[[ build.model.nick ]]

Choose the price, not the parts. Each model is built with the GPUs, CPU, RAM, and storage that maximizes Deep Learning performance per dollar.

[[ build.model.image_alt ]]

[[ build.model.nick ]]

Basic

OS Ubuntu 18.04
GPUs 4x NVIDIA 1080 Ti
CPU Intel Xeon E5-1650 v4
Memory 64 GB memory
STORAGE 2 TB SSD (OS install)
EXTRA 4 TB HDD
NETWORK 1 Gbps ethernet

Premium

OS Ubuntu 18.04
GPUs 4x NVIDIA 1080 Ti
CPU Intel Core i9-7920X
Memory 128 GB memory
STORAGE 2 TB SSD (OS install)
EXTRA 4 TB HDD
NETWORK 10 Gbps ethernet

Max

OS Ubuntu 18.04
GPUs 4x NVIDIA Titan V
CPU Intel Core i9-7920X
Memory 128 GB memory
STORAGE 4 TB SSD (OS install)
EXTRA 4 TB HDD
NETWORK 10 Gbps ethernet

Customize

Not seeing what you want?

Add a Protection Plan

3-year protection for [[ getWarrantyPrice(true) ]]

[[ getSubtotal(true)]]
Talk to an engineer
(650) 479-5530

About the Basic

GPUs

GPUs are the most critical piece of hardware for Deep Learning. The Basic has 4x NVIDIA GTX 1080 Ti GPUs (Pascal Architecture). For Deep Learning in 2018, the 1080 Ti offers the best price/performance trade-off of any GPU on the market. Each 1080 Ti has 11.3 TFLOPs of FP32 performance (the standard precision for Deep Learning training). For most tasks, the 1080 Ti is 95% as fast as the NVIDIA Titan Xp and 70% as fast as the NVIDIA Titan V.

Processor

During training, the CPU preprocesses data and feeds it to the GPUs. A slow processor will cause the GPUs to waste cycles waiting for this data. Core count and PCI-e lane count are important CPU performance factors. More cores means faster data preprocessing; more PCI-e lanes means faster transmission of that data to the GPUs. The Basic uses Intel's Xeon E5-1650 v4 (6 cores, 40x PCI-e lanes). Its core-to-GPU ratio is 1.5, which follows the best practice of at least 1 CPU core per GPU. The Basic's CPU, combined with its PLX-enabled motherboard, provides 16x PCI-e lanes to each GPU (the max possible).

Motherboard

A motherboard's PCI-e lanes have a significant effect on Deep Learning training performance. PCI-e lanes are data pipes connecting the GPUs and CPU. The number of PCI-e lanes connecting each GPU to the CPU varies from 4x to 16x, depending on the motherboard. More PCI-e lanes is better because more lanes = more throughput. The Basic's motherboard has PLX chips, which ensures that each GPU gets 16x PCI-e lanes (the maximum possible as of 2018).

Memory

A Deep Learning computer should have at least as much RAM as GPU memory. For example, a machine with 2x NVIDIA 1080 Ti GPUs should have at least 22 GB of memory (1080 Tis have 11 GB of memory each). The Basic has 4x 1080 Ti GPUs and 64 GB of memory, so it follows this rule of thumb. If you work with large data sets (e.g. many large images), consider upgrading to Premium, which has 128 GB of memory.

Storage

Most datasets do not fit in RAM. In such cases, during model training, subsets must be repeatedly swapped in and out of RAM from nonvolatile storage. Such a pipeline requires fast, solid state storage; without it, the GPUs would waste cycles waiting for their next batch of data. The Basic was designed with this constraint in mind; it has two nonvolatile storage devices: a 2 TB solid state drive (fast) for data you're training on now, and a 4 TB hard disk drive (slower) for everything else. Files located in the /data directory are stored on the HDD; all other files are stored on the SSD.

Network

Network interface speed is irrelevant to Deep Learning performance, unless you're doing multi-node distributed training (in which case, you'll want at least 40 Gbps). With 1 Gbps ethernet, this desktop's network interface is far faster than most ISPs provide. If you frequently copy large files between computers, you may consider upgrading to Premium, which has 10 Gbps.

Who bought a Basic?

About the Premium

GPUs

GPUs are the most critical piece of hardware for Deep Learning. The Premium has 4x NVIDIA GTX 1080 Ti GPUs (Pascal Architecture). For Deep Learning in 2018, the 1080 Ti offers the best price/performance trade-off of any GPU on the market. Each 1080 Ti has 11.3 TFLOPs of FP32 performance (the standard precision for Deep Learning training). For most tasks, the 1080 Ti is 95% as fast as the NVIDIA Titan Xp and 70% as fast as the NVIDIA Titan V.

Processor

During training, the CPU preprocesses data and feeds it to the GPUs. A slow processor will cause the GPUs to waste cycles waiting for this data. Core count and PCI-e lane count are important CPU performance factors. More cores means faster data preprocessing; more PCI-e lanes means faster transmission of that data to the GPUs. The Premium uses Intel's Core i9-7920x (12 cores, 44x PCI-e lanes). Its core-to-GPU ratio is 3, which follows the best practice of at least 1 CPU core per GPU. The Premium's CPU, combined with its PLX-enabled motherboard, provides 16x PCI-e lanes to each GPU (the max possible).

Motherboard

A motherboard's PCI-e lanes have a significant effect on Deep Learning training performance. PCI-e lanes are data pipes connecting the GPUs and CPU. The number of PCI-e lanes connecting each GPU to the CPU varies from 4x to 16x, depending on the motherboard. More PCI-e lanes is better because more lanes = more throughput. The Premium's motherboard has PLX chips, which ensures that each GPU gets 16x PCI-e lanes (the maximum possible as of 2018).

Memory

A Deep Learning computer should have at least as much RAM as GPU memory. For example, a machine with 2x NVIDIA 1080 Ti GPUs should have at least 22 GB of memory (1080 Tis have 11 GB of memory each). The Premium has 4x 1080 Ti GPUs and 128 GB of memory, so it follows this rule of thumb. If you work with large data sets (e.g. many large images), a workstation with 128 GB of memory is standard.

Storage

Most datasets do not fit in RAM. In such cases, during model training, subsets must be repeatedly swapped in and out of RAM from nonvolatile storage. Such a pipeline requires fast, solid state storage; without it, the GPUs would waste cycles waiting for their next batch of data. The Premium was designed with this constraint in mind; it has two nonvolatile storage devices: a 2 TB solid state drive (fast) for data you're training on now, and a 4 TB hard disk drive(slower) for everything else. Files located in the /data directory are stored on the HDD; all other files are stored on the SSD.

Network

Network interface speed is irrelevant to Deep Learning performance, unless you're doing multi-node distributed training (in which case, you'll want at least 40 Gbps). With 10 Gbps ethernet, this desktop's network interface is faster than virtually every ISP. If you frequently copy large files between computers, the Premium's 10 Gbps is an excellent feature.

Who bought a Premium?

About the Max

GPUs

GPUs are the most critical piece of hardware for Deep Learning. The Max has 4x NVIDIA Titan V GPUs (Volta Architecture). The Titan V is powered by the same chip as the NVIDIA Tesla V100. Each Titan V has 13.8 TFLOPs of FP32 performance, the standard precision for Deep Learning training. For most tasks, the Titan V is about 42% faster than the 1080 Ti and 40% faster than the Titan Xp.

Processor

During training, the CPU preprocesses data and feeds it to the GPUs. A slow processor will cause the GPUs to waste cycles waiting for this data. Core count and PCI-e lane count are important CPU performance factors. More cores means faster data preprocessing; more PCI-e lanes means faster transmission of that data to the GPUs. The Premium uses Intel's Core i9-7920x (12 cores, 44x PCI-e lanes). Its core-to-GPU ratio is 3, which follows the best practice of at least 1 CPU core per GPU. The Premium's CPU, combined with its PLX-enabled motherboard, provides 16x PCI-e lanes to each GPU (the max possible).

Motherboard

A motherboard's PCI-e lanes have a significant effect on Deep Learning training performance. PCI-e lanes are data pipes connecting the GPUs and CPU. The number of PCI-e lanes connecting each GPU to the CPU varies from 4x to 16x, depending on the motherboard. More PCI-e lanes is better because more lanes = more throughput. The Max's motherboard has PLX chips, which ensures that each GPU gets 16x PCI-e lanes (the maximum possible as of 2018).

Memory

A Deep Learning computer should have at least as much RAM as GPU memory. For example, a machine with 2x NVIDIA Titan V GPUs should have at least 24 GB of memory (Titan Vs have 12 GB of memory each). The Max has 4x NVIDIA Titan V GPUs and 128 GB of memory, so it follows this rule of thumb. If you work with large data sets (e.g. many large images), a workstation with 128 GB of memory is standard.

Storage

Most datasets do not fit in RAM. In such cases, during model training, subsets must be repeatedly swapped in and out of RAM from nonvolatile storage. Such a pipeline requires fast, solid state storage; without it, the GPUs would waste cycles waiting for their next batch of data. The Max was designed with this constraint in mind; it has two nonvolatile storage devices: a massive 4 TB solid state drive (fast) for data you're training on now, and a 4 TB hard disk drive (slower) for everything else. Files located in the /data directory are stored on the HDD; all other files are stored on the SSD.

Network

Network interface speed is irrelevant to Deep Learning performance, unless you're doing multi-node distributed training (in which case, you'll want at least 40 Gbps). With 10 Gbps ethernet, this desktop's network interface is faster than virtually every ISP. If you frequently copy large files between computers, the Max's 10 Gbps is an excellent feature.

Who bought a Max?

[[ component.name ]]

[[ option.description ]] [[ build.getPriceDiff(component, option)]]