NVIDIA Workstation GPU Buyers Guide
What is an NVIDIA workstation GPU?
The GPU (Graphics Processing Unit) is the central component of a graphics card or GPU-accelerator and is critical for speeding up visualisation and compute workloads. This is achieved by offloading these workloads from the CPU and system memory into the GPU and GPU memory, where the architecture is much more parallel in nature - allowing many tasks to be performed simultaneously. Workstation GPUs are also often referred to as professional or enterprise-grade due to the higher calibre of components used when compared to consumer GPUs.
Unlike NVIDIA GeForce gaming GPUs, NVIDIA workstation GPUs do not have a family name, but simply use the RTX description and a letter / suffix to denote their architectural generation - more on this later. NVIDIA workstation GPUs are designed for the rendering of high resolution images and video concurrently - both hugely parallel workloads, and because GPUs can perform parallel operations on multiple sets of data, they are also perfect for non-graphical tasks such as scientific computing and the development of machine learning and AI models.
What makes NVIDIA workstation GPUs special
NVIDIA workstation GPUs feature a whole host of extra features and capabilities that their consumer counterparts lack.
Certified Drivers
ISVs such as Autodesk, Dassault and Siemens certify their applications, ensuring optimal stability backed by enterprise-class customer support.
Enterprise Class
Enterprise-class components ensure better reliability and resiliency, reducing failure rates especially when used at full load for longer periods of time.
ECC Memory
Error correcting code (ECC) memory acts to protect data from corruption, so any errors are eradicated prior to them affecting the workload being processed.
Extended Memory
Larger onboard frame buffers than consumer GPUs enable larger and more complex renders and compute simulations to be processed.
Security
USB-C ports can be disabled, increasing data integrity when installed in secure environments or when used with sensitive information.
Extended Warranty
The standard warranty provides cover for 3 years in professional environments and can be extended to total of 5 years upon request.
The NVIDIA workstation GPU range
The following table gives an overview of which GPUs are most suitable for different workloads, ranging from architecture, construction and engineering (AEC) to building information management (BIM); computer aided design (CAD) to computer aided engineering (CAE) and computational fluid dynamics (CFD). We also cover media and entertainment (M&E), rendering, virtual reality (VR) and machine learning (ML), deep learning (DL) and artificial intelligence (AI).
| A800 | RTX 6000 Ada | RTX 5000 Ada | RTX A6000 | RTX 4500 Ada | RTX A5500 | RTX A5000 | RTX 4000 Ada | RTX A4000 | RTX 4000 SFF ADA | RTX A2000 | T1000 | T400 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AEC / BIM | No |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
| CAD | No |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
| CAE & CFD | Yes |
Yes |
Yes |
Yes |
No |
No |
No |
No |
No |
No |
No |
No |
No |
| MEDIA & ENTERTAINMENT | No |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
| RENDERING | No |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
No |
No |
No |
No |
| VR | No |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
| ML / DL / AI | Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
No |
No |
No |
No |
Going into more detail, we have ranked each card from highest to lowest performing with additional information on their architecture, cores, memory and performance for a number of tasks. The visualisation score tells you how good a GPU is for design, modelling and how fast it is at rendering; whereas the computational scores are ranked for tasks such as simulation, deep learning and AI workloads. These are split into calculation type - FP64 / TF64 (double precision), FP32 /TF32 (single precision) and FP16 / TF16 (half-precision), as some workloads rely on a specific type.
A800
The A800 is a specialist GPU for accelerating deep learning and AI workloads or CAE and CFD simulations that need to be a high level of mathematical precision, FP64 (aka double precision). It is based on the older Ampere GA100 architecture and features 6,912 CUDA cores and 432 3rd gen Tensor cores, combined with 40GB of HBM2 ECC memory supporting 1.5TB/s bandwidth. It is worth noting that the A800 has no display outputs, as the card is aimed squarely at computational use, so it needs to be combined with another GPU to drive your monitor(s).
CUDA
CUDA cores are the workhorse in the A800 GPU, as the architecture supports many cores accelerating FP64 workloads up to 2x over the previous generation. Also the A800 includes 3-years subscription to NVIDIA AI Enterprise.
HBM2 MEMORY
The A800 has 40GB of high-speed HBM2 memory with a class-leading 1.5 TB/s of memory bandwidth and significantly more on-chip memory, including a 40MB L2 cache to accelerate demanding AI and HPC workloads.
DATA SCIENCE & AI
Third generation Tensor cores boost scientific computing and AI development with up to 11x faster performance compared to the previous generation with hardware-support for structural sparsity.
MIG
Multi-Instance GPU (MIG) fully isolates at the hardware level allowing memory, cache and cores to be partitioned into as many as seven independent instances, giving multiple users access to GPU acceleration.
| VISUALISATION PERFORMANCE | N/A | ||||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 11 | ||||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 11 | ||||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 11 | ||||||||||
Real Time Ray Tracing No
VR Ready No
NVLink Yes
RTX 6000 Ada
The RTX 6000 Ada is the highest performing GPU accelerator based on the Ada Lovelace architecture which supports improved ray tracing and AI performance - it should not be confused with the original RTX 6000 from 2018 which is a much slower GPU and no longer available to buy. The RTX 6000 Ada features 18,176 CUDA cores, 568 4th gen Tensor cores and 142 3rd gen RT cores, combined with 48GB of ultra-reliable ECC memory. Supporting four displays the RTX 6000 Ada is a supremely powerful graphics card for very high-end workstations.
CUDA
CUDA cores are the workhorse in Ada Lovelace GPUs, as the architecture supports many cores and accelerates workloads up to 1.5x (FP32) of the previous Ampere generation.
RAY TRACING
Ada Lovelace GPUs feature third generation RT cores delivering up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Fourth generation Tensor cores boost scientific computing and AI development with up to 3x faster performance compared to Ampere GPUs and support mixed floating-point acceleration.
VR
Ada Lovelace GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 11 | |||||||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 5 | |||||||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 11 | |||||||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 11 | |||||||||||||
Real Time Ray Tracing Yes
VR Ready Yes
NVLink No
RTX 5000 Ada
The RTX 5000 Ada is a high-end GPU accelerator based on the Ada Lovelace architecture which supports improved ray tracing and AI performance. The RTX 5000 Ada features 12,800 CUDA cores, 400 4th Tensor cores and 100 3rd gen RT cores, combined with 32GB of ultra-reliable ECC memory. Supporting four displays the RTX 5000 Ada is an extremely powerful graphics card for very high-end workstations.
CUDA
CUDA cores are the workhorse in Ada Lovelace GPUs, as the architecture supports many cores and accelerates workloads up to 1.5x (FP32) of the previous Ampere generation.
RAY TRACING
Ada Lovelace GPUs feature third generation RT cores delivering up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Fourth generation Tensor cores boost scientific computing and AI development with up to 3x faster performance compared to Ampere GPUs and support mixed floating-point acceleration.
VR
Ada Lovelace GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 10 | ||||||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 5 | ||||||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 10 | ||||||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 10 | ||||||||||||
Real Time Ray Tracing Yes
VR Ready Yes
NVLink No
RTX A6000
The RTX A6000 is a high-end GPU accelerator based on the Ampere architecture which supports ray tracing and accelerated AI. It features 10,752 CUDA cores, 336 3rd gen Tensor cores and 84 2nd gen RT cores, combined with 48GB of ultra-reliable ECC memory. Supporting four displays the RTX A6000 is an extremely powerful graphics card for very high-end workstations.
CUDA
CUDA cores are the workhorse in Ampere GPUs, as the architecture supports many cores and accelerates workloads up to 2.7x (FP32) of the previous Turing generation.
RAY TRACING
Ampere GPUs feature second generation RT cores delivering up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Third generation Tensor cores boost scientific computing and AI development with up to 11x faster performance compared to the previous generation with hardware-support for structural sparsity.
VR
Ampere GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 9 | |||||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 5 | |||||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 9 | |||||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 9 | |||||||||||
Real Time Ray Tracing Yes
VR Ready Yes
NVLink Yes
RTX 4500 Ada
The RTX 4500 Ada is a high-end GPU accelerator based on the Ada Lovelace architecture which supports improved ray tracing and AI performance. The RTX 4500 Ada features 7,680 CUDA cores, 240 4th gen Tensor cores and 60 3rd gen RT cores, combined with 24GB of ultra-reliable ECC memory. Supporting four displays the RTX 4500 Ada is a powerful graphics card for high-end workstations.
CUDA
CUDA cores are the workhorse in Ada Lovelace GPUs, as the architecture supports many cores and accelerates workloads up to 1.5x (FP32) of the previous Ampere generation.
RAY TRACING
Ada Lovelace GPUs feature third generation RT cores delivering up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Fourth generation Tensor cores boost scientific computing and AI development with up to 3x faster performance compared to Ampere GPUs and support mixed floating-point acceleration.
VR
Ada Lovelace GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 9 | |||||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 4 | |||||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 9 | |||||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 9 | |||||||||||
Real Time Ray TracingYes
VR ReadyYes
NVLinkNo
RTX A5500
The RTX A5500 is a high-end GPU accelerator based on the Ampere architecture which supports ray tracing and accelerated AI. It features 10,240 CUDA cores, 320 3rd gen Tensor cores and 80 2nd gen RT cores, combined with 24GB of ultra-reliable ECC memory. Supporting four displays the RTX A5500 is a powerful graphics card for high-end workstations.
CUDA
CUDA cores are the workhorse in Ampere GPUs, as the architecture supports many cores and accelerates workloads up to 2.7x (FP32) of the previous Turing generation.
RAY TRACING
Ampere GPUs feature second generation RT cores delivering up up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Third generation Tensor cores boost scientific computing and AI development with up to 11x faster performance compared to the previous generation with hardware-support for structural sparsity.
VR
Ampere GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 8 | ||||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 4 | ||||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 8 | ||||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 8 | ||||||||||
Real Time Ray Tracing Yes
VR Ready Yes
NVLink Yes
RTX 4000 Ada
The RTX 4000 Ada is a mid-range GPU accelerator based on the Ada Lovelace architecture which supports improved ray tracing and AI performance. The RTX 4000 Ada features 6,144 CUDA cores, 192 4th gen Tensor cores and 48 3rd gen RT cores, combined with 20GB of ultra-reliable ECC memory. Supporting four displays the RTX 4000 Ada is a powerful graphics card for mid-range workstations.
CUDA
CUDA cores are the workhorse in Ada Lovelace GPUs, as the architecture supports many cores and accelerates workloads up to 1.5x (FP32) of the previous Ampere generation.
RAY TRACING
Ada Lovelace GPUs feature third generation RT cores delivering up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Fourth generation Tensor cores boost scientific computing and AI development with up to 3x faster performance compared to Ampere GPUs and support mixed floating-point acceleration.
VR
Ada Lovelace GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 7 | |||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 3 | |||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 7 | |||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 7 | |||||||||
Real Time Ray TracingYes
VR Ready Yes
NVLink No
RTX A5000
The RTX A5000 is a high-end GPU accelerator based on the Ampere architecture which supports ray tracing and accelerated AI. It features 8,192 CUDA cores, 256 3rd gen Tensor cores and 64 2nd gen RT cores, combined with 24GB of ultra-reliable ECC memory. Supporting four displays the RTX A5000 is a powerful graphics card for high-end workstations.
CUDA
CUDA cores are the workhorse in Ampere GPUs, as the architecture supports many cores and accelerates workloads up to 2.7x (FP32) of the previous Turing generation.
RAY TRACING
Ampere GPUs feature second generation RT cores delivering up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Third generation Tensor cores boost scientific computing and AI development with up to 11x faster performance compared to the previous generation with hardware-support for structural sparsity.
VR
Ampere GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 7 | |||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 4 | |||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 7 | |||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 7 | |||||||||
Real Time Ray Tracing Yes
VR Ready Yes
NVLink Yes
RTX A4000
The RTX A4000 is a mid-range GPU accelerator based on the Ampere architecture which supports ray tracing and accelerated AI. It features 6,144 CUDA cores, 192 3rd gen Tensor cores and 48 2nd gen RT cores, combined with 20GB of ultra-reliable ECC memory. Supporting four displays the RTX A4000 is a capable graphics card for mid-range workstations.
CUDA
CUDA cores are the workhorse in Ampere GPUs, as the architecture supports many cores and accelerates workloads up to 2.7x (FP32) of the previous Turing generation.
RAY TRACING
Ampere GPUs feature second generation RT cores delivering up up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
DATA SCIENCE & AI
Third generation Tensor cores boost scientific computing and AI development with up to 11x faster performance compared to the previous generation with hardware-support for structural sparsity.
VR
Ampere GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
| VISUALISATION PERFORMANCE | 5 | |||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 2 | |||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 5 | |||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 5 | |||||||||
Real Time Ray Tracing Yes
VR Ready Yes
NVLink No
RTX 4000 SFF Ada
The RTX 4000 SFF Ada is a compact, half-height version of the RTX 4000 Ada GPU, bringing the benefits of the Ada Lovelace architecture to small form factor workstations. The RTX 4000 SFF Ada features the same 6,144 CUDA cores, 192 4th gen Tensor cores and 48 3rd gen RT cores, combined with 20GB of server-grade ECC memory, but runs at lower clock speeds reducing its performance compared to its full size RTX 4000 Ada brother. Supporting four displays the RTX 4000 SFF Ada is a powerful graphics card for small form factor workstations.
CUDA
CUDA cores are the workhorse in Ada Lovelace GPUs, as the architecture supports many cores and accelerates workloads up to 1.5x (FP32) of the previous Ampere generation.
RAY TRACING
Ada Lovelace GPUs feature third generation RT cores delivering up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
VR
Ada Lovelace GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
SFF
The latest entry-level graphics cards feature a low-profile design that fits in a wide variety of workstation chassis—minimising your desktop workstation footprint without compromising performance.
| VISUALISATION PERFORMANCE | 5 | |||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 2 | |||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 5 | |||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 5 | |||||||||
Real Time Ray TracingYes
VR ReadyYes
NVLink No
RTX A2000
The RTX A2000 is a mid-range workstation GPU, introducing the benefits of the Ampere architecture, such as ray tracing and accelerated AI at a much lower price point than the RTX A4000. It features 3,328 CUDA cores, 104 3rd gen Tensor cores and 26 2nd gen RT cores, combined with 6 or 12GB of server-grade error code correcting (ECC) memory. Supporting four displays the RTX A2000 is a powerful graphics card for small form factor workstations.
CUDA
CUDA cores are the workhorse in Ampere GPUs, as the architecture supports many cores and accelerates workloads up to 2.7x (FP32) of the previous Turing generation.
RAY TRACING
Ampere GPUs feature second generation RT cores delivering up up to double the real-time photorealistic ray-tracing performance of the previous generation GPUs.
VR
Ampere GPUs offer variable rate shading to dynamically change the rate at which different parts of a scene is shaded creating immersive VR experiences.
SFF
The latest entry-level graphics cards feature a low-profile design that fits in a wide variety of workstation chassis—minimising your desktop workstation footprint without compromising performance.
| VISUALISATION PERFORMANCE | 4 | |||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 1 | |||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 4 | |||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 4 | |||||||||
Real Time Ray Tracing Yes
VR ReadyYes
NVLink No
T1000
The T1000 is an entry-level GPU based on a cut down version of the old Turing architecture so it lacks support for ray tracing. It features 896 CUDA cores combined with 4 or 8GB of memory. Supporting four displays the T1000 is a great choice for basic visualisation tasks and small form factor workstations.
CUDA
CUDA cores are the workhorse in Hopper GPUs, as the architecture supports many cores and accelerates workloads up to 1.5x (FP32) of the previous Ampere generation.
SFF
The latest entry-level graphics cards feature a low-profile design that fits in a wide variety of workstation chassis—minimising your desktop workstation footprint without compromising performance.
| VISUALISATION PERFORMANCE | 3 | |||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 1 | |||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 3 | |||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 3 | |||||||||
Real Time Ray Tracing No
VR Ready No
NVLink No
T400
The T400 is an entry-level GPU based on a cut down version of the old Turing architecture so it lacks support for ray tracing. It features 384 CUDA cores combined with 2 or 4GB of memory. Supporting three displays the T400 is a solid choice for basic visualisation tasks and small form factor workstations.
CUDA
CUDA cores are the workhorse in Hopper GPUs, as the architecture supports many cores and accelerates workloads up to 1.5x (FP32) of the previous Ampere generation.
SFF
The latest entry-level graphics cards feature a low-profile design that fits in a wide variety of workstation chassis—minimising your desktop workstation footprint without compromising performance.
| VISUALISATION PERFORMANCE | 1 | |||||||||
| COMPUTE PERFORMANCE (FP64/TF64) | 1 | |||||||||
| COMPUTE PERFORMANCE (FP32/TF32) | 3 | |||||||||
| COMPUTE PERFORMANCE (FP16/TF16) | 3 | |||||||||
Real Time Ray Tracing No
VR Ready No
NVLink No
NVIDIA Professional workstation GPU Summary
The below table summarises each GPUs performance along with their technical specifications.
| A800 | RTX 6000 Ada | RTX 5000 Ada | RTX A6000 | RTX 4500 Ada | RTX A5500 | RTX A5000 | RTX 4000 Ada | RTX A4000 | RTX 4000 SFF Ada | RTX A2000 | T1000 | T400 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RATINGS | |||||||||||||
| Visualisation Performance | 0 | 11 | 10 | 9 | 9 | 8 | 7 | 7 | 5 | 5 | 4 | 3 | 2 |
| Compute Performance (FP64/TF64) | 11 | 5 | 5 | 5 | 4 | 4 | 4 | 3 | 2 | 2 | 1 | 1 | 1 |
| Compute Performance (FP32/TF32) | 11 | 11 | 10 | 9 | 9 | 8 | 7 | 7 | 5 | 5 | 4 | 3 | 2 |
| Compute Performance (FP16/TF16) | 11 | 11 | 10 | 9 | 9 | 8 | 7 | 7 | 5 | 5 | 4 | 3 | 2 |
| Ray Tracing | Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
YES |
No |
No |
| VR Ready | No |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
| NVLink | Yes |
No |
No |
Yes |
No |
Yes |
Yes |
No |
No |
No |
No |
No |
No |
| SPECS | |||||||||||||
| Architecture | Ampere | Ada Lovelace | Ada Lovelace | Ampere | Ada Lovelace | Ampere | Ampere | Ada Lovelace | Ampere | Ada Lovelace | Ampere | Turing | Turing |
| GPU | A100 | AD102 | AD102 | GA102 | AD104 | GA102 | GA102 | AD104 | GA102 | AD104 | GA106 | TU117 | TU117 |
| CUDA Cores | 6,912 | 18,176 | 12,800 | 10,752 | 7,680 | 10,240 | 8,192 | 6,144 | 6,144 | 6,144 | 3,328 | 896 | 384 |
| Tensor Cores | 432 3rd gen | 568 4th gen | 400 4th gen | 336 3rd gen | 240 4th gen | 320 3rd gen | 256 3rd gen | 192 4th gen | 192 3rd gen | 192 4th gen | 104 3rd gen | 0 | 0 |
| RT Cores | 0 | 142 3rd gen | 100 3rd gen | 84 2nd gen | 60 3rd gen | 80 2nd gen | 64 2nd gen | 48 3rd gen | 48 2nd gen | 48 3rd gen | 26 2nd gen | 0 | 0 |
| Memory | 40GB HBM2 | 48GB GDDR6 | 32GB GDDR6 | 48GB GDDR6 | 24GB GDDR6 | 24GB GDDR6 | 24GB GDDR6 | 20GB GDDR6 | 16GB GDDR6 | 20GB GDDR6 | 6/12GB GDDR6 | 4/8GB GDDR6 | 2/4GB GDDR6 |
| ECC Memory | Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
| Memory Controller | 5,120-bit | 384-bit | 256-bit | 384-bit | 192-bit | 384-bit | 384-bit | 160-bit | 256-bit | 160-bit | 192-bit | 128-bit | 64-bit |
| TDP | 240W | 300W | 250W | 300W | 192W | 230W | 230W | 130W | 140W | 70W | 70W | 50W | 40W |
| PRICE |
|
|
|
|
|
|
|
|
|
|
|
|
|
| BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | BUY > | |
Ready to buy?
All NVIDIA workstation GPUs can be purchased standalone or integrated into a workstation from our 3XS Systems division. In addition, we can provide supported pricing for higher education and further education customers.
We hope you’ve found this NVIDIA workstation GPU buyer’s guide helpful, however if you would like further advice on choosing the correct GPU please get in touch on 01204 474747 or at [email protected].