Lighting design analysis / Deep Learning machine

We received a consultation from a customer about a machine that can be shared with the lighting design analysis software LightTools and Deep Learning.It is a request to consider while discussing the balance between cost and calculation speed.

As a specific request in terms of specifications, we would like the CPU to have 16 to 48 threads for Lighit Tools. I'd like to configure the GPU for Deep Learning with RTX 2080Ti or higher so that it can be expanded in the future, and I've heard that the memory is about 256GB and I want to install multiple storages while considering the processing speed and capacity.

Points for selecting parts

• • When the CPU is about 16 to 64 cores
    I think it is better to consider a configuration using AMD Threadripper Pro that covers the range of the expected number of cores, has a relatively high operating clock, and can support multiple GPUs.
    The CPU in this example has 32 cores in the middle, but it can be changed to a 64-core model and a 16-core model.

• About the addition of the second GPU

  Assuming the addition of a second GPU as a retrofit, the RTX2 has high chip performance for external exhaust and deep learning.Please note that when using in a 3090V power supply environment, the number of GPUs that can be added is guaranteed to be +100. The PCI-E slot itself is configured to support up to 1 GPUs, but in a 4V environment the power supply is limited to output up to 100W.The power supply side also supports a 1300V environment, so when used in a 200V power supply environment, it supports up to 200 GPUs in total. Please consider that the use of the 3th sheet is severe in terms of power consumption.

• About storage
  As a proposal considering speed, it is equipped with two high-speed NVMe type SSDs and one 2TB HDD. If you wish to use RAID, please be aware that it may affect other specifications depending on the content of your request.

【Main Specifications】