ASRock Rack Announces EP2C612D24 and 4L: Dual Socket Haswell-EP with 24 DDR4 Slotsby Ian Cutress on May 23, 2015 8:10 AM EST
Two things jumped out at me when I received this press release. Firstly the name, which comes out as a mouthful – it isn’t something you could casually mention in conversation, even if you worked closely with the motherboard. The second is the amount of DRAM slots, which is ultimately what the EP2C612D24 and EP2C612D24-4L are catering for.
The Haswell-EP/Xeon E5 v3 DDR4 memory controllers are designed for up to three DIMMs per channel, similarly to Ivy Bridge-EP (E5 v2) and Sandy Bridge-EP (E5), although in all cases it is usually reserved for more niche systems. With 16GB UDIMMs, this allows for a maximum of 384GB, although moving up to RDIMM, LRDIMMs or 64GB NVDIMMs pushes the max to 1.5TB in a dual socket motherboard. Only those with deep pockets, big budgets or stringent requirements need apply, as the major cost here will be the DRAM.
The motherboard uses a staggered processor arrangement with narrow ILM versions of the LGA2011-3 socket. Combined with the 24 DRAM slots means there is little room for anything else. We get three PCIe 3.0 x8 slots which are open ended, allowing for x16 sized cards to come in, although one slot will be limited to reduced width cards as some of the DRAM slots would encroach a super long co-processor. The ten onboard SATA ports are supported by an M.2 PCIe 3.0 x4 slot and an onboard USB 3.0 Type-A port for in-chassis licensing dongles or Live-USB OSes. One downside to mention, according to the specifications for Haswell-EP, is that when fully populated, the memory should reduce down in speed, from 2133 to 1600 MHz.
Network connectivity is provided by an Intel i350 which gives four gigabit ports on the 4L model, but only two on the regular SKU, but we also get another network port for the AST2400 which provides IPMI 2.0 network management. ASRock Rack is targeting the usual suspects when it comes to large DRAM packages – intensive compute tasks, big data analysis, Hadoop and cloud computing.