New NVIDIA EULA rules for data centres: answering customers’ most frequent questions

Recently have received a lot of questions from our customers about the introduction of the new NVIDIA EULA licence agreement, which prohibits data centres from using video cards from the GeForce line (the only exception is blockchain processing). For this reason, we have decided to answer the most common questions, as well as debunk some myths at the same time. 

One solution to this problem is to refuse to update the drivers. Will this create any problems related to security?

Refusing to update the drivers carries no risk to LeaderGPU customers. We offer bare metal servers, i.e. only you have access to their resources. We do not use virtualisation, which thereby reduces any security risks to zero. Therefore, no matter which version of the driver is used, you get a reliable secure service.

At the moment NVIDIA offers two versions of the new drivers. The version provided by the link does not have any limitations for the data centres:

Other solutions to this problem from LeaderGPU

.In order to expose such restrictions, it is also necessary first to give a clear definition of the term "data centre" - the subject of the agreement. In the NVIDIA licence agreement this definition is missing. 

Also representatives of NVIDIA noted that the use of GeForce and Titan video cards in the new licence agreement is possible for research and non-commercial purposes. Deep training falls into the category of research.

If you need new drivers, you can install them for deep learning using blockchain - this option is allowed by a new agreement from NVIDIA. Note: cryptocurrency mining using LeaderGPU is prohibited! You can use blockchain for other operations (excluding mining), for example, to record the performance of tasks, etc.

We have GPU-servers with P100 / V100 video cards, which have no usage restrictions. You can always rent them from LeaderGPU.

In addition, we are actively working on a solution based on FPGA for deep learning. We will announce it in the near future.

Furthermore, we have AMD video cards, which we actively test with popular software for deep learning. If NVIDIA representatives decide to limit the use of GeForce video cards in future versions of drivers, we will quickly switch to AMD video cards.

P.S. This is not a legal advice and we reject any responsibility for any claims. In disputed situations, we recommend you in contact a lawyer specialising in software licencing. 


Security under control: the detected vulnerability in Intel CPUs did not affect LeaderGPU service levels

In early January, a technological defect was found in Intel's chips, which opens the way for attackers to carry out various malicious activities. The defect is associated with the kernel - the core of the operating system, which, at the most basic level, provides the interaction between the OS and the processor. 

The vulnerability is related to the way the Microsoft Windows kernel interacts with Intel processors. Theoretically, a hacker can use this interaction to bypass standard security measures. The solution to the problem lies in rethinking the interaction between the core and the processor, but the consequences of this approach may be not so good: experts believe that this will directly affect performance.

Why doesn’t this vulnerability affect LeaderGPU clients?

The vulnerability found in Intel processors has significant consequences for companies and ordinary users, but it does not affect LeaderGPU customers. Why?

  1. We offer bare metal servers, which means the CPU and the physical server are completely isolated from any third-party activity. Only you can use the server. No virtualisation, no shared resources.
  2. We use advanced protection techniques against any third-party impact.

You can be assured of the unprecedented security and reliability of LeaderGPU servers. We guarantee it!


LeaderGPU expands its market coverage: now the company's GPU-servers are located in a new facility in Rotterdam

The main goal of LeaderGPU is achieving provision of modern GPU-computing for customers around the world with the highest performance possible. Recently, another important milestone was reached towards this goal: LeaderTelecom, which is accountable for the LeaderGPU project, signed a contract with SmartDC, a data center based in Rotterdam. 

The uniqueness of SmartDC is that the company offers its customers separate rooms with independent cooling systems, power supply, security and admittance with individual protection.

SmartDC data center has Tier 3 certificate (fault tolerance - 99.982%) and ISO 27001 certificate, which is one of the leading international standards of information security management. All this helps to ensure stability and effectiveness of providing LeaderGPU services to its customers.

Expansion to the world market is LeaderGPU top priority. The company persistently expands the range of its services, promptly responding to current trends in the field of GPU-computing. Recently, company offered its customers GPU-servers with Tesla V100 GPUs, abandoning obsolete Tesla P100 cards. The new offer is already available for all existing and prospective users.

Rent GPU-servers with the most up-to-date and proven GPUs at affordable prices in LeaderGPU - assess the impeccable quality of the offered services!


Cutting-edge servers with Tesla V100 video cards – already available from LeaderGPU!

As part of the LeaderGPU project, LeaderTelecom has presented a new service – the rental of ultramodern high-performance servers based on Nvidia Tesla V100 GPU accelerators. It is possible to rent servers with Nvidia Tesla V100 GPU via per-minute payment. This offer currently exists nowhere else on the market – only here.

The training of neural networks for ResNet-50 using Tesla V100 occurs 2.4 times faster than with Tesla P100. Using the Tesla V100 demonstrates unprecedented performance in deep learning tasks: more than 120 TFLOPS! This is a unique result, which you can also evaluate by renting GPU servers with Tesla V100 in LeaderGPU.

Advantages of the Tesla V100:

  • New infrastructure of streaming multiprocessors, which offers a 12-fold increase in performance for training neural networks in comparison with Tesla P100.
  • 16 GB HBM2 memory, guaranteeing a peak bandwidth of 900 GB/sec
  • Optimised software for Volta. Libraries such as cuDNN, cuBLAS and TensorRT use the new Volta GV100 capabilities to achieve maximum performance in the field of deep learning.

Taking into account the significant advantages in performance, LeaderTelecom decided to ditch the obsolete Tesla P100 cards in favour of the more up-to-date Tesla V100, which are available for rent with payment per minute, week and month on

Touch the future together with LeaderGPU - rent the most up-to-date GPU-servers with Tesla V100 for high-performance computing, machine learning, rendering, etc.


LeaderGPU has joined Nvidia Inception Program

The LeaderGPU has become a participant of the Nvidia Inception Program which was established to sustain startups and companies that rely on using modern technologies of deep learning and artificial intelligence in their business activities.

The Nvidia Inception Program offers number of advantages for each participant such as, marketing support, hardware supply, training materials access, and much more. LeaderGPU strongly believes that cooperation with Nvidia will bring a positive outcome as well as opportunity to offer our customers distinguish functionality by expanding our technical infrastructure and reaching out to new markets. 

Nvidia Inception Program offers an extensive range of various marketing tools which would be helpful in managing podcasts, blogs, channels and groups in social networks, holding thematic events, recording video, etc. This marketing tools variety primarily focused on assisting customers with identifying and solving main issues, their company is currently facing, as well as adding appropriate services to their product range, which will affect company earnings.

Nvidia Inception Program provides access to a wide-ranging library of educational materials - great opportunity for our experts to follow up with latest industry trends. Rest assured that our specialists will always provide detailed, technically competent support even for the most difficult issues.

Another vast benefit of the Nvidia Inception Program is the prospect of getting investments from Nvidia within the GPU Ventures project. Financing is carried out for those companies that invest in the development of artificial intelligence, create innovative solutions in this area, change the established paradigms of computation.

Nvidia Inception is one more step forward for the LeaderGPU project. You can be confident that we offer only the most effective, cutting-edge, stable and scalable solutions that can quickly cope with any task. Order a server rental with a GPU right now at favorable prices with per minute, week or month access payment plans!