H100 PCIe vs Tesla T4
Primary details
GPU architecture, market segment, value for money and other general parameters compared.
| Place in the ranking | 200 | not rated |
| Place by popularity | not in top-100 | not in top-100 |
| Power efficiency | 27.30 | no data |
| Architecture | Turing (2018−2022) | Hopper (2022−2023) |
| GPU code name | TU104 | GH100 |
| Market segment | Workstation | Workstation |
| Release date | 13 September 2018 (6 years ago) | 22 March 2022 (2 years ago) |
Detailed specifications
General parameters such as number of shaders, GPU core base clock and boost clock speeds, manufacturing process, texturing and calculation speed. Note that power consumption of some graphics cards can well exceed their nominal TDP, especially when overclocked.
| Pipelines / CUDA cores | 2560 | 7296 |
| Core clock speed | 585 MHz | 1065 MHz |
| Boost clock speed | 1590 MHz | 1650 MHz |
| Number of transistors | 13,600 million | 80,000 million |
| Manufacturing process technology | 12 nm | 4 nm |
| Power consumption (TDP) | 70 Watt | 350 Watt |
| Texture fill rate | 254.4 | 752.4 |
| Floating-point processing power | 8.141 TFLOPS | 24.08 TFLOPS |
| ROPs | 64 | 24 |
| TMUs | 160 | 456 |
| Tensor Cores | 320 | 456 |
| Ray Tracing Cores | 40 | no data |
Form factor & compatibility
Information on compatibility with other computer components. Useful when choosing a future computer configuration or upgrading an existing one. For desktop graphics cards it's interface and bus (motherboard compatibility), additional power connectors (power supply compatibility).
| Interface | PCIe 3.0 x16 | PCIe 5.0 x16 |
| Length | 168 mm | 267 mm |
| Width | 1-slot | 2-slot |
| Supplementary power connectors | None | 8-pin EPS |
VRAM capacity and type
Parameters of VRAM installed: its type, size, bus, clock and resulting bandwidth. Integrated GPUs have no dedicated video RAM and use a shared part of system RAM.
| Memory type | GDDR6 | HBM2e |
| Maximum RAM amount | 16 GB | 80 GB |
| Memory bus width | 256 Bit | 5120 Bit |
| Memory clock speed | 1250 MHz | 1000 MHz |
| Memory bandwidth | 320.0 GB/s | 1,280 GB/s |
Connectivity and outputs
Types and number of video connectors present on the reviewed GPUs. As a rule, data in this section is precise only for desktop reference ones (so-called Founders Edition for NVIDIA chips). OEM manufacturers may change the number and type of output ports, while for notebook cards availability of certain video outputs ports depends on the laptop model rather than on the card itself.
| Display Connectors | No outputs | No outputs |
API compatibility
List of supported 3D and general-purpose computing APIs, including their specific versions.
| DirectX | 12 Ultimate (12_1) | N/A |
| Shader Model | 6.5 | N/A |
| OpenGL | 4.6 | N/A |
| OpenCL | 1.2 | 3.0 |
| Vulkan | 1.2.131 | N/A |
| CUDA | 7.5 | 9.0 |
Synthetic benchmark performance
Non-gaming benchmark results comparison. The combined score is measured on a 0-100 point scale.
GeekBench 5 OpenCL
Geekbench 5 is a widespread graphics card benchmark combined from 11 different test scenarios. All these scenarios rely on direct usage of GPU's processing power, no 3D rendering is involved. This variation uses OpenCL API by Khronos Group.
Pros & cons summary
| Recency | 13 September 2018 | 22 March 2022 |
| Maximum RAM amount | 16 GB | 80 GB |
| Chip lithography | 12 nm | 4 nm |
| Power consumption (TDP) | 70 Watt | 350 Watt |
Tesla T4 has 400% lower power consumption.
H100 PCIe, on the other hand, has an age advantage of 3 years, a 400% higher maximum VRAM amount, and a 200% more advanced lithography process.
We couldn't decide between Tesla T4 and H100 PCIe. We've got no test results to judge.
Should you still have questions concerning choice between the reviewed GPUs, ask them in Comments section, and we shall answer.
Comparisons with similar GPUs
We selected several comparisons of graphics cards with performance close to those reviewed, providing you with more options to consider.
