Radeon HD 6870 vs Tesla T4

When comparing Radeon HD 6870 and Tesla T4, we look primarily at benchmarks and game tests. But it is not only about the numbers. Often you can find third-party models with higher clock speeds, better cooling, or a customizable RGB lighting. Not all of them will have all the features you need. Another thing to consider is the port selection. Most graphics cards have at least one DisplayPort and HDMI interface, but some monitors require DVI. Before you buy, check the TDP of the graphics card - this characteristic will help you estimate the consumption of the graphics card. You may even have to upgrade your PSU to meet its requirements. An important factor when choosing between Radeon HD 6870 and Tesla T4 is the price. Does the additional cost justify the performance hit? Our comparison should help you make the right decision.

Main Specs

  Radeon HD 6870 Tesla T4
Power consumption (TDP) 151 Watt 70 Watt
Interface PCIe 2.0 x16 PCIe 3.0 x16
Supplementary power connectors 2x 6-pin None
Memory type GDDR5 GDDR6
Maximum RAM amount 1 GB 16 GB
Display Connectors 2x DVI, 1x HDMI, 2x mini-DisplayPort No outputs
  Check Price
  • Radeon HD 6870 has 115% more power consumption, than Tesla T4.
  • Radeon HD 6870 is connected by PCIe 2.0 x16, and Tesla T4 uses PCIe 3.0 x16 interface.
  • Tesla T4 has 15 GB more memory, than Radeon HD 6870.
  • Both cards are used in Desktops.
  • Radeon HD 6870 is build with Terascale 2 architecture, and Tesla T4 - with Turing.
  • Radeon HD 6870 is manufactured by 40 nm process technology, and Tesla T4 - by 12 nm process technology.
  • Radeon HD 6870 is 52 mm longer, than Tesla T4.
  • Memory clock speed of Tesla T4 is 8950 MHz higher, than Radeon HD 6870.

Game benchmarks

high / 1080p 10−12 35−40
ultra / 1080p 6−7 21−24
QHD / 1440p 1−2 16−18
4K / 2160p 10−11
low / 720p 24−27 60−65
medium / 1080p 12−14 40−45
The average gaming FPS of Tesla T4 in Assassin's Creed Odyssey is 227% more, than Radeon HD 6870.

Full Specs

  Radeon HD 6870 Tesla T4
Architecture Terascale 2 Turing
Code name RV940 Barts TU104
Type Desktop Workstation
Release date 22 October 2010 13 September 2018
Pipelines 1120 2560
Core clock speed 585 MHz
Boost Clock 900 MHz 1590 MHz
Transistor count 1,700 million 13,600 million
Manufacturing process technology 40 nm 12 nm
Texture fill rate 50.40 254.4
Floating-point performance 2,016.0 gflops
Length 220 mm 168 mm
Memory bus width 256 Bit 256 Bit
Memory clock speed 1050 MHz 10000 MHz
Memory bandwidth 134.4 GB/s 320.0 GB/s
Shared memory -
DirectX 12 Ultimate (12_1)
Shader Model 5.0 6.5
OpenGL 4.4 4.6
OpenCL 1.2 1.2
Vulkan 1.2.131
CUDA 7.5
Bus support PCIe 2.0 x16
HDMI +
Bitcoin / BTC (SHA256) 213 Mh/s
Eyefinity +
Design reference
CrossFire +
  Check Price