DLC CPU V2 Module For Intel Processors (LGA 3647/4189)

We know from experience all the troubles related to positive pressure water cooling systems. Risk of flooding the costly systems is biggest concern for infrastructure operators. This is why we chose state of the art technologies to provide completely safe, 100% leak-proof system. We guarantee complete safety using three technologies:

• One is VCDU negative pressure vacuum based system that guarantees total safety in difference to positive pressure CDU’s. In an unlikely event of integrity loss, Cool-Flo® technology works with negative pressure on both supply and return, so if anything happens air will flow into the system instead of coolant leaking out. The “leak” does not stop the operation and electronic components are safe from water damage. The system also automatically evacuates coolant from a server when it is disconnected from a liquid cooling system for maintenance.
• Second feature: proprietary nanoparticles enhanced engineered fluid. We use two types of dielectric fluids for DLC and ILC cooling. Single or dual phase dielectric fluid option provides another layer of protection.
• Third component: over engineering policy. All modules, ports, tubing, dry break quick disconnects and couplings are more robust than necessary. We use high pressure rated  components in our low-pressure system. We choose best in class simple solutions – e.g. flat tubing so the loop is twice as much durable than corrugated. Our tube can withstand over 650 PSI, when typical pressure in our system is up to 10 PSI. Because safety & uptime matters most.

Liquid cooling technology provide massive saving for hyperscale, enterprise and smb installations. Even individual users see significant cost cut. Savings can be measured in CAPEX and OPEX so ROI is immediate. In terms of CAPEX there are three biggest factors:

• cost of liquid cooling systems is usually lower than air cooling infrastructure of the same capacity. In most cases it’s 15-20% less.
• liquid cooling renders the need for costly HVAC infrastructure which effects in fever pieces of critical equipment in data hall space. Liquid cooled servers require only 20% of previous airflow with allows for free-cooling during all seasons. Of course immersion cooling requires virtually zero airflow thus providing another discount.
• Increase in  rack power density (from 10-20 kW to 100 kW and beyond) allows to lower data center footprint, resign from air exchange plenum, stack data center vertically. Fewer server racks and interconnects means another reduction in capital cost.

Both of this factors results in reduced site & structural construction, compared to traditional build. Simplified electrical and mechanical topology and faster go to market gives tremendous advantage to data centers outfitted with liquid cooling, over air cooled designs. For new data center projects, the cost savings is even more dramatic as capital expenditures can be cut nearly in half. 

The biggest advantages however starts with OPEX:

• For most installations we observe an overall reduction in average data center power consumption by up to 45%. The savings comes from a combination of HVAC elimination, reduced infrastructure footprint, and reduced fan power consumption. This cuts lot’s of operational cost associated with power cost.
• Increase in computing power of liquid cooled processors and gpu’s effects in 15-20% more performance. For specific application it can reach with additional fine tuning and overclocking even 30% more compared to air cooled server rooms. This means we can spend 20% less on computing infrastructure having more performance for less.
• Increased reliability of equipment through elimination of the most of airflow. Liquid protects IT devices from harsh environment including high temperature, humidity, vibration, dust, air contamination extends MTBF of ICT infrastructure and extends lifespan of systems.
• Riddance of most of aircooling critical infrastructure that must be maintained on regular basis, less maintenance overhead, means reduced maintenance and personel cost

Currently available liquid cooling systems are highly customised – “boutique” solutions. Water cooling coldplates and tubing length must be carefully prepared for one and one only specific server model. Even the size of the rack which holds the device must be taken into consideration, as tubing length between the coldplates and manifold must be carefully measured. All recognised hardware vendors propose liquid cooling for one single server in their portfolio – in the same way as IBM in Manframe s/360 servers in 1965.  In case of immersion cooling available solutions support only some types of servers, switches and storage systems because of dimensions of immersion baths. Summing up – all solutions we saw and tested were unique and proprietary.

Industry however requires open solutions, and typical cloud / colo infrastructure is usually diversified with many server vendors and types. In case of immersion cooling our universal enclosures and computing enclosures can accommodate variety of servers, switches and storage systems from a-brands and oem vendors. DCX believes in open systems and our liquid cooling systems support over 95% of available hardware without specific customisation. In case od direct chip cooling – DCX patent pending Next Gen DLC system features Push To Lock quick disconnect ports on cooling module / rack level. Customer can purchase Next Gen DLC components as every other third party product: memory chip, disk drive or gpu card. Administrators can link DLC modules with our proprietary tubing and create the loop for every server and platform worldwide – without being forced to order custom solution for each device.

The best is that using provided leak- proof tubing & PTL quick disconnects the Customer can reconfigure the LOOP if needed and move to the next generation server, utilising the same socket, cooling modules, and LDUs (liquid distribution units). This extends lifetime and ROI of Next Gen Liquid Cooling components  three times comparing to current DLC offering.

Modern, energy efficient CPU’s, GPU’s and memory chips are subject to thermal throttling. Vendors know that overheating can cause errors and accelerate component failure. This is why all existing hardware operate with lower performance than advertised if reaching certain thermal point. In Intel Skylake architecture AVX-512 and heavy AVX2 instructions throttle the CPU’s frequency. This is why Intel diversifies processor models to “thermally optimised” and not In case of memory – command rates are reduced if system works over certain limits.

One need to realise that in real case application most of the customers get 20% less performance from their chips – if CPU temperature reaches over 55C/130F – cpu frequency will be reduced and performance hit may be even higher than 20%. For most of Nvidia GPU’s – thermal throttle point starts at 50C/122F and clock frequency will drop – step by step to 50% of base Mhz.

To put it simply – you pay 100%  price for the chip and get 80% of performance in real world applications. There are two ways to cope with that issue: one is to use power hungry chillers to run the systems in cold air. The second is to use direct liquid cooling and extract the heat with warm fluid – at the source and keep the chips at optimum 50C/120F temperature. Liquid cooling allows also, with additional fine tuning and overclocking, to turbo boost your chips getting additional increase in performance of  15% to 25%. Without any compressed cooling.

Sharp increases in energy prices have forced many IT pros to look at how inefficient existing cooling practices are. Traditionally only half the total data center energy is used at the equipment with typically about 30-45% or even over 50% of the total data center energy is consumed by the cooling infrastructure. Most of this is consumed by the site chiller plant, used to provide chilled water to the data center, and by computer room air conditioners (CRAC) and air handlers (CRAH), used to cool the computer room. With average PUE of 1.89 for many datacentres over 50% of energy consumption and carbon footprint is not caused by computing but by powering the necessary cooling systems to keep the processors from overheating.

Datacenters use currently 3% of the world’s energy (around 420 terawatts) which is around 45% more than the entire United Kingdom energy spend. And this consumption will double every four years. Researchers predicts that by 2025, data centres will amount to largest share of global electricity production at 33% This is why EU Commission issued EU Code of Conduct on Data Centre Energy Efficiency along with best practice guidelines.

DCX systems save over 50 % of typical data center’s energy consumption with hot liquid cooling and 30% save can be expected in small scale consumer systems. For most installations we observe an overall reduction in average data center power consumption by up to 45%. The savings comes from a combination of HVAC elimination, reduced infrastructure footprint, and reduced fan power consumption. This cuts lot’s of operational cost associated with power cost. Increase in computing power of liquid cooled processors and gpu’s effects also in 15-20% more performance. For specific application it can reach with additional fine tuning and overclocking even 30% more compared to air cooled server rooms. This means we can do the same having 15-20% less servers.

Moreover, high-grade heat at the output can be used for such needs as heating building spaces. In our liquid cooling installations, with 50-60*C hot fluid output we reuse from 64% to over 80% of usually wasted heat. This is why liquid cooling sites demonstrate the extreme energy efficiency and can deliver heat to local community or office space.

Climate change is recognised as one of the key challenges humankind is facing.
The Information and Communication Technology (ICT) sector including data centres generates up to 2% of the global CO2 emissions, a number on par to the aviation sector contribution. Problem with supply of renewable sourced energy will make data centres one of the biggest polluters in just seven years.

Additionally data centres are estimated to have the fastest growing carbon footprint from across the whole ICT sector, mainly due to technological advances such as the cloud computing and the rapid growth of the use of Internet services. ICT industry is posed to be responsible for up to 3.5% of global emissions by 2020, with this value potentially escalating to 14% by 2040, according to Climate Change News. Researchers say this will be directly related to the fact that the data centre sector could be using 20% of all available electricity in the world by 2025 on the back of the large amounts of data being created at a fastest speed than ever before seen.

Our solutions render the need to use high GWP refrigerant associated with chilling plants, reduce 3 tons carbon dioxide emission per kW of ICT equipment in one year and reduce power consumption on average by 50%. Simple as that. Moreover, high-grade heat at the output can be used for such needs as heating building spaces. In our liquid cooling installations, with 50-60*C hot fluid output we reuse from 64% to over 80% of usually wasted heat. This is why liquid cooling sites demonstrate the extreme energy efficiency and can deliver heat to local community or office space.