NVIDIA H100 PCIe contre NVIDIA H100 SXM5. Spécifications, performances, tests

This section provides a detailed comparison of the fundamental technical specifications of the graphics cards NVIDIA H100 PCIe et NVIDIA H100 SXM5. It includes key information such as the GPU chip architecture, the number of processing units, and other core features that directly impact the overall performance of the cards in various applications, including gaming and professional workloads.

• Puce GPU
  GH100 GH100
• Unités d'exécution
  7296 8448
• Unités de sortie de rendu
  24 24
• Unités de texture
  456 528

This section provides a comparison of the memory configurations of NVIDIA H100 PCIe et NVIDIA H100 SXM5. It includes details such as the memory size, type (e.g., GDDR6, HBM2), and bandwidth, which are key factors that determine how efficiently the card handles large textures, data sets, and high-resolution tasks.Larger memory sizes are typically beneficial for demanding applications like 4K gaming and video editing.

• Taille de la mémoire
  80 GB 80 GB
• Type de mémoire
  HBM2e HBM3
• Memory Clock
  1 GHz 1.5 GHz
• Vitesse de la mémoire
  2 Gbps 3 Gbps
• Bande de mémoire
  1,280 GB/s 1,920 GB/s
• Interface mémoire
  5120 bit 5120 bit

This section focuses on comparing the clock speeds of NVIDIA H100 PCIe et NVIDIA H100 SXM5. It includes both the base and boost clock speeds, which directly influence the performance of the GPU during intensive tasks. Faster clock speeds typically translate to higher frame rates in games and quicker processing in compute-heavy applications.

• Fréquence de base
  1.065 GHz 1.065 GHz
• Fréquence maximale
  1.65 GHz 1.78 GHz

This section examines the thermal design of NVIDIA H100 PCIe et NVIDIA H100 SXM5. It includes details such as power consumption (in watts) and thermal output during regular operation and overclocking. Efficient thermal management is essential for maintaining stable performance during long gaming sessions or other demanding tasks.

• TDP
  350 W 700 W
• Power Connectors
  8-pin EPS 8-pin EPS
• Suggested PSU
  750 W 1100 W

This section compares the technical specifications of NVIDIA H100 PCIe et NVIDIA H100 SXM5. It highlights key features such as API support for DirectX, OpenGL, and Vulkan, which are crucial for running modern games and applications efficiently.These specifications ensure compatibility and optimal performance across a wide range of software environments.

• Max. Affichages
  1 1
• DirectX
  Non Non
• OpenGL
  Non Non
• OpenCL
  3.0 3.0
• Vulkan
  Non Non
• Shader Model
  Non Non
• CUDA
  9.0 9.0

This section provides additional information about NVIDIA H100 PCIe et NVIDIA H100 SXM5. It covers aspects such as release dates, interface compatibility, and other unique features that might not be covered in other sections.This helps users make a more informed decision about the suitability of each card for their specific use case.

• Architecture
  Hopper Hopper
• Interface GPU
  PCIe 5.0 x16 PCIe 5.0 x16
• Processus technologique
  4 nm 4 nm
• Date de sortie
  Q2/2022 Q2/2022

• Pixel Rate
  39.60 GPixel/s 42.72 GPixel/s
• Texture Rate
  752.4 GTexel/s 939.8 GTexel/s
• FP16 (half)
  96.31 TFLOPS 120.3 TFLOPS
• FP32 (float)
  24.08 TFLOPS 30.07 TFLOPS
• FP64 (double)
  12.04 TFLOPS 15.04 TFLOPS