8 new pro-grade NVIDIA GPUs tackle massive demand for 3D design collaboration capable systems
8 new pro-grade NVIDIA GPUs tackle massive demand for 3D design collaboration capable systems
With the rise of digital artists and creators designing new virtual photorealistic worlds, be it for a new building project, a virtual concert experience or even aiding the creation of the metaverse, 3D design collaboration is now exploding. Coupled with hybrid work environment needs and re-kitting the ‘office’ with suitable work tools, the professional visualization business is a key growth area for NVIDIA.
Today, NVIDIA launched a slew of mobile and desktop professional visualization class workstation GPUs and graphics cards to meet this demand. While NVIDIA has previously launched Ampere architecture based RTX A4000 and A5000 and for both desktops and mobile, they weren’t enough in addressing the market demand and overcoming the parts shortage. For example, the jump between the RTX A5000 and RTX A6000 was a very big leap in cost and many might not be ready for the top of the line model, but appreciate a little extra processing power and more importantly, the right frame buffer size to get the job done. This is why NVIDIA is launching the RTX A5500 and RTX A4500 Is sit in between their current desktop pro-viz lineup and compared to the top of the line counterpart from the previous generation:-
Graphics Card | RTX A6000 | RTX A5500 | RTX A5000 | RTX A4500 | RTX A4000 | Quadro RTX 6000 |
---|---|---|---|---|---|---|
GPU | Ampere (GA102) |
Ampere (GA102) |
Ampere (GA103) |
Ampere (GA104) |
Ampere (GA104) |
Turing (TU102) |
Process | 8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
12nm FinFET |
Die Size (mm2) | 628 | 628 | 628 | 392 | 392 | 754 |
Transistors | 28 billion | 28 billion | 28 billion | 17.4 billion | 17.4 billion | 18.6 billion |
CUDA cores | 10752 | 10240 | 8192 | 7168 | 6144 | 4608 |
Tensor Cores | 336 | 320 | 256 | 224 | 192 | 576 |
Tensor Performance | 309.7 TFLOPS | 272.8 TFLOPS | 222.2 TFLOPS | 189.2 TFLOPS | 153.4 TFLOPS | 130 TFLOPS |
RT Cores | 84 | 80 | 64 | 56 | 48 | 72 |
RT Performance | 58 TFLOPS | 66.6 TFLOPS | 54.2 TFLOPS | 46.2 TFLOPS | 37.4 TFLOPS | ? |
GPU base / boost clock speeds | 1455MHz / 1860MHz | - | - | - | - | 1400MHz / 1770MHz |
Memory | 48GB GDDR6 with ECC | 24GB GDDR6 with ECC | 24GB GDDR6 with ECC | 20GB GDDR6 with ECC | 16GB GDDR6 with ECC | 24GB GDDR6 |
Memory clock speed | 2.0Gbps | 2.0Gbps | 2.0Gbps | 2.0Gbps | 1.75Gbps | 14,000MHz |
Memory bus width | 384-bit | 384-bit | 384-bit | 320-bit | 256-bit | 384-bit |
Memory bandwidth | 768GB/s | 768GB/s | 768GB/s | 640GB/s | 448GB/s | 672GB/s |
TDP | 300W | 230W | 230W | 200W | 140W | 295W |
Price | US$4,694 | -- | -- | -- | -- | US$6,300 |
Besides qualifying more silicon for different SKUs, a differentiated board design further helps optimize board design components sourced and overall cost of the graphics cards. The result is that even early customers of the RTX A5500 were delighted to get their hands on the much needed new SKUs to get the job done:-
My team can handle larger CAD and CAE datasets than before and, critically, we can interact with and iterate these larger datasets simultaneously by making use of the extra GPU memory headroom and compute capabilities of this new card -- Brian Rotty, senior engineer at M4 Engineering.
The performance we get with the NVIDIA RTX A5500 is unprecedented. Previously, our point cloud processing took too long and created a bottleneck for designing and analyzing current site conditions. Using the NVIDIA RTX A5500, in less than 20 minutes we can integrate a fully processed, geolocated, classified point cloud into our 3D models -- Hiram Rodriguez, design technology specialist at AS+GG Architecture.
According to NVIDIA, the new NVIDIA RTX A5500 desktop GPU combines the latest-generation RT Cores, Tensor Cores and CUDA cores with 24GB of memory for incredible rendering performance that’s up two times faster than the previous generation and its motion blur rendering performance is up to 9x faster.
For creators on the-the-move
Over on the mobile workstation laptop solutions, NVIDIA fielded a much larger array of new RTX A-series pro-viz GPUs to better satiate the needs of the modern creator on the move with various SKUs to match different price points and the needs of their work. With the RTX A5550, A4500, A3000 12GB, A2000 8GB, A1000 and the A500 laptop GPUs, NVIDIA’s bringing more professional GPU options that boast all of their latest hardware features and capabilities of the Ampere architecture to creator laptops. Some of these capabilities include taking advantage of second-gen RT cores for ray-tracing and pixel shading needs, third-gen Tensor cores for massive boost in AI inferencing and deep learning tasks, and taking advantage of the more advanced GPU process technology with speedier GDDR6 memory and speedy PCIe 4.0 connectivity.
Here's how the new mobile GPU parts sit within the existing lineup:-
Graphics Card | RTX A5500 Laptop | RTX A5000 Laptop | RTX A4500 Laptop | RTX A4000 Laptop |
---|---|---|---|---|
GPU | Ampere | Ampere (GA104) |
Ampere (GA104) |
Ampere (GA104) |
Process | 8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
CUDA cores | 7424 | 6144 | 5888 | 5120 |
Tensor Cores | 232 | 192 | 184 | 160 |
Tensor Performance | 197.8 TFLOPS | 174 TFLOPS | 148.4 TFLOPS | 142.5 TFLOPS |
RT Cores | 58 | 48 | 46 | 40 |
RT Performance | - | 75.6 TFLOPS | - | 34.8 TFLOPS |
Memory | 16GB GDDR6 | 16GB GDDR6 | 16GB GDDR6 | 8GB GDDR6 |
Memory clock speed | 2.0Gbps | 1.75Gbps | 2.0Gbps | 1.5Gbps |
Memory bus width | 256-bit | 256-bit | 256-bit | 256-bit |
Memory bandwidth | 512GB/s | 448GB/s | 512GB/s | 384GB/s |
TGP | - | 80 - 165W | - | 80 - 140W |
Graphics Card | RTX A3000 12GB Laptop | RTX A3000 Laptop | RTX A2000 8GB Laptop | RTX A2000 Laptop | RTX A1000 Laptop | RTX A500 Laptop |
---|---|---|---|---|---|---|
GPU | Ampere (GA106) |
Ampere (GA106) |
Ampere (GA107) |
Ampere (GA107) |
Ampere (GA107) |
Ampere (GA107) |
Process | 8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
8nm (Samsung) |
CUDA cores | 4096 | 4096 | 2,560 | 2,560 | 2,048 | 2,048 |
Tensor Cores | 128 | 128 | 80 | 80 | 64 | 64 |
Tensor Performance | 113 TFLOPS | 102 TFLOPS | 74 TFLOPS | 74 TFLOPS | 59.7 TFLOPS | 58.2 TFLOPS |
RT Cores | 32 | 32 | 20 | 20 | 16 | 16 |
RT Performance | - | 25 TFLOPS | - | 18.2 TFLOPS | - | - |
Memory | 12GB GDDR6 | 6GB GDDR6 | 8GB GDDR6 | 4GB GDDR6 | 4GB GDDR6 | 4GB GDDR6 |
Memory clock speed | 1.75Gbps | 1.375Gbps | 1.75Gbps | 1.5Gbps | 1.75Gbps | 875Mbps |
Memory bus width | 192-bit | 192-bit | 128-bit | 128-bit | 128-bit | 128-bit |
Memory bandwidth | 336GB/s | 264GB/s | 224GB/s | 192GB/s | 224GB/s | 112GB/s |
TGP | - | 60 -130W | - | 35 - 95W | - | - |
The new NVIDIA RTX A5500 desktop GPUs are available today from channel partners and through global system builders starting in Q2.
The new NVIDIA RTX laptop GPUs will be available in mobile workstations from global OEM partners including Acer, ASUS, BOXX Technologies, Dell Technologies, HP, Lenovo, and MSI starting this spring. You can contact these partners for specific system configuration details and availability.
Here are all the other key highlights from NVIDIA GTC 2022
- NVIDIA’s 80-billion transistor H100 GPU and Hopper Architecture will drive world’s AI Infrastructure
- Here are NVIDIA's key healthcare AI advancements for medical computing
- NVIDIA unveils Drive Hyperion 9 autonomous vehicle platform, targeted for use in 2026 cars
- NVIDIA brings the power of Omniverse to the cloud and empowers millions to collaborate through any device
- NVIDIA's new Jetson AGX Orin dev-kit spurs AI development at the edge
- Grace superchip announced, coming in 2023