NVIDIA new miner

mining machinery may focus on different projects. Taking IPFs mining machine configuration as an example, the server level CPU such as amd Xiaolong 7302 is generally used; Most graphics cards are made of NVIDIA 2080ti, which can be proced in different factories; Seagate enterprise hard disk is the majority of hard disks, with large capacity, reliability and long-term operation of 7 * 24 hours

"Negative optimization" is just a relative term. Amd graphics card has been using GCN architecture for many years, and there has been no big change in minor repairs. At the same time, it also has a certain foundation in hard power. NVIDIA graphics card, from Fermi to Kepler to Maxwell, has a large and frequent architecture change. One driver may not be able to take care of all graphics cards under all architectures, Therefore, the performance of GCN based a-card is not as good as that of GCN based a-card when it is first launched. However, with the optimization and catching up, it does not mean that the performance of GCN based a-card will decline with the driver update. It is just that the improvement rate of GCN based a-card in the new game is not as good as that of GCN based a-card in the early stage. Since AMD will give up the GCN architecture after Vega, it's hard to say who will be optimized in the future (AMD's hd6000 series will no longer enjoy the benefits of chicken blood drive). Now when buying a graphics card, you don't have to pay too much attention to the problem of "negative optimization"

Appendix 1: evaluation of driving house

1. Geforce GTX 1660 super and geforce GTX 1650 super

these two procts are designed to bring more powerful performance to today's most popular games such as fortnite, pubg and apex legends

compared with the previous generation of GTX 10 series graphics card, the performance of gtx16 Series graphics card is improved by more than 50%. The new graphics card adopts Turing architecture and uses ultra fast gddr6 memory

2. Gtx1660 Ti graphics card

from the perspective of proct positioning, geforce GTX 1660 Ti mainly focuses on the update of the previous generation graphics cards geforce GTX 1070 and 1060, and still adopts the latest Turing architecture of NVIDIA. However, e to the removal of RT cores and tensor cores, it can not play the latest anti saw technology and ray tracing effect of RTX Series graphics cards, and the proct name is restored to the previous generation GTX

GTX 1660 ti is equipped with 6GB gddr6 memory, 1536 CUDA cores, frequency up to 1.8GHz and supports overclocking. Compared with the previous GTX 1060 Ti graphics card, the performance of the new proct is 1.5 times that of the latter, and the energy efficiency is also improved by 1.4 times

3. GTX 1650 graphics card

GTX 1650 is based on Turing architecture, equipped with tu117-300 core, 1024 CUDA core, 4GB gddr5 display memory, and supports 128bit memory bus, corresponding to 128GB memory transmission bandwidth per second. The basic operating frequency is 1485mhz, while the overclocking mode can reach 1665mhz

at the same time, GTX 1650 and GTX 1660 ti will also be included in the max-q lightweight design, so that the entry-level lightweight laptop procts can also play a better game performance

extended data

NVIDIA's popular graphics card:

1, NVIDIA gtx770. Although the gk104 core number of GTX 770 has become "gk104-425-a2", it has new GPU boost 2.0, 7GHz high video memory frequency, 1536 cudacores and 2GB / 256bit video memory bit width. In terms of specifications, it is not much different from GTX 680, so a new GPU boost 2.0 is added

2、NVIDIA GTX780Ti The memory capacity of this graphics card is 3gb, which is concive to improving the running speed. Its memory bit width reaches 384bit, the memory type supports gddr5, the core frequency is up to 876mhz, and supports HDCP. Moreover, the proct and specification are reasonable

3、NVIDIA GTX650 GTX 650 ti is the detailed specification of GTX 650 ti. It will shield two groups of SMX units and a group of video memory controller on the basis of gk106. That is to say, it has 576 CUDA cores, 48 texture units and 16 ROP raster units. The video memory bit width is 128bit, with 1GB gddr5 video memory. Reference price as low as 2599, is the main cost-effective procts

4、NVIDIA Quadro 4000 With NVIDIA Maxwell architecture, powerful performance and single slot design flexibility, you can easily handle the most challenging visual computing work. Equipped with super fast bandwidth of 8GB drawing memory, you can create and render large, complex 3D models and calculate a large amount of data

In addition, Faraday announced that it will carry NVIDIA drive Orin platform on FF 91 in the future. In the future, FF 81 and FF 71 models will be equipped with more advanced automatic driving and parking functions. The two models also equipped with NVIDIA drive origin automatic driving platform are expected to be launched in 2023 and 2024 respectively