NVIDIA Quadro M6000 24 GB Mining: Hashrate | Benchmark | 250 Watt (K6000)

NVIDIA Quadro M6000 24 GB Mining: Hashrate | Benchmark | 250 Watt (K6000) – Quadro M6000 24 GB graphics card released by NVIDIA; release date: 5 March 2016. At time of release, the graphics card was priced at $ 4,999. The video card is designed for workstation computers and is built on the Maxwell 2.0 architecture, codenamed GM200.

The core frequency is 988 MHz. The core frequency in Boost mode is 1114 MHz. Texturing speed – 285.2 GTexel / s. Number of shader processors – 3072. Floating point performance – 6,844 gflops. Technological process – 28 nm. Transistors count – 8,000 million. Power Consumption (TDP) – 250 Watt.

Memory type: GDDR5. The maximum memory size is 24 GB. The memory bus width is 384 Bit. The memory frequency is 6612 MHz. Memory bandwidth – 317.4 GB / s.

NVIDIA presented the professional Quadro M6000. It also builds on the second generation “Maxwell” architecture and the GM200 GPU, but is aimed at professional users who care about single precision computing performance. The Quadro M6000 delivers 7 TFLOPS, which is a 30 percent increase over the fastest Kepler-based Quadro graphics card. Keep in mind, though, that the GPU uses 3,072 stream processors. NVIDIA does not speak about FP64 units in this case, since the double precision performance is very low – 291 GFLOPS.

Compared to the K6000, the memory configuration and its connection have changed little. The 12 GB ECC protected memory is connected via a 384-bit bus. By increasing the clock speed, NVIDIA was able to increase the memory bandwidth.

The parallels with the GeForce GTX Titan X are obvious. NVIDIA used the same PCB layout (with minor changes) and cooling system. Except that the cooler now has the Quadro logo, the color scheme has also been changed. In the segment of workstations, the maximum power consumed by NVIDIA is important – 250 watts. Up to four displays are connected to the Quadro M6000 graphics card. Accordingly, if three or even four video cards are installed in a workstation, then it becomes possible to work with 16 displays.

NVIDIA Quadro M6000 24 GB Mining Specification: Quadro K6000 vs Quadro M6000 vs FirePro W9100

For comparison, we took the AMD FirePro W9100, AMD’s most powerful single-GPU workstation graphics card. All of the above models are aimed at one area, but the difference in computational performance is clearly noticeable between them. Depending on the needs of the user, namely the high speed of calculations with single or double precision, one or another model can be recommended. Of course, the applications used also influence the choice. The advantages of NVIDIA include support for CUDA and OpenCL, which gives leadership in a number of tests:

NVIDIA Quadro M6000 Benchmarks

If we compare GPU + CPU with pure calculations on the CPU, then in the first case we get a significant gap. On the GPU, it turns out to be much faster to calculate parallel tasks; nevertheless, several thousand stream processors work in the GPU.

NVIDIA Quadro M6000 PassMark – Frames Per Second (Fps)

Of course, NVIDIA itself chose the FirePro W91000 for comparison, and also provided the results of those tests that are beneficial. But in other applications the situation may be different, we have not yet received the video card for our test lab, so we cannot present our own results.

It makes sense to compare the GeForce GTX Titan X and the Quadro M6000, since the developer must clearly understand what exactly he will pay a few thousand euros for. What are the advantages of Quadro M6000 over GeForce GTX Titan X. In addition to ECC memory support, driver optimization can also be noted. They recognize the installed graphics card and then speed up certain scenarios. In addition, the Quadro M6000 is designed with four years of continuous operation in mind.

The Quadro M6000 will soon go on sale at a price of 4.999 euros.

In recent years, NVIDIA has worked closely with Honda to render all vehicle components in real time. 192 GPUs are used for this task, the entire hardware complex is estimated at 2 million US dollars. Now, with a new generation of workstation graphics cards, NVIDIA plans to bring the technology to the wider market, as well as promote the Iray project. Iray is a technology for displaying rendering scenes as they are calculated. Iray can be called a view from the camera to render without making any modifications. In Iray 2015, there are more options to change the behavior of materials and lighting technologies. The technology has been tested in various rendering environments throughout the year. Below are a few examples of renders calculated on the NVIDIA Quadro M6000:

Quadro K1200

The Quadro K1200 card was also presented today. At first glance, the “junior” video card “Maxwell” with 512 stream processors and 4 GB of video memory is hardly relevant, but NVIDIA considers “junior” video cards for workstations an important market. For the price of 359 euros, developers will get all the features of the Quadro graphics card, even if they have to make sacrifices in terms of performance. From the technical specifications, it is worth noting a 128-bit memory interface, 80 GB / s memory bandwidth and a maximum power consumption of 45 watts.

Quadro VCA

At GTC 2015, a Quadro VCA was shown with eight Quadro M6000 graphics cards that were specially adapted. A single VCA system uses 24,576 stream processors, complemented by 20 Intel CPU cores, 256GB of RAM and a 2TB SSD. The system can be connected to a wide variety of other systems in the corporate network through different interfaces, namely 2x Gigabit Ethernet, 10GigE 2x, or 1x InfiniBand. Designers will be able to order similar systems for $ 50,000 in the near future.