FAQ

Q: How do the applications of each product differ?

dg4F: Positioned as the flagship GPU server in deep gadget series, dg4F boasts twice the cooling power, enabling efficient cooling of a larger number of GPUs at lower temperatures. Ideal for both server rack installation and standalone use with wheels, it supports up to 16 GPUs, outperforming multiple GPU servers single-handedly. dg4R: Representing over a decade of expertise, dg4R is a mainstream server designed to maintain 8 GPUs at optimal temperatures. Well-suited for server rack deployment, it offers an effective balance between price and performance. dg4W: dg4W is a workstation that can operate independently without the need for a server rack, making it perfect for quiet office environments. With the ability to accommodate up to seven GPUs, it delivers impressive performance with workstation-class CPUs while maintaining quiet operation. dg4A: dg4A is a personal high-performance liquid cooling PC developed by ManyCoreSoft after years of research and development. It excels in various tasks such as simple deep learning research, graphics work, and high-spec gaming. Designed for silent operation, it is suitable for both office and home use.

Q: Is there any additional management required for liquid-cooled servers compared to other servers?

Compared to other servers, deep gadget does not require additional management. The Liquid Cooling System of deep gadget is built-in and operates within a fully sealed circulation structure inside the server, allowing for semi-permanent use. You can confidently utilize it like any other server.

Q. Is it possible to configure the system with specifications not listed in the estimate?

Certainly. If necessary, we can accommodate other specifications not included in the estimate. The specifications listed in the estimate have undergone thorough compatibility and stability tests beforehand. However, configuring with different specifications may affect the delivery timeline due to factors such as additional parts supply and compatibility testing. Please feel free to contact us, and we will provide you with comprehensive assistance and guidance.

Q. How long can I expect delivery to take after placing an order?

Generally, deep gadget is delivered within a minimum of two weeks and up to four weeks after ordering. For precise delivery timing, please consult with your official partner.

Q: What happens if there is a leak in the Liquid Cooling System?

deep gadget is engineered as a fully sealed structure, minimizing the risk of leakage under normal circumstances. However, we do have precautionary measures in place. Our system is equipped to shut down automatically in the event of a critical failure such as a leak caused by physical shock or other factors. Additionally, the amount of coolant that may leak due to our liquid cooling system design is minimal, typically not exceeding tens of milliliters. This level of leakage has negligible impact on other devices and minimal effect on deep gadget itself. In the event of any malfunction, our engineers will promptly address the issue, replacing the problematic part and reinstalling it as needed.

Q: Is it possible to purchase the product in several parts?

Absolutely. If you prefer to purchase the product in multiple installments, we can accommodate that. Simply reach out to your official partner, and they will provide you with comprehensive information tailored to your needs.

Q. Is it possible to add GPUs after purchasing?

Yes, it's possible. In this case, please note that there may be a small fee.

Q: Is it possible to configure a cluster by grouping several servers?

Absolutely. Our system allows for the configuration of clusters by grouping multiple servers. You have the flexibility to set it up as a standard 1G Ethernet network, or opt for higher-speed options like 10G Ethernet or InfiniBand. Additionally, we can integrate task schedulers such as SLURM to optimize cluster performance. Feel free to reach out to us for further details and personalized assistance.

Q. What kind of software installation is possible?

We offer a comprehensive range of software installation options tailored to your needs. If you require software for deep learning purposes, simply let us know, and we will ensure it is installed prior to shipment. Additionally, if you request not only the operating system but also frameworks such as CUDA, PyTorch, and TensorFlow, we will complete the setup and deliver it promptly. Please remember to specify your software requirements when placing your product order.

Q. What are the advantages of a liquid cooling system over traditional cooling methods?

With the increasing performance of modern CPUs and GPUs, system heating has become a pressing concern. Instances of unstable system behavior and thermal throttling (temperature degradation when CPU or GPU exceeds a certain temperature) are becoming more frequent. Even in large data centers equipped with costly air conditioning, high-performance GPU servers encounter these issues. Traditional servers relying on fans for air cooling often struggle to efficiently cool high-performance devices mounted on them. In this context, liquid cooling has transitioned from being an option to a necessity for high-performance systems. ManyCoreSoft's liquid cooling system, honed over more than a decade of supercomputer cooling, consistently maintains devices at low temperatures, ensuring uninterrupted operation at optimal performance levels.

Q. Is a PCIe switch necessary for a multi-GPU server?

In multi-GPU setups, each computer system utilizes the PCI Express (PCIe) interface to connect and communicate with various devices such as CPUs, GPUs, HDDs, SSDs, and NICs. However, the total number of PCIe lanes available in a system is limited by the CPU's specifications. PCIe switches are commonly employed in multi-GPU configurations to share this limited number of PCIe lanes, allowing for the installation of multiple devices within the same lane count. While PCIe switches offer the advantage of accommodating multiple devices, they can also introduce potential issues such as bottlenecks and signal integrity problems due to simultaneous communication across multiple devices. It's worth noting that not all systems in deep gadget incorporate PCIe switches. Instead, direct connections between the CPU and GPU are utilized, thereby conserving PCIe lanes that are crucial for high-performance computing within the server. In these configurations, 8 GPUs are connected via PCIe Gen4 x16, while 16 GPUs are connected via PCIe Gen4 x8.

Q. Is it easy to change the HDD/SSD storage configuration?

Yes, deep gadget simplifies the process of adjusting your storage configuration. With a range of externally hot-swappable HDD/SSD bays, adding or removing storage is straightforward and accessible for users of all levels.

Q. Is deep gadget as noisy as other servers?

Traditional air-cooled GPU servers often produce significant noise levels as their fans work to cool heated components. Noise levels exceeding 80dB can not only be inconvenient but may also pose serious health risks. deep gadget, on the other hand, utilizes a high-efficiency liquid cooling system, resulting in significantly reduced noise levels even under full load compared to air-cooled servers. Additionally, the cooling system operation is adaptively controlled based on the usage environment and load, ensuring quiet operation even during light workloads.