The nvidia gp100 gpu reaches 4 tflops dpfp
The Nvidia Pascal architecture promises to make a great leap forward in several aspects, one of them is the power of double precision floating point calculation (DPFP), the Nvidia GP100 GPU will be a real monster in this regard.
The Nvidia GP100 chip will be the maximum exponent, at least for now, of the promising Pascal architecture. This chip offers a double precision floating point calculation performance of 4 TFLOPs, a figure that if it does not tell you anything we tell you that it is three times greater than the power offered by the Kepler GK110 based Tesla K20 card and that it is the most powerful in this regard that Nvidia currently has.
The same goes for the simple precision floating point computing power, in this case the figure amounts to 12 TFLOPs compared to the 3 TFLOPs offered by approximately the Kepler-based GK110 chip.
Pascal is the new Nvidia graphics architecture that will arrive manufactured in 16nm FinFET and that will succeed the successful Maxwell. Pascal will use the high-performance memory HBM2 and also the GDDR5X memory, both represent a great leap forward compared to the GDDR5 and are one of the keys to the great performance that Pascal is expected to offer.
Source: techpowerup
Intel stratix 10 fpga processor reaches 10 tflops of power
Intel Stratix 10 FPGA is a specialized floating point operations processor capable of reaching a power of 10 TFLOPs.
Galax kfa 2 gtx 1080 hof reaches 2.5 ghz on its pascal gpu
The GALAX GTX 1080 HOF has been able to achieve a voltage of 1.3V and with it an operating frequency on its 2.5GHz Pascal GP104 GPU.
Amd radeon rx 480 reaches 1.5 ghz on its gpu
The AMD Radeon RX 480 graphics card is capable of reaching 1.5 GHz at its core in the reference version with a slight increase in voltage.
