Procurement of a Double-Precision General-Purpose Graphics Processing Unit (GPGPU) Server

    
    COMBINED SYNOPSIS / SOLICITATION WORKFORM

(1)    Action Code:  Combined Synopsis/Solicitation

(2)    Date: 7/24/2021

(3)    Year: 2021

(4)    Contracting Office Zip Code: 20892

(5)    Project Service Code: 7B22 - IT AND TELECOM - COMPUTE: SERVERS (HARDWARE AND PERPETUAL LICENSE SOFTWARE)

(6)    Contracting Office Address:
6710B Rockledge Drive
Bethesda, MD 20892

(7)    Subject/Title: Procurement of a Double-Precision General-Purpose Graphics Processing Unit (GPGPU) Server

(8)    Proposed Solicitation Number: NICHD-21-215

(9)    Closing Response Date: 8/2/2021

(10)    Contact Point: Amber Harris

(11)    Contract Award and Solicitation Number: TBD

(12)    Contract Award Dollar Amount: TBD

(13)    Contract Line Item Number(s): TBD

(14)    Contractor Award Date: TBD

(15)    Contractor Name: TBD

(16)    Description: (Background)

(i)    “This is a combined synopsis/solicitation for commercial items prepared in accordance with the format in Subpart 12.6 as supplemented with additional information included in this notice.  This announcement constitutes the only solicitation; quotes are being requested and a written solicitation will not be issued.” 

(ii)    The solicitation number is NICHD-21-215 and the solicitation is issued as a request for quote (RFQ).

(iii)    The solicitation document, the incorporated provisions and clauses are those in effect through Federal Acquisition Circular  2021-06 Effective July 12, 2021.

(iv)    The associated NAICS code is 334118, COMPUTER TERMINAL AND OTHER COMPUTER PERIPHERAL EQUIPMENT MANUFACTURING, and the small business size standard is 1,000.  The acquisition is being conducted in accordance with the procedures of FAR Part 12 and FAR Part 13. 

(v)    The Contractor Requirements are listed below:

1.0    PURCHASE ORDER TITLE

Procurement of a Double-Precision General-Purpose Graphics Processing Unit (GPGPU) Server
        
2.0    BACKGROUND

We develop methods for analyzing biomedical data for finding predictive computational models for diseases. Deep learning algorithms are playing a larger role in image analysis and large-scale data analysis and comparing deep learning analysis techniques to mechanistic modeling is a major part of how we approach disease modeling. We are developing algorithms that can provide mechanistic understanding of biomedical disease states with accuracy as good or better than black-box neural network approaches. An example that requires double-precision accuracy in computations is the calculation of effective quantum Hamiltonians from short duration molecular dynamics simulation data.

In order to do this, we require a GPGPU machine with interconnected GPU cards. GPGPU units have until recently been inefficient at computing at double-precision. However, advances in technology have made double-precision GPGPU servers available now. With a state-of-the-art GPGPU machine with interconnected GPU cards, we aim to efficiently solve the problems of interest for NIDDK’s goals. This double-precision GPGPU machine will enable us to develop more accurate models for (1) calculating effective quantum Hamiltonians to model protein dynamics in health and disease; (2) develop reservoir computing models for efficient neural network learning of classification models to distinguish between disease and healthy states from large-scale datasets.

3.0    OBJECTIVE

The objective is to acquire a state-of-the-art GPGPU machine with interconnected GPU cards to efficiently solve the problems of interest.  As GPGPU units have until recently been inefficient at computing at double-precision. Advances in technology have made double-precision GPGPU servers available now.

4.0     SALIENT / REQUIRED FEATURES AND SPECIFICATIONS

Double-precision GPGPU server:
•    64 core CPU, 2.0 GHz
•    8 x 64GB ECC DDR4 3200MHz RAM
•    2 x 240 GB or larger DataCenter or Enterprise Class SATA SSD for 

Operating System
•    2 x 4 TB NVMe data drives
•    Software RAID1 for OS drives
•    Software install: Linux OS and GPGPU computation libraries
•    2 x 40 GB double-precision GPGPU cards
•    Bridge linking 2-slot GPGPU cards 

Towards NIDDK’s goals, this double-precision GPGPU machine will enable us to develop more accurate models for (1) calculating effective quantum Hamiltonians to model protein dynamics in health and disease; and (2) develop reservoir computing models for efficient neural network learning of classification models to distinguish between disease and healthy states from large-scale datasets.

5.0     TASKS AREAS

Independently, and not as an agent of the Government, the contractor shall perform all of the detailed Task Areas listed below:

5.1    Task Area 1 – The vendor shall provide a computer with the stated characteristics no later than August 30, 2021. The vendor will provide a pre-configured system that can be installed by CTB personnel.

5.2    Task Area 2 – The vendor shall ship the server to the NIDDK CTB Unix group for installation. Shipping address: Errin Frahm (NIDDK Hosting Customer), 5601 Fishers Lane, Rockville, MD 20892, (Attention: MelLoyd Graves).

5.3    Task Area 3 – The server will be rack-ready for installation with partitions set up according to CTB requirements.

5.4    Task Area 4 – The vendor will install a Linux distribution specified by NIDDK CTB and GPGPU computing libraries suited to the GPGPU cards in the server.

5.5    Task Area 5 – No training requirements.

6.0    PURCHASE ORDER TYPE

This Purchase Order is to be a firm fixed price. 

7.0    INSPECTION AND ACCEPTANCE

Inspection and acceptance will be performed at the place of installation/upon delivery of service - National Institutes of Health, NIDDK, NIH main campus, in Bethesda, MD 20892.

(vi)    Period of Performance:  

Delivery by August 30, 2021

(viii)    The provision at FAR clause 52.212-1, Instructions to Offerors – Commercial Items, applies to this acquisition.

(vii)    The provision at FAR clause 52.212-2, Evaluation – Commercial Items, is applicable.  Award will be made to the contractor providing the most advantageous quote to the Government.  Technical factors, past performance, and price considered with technical factors outweighing price.  

Specifically, the following technical factors will be evaluated:

i.    The vendor will be evaluated on their quoted services meeting the specifications under section 4.0 SALIENT / REQUIRED FEATURES AND SPECIFICATIONS

(x)    FAR clause 52.212-3, Offeror Representations and Certifications – Commercial Items, is applicable.  An offeror shall complete only paragraphs (b) of this provision if the offeror has completed the annual representations and certificates electronically via http://www.acquisition.gov. If an offeror has not completed the annual representations and certifications electronically at the System for Award Management (SAM) website, the offeror shall complete only paragraphs (c) through (p) of this provision.
 
(xi)    FAR clause at 52.212-4, Contract Terms and Conditions – Commercial Items, applies to this acquisition.

(xii)    FAR clause at 52.212-5, Contract Terms and Conditions Required to Implement Statutes or Executive Orders – Commercial Items, applies to this acquisition.  Including the following clauses and Prohibitions:

i.    FAR - 52.204-25 - Prohibition on Contracting for Certain Telecommunications and Video Surveillance Services or Equipment. (AUG 2020).
ii.    52.219-28, Post Award Small Business Program Representation
iii.    52.222-19, Child Labor—Cooperation with Authorities and Remedies (Jan 2014)
iv.    52.222-21, Prohibition of Segregated Facilities
v.    52.222-26, Equal Opportunity
vi.    52.222-36, Equal Opportunity for Workers with Disabilities
vii.    52.222-50 Combating Trafficking in Persons
viii.    52.223-18, Encouraging Contractor Policies to Ban Text Messaging while Driving 
ix.    52.225-13, Restrictions on Certain Foreign Purchases

Additional Clauses:

i.    52.232-33, Payment by Electronic Funds Transfer— System for Award Management (Jul 2013) (31 U.S.C. 3332)
ii.    52.232-39, Unenforceability of Unauthorized Obligations.

(xiii)    Additional Contract Requirements: None

(xiv)    The Defense Priorities and Allocations System (DPAS) and assigned rating are not applicable. 

(xv)    Offers shall be submitted via email to amber.harris@nih.gov 

(17)    Place of Contract Performance: 
National Institute of Health, NIDDK
5 Memorial Drive
Bethesda, Maryland 20892

(18)    Set-aside Status: This requirement is not being set aside for small business.