NVIDIA RTX 6000 Ada vs NVIDIA H100 PCIe. Specifications, performance, tests

Comparison of basic technical data of graphics cards NVIDIA RTX 6000 Ada and NVIDIA H100 PCIe, chip, information processing units.

• GPU Chip
  AD102 GH100
• Execution units
  18176 7296
• Render Output Units
  192 24
• Texture Units
  568 456

Comparison of the amount of memory on board graphics cards. The more the better.

• Memory Size
  48 GB 80 GB
• Memory Type
  GDDR6 HBM2e
• Memory Clock
  2.0 GHz 1 GHz
• Memory Speed
  16 Gbps 2 Gbps
• Memory Bandwith
  768.0 GB/s 1,280 GB/s
• Memory Interface
  384 bit 5120 bit

Let's compare the memory frequency of graphics cards NVIDIA RTX 6000 Ada and NVIDIA H100 PCIe. The higher the better.

• Base Clock
  2.175 GHz 1.065 GHz
• Boost Clock
  2.535 GHz 1.65 GHz

Connectors, the number of thermal watts emitted in normal mode and at overclocking.

• TDP
  300 W 350 W
• Power Connectors
  1x 16-pin 8-pin EPS
• Suggested PSU
  700 W 750 W

Technical data that is used to its full potential in computer games.

• DirectX
  12.2 No
• OpenGL
  4.6 No
• OpenCL
  3.0 3.0
• Vulkan
  1.3 No
• CUDA
  8.9 9.0
• Shader Model
  6.7 No

Comparison of interfaces and release dates for GPU NVIDIA H100 PCIe and NVIDIA RTX 6000 Ada.

• Architecture
  Ada Lovelace Hopper
• GPU Interface
  PCIe 4.0 x16 PCIe 5.0 x16
• Structure size
  4 nm 4 nm
• Release date
  Q4/2022 Q2/2022

• Pixel Rate
  486.7 GPixel/s 39.60 GPixel/s
• Texture Rate
  1,440 GTexel/s 752.4 GTexel/s
• FP16 (half)
  92.15 TFLOPS (1:1) 96.31 TFLOPS
• FP32 (float)
  92.15 TFLOPS 24.08 TFLOPS
• FP64 (double)
  1,440 GFLOPS (1:64) 12.04 TFLOPS