![NVIDIA H100 PCIe](/upload/resize_cache/iblock/f52/340_345_0/NVIDIA.png)
![AMD Radeon Pro W6800](/upload/resize_cache/iblock/a7e/340_345_0/AMD.png)
Comparision NVIDIA H100 PCIe vs AMD Radeon Pro W6800
Grade
Top specs and features
GPU base clock speed
RAM
Memory bandwidth
Gpu memory speed
FLOPS
Description
The NVIDIA H100 PCIe video card is based on the Hopper architecture. AMD Radeon Pro W6800 on the RDNA 2.0 architecture. The first has 80000 million transistors. The second is 26800 million. NVIDIA H100 PCIe has a transistor size of 4 nm versus 7.
The base clock speed of the first video card is 1065 MHz versus 2075 MHz for the second.
Let's move on to memory. NVIDIA H100 PCIe has 80 GB. AMD Radeon Pro W6800 has 80 GB installed. The bandwidth of the first video card is 1.28 Gb/s versus 512 Gb/s of the second.
FLOPS of NVIDIA H100 PCIe is 47.14. At AMD Radeon Pro W6800 18.5.
Goes to tests in benchmarks. In the Passmark benchmark, NVIDIA H100 PCIe scored There is no data points. And here is the second card 17319 points. In 3DMark, the first model scored There is no data points. Second There is no data points.
In terms of interfaces. The first video card is connected using There is no data. The second is PCIe 4.0 x16. Video card NVIDIA H100 PCIe has Directx version There is no data. Video card AMD Radeon Pro W6800 -- Directx version - 12.2.
Regarding cooling, NVIDIA H100 PCIe has 350W heat dissipation requirements versus 250W for AMD Radeon Pro W6800.
Why AMD Radeon Pro W6800 is better than NVIDIA H100 PCIe
- RAM 80 GB против 32 GB, more on 150%
- FLOPS 47.14 TFLOPS против 18.5 TFLOPS, more on 155%
- Technological process 4 nm против 7 nm, less by -43%
NVIDIA H100 PCIe vs AMD Radeon Pro W6800: highlights
![NVIDIA H100 PCIe](/upload/resize_cache/iblock/f52/340_345_0/NVIDIA.png)
![AMD Radeon Pro W6800](/upload/resize_cache/iblock/a7e/340_345_0/AMD.png)
Performance
Memory
General information
Functions
FAQ
How does the NVIDIA H100 PCIe processor perform in benchmarks?
Passmark NVIDIA H100 PCIe scored There is no data points. The second video card scored 17319 points in Passmark.
What FLOPS do video cards have?
FLOPS NVIDIA H100 PCIe is 47.14 TFLOPS. But the second video card has FLOPS equal to 18.5 TFLOPS.
What power consumption?
NVIDIA H100 PCIe 350 Watt. AMD Radeon Pro W6800 250 Watt.
How fast are NVIDIA H100 PCIe and AMD Radeon Pro W6800?
NVIDIA H100 PCIe operates at 1065 MHz. In this case, the maximum frequency reaches 1650 MHz. The clock base frequency of AMD Radeon Pro W6800 reaches 2075 MHz. In turbo mode it reaches 2320 MHz.
What kind of memory do graphics cards have?
NVIDIA H100 PCIe supports GDDRThere is no data. Installed 80 GB of RAM. Throughput reaches 1.28 GB/s. AMD Radeon Pro W6800 works with GDDR6. The second one has 32 GB of RAM installed. Its bandwidth is 1.28 GB/s.
How many HDMI connectors do they have?
NVIDIA H100 PCIe has There is no data HDMI outputs. AMD Radeon Pro W6800 is equipped with There is no data HDMI outputs.
What power connectors are used?
NVIDIA H100 PCIe uses There is no data. AMD Radeon Pro W6800 is equipped with There is no data HDMI outputs.
What architecture are video cards based on?
NVIDIA H100 PCIe is built on Hopper. AMD Radeon Pro W6800 uses the RDNA 2.0 architecture.
What graphics processor is being used?
NVIDIA H100 PCIe is equipped with GH100. AMD Radeon Pro W6800 is set to Navi 21.
How many PCIe lanes
The first graphics card has 16 PCIe lanes. And the PCIe version is 4. AMD Radeon Pro W6800 16 PCIe lanes. PCIe version 4.
How many transistors?
NVIDIA H100 PCIe has 80000 million transistors. AMD Radeon Pro W6800 has 26800 million transistors