NVIDIA RTX A1000 & RTX 400 GPUs are small
Following the release of RTX 2000 ADA, NVIDIA has new graphics cards featuring even lower specs.
NVIDIA is launching new Ampere-based RTX GPUs for professional use. The new models are called RTX A1000 and RTX 400, and both share a minimalist design. These models arrive two and a half years after the RTX A2000 12GB, which was the last model for the Ampere series, and nearly three years after the previous major update for Ampere RTX workstation models.
As shown in official renders, the GPUs feature a low-profile and single-slot design. In fact, both cards do not require any external power connections as their power is limited to 50W. This is the lowest TDP among all modern NVIDIA GPUs for the ProViz market.
The RTX A1000 appears to be based on the GA106 GPU and offers 2304 CUDA cores, 72 Tensor Cores, and 18 RT Cores. It can deliver up to 6.74 TFLOPS of FP32 compute performance, which translates to a boost clock of 1463 MHz. The card is equipped with 8GB of GDDR6 memory across a 128-bit memory bus. Officially, the card has a maximum bandwidth of 192 GB/s thanks to the 12 Gbps memory speed.
RTX 400 and RTX A1000, Source: NVIDIA
NVIDIA has not confirmed the GPU for the RTX A400, but it seems to be GA107 based while offering 768 CUDA cores, 24 Tensor Cores, and 6 RT cores. This model has 2.7 TFLOPS of FP32 compute, which translates to around a 1758 MHz clock. The memory capacity is further reduced on this card to 4GB of GDDR6 with 64-bit memory bus specs and 96 GB/s bandwidth.
Next-Generation Features and Accelerated Performance
The NVIDIA RTX A400 and A1000 GPUs are equipped with features designed to supercharge everyday workflows, including:
- Second-generation RT Cores: Real-time ray tracing, photorealistic, physically based rendering and visualization for all professional workflows, including architectural drafting, 3D design and content creation, where accurate lighting and shadow simulations can greatly enhance the quality of work.
- Third-generation Tensor Cores: Accelerates AI-augmented tools and applications such as generative AI, image rendering denoising and deep learning super sampling to improve image generation speed and quality.
- Ampere architecture-based CUDA cores: Up to 2x the single-precision floating point throughput of the previous generation for significant speedups in graphics and compute workloads.
- 4GB or 8GB of GPU memory: 4GB of GPU memory with the A400 GPU and 8GB with the A1000 GPU accommodate a range of professional needs, from basic graphic design and photo editing to more demanding 3D modeling with textures or high-resolution editing and data analyses. The GPUs also feature increased memory bandwidth over the previous generation for quicker data processing and smoother handling of larger datasets and scenes.
- Encode and decode engines: With seventh-generation encode (NVENC) and fifth-generation decode (NVDEC) engines, the GPUs offer efficient video processing to support high-resolution video editing, streaming and playback with ultra-low latency. Inclusion of AV1 decode enables higher efficiency and smoother playback of more video formats.
Unlike ADA-based models, these cards lack AV1 encoding support; however, they do support decoding and encoding for other codecs like H264 or H265. NVIDIA has not confirmed pricing for these cards, but it is confirmed that two board partners (PNY and Ryoyo Electric) will be selling them. The RTX A1000 should already be available, while the RTX 400 will start shipping next month.
| NVIDIA RTX Workstation GPUs | |||||
|---|---|---|---|---|---|
| VideoCardz.com | Picture | GPU | CUDA Cores | Memory | TGP |
| ADA LOVELACE | |||||
| RTX 6000 ADA |  | AD102 | |||
| RTX 5880 ADA | | AD102 | |||
| RTX 5000 ADA |  | AD104 | |||
| RTX 4500 ADA | | AD104 | |||
| RTX 4000 ADA |  | AD104 | |||
| RTX 4000 SFF ADA |  | AD104 | |||
| RTX 2000 ADA | | AD107 | |||
| AMPERE | |||||
| RTX A6000 |  | GA102 | |||
| RTX A5500 |  | GA102 | |||
| RTX A5000 | | GA102 | |||
| RTX A4500 | | GA102 | |||
| RTX A4000 |  | GA104 | |||
| RTX A2000 |  | GA106 | |||
| RTX A1000 |  | GA106 | |||
| RTX A400 | | GA107 | |||
Source: NVIDIA
