Blog

News

Exxact Offers NVIDIA RTX-Powered Systems

January 19, 2022
10 min read
blog-rtx-product-update.jpg

What's New with the NVIDIA RTX Product Line Update?

With remote work becoming more popular and demand for GPU-accelerated workloads growing, Exxact solutions powered by NVIDIA RTX™ GPUs pack powerful real-time ray tracing and AI acceleration capabilities, transforming design and visualization workflows for the most complex tasks, like designing airplanes and automobiles, visual effects in movies and large-scale architectural design. NVIDIA recently added four new products to their RTX™ family of professional GPUs. Most of them may be familiar and can already be found in Exxact workstations, servers and rendering systems, but they now pack twice the GPU memory as the previous generation. NVIDIA T400 bumped GPU memory from 2GB GDDR6 to 4GB GDDR6, NVIDIA T1000 from 4GB GDDR6 to 8GB GDDR6, and NVIDIA RTX A2000 from 6GB GDDR6 to 12 GB GDDR6. New to the NVIDIA RTX lineup is the NVIDIA RTX A4500 GPU, which packs 20GB GDDR6.

The increased GPU memory in the updates and new releases provides more performance by allowing more data to reside entirely on it, making it easier and faster for the graphics processor to access. Increased functionality is achieved by allowing users to run more applications simultaneously, use more plugins/tools, run higher fidelity calculations, and work with higher-resolution models and images. Greater productivity is also achieved since running multiple graphics/compute/AI intensive applications across multiple high-resolution displays is common, and significant GPU memory is required just to drive the displays.

Exxact deep learning workstations combine the benefits of increased GPU memory with state-of-the-art PCIe peer-to-peer interconnect technology and a fully pre-installed suite of deep learning software to provide groundbreaking performance to accelerate deep learning research. Exxact AI servers are better able to train massive deep neural networks, achieve state-of-the-art results, or run production grade AI applications. Exxact rendering systems boost rendering performance, accelerate offline rendering, and provision high-performance virtual workstations.

New Professional GPUs Powered by the NVIDIA Turing GPU Architecture

The NVIDIA T400 4GB and NVIDIA T1000 8GB are built on the NVIDIA Turing GPU architecture. The powerful, low-profile solutions deliver full-size features, performance, and capabilities in a compact package so professionals can tackle a range of multi-app workflows with ease. Helping Exxact workstation users create crisp, clear, lifelike images and special effects, Turing GPUs are highly compact, as well as cost- and power-efficient.

NVIDIA T1000 8GB Performance Advantage over NVIDIA T1000 4GB

Image Source: NVIDIA.com

NVIDIA T1000 8GB Performance Advantage over NVIDIA T600 4GB

Image Source: NVIDIA.com


Looking for a new NVIDIA Powered Workstation or Server?
Shop NVIDIA Systems


New Professional GPUs Powered by the NVIDIA Ampere GPU Architecture

The NVIDIA RTX A2000 12GB and NVIDIA RTX A4500 are built on the NVIDIA Ampere architecture, successor to the Turing architecture. These GPUs consist of thousands of CUDA cores versus the hundreds found in the above Turing-based GPUs, second-generation RT Cores for more visually accurate renders, third-generation Tensor Cores that allow Exxact AI workstations to boost AI and data science model training performance, and a PCI-E 4.0 x16 interface to improve data transfer speeds from CPU memory for data-intensive tasks.

NVIDIA RTX A2000 Graphics Performance Comparison

The NVIDIA RTX A2000 12GB GPU provides similar graphics performance compared to the NVIDIA RTX A2000 6GB, but the real advantage lies in increased functionality and higher fidelity, as well as 3D scene rendering where simpler looking scenes can take up more memory in the GPU.


NVIDIA RTX A2000 6GB vs 12GB Ansys Discovery Simulation Comparison

Image Source: NVIDIA.com

NVIDIA RTX A2000 6GB

NVIDIA RTX A2000 12GB

Simulation is run at lower resolution

Simulation is run at much higher resolution

Accuracy of the physics results limited by simulation resolution, which depends on GPU memory.

Higher GPU memory allows simulations at much higher fidelity, which leads to more accurate results with lower % error.

Many modern innovations such as space exploration, autonomous vehicles and 5G networks would not exist without simulation. Exxact is one of the six listed on-premises hardware partners on the Ansys website that have performed testing to determine what hardware runs Ansys simulations best. Exxact workstations featuring NVIDIA RTX GPUs for Ansys applications allow users to accelerate design cycles with advanced physics modeling and industry-leading accuracy, which makes choosing the appropriate GPU memory crucial.

NVIDIA RTX A2000 3D Scene Rendering Comparison

Image Source: NVIDIA.com

Exxact rendering systems with NVIDIA RTX GPUs are built to work with popular rendering and real-time rendering applications, such as Blender, Maya, 3ds Max, Unity, and more. A scene’s texture, lighting, shadows, and geometry, among other aspects, can easily fill up a GPU’s available memory, hence a larger GPU memory is greatly beneficial.

NVIDIA RTX A4500 Graphics Performance Comparison with RTX A4000

The NVIDIA RTX A4500 clearly outperforms its previous generation for rendering, AI training and inference, and workflow, as seen in the following comparisons.

NVIDIA RTX A4500 Rendering Performance Advantage over RTX A4000

Image Source: NVIDIA.com


NVIDIA RTX A4500 AI Training & Inference Advantage over RTX A4000

Image Source: NVIDIA.com


Exxact's deep learning infrastructure featuring NVIDIA RTX GPUs with increased memory significantly accelerates AI training, resulting in deeper insights in less time and faster time to ROI. A single Exxact inference server with increased GPU memory can replace multiple commodity CPU servers for deep learning inference applications and services, which reduces energy requirements and delivers both CapEx and OpEx savings.


NVIDIA RTX A4500 Workflow Comparison Example

Image Source: NVIDIA.com

The NVIDIA RTX A4500 GPU can realize a 30 percent performance advantage for AI training over the NVIDIA RTX A4000. Even a small performance advantage can result in huge productivity gains where training time can be reduced by minutes, hours, and even days.

Another advantage of the NVIDIA RTX A4500 over the NVIDIA RTX A4000 is NVLink support—touted as the world’s first high-speed GPU interconnect that offers a significantly faster alternative for multi-GPU systems than traditional PCIe-based solutions. NVLink helps future-proof a user’s investment by letting them expand GPU memory for even larger models and data sets of tomorrow.

To further help you choose which GPU to future proof for tomorrow’s challenges, the following chart shows comparisons to the previous generation line-up.


NVIDIA GPU Specifications Comparison

NVIDIA T400NVIDIA T600NVIDIA T1000NVIDIA RTX A2000NVIDIA RTX A4000NVIDIA RTX A4500NVIDIA RTX A5000NVIDIA RTX A6000
ArchitectureTuringTuringTuringAmpereAmpereAmpereAmpereAmpere
GPU Memory2/4 GB GDDR64 GB GDDR64/8 GB GDDR66/12 GB GDDR616 GB GDDR620 GB GDDR624 GB DDR648 GB GDDR6
ECC MemoryNoNoNoYesYesYesYesYes
CUDA Cores3846408963,3286,1447,1688,19210,752
Tensor Cores104192224256336
RT Cores



2648566484
SP perf1.09 TFLOPS1.71 TFLOPS2.5 TFLOPS8 TFLOPS19.2 TFLOPS23.7 TFLOPS27.8 TFLOPS38.7 TFLOPS
RT Core perf





15.6 TFLOPS37.4 TFLOPS46.2 TFLOPS54.2 TFLOPS75.6 TFLOPS
Tensor perf





63.9 TFLOPS153.4 TFLOPS189.2 TFLOPS222.2 TFLOPS309.7 TFLOPS
Max Power30W40W50W70W140W200W230W300W
Graphic BusPCI-E 3.0 x16PCI-E 3.0 x16PCI-E 3.0 x16PCI-E 4.0 x16PCI-E 4.0 x16PCI-E 4.0 x16PCI-E 4.0 x16PCI-E 4.0 x16
Connectors3x mDP 1.44x mDP 1.44x mDP 1.44x mDP 1.44x DP 1.44x DP 1.44x DP 1.44x DP 1.4
Form FactorSingle SlotSingle SlotSingle SlotDual SlotSingle SlotDual SlotDual SlotDual Slot
vGPU SoftwareN/AN/AN/A N/AN/AN/ANVIDIA RTX vWSNVIDIA RTX vWS
NVLinkN/AN/AN/A N/AN/A2x RTX A45002x RTX A50002x RTX A6000
Power ConnectorN/AN/AN/A N/A1x 6-pin PCIe1x 8-pin PCIe1x 8-pin PCIe1x 8-pin CPU

Final Thoughts

All NVIDIA RTX GPUs are certified with a broad range of professional applications, tested by leading independent software vendors and workstation manufacturers like Exxact, and backed by a global team of support specialists to give you peace of mind when it comes to a visual computing platform for your mission-critical business.

However, larger GPU memory boosts workflows that consist of multiple simultaneous applications across multiple high-resolution displays, and provide the required memory to run advanced graphics, AI-augmented tools, CUDA-accelerated applications, and rendering. The new GPUs offer great value in regard to time-to-solution, time-to-market, and meeting the demands of future challenges.



Have any questions?
Contact Exxact Today


blog-rtx-product-update.jpg
News

Exxact Offers NVIDIA RTX-Powered Systems

January 19, 202210 min read

What's New with the NVIDIA RTX Product Line Update?

With remote work becoming more popular and demand for GPU-accelerated workloads growing, Exxact solutions powered by NVIDIA RTX™ GPUs pack powerful real-time ray tracing and AI acceleration capabilities, transforming design and visualization workflows for the most complex tasks, like designing airplanes and automobiles, visual effects in movies and large-scale architectural design. NVIDIA recently added four new products to their RTX™ family of professional GPUs. Most of them may be familiar and can already be found in Exxact workstations, servers and rendering systems, but they now pack twice the GPU memory as the previous generation. NVIDIA T400 bumped GPU memory from 2GB GDDR6 to 4GB GDDR6, NVIDIA T1000 from 4GB GDDR6 to 8GB GDDR6, and NVIDIA RTX A2000 from 6GB GDDR6 to 12 GB GDDR6. New to the NVIDIA RTX lineup is the NVIDIA RTX A4500 GPU, which packs 20GB GDDR6.

The increased GPU memory in the updates and new releases provides more performance by allowing more data to reside entirely on it, making it easier and faster for the graphics processor to access. Increased functionality is achieved by allowing users to run more applications simultaneously, use more plugins/tools, run higher fidelity calculations, and work with higher-resolution models and images. Greater productivity is also achieved since running multiple graphics/compute/AI intensive applications across multiple high-resolution displays is common, and significant GPU memory is required just to drive the displays.

Exxact deep learning workstations combine the benefits of increased GPU memory with state-of-the-art PCIe peer-to-peer interconnect technology and a fully pre-installed suite of deep learning software to provide groundbreaking performance to accelerate deep learning research. Exxact AI servers are better able to train massive deep neural networks, achieve state-of-the-art results, or run production grade AI applications. Exxact rendering systems boost rendering performance, accelerate offline rendering, and provision high-performance virtual workstations.

New Professional GPUs Powered by the NVIDIA Turing GPU Architecture

The NVIDIA T400 4GB and NVIDIA T1000 8GB are built on the NVIDIA Turing GPU architecture. The powerful, low-profile solutions deliver full-size features, performance, and capabilities in a compact package so professionals can tackle a range of multi-app workflows with ease. Helping Exxact workstation users create crisp, clear, lifelike images and special effects, Turing GPUs are highly compact, as well as cost- and power-efficient.

NVIDIA T1000 8GB Performance Advantage over NVIDIA T1000 4GB

Image Source: NVIDIA.com

NVIDIA T1000 8GB Performance Advantage over NVIDIA T600 4GB

Image Source: NVIDIA.com


Looking for a new NVIDIA Powered Workstation or Server?
Shop NVIDIA Systems


New Professional GPUs Powered by the NVIDIA Ampere GPU Architecture

The NVIDIA RTX A2000 12GB and NVIDIA RTX A4500 are built on the NVIDIA Ampere architecture, successor to the Turing architecture. These GPUs consist of thousands of CUDA cores versus the hundreds found in the above Turing-based GPUs, second-generation RT Cores for more visually accurate renders, third-generation Tensor Cores that allow Exxact AI workstations to boost AI and data science model training performance, and a PCI-E 4.0 x16 interface to improve data transfer speeds from CPU memory for data-intensive tasks.

NVIDIA RTX A2000 Graphics Performance Comparison

The NVIDIA RTX A2000 12GB GPU provides similar graphics performance compared to the NVIDIA RTX A2000 6GB, but the real advantage lies in increased functionality and higher fidelity, as well as 3D scene rendering where simpler looking scenes can take up more memory in the GPU.


NVIDIA RTX A2000 6GB vs 12GB Ansys Discovery Simulation Comparison

Image Source: NVIDIA.com

NVIDIA RTX A2000 6GB

NVIDIA RTX A2000 12GB

Simulation is run at lower resolution

Simulation is run at much higher resolution

Accuracy of the physics results limited by simulation resolution, which depends on GPU memory.

Higher GPU memory allows simulations at much higher fidelity, which leads to more accurate results with lower % error.

Many modern innovations such as space exploration, autonomous vehicles and 5G networks would not exist without simulation. Exxact is one of the six listed on-premises hardware partners on the Ansys website that have performed testing to determine what hardware runs Ansys simulations best. Exxact workstations featuring NVIDIA RTX GPUs for Ansys applications allow users to accelerate design cycles with advanced physics modeling and industry-leading accuracy, which makes choosing the appropriate GPU memory crucial.

NVIDIA RTX A2000 3D Scene Rendering Comparison

Image Source: NVIDIA.com

Exxact rendering systems with NVIDIA RTX GPUs are built to work with popular rendering and real-time rendering applications, such as Blender, Maya, 3ds Max, Unity, and more. A scene’s texture, lighting, shadows, and geometry, among other aspects, can easily fill up a GPU’s available memory, hence a larger GPU memory is greatly beneficial.

NVIDIA RTX A4500 Graphics Performance Comparison with RTX A4000

The NVIDIA RTX A4500 clearly outperforms its previous generation for rendering, AI training and inference, and workflow, as seen in the following comparisons.

NVIDIA RTX A4500 Rendering Performance Advantage over RTX A4000

Image Source: NVIDIA.com


NVIDIA RTX A4500 AI Training & Inference Advantage over RTX A4000

Image Source: NVIDIA.com


Exxact's deep learning infrastructure featuring NVIDIA RTX GPUs with increased memory significantly accelerates AI training, resulting in deeper insights in less time and faster time to ROI. A single Exxact inference server with increased GPU memory can replace multiple commodity CPU servers for deep learning inference applications and services, which reduces energy requirements and delivers both CapEx and OpEx savings.


NVIDIA RTX A4500 Workflow Comparison Example

Image Source: NVIDIA.com

The NVIDIA RTX A4500 GPU can realize a 30 percent performance advantage for AI training over the NVIDIA RTX A4000. Even a small performance advantage can result in huge productivity gains where training time can be reduced by minutes, hours, and even days.

Another advantage of the NVIDIA RTX A4500 over the NVIDIA RTX A4000 is NVLink support—touted as the world’s first high-speed GPU interconnect that offers a significantly faster alternative for multi-GPU systems than traditional PCIe-based solutions. NVLink helps future-proof a user’s investment by letting them expand GPU memory for even larger models and data sets of tomorrow.

To further help you choose which GPU to future proof for tomorrow’s challenges, the following chart shows comparisons to the previous generation line-up.


NVIDIA GPU Specifications Comparison

NVIDIA T400NVIDIA T600NVIDIA T1000NVIDIA RTX A2000NVIDIA RTX A4000NVIDIA RTX A4500NVIDIA RTX A5000NVIDIA RTX A6000
ArchitectureTuringTuringTuringAmpereAmpereAmpereAmpereAmpere
GPU Memory2/4 GB GDDR64 GB GDDR64/8 GB GDDR66/12 GB GDDR616 GB GDDR620 GB GDDR624 GB DDR648 GB GDDR6
ECC MemoryNoNoNoYesYesYesYesYes
CUDA Cores3846408963,3286,1447,1688,19210,752
Tensor Cores104192224256336
RT Cores



2648566484
SP perf1.09 TFLOPS1.71 TFLOPS2.5 TFLOPS8 TFLOPS19.2 TFLOPS23.7 TFLOPS27.8 TFLOPS38.7 TFLOPS
RT Core perf





15.6 TFLOPS37.4 TFLOPS46.2 TFLOPS54.2 TFLOPS75.6 TFLOPS
Tensor perf





63.9 TFLOPS153.4 TFLOPS189.2 TFLOPS222.2 TFLOPS309.7 TFLOPS
Max Power30W40W50W70W140W200W230W300W
Graphic BusPCI-E 3.0 x16PCI-E 3.0 x16PCI-E 3.0 x16PCI-E 4.0 x16PCI-E 4.0 x16PCI-E 4.0 x16PCI-E 4.0 x16PCI-E 4.0 x16
Connectors3x mDP 1.44x mDP 1.44x mDP 1.44x mDP 1.44x DP 1.44x DP 1.44x DP 1.44x DP 1.4
Form FactorSingle SlotSingle SlotSingle SlotDual SlotSingle SlotDual SlotDual SlotDual Slot
vGPU SoftwareN/AN/AN/A N/AN/AN/ANVIDIA RTX vWSNVIDIA RTX vWS
NVLinkN/AN/AN/A N/AN/A2x RTX A45002x RTX A50002x RTX A6000
Power ConnectorN/AN/AN/A N/A1x 6-pin PCIe1x 8-pin PCIe1x 8-pin PCIe1x 8-pin CPU

Final Thoughts

All NVIDIA RTX GPUs are certified with a broad range of professional applications, tested by leading independent software vendors and workstation manufacturers like Exxact, and backed by a global team of support specialists to give you peace of mind when it comes to a visual computing platform for your mission-critical business.

However, larger GPU memory boosts workflows that consist of multiple simultaneous applications across multiple high-resolution displays, and provide the required memory to run advanced graphics, AI-augmented tools, CUDA-accelerated applications, and rendering. The new GPUs offer great value in regard to time-to-solution, time-to-market, and meeting the demands of future challenges.



Have any questions?
Contact Exxact Today