D rendering. It stands as a top-tier solution for demanding high-performance computing workloads that require extreme reliability and vast data capacity.

Specifications

LAUNCH

Released: 2020

GAMING PERFORMANCE

Gaming Performance Index (GPI): Configuration onlyComino GRANDO RM-S / liquid-cooled RTX A6000: 104 Estimated · Three-anchor Time Spy proxy · Input years: 2023Physical desktop reference-style card; exact part number not disclosed: 184 Estimated · Three-anchor Time Spy proxy · Input years: unknownEstimated · No generic GPU scorePS5 reference: 2024 = 100 · Input years: unknown · Estimates apply only to the named device and mode; no generic GPU score. Standard PS5 reference = 100.

GRAPHICS BENCHMARKS

3DMark Score: -

MEMORY & STORAGE

Graphics Memory: 48 GB GDDR6
Memory Bandwidth: 768 GB/sVendor-specified memory bandwidth
Bus Width: 384 bit

ARCHITECTURE & COMPUTE

* FP32 describes floating-point throughput, not measured gaming performance. Architecture, clock and method differences affect comparison.

Architecture: Ampere
Process: 8 nm
Platform: Desktop
CUDA Cores: 10,752
FP32 TFLOPS*: Conditions unconfirmed38.7
Test conditions & details
Test: FP32
Score unit: TFLOPS
Version not recorded
Recorded test: Unknown
Bars align retained values of the same metric and unit. Test conditions may be unconfirmed. The endpoint is a display scale, not a test maximum, comparable device measurement, game FPS ratio or winner.
FP16 (AI) TFLOPS: -
Resolution & frame-rate reference: 1440p / 60 fps

POWER & BATTERY

Power: 300 W · Total board power

OTHER SPECIFICATIONS

Manufacturer: NVIDIA