NVIDIA A100 (80 GB) vs NVIDIA A100 (40 GB)
We do not have any performance data for the A100 (80 GB) and the A100 (40 GB) for the 3DMark 11 Performance GPU benchmark.
We do not have any performance per dollar data for the A100 (80 GB) and the A100 (40 GB) for the 3DMark 11 Performance GPU benchmark.
Summary
About the NVIDIA A100 (80 GB) GPU
The NVIDIA A100 (80 GB) is a workstation graphics card that launched in Q2 2021. It is built on the Ampere GPU microarchitecture (codename GA100) and is manufactured on a 7 nm process.
Memory
The A100 has 80 GB of HBM2e memory, with a 1,512 MHz memory clock and a 5,120 bit interface. This gives it a memory bandwidth of 1.94 Tb/s, which affects how fast it can transfer data to and from memory. GPU memory stores temporary data that helps the GPU with complex math and graphics operations. More memory is generally better, as not having enough can cause performance bottlenecks.
Cores and Clock Speeds
The A100 includes 6,912 CUDA cores, the processing units for handling parallel computing tasks. The GPU operates at a core clock speed of 1,065 MHz and can dynamically boost its clock speed up to 1,410 MHz. Complementing the processing units are 432 texture mapping units (TMUs) for efficient texture filtering and 160 render output units (ROPs) for pixel processing. Additionally, the GPU features 432 tensor cores optimized for AI-accelerated workloads.
Compatibility & Power Consumption
The A100 occupies 2 PCIe expansion slots. NVIDIA recommends a power supply of at least 700 W to handle the GPU's thermal design power (TDP) of 300 W.
About the NVIDIA A100 (40 GB) GPU
The NVIDIA A100 (40 GB) is a workstation graphics card that launched in Q2 2020. It is built on the Ampere GPU microarchitecture (codename GA100) and is manufactured on a 7 nm process.
Memory
The A100 has 40 GB of HBM2e memory, with a 1,215 MHz memory clock and a 5,120 bit interface. This gives it a memory bandwidth of 1.56 Tb/s, which affects how fast it can transfer data to and from memory. GPU memory stores temporary data that helps the GPU with complex math and graphics operations. More memory is generally better, as not having enough can cause performance bottlenecks.
Cores and Clock Speeds
The A100 includes 6,912 CUDA cores, the processing units for handling parallel computing tasks. The GPU operates at a core clock speed of 765 MHz and can dynamically boost its clock speed up to 1,410 MHz. Complementing the processing units are 432 texture mapping units (TMUs) for efficient texture filtering and 160 render output units (ROPs) for pixel processing. Additionally, the GPU features 432 tensor cores optimized for AI-accelerated workloads.
Compatibility & Power Consumption
The A100 occupies 2 PCIe expansion slots. NVIDIA recommends a power supply of at least 600 W to handle the GPU's thermal design power (TDP) of 250 W.
General Info
General overview of the GPU, including details like its manufacturer, release date, launch price, and current production status.
| Info | A100 (80 GB) | A100 (40 GB) |
|---|---|---|
| Manufacturer | NVIDIA | NVIDIA |
| Architecture | Ampere | Ampere |
| Market Segment | Workstation | Workstation |
| Release Date | Q2 2021 | Q2 2020 |
| Production Status | Active | Active |
| Shop | Check Price | Check Price |
Gaming Performance
Benchmark Performance
Technical Specs
Graphics Processor
General information about the graphics processing unit like their architecture, manufacturing process size, and transistor count. Newer GPU architectures generally bring efficiency improvements and may introduce technologies that enhance graphical capabilities.
| Spec | A100 (80 GB) | A100 (40 GB) |
|---|---|---|
| Codename | GA100 | GA100 |
| Architecture | Ampere | Ampere |
| Process Size | 7 nm | 7 nm |
| Transistors | 54,200 million | 54,200 million |
Memory Details
Memory specifications like their capacity, bandwidth, and clock speeds. GPU memory stores graphics data like frames, textures, and shadows which helps display rendered images. These specs are crucial for graphics-intense applications like gaming and 3D modeling.
| Spec | A100 (80 GB) | A100 (40 GB) |
|---|---|---|
| Memory Size | 80 GB | 40 GB |
| Memory Type | HBM2e | HBM2e |
| Memory Bandwidth | 1.94 Tb/s | 1.56 Tb/s |
| Memory Clock | 1,512 MHz | 1,215 MHz |
| Memory Interface | 5,120 bit | 5,120 bit |
| L1 Cache | 192 KB | 192 KB |
| L2 Cache | 80 MB | 40 MB |
Board Compatibility
Compatibility information like their slot size, bus interface, power consumption, and display support. These specs are useful for verifying compatibility with your motherboard, power supply, and monitor.
| Spec | A100 (80 GB) | A100 (40 GB) |
|---|---|---|
| Slots | 2 slots | 2 slots |
| Bus Interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
| Thermal Design Power (TDP) | 300 W | 250 W |
| Suggested PSU | 700 W | 600 W |
| Power Connectors | 8-pin EPS | 8-pin EPS |
| Outputs | No outputs | No outputs |
Cores & Clock Speeds
Processing power information like its cores and clock speed. These specs impact how fast they can process graphics. Each type of core or component serves a specific computational purpose.
| Spec | A100 (80 GB) | A100 (40 GB) |
|---|---|---|
| CUDA Cores | 6,912 | 6,912 |
| Stream Multiprocessors (SM) | 108 | 108 |
| Texture Mapping Units (TMU) | 432 | 432 |
| Render Output Units (ROP) | 160 | 160 |
| Tensor Cores | 432 | 432 |
| Core Clock Speed | 1,065 MHz | 765 MHz |
| Core Clock Speed (Boost) | 1,410 MHz | 1,410 MHz |
Theoretical Performance
Theoretical performance numbers derived from the raw specifications of the different components like core count and clock speeds. While these provide a glimpse into peak processing power, they do not represent real-world performance.
| Spec | A100 (80 GB) | A100 (40 GB) |
|---|---|---|
| Pixel Fill Rate | 225.6 GPixel/s | 225.6 GPixel/s |
| Texture Fill Rate | 609.1 GTexel/s | 609.1 GTexel/s |
| FP32 Performance | 19.49 TFLOPS | 19.49 TFLOPS |
| FP64 Performance | 9.75 TFLOPS | 9.75 TFLOPS |
API Support
Graphics API versions supported by these graphics cards. APIs evolve over time, introducing new features and functionalities. Older GPUs may not support recent versions.
| Spec | A100 (80 GB) | A100 (40 GB) |
|---|---|---|
| OpenCL | 3.0 | 3.0 |
* Performance rating, performance per dollar, and rankings are based on the 3DMark 11 Performance GPU benchmark and MSRP.