(GTC 2021) New NVIDIA RTX GPUs for Professionals (RTX A5000, RTX A4000, RTX A3000, RTX A2000)

In an AI-powered keynote (with a touch of HairWorks too), NVIDIA’s CEO announced yesterday in his anti-covid19 kitchen the new GPUs for professional needs. These new GPUs based on Ampere architecture are more or less the professional versions of gaming GeForce GPUs.

 
The high-end RTX A6000 (10572 cores, 48GB VRAM) has been announced last year (see this news), the new RTX pro GPUs are the RTX A5000, RTX A4000, RTX A3000 and RTX A2000 for desktops and laptops.

GTC — NVIDIA today announced a range of eight new NVIDIA Ampere architecture GPUs for next-generation laptops, desktops and servers that make it possible for professionals to work from wherever they choose, without sacrificing quality or time.

Using the new NVIDIA RTX™ GPUs, artists can create dazzling 3D scenes, designers can iterate on stunning building architecture in real time, and engineers can create groundbreaking products on any system.

“Hybrid work is the new normal,” said Bob Pette, vice president of Professional Visualization at NVIDIA. “RTX GPUs, based on the NVIDIA Ampere architecture, provide the performance for demanding workloads from any device so people can be productive from wherever they need to work.”

For desktops, the new NVIDIA RTX A5000 and NVIDIA RTX A4000 GPUs feature new RT Cores, Tensor Cores and CUDA® cores to speed AI, graphics and real-time rendering up to 2x faster than previous generations.

For professionals on the go needing thin and light devices, the new NVIDIA RTX A2000, NVIDIA RTX A3000, RTX A4000 and RTX A5000 laptop GPUs deliver accelerated performance without compromising mobility. They include the latest generations of Max-Q and RTX technologies and are backed by the NVIDIA Studio™ ecosystem, which includes exclusive driver technology that enhances creative apps for optimal levels of performance and reliability.

For the data center, there are the new NVIDIA A10 GPU and A16 GPU. The A10 provides up to 2.5x the virtual workstation performance of the previous generation for designers and engineers, while the A16 GPU provides up to 2x user density with lower total cost of ownership and an enhanced virtual desktop infrastructure experience over the previous generation.

Combined with NVIDIA RTX Virtual Workstation (vWS) and NVIDIA Virtual PC (vPC) software, the A10 and A16 GPUs deliver more power, memory and speed to boost any workflow, from graphics and AI to VDI, so users can focus more time on working efficiently and productively.

Complete press release is available HERE.

Let’s detail these new graphics cards.

 

NVIDIA RTX A5000 (desktop)

– RTX A5000 homepage

• GPU: GA102, process: Samsung 8nm
• CUDA cores: 8192
• SMs: 64
• RT cores: 64
• Tensor cores: 256
• TMUs: 256
• ROPs: 96
• Memory: 24GB GDDR6 with error-correcting code (ECC), 384-bit
• TDP: 230W
• FP32 performance: 27.8 TFLOPS

 

NVIDIA RTX A4000 (desktop)

– RTX A4000 homepage

• GPU: GA104, process: Samsung 8nm
• CUDA cores: 6144
• SMs: 48
• RT cores: 48
• Tensor cores: 192
• TMUs: 192
• ROPs: 64
• Memory: 16GB GDDR6 with error-correcting code (ECC), 256-bit
• TDP: 140W
• FP32 performance: 19.2 TFLOPS

 

NVIDIA RTX A5000 (notebook)

• GPU: GA102, process: Samsung 8nm
• CUDA cores: 6144
• SMs: 48
• RT cores: 48
• Tensor cores: 192
• TMUs: 192
• ROPs: 64
• Memory: 16GB GDDR6 with error-correcting code (ECC), 256-bit
• TDP: 80 – 165W
• FP32 performance: 21.7 TFLOPS

 

NVIDIA RTX A4000 (notebook)

• GPU: GA104, process: Samsung 8nm
• CUDA cores: 5120
• SMs: 40
• RT cores: 40
• Tensor cores: 160
• TMUs: 160
• ROPs: 64
• Memory: 8GB GDDR6 with error-correcting code (ECC), 256-bit
• TDP: 80 – 140W
• FP32 performance: 17.8 TFLOPS

 

NVIDIA RTX A3000 (notebook)

• GPU: GA104, process: Samsung 8nm
• CUDA cores: 4096
• SMs: 32
• RT cores: 32
• Tensor cores: 128
• TMUs: 128
• ROPs: 48
• Memory: 6GB GDDR6 with error-correcting code (ECC), 192-bit
• TDP: 60 – 130W
• FP32 performance: 12.8 TFLOPS

 

NVIDIA RTX A2000 (notebook)

• GPU: GA106, process: Samsung 8nm
• CUDA cores: 2560
• SMs: 20
• RT cores: 20
• Tensor cores: 80
• TMUs: 80
• ROPs: 32
• Memory: 4GB GDDR6 with error-correcting code (ECC), 128-bit
• TDP: 35 – 95W
• FP32 performance: 9.3 TFLOPS