March 30, 2016
Writing and Submitting a Successful XSEDE Proposal - Bridges
Ken Hackworth
Overview • References&Terms• GuidelinesforXSEDEResearch(XRAC)Request• OtherDocuments• ReviewCriteria• OverviewofProposal(Request)TypesandAcDons• XSEDEAwards(AllocaDons)• XSEDESystems(Resources)• ProceduresforsubmiFngAllocaDonrequest• Whyrequestsarereduced?• Metrics
Terminology
• Startup Development/tes,ng/por,ng/benchmarking
• EducaDon Classroom,Training• Research Produc,onwork
• PI PrincipalInves,gator• XRAS XSEDEResourceAlloca,onsSystem• XRAC XSEDEResourceAlloca,onsCommiDee• SU ServiceUnit=1Core-hour
Allocation Request Types
3 Types of XSEDE Projects
“Traditional” v. Community
• XRACproposalsareacceptedinthreegeneralcategoriesofresearchacDviDes– SinglePrincipalInvesDgator– LargeresearchCollaboraDons– CommunityServices(e.g.,XSEDEGateways)
• Thegeneralrequirementsforproposalsofalltypesremainlargelythesame.– WhetherrequesDngcompute,storage,visualizaDon,orextendedcollaboraDvesupportorsomecombinaDon
General Proposal Outline I. ResearchObjecDvesII. ComputaDonalmethodology(ApplicaDons/Codes)III. ApplicaDonefficienciesIV. ComputaDonalResearchPlanV. JusDficaDonforSUs(TB)requestedVI. AddiDonalconsideraDons
Note:Sec,onsIIIandIVareoLenintegrated.
I. Research Objectives • TradiDonalproposals
– DescribetheresearchacDviDestobepursued• Communityproposals
– DescribetheclassesofresearchacDviDesthattheproposedeffortwillsupport.
• Keepitshort:YouonlyneedenoughdetailtosupportthemethodsandcomputaDonalplanbeingproposed.
• TIP—Reviewersdon’twanttoreadtheproposalyousubmi^edtoNSF/NIH/etc.,buttheyneedtoseeifyouhavemerit-reviewed(grant)funding.
II. Computational Methods (and Applications/Codes used) • VerysimilarbetweentradiDonalandcommunityproposals.• Forcomputerequests
– DescribetheApplicaDonsandcomponentsyouwilluse.– Describethemethods/algorithmsemployedinyourcomputaDonal
research– Describecodedevelopment/features/advances‘home-grown’codes.
• Forstoragerequests– ProvidedescripDonofdatatobestored(organizaDon,formats,
collecDonmechanisms,permissionsgrantedorreceived)– Describetheamountandexpectedgrowthofdatatobestored.
III. Application Efficiencies • VerysimilarbetweentradiDonalandcommunityproposals.• Forcomputerequests
– ExplainwhyyouchosespecificresourcesforyourapplicaDons.– Provideperformanceandscalingdetailsonproblemsandtestcases
similartothosebeingpursued.(Whatistheappropriatescaleforyourproblem?)
– Ideally,provideperformanceandscalingdatacollectedbyyouforthespecificresource(s)youarerequesDng.Mandatorydocument!
• Forstoragerequests– Explaintheefficiencyofyourstoragealgorithmsandprotocols.– DescribeandesDmatetheexpectedcostsofscalingtolargerdatasets
andalargernumberofclients.
IV. Computational Research Plan • TradiDonalproposals
– Explicitlydescribetheproblemcasesyouwillexamine• BAD:“…adozenorsoimportantproteinsundervariouscondiDons…”• GOOD:“…7proteins[listedhere;includescienDficimportanceoftheseselecDonssomewhere,too].Eachproteinwillrequire[X]numberofruns,varying[x]parameters[listedhere][inveryspecificandscienDficallymeaningfulways]…”
• Communityproposals– Explicitlydescribethetypicaluse-case(s)thatthegatewaysupports
andthetypeofrunsthatyouexpectuserstomake– Describehowyouwillhelpensurethatthecommunitywillmake
scienDficallymeaningfulruns(ifapplicable)• BAD:“…thegatewayletsusersrunNAMDonXSEDEresources…”• BETTER:“…userswillrunNAMDjobson[biologicalsystemslikethis]…”• BETTERSTILL:“…thegatewayallowsuserstorunNAMDjobsonupto128processorsonproblemsizeslimited[insomefashion]…”
V. Justification of SUs, TBs
• TradiDonal,Researchproposals– Ifyou’vedonesecDonsII,IIIandIVwell,thissecDonshouldbeastraighmorwardmathproblem
– Foreachresearchproblem,calculatetheSUsrequiredbasedonruns(baseunits)definedinIVandtheDmingsinsecDonIII,brokenoutappropriatelybyresource
• ReasonablescalingesDmatesfromtest-caseDmingrunstofull-scaleproducDonrunsareacceptable.
– ClearpresentaDonherewillallowreviewerstoawardDmeorstorageinaraDonalfashion
– AnalogouscalculaDonsshouldapplyforstoragerequests
V. Justification of SUs, TBs • Community(gateway-type)proposals
– Thefirstbigtrick:CalculaDngSUswhenyoudon’tknowthepreciserunstobemadeapriori.
– InYear2andbeyond• StartwithanesDmateoftotalusagebasedonprioryear’susagepa^ernsandesDmateforcomingyear’susagepa^erns.
• FromthisinformaDon,alongwithdatafromsecDonsIVandIII,youcancomeupwithatabulaDonofSUesDmates.
– Year1requiresbootstrapping• PickconservaDvevalues(andjusDfythem)forthesizeofthecommunityandrunstobemade,andcalculateSUs.
• TIP—Startmodestly.Ifyouhave~0users,don’texpectthereviewerstobelievethatyouwillgetthousands(orevenhundreds)nextyear.
– AnalogouscalculaDonsforTBsofstorageneeded
VI. Additional Review Considerations ! Abilitytocompletetheworkplandescribed
(moresignificantforlargerrequests)– Sufficientmerit-reviewedfunding– Staff,bothnumberandexperience
! LocalcompuDngenvironment! SpecialNeeds! OtheraccesstoHPCresources
– (e.g.,Campuscenters,DOEcenters,etc.)
VI. Additional Considerations Community(gateway)proposalsthesecomponentscanprovidekeydetails:– CommunitySupportandManagementPlan
• Describethegatewayinterface—intermsofhowithelpscommunityburnSUsoraccessTBs.
• Describeplansforgrowingtheusercommunity,“graduaDng”userstoResearchallocaDonawards,regulaDng“gatewayhogs”
– Progressreport• Theactualusercommunityandusagepa^erns• Manuscriptsthankingthisservice,orlistarDclesreferencingXSEDE.
– LocalcompuDngenvironment– OtherHPCresources
Renewals require a Progress Report
• ForResearchProjectRenewalandSupplementRequests– SummaryofScien,ficDiscoveries– AccomplishmentsofComputa,onPlan
• Usage• AchievementsoftheComputaDons(moredetailthansummary).
– Specifythenumberofpublica,ons,conferences,reportsthatresultfromXSEDEsupport.
– Contribu,onstootherresearchefforts.(experimental/computaDonal/instrumental,etc.).
Other Documents:
Required:• CVsforPIsandCo-PIs(2pages)• AddingpublicaDonsintheXRASsubmissionformpullingfrom
portalprofileOpDonal:– SpecialRequirements– References(Iftheywon’tfitinMainDoc.)– Other
Review Criteria • AppropriatenessofMethodology
– ApplicaDons,methods,algorithmsandtechniquesreasonablydescribedandmoDvated?
– Datausageandaccessmethodsclearlydescribed?
• AppropriatenessofResearchPlan– AreresearchobjecDvesexplainedandhowtheywillbeachieved?– DoproposedcomputaDonslistsimulaDonparametersthatareneededto
obtainresults?– Detailsofhumanresourcesdevotedtoaccomplishthesetasks?
• EfficientUseofResources– Areresourcesselectedbeingusedefficientlyaspossibleandinaccordance
withtherecommendeduseguidelinesoftheresources?– Isperformanceandscalingdatepresentedfortheresourcerequested?– AretheredetailsoftheeffortstoimprovetheopDmizaDonand/or
parallelizaDonoftheapplicaDons?
XSEDE Projects AnXSEDEProjectislikeabankaccountforallocaDons.
– Itispermanent,onlyoneperPI.– Itholdsayear’sworthofallocaDon(on1ormoresystems)– PI’srequestanallocaDonrenewaleachyearthereaqer.– AnAllocaDonawardedtoaNewRequestcreatesanXSEDEProject.
API’sComputaDonalProjectsevolveovertheyears.– ComputaDonalProjectsbegin,endandextend.– InsubsequentyearssuccessfulRenewalRequestsprovideallocaDonsfornewComputaDonalProjectsunderthesameXSEDEProject.YourXSEDEProjectremainsthesame.
– ARenewalRequestsisjustlikeNewRequest,butmustcontainaProgressReportoflastyear’sComputaDonalProjectsandlistofpublicaDonsfrompastyear’sallocaDon.
Eligibility
• PrincipalinvesDgator(PI)mustbearesearcheroreducatorataU.S.-basedinsDtuDon,includingfederalresearchlabsorcommercialorganizaDons,(Commercialrequestsmustguaranteethattheirresultsarepublicallyavailable,andworkmustbeincollaboraDonwithanopenscienceorganizaDon.)
• ApostdoctoralresearcheriseligibletobeaPI.
• AqualifiedadvisormayapplyforanallocaDonforhisorherclass;butahighschool,undergraduateorgraduatestudentmaynotbeaPI.
Overview: Research Request
• Webforms:InvesDgator,Grants,ResourceRequest,…• RequiresMainDoc.=“proposal”(pdfupload)&CV• ReviewedbyexpertsinsameFieldofScience• 2.5monthsfromdeadlinetoawardavailability• Details:
– Alloca,onSize:Unlimited– Reviewed: Quarterly– Deadlines: 15thofOctober,January,April,July– AwardsBegin:1stofJanuary,April,July,October
portal.xsede.org "Allocations"Submit/Review Request **
Overview: Startup/Education Requests
• Webforms:InvesDgator,ResourceRequest,…• RequiresonlyanabstractandCV• ReviewedbyaXSEDEStaff(StartupAllocaDonsCommi^ee)• 2weeksfromsubmissiontoawardavailability• Forcodedevel/performanceeval/small-scalingcomputa,ons/
classroom&traininginstruc,on• Details:
– Requestlimit: 200,000SUstotalorcombina,onofallresourcesrequested– Reviewed:within2weeksofsubmission– Deadlines:None– AwardsBegin:within2weeksofsubmission
portal.xsede.org "Allocations"Submit/Review Request **
Proposal Document(s)
Keytoasuccessfulreview:• Adheretopagelimits!• “JusDfy”allocaDonrequest.
https://www.xsede.org/web/xup/allocation-policies**
PageLimit ProposalDocument3 Progressreport10 NeworRenewal15 Over10MillionSUs
Pg. limit: DOES NOT INCLUDE FIGURES & TABLES.
• CV(s)requiredforallrequests.• Abstractforstartup/educaDonrequest
(informs,orasaPDFdocument)• Proposal“MainDocument”forResearchrequest
(renewals/supplements)
The Award = Allocation • OneperPI(generally)• 1-yearduraDon• UnusedSUsareforfeitedatthe
endofanawardperiod• Progressreportrequiredfor
renewalrequests.• AdduserstoagrantviaXSEDE
UserPortal4 quarters = 1 yr allocation period
Submission Review Award
Advance
Time to renew
The Resources: Compute https://www.xsede.org/resources/overview
HPC Systems: (Bridges(PSC), Comet(SDSC), Gordon(SDSC), Stampede(TACC), SuperMIC(LSU) and Wrangler(TACC), Xstream(Stanford)) Large Memory Systems: (Bridges(PSC) and Comet(SDSC)) VMs: (Jetstream(IU/TACC)) Advanced VIS Systems: (Maverick(TACC)) HTC Systems: (OSG) Storage Systems: (Data Oasis(SDSC), Jetstream(IU/TACC), Pylon(PSC), Ranch(TACC)) *Ranch is the only storage resources that can be requested without compute resources.
The Resources: Extended Collaborative Support(ECS)
• Dedicated,butlimited,XSEDEstaffassistance• 5QuesDonswhicharepartofresourcerequestsecDonof
applicaDon• ReviewersrateneedforECS
https://www.xsede.org/ecss
Login at portal.xsede.org
Example Form: New Project
Example Form: Getting Info in order
Example Form: PI entry page, populating with Portal information
Example proposal submission: Personnel/Roles
Example proposal submission: Title/Abstract/FOS
Example proposal submission: Resource request
Example proposal submission: Resources request
Example proposal submission: Resource Request
Example proposal submission: ECS
Example proposal submission: Uploading documents
Example proposal submission: Uploading documents
Example proposal submission: Uploading documents
Example proposal submission: Supporting Grants
Example proposal submission: Supporting Grants
Example proposal submission: Supporting Grants
Example proposal submission: Publications
Example proposal submission: Saving and Final Submission
Example proposal submission: Successful submission
Pending Request
Notification of successful submission
Requesting an Advance
Why a Research Request may be reduced or rejected: • PoorornoScaling/CodeperformanceinformaDon• Scaling/CodeperformanceinformaDonnotfromresourcebeingrequested
• Non-disclosureofaccesstootherresource(s)• DisclosureofaccesstootherHPCresource(s)butnodetailsofdifferenceofresearch
• MulDplerequestsfromsameresearchlab• Pagelimits• PoorjusDficaDonforresources• LackofdescripDonofresearchteam• LowproducDvityfrompriorXSEDEResearchawardwithnodetailsinprogressreport
Example Proposals
Interesting Facts
• ~800researchrequestsperyear• ~1,200otherrequests• ~4.0BSUsrequested(3.7Bareresearchrequests)
• ~1.4BSUsawarded(1.0Bareresearchawards)