HPCMPUG2011 cray tutorial

 Review of XT6 Architecture
 AMD Opteron
 Cray Networks
 Lustre Basics
 Programming Environment
 PGI Compiler Basics
 The Cray Compiler Environment
 Cray Scientific Libraries
 Cray Message Passing Toolkit
 Cray Performance Analysis Tools
 ATP
 CCM
 Optimizations
 CPU
 Communication
 I/O
2011 HPCMP User Group © Cray Inc. June 20, 2011 2

AMD CPU Architecture
Cray Architecture
Lustre Filesystem Basics


2003 2005 2007 2008 2009 2010
AMD AMD
“Barcelona” “Shanghai” “Istanbul” “Magny-Cours”
Opteron™ Opteron™

Mfg.
130nm SOI 90nm SOI 65nm SOI 45nm SOI 45nm SOI 45nm SOI
Process
K8 K8 Greyhound Greyhound+ Greyhound+ Greyhound+
CPU Core

L2/L3 1MB/0 1MB/0 512kB/2MB 512kB/6MB 512kB/6MB 512kB/12MB

Hyper
Transport™ 3x 1.6GT/.s 3x 1.6GT/.s 3x 2GT/s 3x 4.0GT/s 3x 4.8GT/s 4x 6.4GT/s
Technology

Memory 2x DDR1 300 2x DDR1 400 2x DDR2 667 2x DDR2 800 2x DDR2 800 4x DDR3 1333


12 cores
1.7-2.2Ghz
1 4 7 10 105.6Gflops

8 cores
5 11 1.8-2.4Ghz
2 8 76.8Gflops

Power (ACP)
3 6 9 12 80Watts

Stream
27.5GB/s

Cache
12x 64KB L1
12x 512KB L2
12MB L3


L3 cache

HT Link
HT Link
HT Link
HT Link

L2 cache L2 cache L2 cache L2 cache

MEMORY CONTROLLER
Core 2 MEMORY CONTROLLER Core 5 Core 8 Core 11

HT Link

HT Link
HT Link

HT Link

Core 1 Core 4 Core 7 Core 10


Core 0 Core 3 Core 6 Core 9


 A cache line is 64B
 Unique L1 and L2 cache attached to each core
 L1 cache is 64 kbytes
 L2 cache is 512 kbytes
 L3 Cache is shared between 6 cores
 Cache is a “victim cache”
 All loads go to L1 immediately and get evicted down the caches
 Hardware prefetcher detects forward and backward strides through
memory
 Each core can perform a 128b add and 128b multiply per clock cycle
 This requires SSE, packed instructions
 “Stride-one vectorization”
 6 cores share a “flat” memory
 Non-uniform-memory-access (NUMA) beyond a node

Processor Frequency Peak Bandwidth Balance
(Gflops) (GB/sec) (bytes/flop
)
Istanbul
2.6 62.4 12.8 0.21
(XT5)
2.0 64 42.6 0.67

MC-8 2.3 73.6 42.6 0.58

2.4 76.8 42.6 0.55

1.9 91.2 42.6 0.47

MC-12 2.1 100.8 42.6 0.42

2.2 105.6 42.6 0.40


Gemini (XE-series)


 Microkernel on Compute PEs,
full featured Linux on Service
PEs.
 Service PEs specialize by
function
Compute PE
 Software Architecture
Login PE eliminates OS “Jitter”
Network PE  Software Architecture enables
reproducible run times
System PE
 Large machines boot in under
I/O PE 30 minutes, including
filesystem
Service Partition
Specialized
Linux nodes


XE6
System

External
Login Server

Boot RAID 10 GbE
IB QDR


6.4 GB/sec direct connect
Characteristics HyperTransport
Number of 16 or 24 (MC)
Cores 32 (IL)
Peak 153 Gflops/sec
Performance
MC-8 (2.4)
Peak 211 Gflops/sec
Performance
MC-12 (2.2)
Memory Size 32 or 64 GB per
node
Memory 83.5 GB/sec
Bandwidth 83.5 GB/sec direct
connect memory
Cray
SeaStar2+
Interconnect


Greyhound Greyhound
Greyhound Greyhound DDR3 Channel
DDR3 Channel 6MB L3 HT3 6MB L3
Greyhound Greyhound
Cache Greyhound Cache Greyhound
Greyhound Greyhound

DDR3 Channel Greyhound Greyhound DDR3 Channel

HT3

HT3
Greyhound H Greyhound

DDR3 Channel 6MB L3
Greyhound
Greyhound
T3 6MB L3
Greyhound
Greyhound
DDR3 Channel

Cache Greyhound Cache Greyhound
Greyhound Greyhound
Greyhound HT3 Greyhound DDR3 Channel
DDR3 Channel

To Interconnect
HT1 / HT3
 2 Multi-Chip Modules, 4 Opteron Dies
 8 Channels of DDR3 Bandwidth to 8 DIMMs
 24 (or 16) Computational Cores, 24 MB of L3 cache
 Dies are fully connected with HT3
 Snoop Filter Feature Allows 4 Die SMP to scale well


Without snoop filter, a streams test
shows 25MB/sec out of a possible
51.2 GB/sec or 48% of peak
bandwidth


With snoop filter, a streams test
shows 42.3 MB/sec out of a
possible 51.2 GB/sec or 82% of
peak bandwidth

This feature will be key for two-
socket Magny Cours Nodes which
are the same architecture-wise

 New compute blade with 8 AMD
Magny Cours processors
 Plug-compatible with XT5 cabinets
and backplanes
 Upgradeable to AMD’s
“Interlagos” series
 XE6 systems ship with the current
SIO blade


 Supports 2 Nodes per ASIC
 168 GB/sec routing capacity
 Scales to over 100,000 network
endpoints
 Link Level Reliability and Adaptive Hyper Hyper
Routing Transport Transport
3 3
 Advanced Resiliency Features
 Provides global address NIC 0 Netlink NIC 1
SB
space Block Gemini
LO
 Advanced NIC designed to Processor
efficiently support 48-Port
 MPI YARC Router
 One-sided MPI
 Shmem
 UPC, Coarray FORTRAN

Cray Baker Node
Characteristics
Number of 16 or 24
10 12X Gemini Cores
Channels
Peak 140 or 210 Gflops/s
(Each Gemini High Radix
YARC Router
Performance
acts like two
nodes on the 3D with adaptive Memory Size 32 or 64 GB per
Torus) Routing
node
168 GB/sec
capacity Memory 85 GB/sec
Bandwidth

Module with
SeaStar

Z

Y

X

Module with
Gemini

net rsp
net req

LB
ht treq p
net LB Ring
ht treq np FMA req T net
ht trsp net net
A req S req req vc0
ht p req
net R S
req O
ht np req B I
BTE R net
D rsp
B vc1

Router Tiles
HT3 Cave

NL
ht irsp
NPT vc1

ht np net
ireq rsp
net req CQ NAT
ht np req
H
ht p req net rsp headers
ht p
A AMO net
ht p req
ireq R net req req net req vc0
B RMT
ht p req
RAT net rsp

LM
CLM

 FMA (Fast Memory Access)
 Mechanism for most MPI transfers
 Supports tens of millions of MPI requests per second
 BTE (Block Transfer Engine)
 Supports asynchronous block transfers between local and remote memory,
in either direction
 For use for large MPI transfers that happen in the background

 Two Gemini ASICs are
packaged on a pin-compatible
mezzanine card
 Topology is a 3-D torus
 Each lane of the torus is
composed of 4 Gemini router
“tiles”
 Systems with SeaStar
interconnects can be upgraded
by swapping this card
 100% of the 48 router tiles on
each Gemini chip are used


Name Architecture Processor Network # Cores Memory/Core

Jade XT-4 AMD Seastar 2.1 8584 2GB DDR2-800
Budapest (2.1
Ghz)
Einstein XT-5 AMD Seastar 2.1 12827 2GB (some
Shanghai (2.4 nodes have
Ghz) 4GB/core)
DDR2-800
MRAP XT-5 AMD Seastar 2.1 10400 4GB DDR2-800
Barcelona (2.3
Ghz)
Garnet XE-6 Magny Cours Gemini 1.0 20160 2GB DDR3-1333
8 core 2.4 Ghz
Raptor XE-6 Magny Cours Gemini 1.0 43712 2GB DDR3-1333
8 core 2.4 Ghz
Chugach XE-6 Magny Cours Gemini 1.0 11648 2GB DDR3 -1333
8 core 2.3 Ghz


Low Velocity Airflow

High Velocity Airflow

Low Velocity Airflow

High Velocity Airflow

2011 HPCMP User Group © Cray Inc.
June 20, 2011
33 Low Velocity Airflow

Cool air is released into the computer room

Liquid Liquid/Vapor
in Mixture out

Hot air stream passes through evaporator, rejects
heat to R134a via liquid-vapor phase change
(evaporation).

R134a absorbs energy only in the presence of heated air.
Phase change is 10x more efficient than pure water
cooling.


R134a piping Exit Evaporators

Inlet Evaporator


Term Meaning Purpose
MDS Metadata Server Manages all file metadata for
filesystem. 1 per FS
OST Object Storage Target The basic “chunk” of data written
to disk. Max 160 per file.
OSS Object Storage Server Communicates with disks,
manages 1 or more OSTs. 1 or
more per FS
Stripe Size Size of chunks. Controls the size of file chunks
stored to OSTs. Can’t be changed
once file is written.
Stripe Count Number of OSTs used per Controls parallelism of file. Can’t
file. be changed once file is writte.


2011 HPCMP User Group © Cray Inc. une 20, 2011
J 38

2011 HPCMP User Group © Cray Inc. une 20, 2011
J 39

 32 MB per OST (32 MB – 5 GB) and 32 MB Transfer Size
 Unable to take advantage of file system parallelism
 Access to multiple disks adds overhead which hurts performance

Single Writer
Write Performance
120

100

80
Write (MB/s)

1 MB Stripe
60
32 MB Stripe

40
Lustre
20

0
1 2 4 16 32 64 128 160
Stripe Count

40 2011 HPCMP User Group © Cray Inc. une 20, 2011
J

 Single OST, 256 MB File Size
 Performance can be limited by the process (transfer size) or file system
(stripe size)

Single Writer
Transfer vs. Stripe Size
140

120

100
Write (MB/s)

80
32 MB Transfer
60
8 MB Transfer
1 MB Transfer
40 Lustre
20

0
1 2 4 8 16 32 64 128
Stripe Size (MB)

41 2011 HPCMP User Group © Cray Inc. une 20, 2011
J

 Use the lfs command, libLUT, or MPIIO hints to adjust your stripe count and
possibly size
 lfs setstripe -c -1 -s 4M <file or directory> (160 OSTs, 4MB stripe)
 lfs setstripe -c 1 -s 16M <file or directory> (1 OST, 16M stripe)
 export MPICH_MPIIO_HINTS=‘*: striping_factor=160’
 Files inherit striping information from the parent directory, this cannot be
changed once the file is written
 Set the striping before copying in files


Available Compilers
Cray Scientific Libraries
Cray Message Passing Toolkit


 Cray XT/XE Supercomputers come with compiler wrappers to simplify
building parallel applications (similar the mpicc/mpif90)
 Fortran Compiler: ftn
 C Compiler: cc
 C++ Compiler: CC
 Using these wrappers ensures that your code is built for the compute
nodes and linked against important libraries
 Cray MPT (MPI, Shmem, etc.)
 Cray LibSci (BLAS, LAPACK, etc.)
 …
 Choosing the underlying compiler is via the PrgEnv-* modules, do not call
the PGI, Cray, etc. compilers directly.
 Always load the appropriate xtpe-<arch> module for your machine
 Enables proper compiler target
 Links optimized math libraries


…from Cray’s Perspective

 PGI – Very good Fortran and C, pretty good C++
 Good vectorization
 Good functional correctness with optimization enabled
 Good manual and automatic prefetch capabilities
 Very interested in the Linux HPC market, although that is not their only focus
 Excellent working relationship with Cray, good bug responsiveness
 Pathscale – Good Fortran, C, possibly good C++
 Outstanding scalar optimization for loops that do not vectorize
 Fortran front end uses an older version of the CCE Fortran front end
 OpenMP uses a non-pthreads approach
 Scalar benefits will not get as much mileage with longer vectors
 Intel – Good Fortran, excellent C and C++ (if you ignore vectorization)
 Automatic vectorization capabilities are modest, compared to PGI and CCE
 Use of inline assembly is encouraged
 Focus is more on best speed for scalar, non-scaling apps
 Tuned for Intel architectures, but actually works well for some applications on
AMD

…from Cray’s Perspective

 GNU so-so Fortran, outstanding C and C++ (if you ignore vectorization)
 Obviously, the best for gcc compatability
 Scalar optimizer was recently rewritten and is very good
 Vectorization capabilities focus mostly on inline assembly
 Note the last three releases have been incompatible with each other (4.3, 4.4,
and 4.5) and required recompilation of Fortran modules
 CCE – Outstanding Fortran, very good C, and okay C++
 Very good vectorization
 Very good Fortran language support; only real choice for Coarrays
 C support is quite good, with UPC support
 Very good scalar optimization and automatic parallelization
 Clean implementation of OpenMP 3.0, with tasks
 Sole delivery focus is on Linux-based Cray hardware systems
 Best bug turnaround time (if it isn’t, let us know!)
 Cleanest integration with other Cray tools (performance tools, debuggers,
upcoming productivity tools)
 No inline assembly support


 PGI
 -fast –Mipa=fast(,safe)
 If you can be flexible with precision, also try -Mfprelaxed
 Compiler feedback: -Minfo=all -Mneginfo
 man pgf90; man pgcc; man pgCC; or pgf90 -help
 Cray
 <none, turned on by default>
 Compiler feedback: -rm (Fortran) -hlist=m (C)
 If you know you don’t want OpenMP: -xomp or -Othread0
 man crayftn; man craycc ; man crayCC
 Pathscale
 -Ofast Note: this is a little looser with precision than other compilers
 Compiler feedback: -LNO:simd_verbose=ON
 man eko (“Every Known Optimization”)
 GNU
 -O2 / -O3
 Compiler feedback: good luck
 man gfortran; man gcc; man g++
 Intel
 -fast
 Compiler feedback:
 man ifort; man icc; man iCC


 Traditional (scalar) optimizations are controlled via -O# compiler flags
 Default: -O2
 More aggressive optimizations (including vectorization) are enabled with
the -fast or -fastsse metaflags
 These translate to: -O2 -Munroll=c:1 -Mnoframe -Mlre
–Mautoinline -Mvect=sse -Mscalarsse
-Mcache_align -Mflushz –Mpre
 Interprocedural analysis allows the compiler to perform whole-program
optimizations. This is enabled with –Mipa=fast
 See man pgf90, man pgcc, or man pgCC for more information about
compiler options.


 Compiler feedback is enabled with -Minfo and -Mneginfo
 This can provide valuable information about what optimizations were
or were not done and why.
 To debug an optimized code, the -gopt flag will insert debugging
information without disabling optimizations
 It’s possible to disable optimizations included with -fast if you believe one
is causing problems
 For example: -fast -Mnolre enables -fast and then disables loop
redundant optimizations
 To get more information about any compiler flag, add -help with the
flag in question
 pgf90 -help -fast will give more information about the -fast
flag
 OpenMP is enabled with the -mp flag


Some compiler options may effect both performance and accuracy. Lower
accuracy is often higher performance, but it’s also able to enforce accuracy.

 -Kieee: All FP math strictly conforms to IEEE 754 (off by default)
 -Ktrap: Turns on processor trapping of FP exceptions
 -Mdaz: Treat all denormalized numbers as zero
 -Mflushz: Set SSE to flush-to-zero (on with -fast)
 -Mfprelaxed: Allow the compiler to use relaxed (reduced) precision to
speed up some floating point optimizations
 Some other compilers turn this on by default, PGI chooses to favor
accuracy to speed by default.


 Cray has a long tradition of high performance compilers on Cray
platforms (Traditional vector, T3E, X1, X2)
 Vectorization
 Parallelization
 Code transformation
 More…
 Investigated leveraging an open source compiler called LLVM

 First release December 2008


Fortran Source C and C++ Source C and C++ Front End
supplied by Edison Design
Group, with Cray-developed
Fortran Front End C & C++ Front End code for extensions and
interface support

Interprocedural Analysis
Cray Inc. Compiler
Technology
Compiler

Optimization and
Parallelization

X86 Code Cray X2 Code
Generator Generator
X86 Code Generation from
Open Source LLVM, with
Object File
additional Cray-developed
optimizations and interface
support

 Standard conforming languages and programming models
 Fortran 2003
 UPC & CoArray Fortran
 Fully optimized and integrated into the compiler
 No preprocessor involved
 Target the network appropriately:
 GASNet with Portals
 DMAPP with Gemini & Aries

 Ability and motivation to provide high-quality support for custom
Cray network hardware
 Cray technology focused on scientific applications
 Takes advantage of Cray’s extensive knowledge of automatic
vectorization
 Takes advantage of Cray’s extensive knowledge of automatic
shared memory parallelization
 Supplements, rather than replaces, the available compiler
choices

 Make sure it is available
 module avail PrgEnv-cray
 To access the Cray compiler
 module load PrgEnv-cray
 To target the various chip
 module load xtpe-[barcelona,shanghi,mc8]
 Once you have loaded the module “cc” and “ftn” are the Cray
compilers
 Recommend just using default options
 Use –rm (fortran) and –hlist=m (C) to find out what happened
 man crayftn


 Excellent Vectorization
 Vectorize more loops than other compilers
 OpenMP 3.0
 Task and Nesting
 PGAS: Functional UPC and CAF available today
 C++ Support
 Automatic Parallelization
 Modernized version of Cray X1 streaming capability
 Interacts with OMP directives
 Cache optimizations
 Automatic Blocking
 Automatic Management of what stays in cache
 Prefetching, Interchange, Fusion, and much more…


 Loop Based Optimizations
 Vectorization
 OpenMP
 Autothreading
 Interchange
 Pattern Matching
 Cache blocking/ non-temporal / prefetching
 Fortran 2003 Standard; working on 2008
 PGAS (UPC and Co-Array Fortran)
 Some performance optimizations available in 7.1
 Optimization Feedback: Loopmark
 Focus


 Cray compiler supports a full and growing set of directives
and pragmas

!dir$ concurrent
!dir$ ivdep
!dir$ interchange
!dir$ unroll
!dir$ loop_info [max_trips] [cache_na] ... Many more
!dir$ blockable

man directives
man loop_info

 Compiler can generate an filename.lst file.
 Contains annotated listing of your source code with letter indicating important
optimizations
%%% L o o p m a r k L e g e n d %%%
Primary Loop Type Modifiers
------- ---- ---- ---------
a - vector atomic memory operation
A - Pattern matched b - blocked
C - Collapsed f - fused
D - Deleted i - interchanged
E - Cloned m - streamed but not partitioned
I - Inlined p - conditional, partial and/or computed
M - Multithreaded r - unrolled
P - Parallel/Tasked s - shortloop
V - Vectorized t - array syntax temp used
W - Unwound w - unwound


• ftn –rm … or cc –hlist=m …
29. b-------< do i3=2,n3-1
30. b b-----< do i2=2,n2-1
31. b b Vr--< do i1=1,n1
32. b b Vr u1(i1) = u(i1,i2-1,i3) + u(i1,i2+1,i3)
33. b b Vr > + u(i1,i2,i3-1) + u(i1,i2,i3+1)
34. b b Vr u2(i1) = u(i1,i2-1,i3-1) + u(i1,i2+1,i3-1)
35. b b Vr > + u(i1,i2-1,i3+1) + u(i1,i2+1,i3+1)
36. b b Vr--> enddo
37. b b Vr--< do i1=2,n1-1
38. b b Vr r(i1,i2,i3) = v(i1,i2,i3)
39. b b Vr > - a(0) * u(i1,i2,i3)
40. b b Vr > - a(2) * ( u2(i1) + u1(i1-1) + u1(i1+1) )
41. b b Vr > - a(3) * ( u2(i1-1) + u2(i1+1) )
42. b b Vr--> enddo
43. b b-----> enddo
44. b-------> enddo


ftn-6289 ftn: VECTOR File = resid.f, Line = 29
A loop starting at line 29 was not vectorized because a recurrence was found on "U1" between lines
32 and 38.
ftn-6049 ftn: SCALAR File = resid.f, Line = 29
A loop starting at line 29 was blocked with block size 4.
A loop starting at line 30 was not vectorized because a recurrence was found on "U1" between lines 32
and 38.
A loop starting at line 30 was blocked with block size 4.
A loop starting at line 31 was unrolled 4 times.
A loop starting at line 31 was vectorized.
A loop starting at line 37 was unrolled 4 times.
A loop starting at line 37 was vectorized.

 -hbyteswapio
 Link time option
 Applies to all unformatted fortran IO
 Assign command
 With the PrgEnv-cray module loaded do this:
setenv FILENV assign.txt
assign -N swap_endian g:su
assign -N swap_endian g:du

 Can use assign to be more precise


 OpenMP is ON by default
 Optimizations controlled by –Othread#
 To shut off use –Othread0 or –xomp or –hnoomp

 Autothreading is NOT on by default;
 -hautothread to turn on
 Modernized version of Cray X1 streaming capability
 Interacts with OMP directives

If you do not want to use OpenMP and have OMP directives
in the code, make sure to make a run with OpenMP shut
off at compile time


 Cray have historically played a role in scientific library
development
 BLAS3 were largely designed for Crays
 Standard libraries were tuned for Cray vector processors
(later COTS)
 Cray have always tuned standard libraries for Cray
interconnect
 In the 90s, Cray provided many non-standard libraries
 Sparse direct, sparse iterative
 These days the goal is to remain portable (standard APIs)
whilst providing more performance
 Advanced features, tuning knobs, environment variables


FFT Dense Sparse
BLAS
CRAFFT CASK
LAPACK

FFTW ScaLAPACK PETSc

IRT
P-CRAFFT Trilinos
CASE

IRT – Iterative Refinement Toolkit
CASK – Cray Adaptive Sparse Kernels
CRAFFT – Cray Adaptive FFT
CASE – Cray Adaptive Simple Eigensolver

 There are many libsci libraries on the systems
 One for each of
 Compiler (intel, cray, gnu, pathscale, pgi )
 Single thread, multiple thread
 Target (istanbul, mc12 )
 Best way to use libsci is to ignore all of this
 Load the xtpe-module (some sites set this by default)
 E.g. module load xtpe-shanghai / xtpe-istanbul / xtpe-mc8
 Cray’s drivers will link the library automatically
 PETSc, Trilinos, fftw, acml all have their own module
 Tip : make sure you have the correct library loaded e.g.
–Wl, -ydgemm_

 Perhaps you want to link another library such as ACML
 This can be done. If the library is provided by Cray, then load
the module. The link will be performed with the libraries in the
correct order.
 If the library is not provided by Cray and has no module, add it
to the link line.
 Items you add to the explicit link will be in the correct place
 Note, to get explicit BLAS from ACML but scalapack from libsci
 Load acml module. Explicit calls to BLAS in code resolve
from ACML
 BLAS calls from the scalapack code will be resolved from
libsci (no way around this)

 Threading capabilities in previous libsci versions were poor
 Used PTHREADS (more explicit affinity etc)
 Required explicit linking to a _mp version of libsci
 Was a source of concern for some applications that need
hybrid performance and interoperability with openMP
 LibSci 10.4.2 February 2010
 OpenMP-aware LibSci
 Allows calling of BLAS inside or outside parallel region
 Single library supported (there is still a single thread lib)
 Usage – load the xtpe module for your system (mc12)

GOTO_NUM_THREADS outmoded – use OMP_NUM_THREADS


 Allows seamless calling of the BLAS within or without a parallel
region

e.g. OMP_NUM_THREADS = 12

call dgemm(…) threaded dgemm is used with 12 threads
!$OMP PARALLEL DO
do
call dgemm(…) single thread dgemm is used
end do

Some users are requesting a further layer of parallelism here (see
later)


120

Libsci DGEMM efficiency
100

80
GFLOPs

1thread
60
3threads
6threads
40 9threads
12threads

20

0

Dimension (square) Inc.
2011 HPCMP User Group © Cray June 20, 2011 74

140
Libsci-10.5.2 performance on 2 x MC12 2.0 GHz K=64
120 (Cray XE6)
K=128

100 K=200

K=228
80
GFLOPS

K=256

60 K=300

K=400
40
K=500
20
K=600

0 K=700
1 2 4 8 12 16 20 24
K=800
Number of threads

 All BLAS libraries are optimized for rank-k update

* =

 However, a huge % of dgemm usage is not from solvers but explicit calls
 E.g. DCA++ matrices are of this form

* =

 How can we very easily provide an optimization for these types of
matrices?

 Cray BLAS existed on every Cray machine between Cray-2 and Cray
X2
 Cray XT line did not include Cray BLAS
 Cray’s expertise was in vector processors
 GotoBLAS was the best performing x86 BLAS
 LibGoto is now discontinued
 In Q3 2011 LibSci will be released with Cray BLAS


1. Customers require more OpenMP features unobtainable
with current library
2. Customers require more adaptive performance for
unusual problems .e.g. DCA++
3. Interlagos / Bulldozer is a dramatic shift in
ISA/architecture/performance
4. Our auto-tuning framework has advanced to the point
that we can tackle this problem (good BLAS is easy,
excellent BLAS is very hard)
5. Need for Bit-reproducable BLAS at high-performance


"anything that can be represented in C, Fortran or ASM
code can be generated automatically by one instance
of an abstract operator in high-level code“

In other words, if we can create a purely general model
of matrix-multiplication, and create every instance of
it, then at least one of the generated schemes will
perform well


 Start with a completely general formulation of the BLAS
 Use a DSL that expresses every important optimization
 Auto-generate every combination of orderings, buffering, and
optimization
 For every combination of the above, sweep all possible sizes
 For a given input set ( M, N, K, datatype, alpha, beta ) map the
best dgemm routine to the input
 The current library should be a specific instance of the above
 Worst-case performance can be no worse than current library
 The lowest level of blocking is a hand-written assembly kernel


7.5

7.45

7.4

7.35

7.3
bframe GFLOPS
7.25
libsci
7.2

7.15

7.1

7.05

143
72
12
17
22
27

62

133
67
37
42

57

105
2
7

47

100

128

138
95
32

52


 New optimizations for Gemini network in the ScaLAPACK LU and Cholesky
routines

1. Change the default broadcast topology to match the Gemini network

2. Give tools to allow the topology to be changed by the user

3. Give guidance on how grid-shape can affect the performance


 Parallel Version of LAPACK GETRF
 Panel Factorization
 Only single column block is involved
 The rest of PEs are waiting
 Trailing matrix update
 Major part of the computation
 Column-wise broadcast (Blocking)
 Row-wise broadcast (Asynchronous)
 Data is packed before sending using PBLAS
 Broadcast uses BLACS library
 These broadcasts are the major communication
patterns

 MPI default
 Binomial Tree + node-aware broadcast
 All PEs makes implicit barrier to make sure the completion
 Not suitable for rank-k update

 Bidirectional-Ring broadcast
 Root PE makes 2 MPI Send calls to both of the directions
 The immediate neighbor finishes first
 ScaLAPACK’s default
 Better than MPI


 Increasing Ring Broadcast (our new default)
 Root makes a single MPI call to the immediate neighbor
 Pipelining
 Better than bidirectional ring

 Multi-Ring Broadcast (2, 4, 8 etc)
 The root PE sends to multiple sub-rings
 Can be done with tree algorithm

 2 rings seems the best for row-wise broadcast of LU


 Hypercube
 Behaves like MPI default
 Too many collisions in the message traffic
 Decreasing Ring
 The immediate neighbor finishes last
 No benefit in LU
 Modified Increasing Ring
 Best performance in HPL
 As good as increasing ring


XDLU performance: 3072 cores, size=65536
10000
9000
8000
7000
6000
Gflops

5000
4000
3000 SRING
IRING
2000
1000
0
32 64 32 64 32 64 32 64 32 64

48 48 24 24 12 12 32 32 16 16

64 64 128 128 256 256 96 96 192 192
NB / P / Q

XDLU performance: 6144 cores, size=65536
14000

12000

10000

8000
Gflops

6000
SRING
4000
IRING
2000

0
32 64 32 64 32 64 32 64 32 64

48 48 24 24 12 12 64 64 32 32

128 128 256 256 512 512 96 96 192 192
NB / P / Q

 Row Major Process Grid puts adjacent PEs in the same row
 Adjacent PEs are most probably located in the same node
 In flat MPI, 16 or 24 PEs are in the same node
 In hybrid mode, several are in the same node
 Most MPI sends in I-ring happen in the same node
 MPI has good shared-memory device
 Good pipelining

Node 0 Node 1 Node 2


 For PxGETRF:  The variables let users to choose
 SCALAPACK_LU_CBCAST broadcast algorithm :
 SCALAPACK_LU_RBCAST  IRING increasing ring
 For PxPOTRF:
(default value)
 DRING decreasing ring
 SCALAPACK_LLT_CBCAST
 SRING split ring (old default
 SCALAPCK_LLT_RBCAST
value)
 SCALAPACK_UTU_CBCAST
 MRING multi-ring
SCALAPACK_UTU_RBCAST
 HYPR hypercube
 MPI mpi_bcast
 TREE tree
 There is also a set function, allowing
 FULL full connected
the user to change these on the fly


 Grid shape / size
 Square grid is most common
 Try to use Q = x * P grids, where x = 2, 4, 6, 8
 Square grids not often the best
 Blocksize
 Unlike HPL, fine-tuning not important.
 64 usually the best
 Ordering
 Try using column-major ordering, it can be better
 BCAST
 The new default will be a huge improvement if you can make your grid
the right way. If you cannot, play with the environment variables.


 Full MPI2 support (except process spawning) based on ANL MPICH2
 Cray used the MPICH2 Nemesis layer for Gemini
 Cray-tuned collectives
 Cray-tuned ROMIO for MPI-IO

 Current Release: 5.3.0 (MPICH 1.3.1)
 Improved MPI_Allreduce and MPI_alltoallv
 Initial support for checkpoint/restart for MPI or Cray SHMEM on XE
systems
 Improved support for MPI thread safety.
 module load xt-mpich2
 Tuned SHMEM library
 module load xt-shmem


MPI_Alltoall with 10,000 Processes
Comparing Original vs Optimized Algorithms
on Cray XE6 Systems
25000000

20000000
Microseconds

15000000

Original Algorithm
10000000
Optimized Algorithm

5000000

0
256 512 1024 2048 4096 8192 16384 32768
MessageHPCMP User Group © Cray Inc.
2011
Size (in bytes) June 20, 2011 95

8-Byte MPI_Allgather and MPI_Allgatherv Scaling
Comparing Original vs Optimized Algorithms
45000 on Cray XE6 Systems
40000
MPI_Allgather and
35000 MPI_Allgatherv algorithms
optimized for Cray XE6.
30000
Microseconds

Original Allgather
25000
20000 Optimized Allgather

15000 Original Allgatherv

10000 Optimized Allgatherv

5000
0
1024p 2048p 4096p 8192p 16384p 32768p
Number ofUser Group © Cray Inc. June 20, 2011
2011 HPCMP Processes 96

 Default is 8192 bytes
 Maximum size message that can go through the eager protocol.
 May help for apps that are sending medium size messages, and do better
when loosely coupled. Does application have a large amount of time in
MPI_Waitall? Setting this environment variable higher may help.
 Max value is 131072 bytes.
 Remember for this path it helps to pre-post receives if possible.
 Note that a 40-byte CH3 header is included when accounting for the
message size.


 Default is 64 32K buffers ( 2M total )
 Controls number of 32K DMA buffers available for each rank to use in the
Eager protocol described earlier
 May help to modestly increase. But other resources constrain the usability
of a large number of buffers.


 What do I mean by PGAS?
 Partitioned Global Address Space
 UPC
 CoArray Fortran ( Fortran 2008 )
 SHMEM (I will count as PGAS for convenience)
 SHMEM: Library based
 Not part of any language standard
 Compiler independent
 Compiler has no knowledge that it is compiling a PGAS code and
does nothing different. I.E. no transformations or optimizations


 UPC
 Specification that extends the ISO/IEC 9899 standard for C
 Integrated into the language
 Heavily compiler dependent
 Compiler intimately involved in detecting and executing remote
references
 Flexible, but filled with challenges like pointers, a lack of true
multidimensional arrays, and many options for distributing data
 Fortran 2008
 Now incorporates coarrays
 Compiler dependent
 Philosophically different from UPC
 Replication of arrays on every image with “easy and obvious” ways
to access those remote locations.


 Translate the UPC source code into hardware executable operations that
produce the proper behavior, as defined by the specification
 Storing to a remote location?
 Loading from a remote location?
 When does the transfer need to be complete?
 Are there any dependencies between this transfer and anything else?
 No ordering guarantees provided by the network, compiler is
responsible for making sure everything gets to its destination in the
correct order.


for ( i = 0; i < ELEMS_PER_THREAD; i+=1 ) {
local_data[i] += global_2d[i][target];
}

temp = pgas_get(&global_2d[i]); // Initiate the get
pgas_fence(); // makes sure the get is complete
local_data[i] += temp; // Use the local location to complete the operation
}
 The compiler must
 Recognize you are referencing a shared location
 Initiate the load of the remote data
 Make sure the transfer has completed
 Proceed with the calculation
 Repeat for all iterations of the loop


temp = pgas_get(&global_2d[i]); // Initiate the get
pgas_fence(); // makes sure the get is complete
local_data[i] += temp; // Use the local location to complete the operation
}

 Simple translation results in
 Single word references
 Lots of fences
 Little to no latency hiding
 No use of special hardware
 Nothing here says “fast”

2011 HPCMP User 105
June 20, 2011 Group © Cray Inc.

Want the compiler to generate code that will run as fast as possible given
what the user has written, or allow the user to get fast performance with
simple modifications.
 Increase message size
 Do multi / many word transfers whenever possible, not single word.
 Minimize fences
 Delay fence “as much as possible”
 Eliminate the fence in some circumstances
 Use the appropriate hardware
 Use on-node hardware for on-node transfers
 Use transfer mechanism appropriate for this message size
 Overlap communication and computation
 Use hardware atomic functions where appropriate


Primary Loop Type Modifiers
A - Pattern matched a - atomic memory operation
b - blocked
C - Collapsed c - conditional and/or computed
D - Deleted
E - Cloned f - fused
G - Accelerated g - partitioned
I - Inlined i - interchanged
M - Multithreaded m - partitioned
n - non-blocking remote transfer
p - partial
r - unrolled
s - shortloop
V - Vectorized w - unwound


15. shared long global_1d[MAX_ELEMS_PER_THREAD * THREADS];
…
83. 1 before = upc_ticks_now();
84. 1 r8------< for ( i = 0, j = target; i < ELEMS_PER_THREAD ;
85. 1 r8 i += 1, j += THREADS ) {
86. 1 r8 n local_data[i]= global_1d[j];
87. 1 r8------> }
88. 1 after = upc_ticks_now();

 1D get BW= 0.027598 Gbytes/s

2011 HPCMP User 109

15. shared long global_1d[MAX_ELEMS_PER_THREAD * THREADS];
…
101. 1 before = upc_ticks_now();
102. 1 upc_memget(&local_data[0],&global_1d[target],8*ELEMS_PER_THREAD);
103. 1
104. 1 after = upc_ticks_now();

 1D upc_memget BW= 4.972960 Gbytes/s

 upc_memget is 184 times faster!!

2011 HPCMP User 110

16. shared long global_2d[MAX_ELEMS_PER_THREAD][THREADS];
…
121. 1 A-------< for ( i = 0; i < ELEMS_PER_THREAD; i+=1) {
122. 1 A local_data[i] = global_2d[i][target];
123. 1 A-------> }

 1D upc_memget BW= 4.972960 Gbytes/s
 2D get time BW= 4.905653 Gbytes/s
 Pattern matching can give you the same
performance as if using upc_memget

2011 HPCMP User 111

 PGAS data references made by the single statement immediately following the pgas
defer_sync directive will not be synchronized until the next fence instruction.
 Only applies to next UPC/CAF statement
 Does not apply to upc “routines”
 Does not apply to shmem routines

 Normally the compiler synchronizes the references in a statement as late as
possible without violating program semantics. The purpose of the defer_sync
directive is to synchronize the references even later, beyond where the compiler
can determine it is safe.

 Extremely powerful!
 Can easily overlap communication and computation with this statement
 Can apply to both “gets” and “puts”
 Can be used to implement a variety of “tricks”. Use your imagination!


CrayPAT


 Future system basic characteristics:
 Many-core, hybrid multi-core computing

 Increase in on-node concurrency
 10s-100s of cores sharing memory
 With or without a companion accelerator
 Vector hardware at the low level

 Impact on applications:
 Restructure / evolve applications while using existing programming
models to take advantage of increased concurrency

 Expand on use of mixed-mode programming models (MPI + OpenMP +
accelerated kernels, etc.)


 Focus on automation (simplify tool usage, provide feedback based on
analysis)

 Enhance support for multiple programming models within a program (MPI,
PGAS, OpenMP, SHMEM)

 Scaling (larger jobs, more data, better tool response)

 New processors and interconnects

 Extend performance tools to include pre-runtime optimization information
from the Cray compiler


 New predefined wrappers (ADIOS, ARMCI, PetSc, PGAS libraries)
 More UPC and Co-array Fortran support
 Support for non-record locking file systems
 Support for applications built with shared libraries
 Support for Chapel programs
 pat_report tables available in Cray Apprentice2


 Enhanced PGAS support is available in perftools 5.1.3 and later
 Profiles of a PGAS program can be created to show:
 Top time consuming functions/line numbers in the code
 Load imbalance information
 Performance statistics attributed to user source by default
 Can expose statistics by library as well
 To see underlying operations, such as wait time on barriers
 Data collection is based on methods used for MPI library
 PGAS data is collected by default when using Automatic Profiling Analysis
(pat_build –O apa)
 Predefined wrappers for runtime libraries (caf, upc, pgas) enable attribution of
samples or time to user source
 UPC and SHMEM heap tracking coming in subsequent release
 -g heap will track shared heap in addition to local heap

June 20, 2011 2011 HPCMP User Group © Cray Inc. 118

Table 1: Profile by Function

Samp % | Samp | Imb. | Imb. |Group
| | Samp | Samp % | Function
| | | | PE='HIDE'

100.0% | 48 | -- | -- |Total
|------------------------------------------
| 95.8% | 46 | -- | -- |USER
||-----------------------------------------
|| 83.3% | 40 | 1.00 | 3.3% |all2all
|| 6.2% | 3 | 0.50 | 22.2% |do_cksum
|| 2.1% | 1 | 1.00 | 66.7% |do_all2all
|| 2.1% | 1 | 0.50 | 66.7% |mpp_accum_long
|| 2.1% | 1 | 0.50 | 66.7% |mpp_alloc
||=========================================
| 4.2% | 2 | -- | -- |ETC
||-----------------------------------------
|| 4.2% | 2 | 0.50 | 33.3% |bzero
|==========================================


Table 2: Profile by Group, Function, and Line

Samp % | Samp | Imb. | Imb. |Group
| | Samp | Samp % | Function
| | | | Source
| | | | Line
| | | | PE='HIDE'

100.0% | 48 | -- | -- |Total
|--------------------------------------------
| 95.8% | 46 | -- | -- |USER
||-------------------------------------------
|| 83.3% | 40 | -- | -- |all2all
3| | | | | mpp_bench.c
4| | | | | line.298
|| 6.2% | 3 | -- | -- |do_cksum
3| | | | | mpp_bench.c
||||-----------------------------------------
4||| 2.1% | 1 | 0.25 | 33.3% |line.315
4||| 4.2% | 2 | 0.25 | 16.7% |line.316
||||=========================================


Table 1: Profile by Function and Callers, with Line Numbers
Samp % | Samp |Group
| | Function
| | Caller
| | PE='HIDE’
100.0% | 47 |Total
|---------------------------
| 93.6% | 44 |ETC
||--------------------------
|| 85.1% | 40 |upc_memput
3| | | all2all:mpp_bench.c:line.298
4| | | do_all2all:mpp_bench.c:line.348
5| | | main:test_all2all.c:line.70
|| 4.3% | 2 |bzero
3| | | (N/A):(N/A):line.0
|| 2.1% | 1 |upc_all_alloc
3| | | mpp_alloc:mpp_bench.c:line.143
|| 2.1% | 1 |upc_all_reduceUL
3| | | mpp_accum_long:mpp_bench.c:line.185
4| | | do_cksum:mpp_bench.c:line.317
5| | | do_all2all:mpp_bench.c:line.341
||==========================


Table 1: Profile by Function and Callers, with Line Numbers

Time % | Time | Calls |Group
| | | Function
| | | Caller
| | | PE='HIDE'

100.0% | 0.795844 | 73904.0 |Total
|-----------------------------------------
| 78.9% | 0.628058 | 41121.8 |PGAS
||----------------------------------------
|| 76.1% | 0.605945 | 32768.0 |__pgas_put
3| | | | all2all:mpp_bench.c:line.298
4| | | | do_all2all:mpp_bench.c:line.348
5| | | | main:test_all2all.c:line.70
|| 1.5% | 0.012113 | 10.0 |__pgas_barrier
3| | | | (N/A):(N/A):line.0
…


…
||========================================
| 15.7% | 0.125006 | 3.0 |USER
||----------------------------------------
|| 12.2% | 0.097125 | 1.0 |do_all2all
|| 3.5% | 0.027668 | 1.0 |main
3| | | | (N/A):(N/A):line.0
||========================================
| 5.4% | 0.042777 | 32777.2 |UPC
||----------------------------------------
|| 5.3% | 0.042321 | 32768.0 |upc_memput
3| | | | all2all:mpp_bench.c:line.298
4| | | | do_all2all:mpp_bench.c:line.348
|=========================================


New text
table icon

Right click
for table
generation
options


 Scalability
 New .ap2 data format and client / server model
 Reduced pat_report processing and report generation times
 Reduced app2 data load times
 Graphical presentation handled locally (not passed through ssh
connection)
 Better tool responsiveness
 Minimizes data loaded into memory at any given time
 Reduced server footprint on Cray XT/XE service node
 Larger jobs supported

 Distributed Cray Apprentice2 (app2) client for Linux
 app2 client for Mac and Windows laptops coming later this year


 CPMD
 MPI, instrumented with pat_build –u, HWPC=1
 960 cores
Perftools 5.1.3 Perftools 5.2.0
.xf -> .ap2 88.5 seconds 22.9 seconds
ap2 -> report 1512.27 seconds 49.6 seconds

 VASP
 MPI, instrumented with pat_build –gmpi –u, HWPC=3
 768 cores

Perftools 5.1.3 Perftools 5.2.0
.xf -> .ap2 45.2 seconds 15.9 seconds
ap2 -> report 796.9 seconds 28.0 seconds


‘:’ signifies
 From Linux desktop – a remote
host
instead of
 % module load perftools
ap2 file

 % app2
 % app2 kaibab:
 % app2 kaibab:/lus/scratch/heidi/swim+pat+10302-0t.ap2

 File->Open Remote…


 Optional app2 client for Linux desktop available as of 5.2.0

 Can still run app2 from Cray service node

 Improves response times as X11 traffic is no longer passed through the ssh
connection

 Replaces 32-bit Linux desktop version of Cray Apprentice2

 Uses libssh to establish connection

 app2 clients for Windows and Mac coming in subsequent release


Linux desktop All data from Cray XT login Collected Compute nodes
my_program.ap2 + performance
X Window
X11 protocol data
app2
System
application my_program.ap2 my_program+apa

 Log into Cray XT/XE login node
 % ssh –Y seal

 Launch Cray Apprentice2 on Cray XT/XE login node
 % app2 /lus/scratch/mydir/my_program.ap2
 User Interface displayed on desktop via ssh trusted X11 forwarding
 Entire my_program.ap2 file loaded into memory on XT login node (can
be Gbytes of data)

Linux desktop User requested data Cray XT login Collected
Compute nodes
from
X Window performance
my_program.ap2 app2 server
System data
application
my_program.ap2
my_program+apa
app2 client

 Launch Cray Apprentice2 on desktop, point to data
 % app2 seal:/lus/scratch/mydir/my_program.ap2

 User Interface displayed on desktop via X Windows-based software
 Minimal subset of data from my_program.ap2 loaded into memory on
Cray XT/XE service node at any given time
 Only data requested sent from server to client


 Major change to the way HW counters are collected starting with CPMAT
5.2.1 and CLE 4.0 (In conjunction with Interlagos support)

 Linux has officially incorporated support for accessing counters through a
perf_events subsystem. Until this, Linux kernels have had to be patched to
add support for perfmon2 which provided access to the counters for PAPI
and for CrayPat.

 Seamless to users except –
 Overhead incurred when accessing counters has increased
 Creates additional application perturbation
 Working to bring this back in line with perfmon2 overhead


 When possible, CrayPat will identify dominant communication grids
(communication patterns) in a program
 Example: nearest neighbor exchange in 2 or 3 dimensions
 Sweep3d uses a 2-D grid for communication

 Determine whether or not a custom MPI rank order will produce a
significant performance benefit

 Custom rank orders are helpful for programs with significant point-to-point
communication

 Doesn’t interfere with MPI collective communication optimizations


 Focuses on intra-node communication (place ranks that communication
frequently on the same node, or close by)
 Option to focus on other metrics such as memory bandwidth

 Determine rank order used during run that produced data
 Determine grid that defines the communication

 Produce a custom rank order if it’s beneficial based on grid size, grid order
and cost metric

 Summarize findings in report
 Describe how to re-run with custom rank order


For Sweep3d with 768 MPI ranks:

This application uses point-to-point MPI communication between nearest
neighbors in a 32 X 24 grid pattern. Time spent in this communication
accounted for over 50% of the execution time. A significant fraction (but
not more than 60%) of this time could potentially be saved by using the
rank order in the file MPICH_RANK_ORDER.g which was generated along
with this report.

To re-run with a custom rank order …


 Assist the user with application performance analysis and optimization
 Help user identify important and meaningful information from
potentially massive data sets
 Help user identify problem areas instead of just reporting data
 Bring optimization knowledge to a wider set of users

 Focus on ease of use and intuitive user interfaces
 Automatic program instrumentation
 Automatic analysis

 Target scalability issues in all areas of tool development
 Data management
 Storage, movement, presentation


 Supports traditional post-mortem performance analysis
 Automatic identification of performance problems
 Indication of causes of problems
 Suggestions of modifications for performance improvement

 CrayPat
 pat_build: automatic instrumentation (no source code changes needed)
 run-time library for measurements (transparent to the user)
 pat_report for performance analysis reports
 pat_help: online help utility

 Cray Apprentice2
 Graphical performance analysis and visualization tool


 CrayPat
 Instrumentation of optimized code
 No source code modification required
 Data collection transparent to the user
 Text-based performance reports
 Derived metrics
 Performance analysis

 Cray Apprentice2
 Performance data visualization tool
 Call tree view
 Source code mappings


 When performance measurement is triggered
 External agent (asynchronous)
 Sampling
 Timer interrupt
 Hardware counters overflow
 Internal agent (synchronous)
 Code instrumentation
 Event based
 Automatic or manual instrumentation
 How performance data is recorded
 Profile ::= Summation of events over time
 run time summarization (functions, call sites, loops, …)
 Trace file ::= Sequence of events over time


 Millions of lines of code
 Automatic profiling analysis
 Identifies top time consuming routines
 Automatically creates instrumentation template customized to your
application
 Lots of processes/threads
 Load imbalance analysis
 Identifies computational code regions and synchronization calls that
could benefit most from load balance optimization
 Estimates savings if corresponding section of code were balanced
 Long running applications
 Detection of outliers


 Important performance statistics:

 Top time consuming routines

 Load balance across computing resources

 Communication overhead

 Cache utilization

 FLOPS

 Vectorization (SSE instructions)

 Ratio of computation versus communication


 No source code or makefile modification required
 Automatic instrumentation at group (function) level
 Groups: mpi, io, heap, math SW, …

 Performs link-time instrumentation
 Requires object files
 Instruments optimized code
 Generates stand-alone instrumented program
 Preserves original binary
 Supports sample-based and event-based instrumentation


 Analyze the performance data and direct the user to meaningful
information

 Simplifies the procedure to instrument and collect performance data for
novice users

 Based on a two phase mechanism
1. Automatically detects the most time consuming functions in the
application and feeds this information back to the tool for further
(and focused) data collection

2. Provides performance information on the most significant parts of the
application


 Performs data conversion

 Combines information from binary with raw performance
data

 Performs analysis on data

 Generates text report of performance results

 Formats data for input into Cray Apprentice2


 Craypat / Cray Apprentice2 5.0 released September 10, 2009

 New internal data format
 FAQ
 Grid placement support
 Better caller information (ETC group in pat_report)
 Support larger numbers of processors
 Client/server version of Cray Apprentice2
 Panel help in Cray Apprentice2


 Access performance tools software

% module load perftools

 Build application keeping .o files (CCE: -h keepfiles)

% make clean
% make

 Instrument application for automatic profiling analysis
 You should get an instrumented program a.out+pat

% pat_build –O apa a.out

 Run application to get top time consuming routines
 You should get a performance file (“<sdatafile>.xf”) or
multiple files in a directory <sdatadir>

% aprun … a.out+pat (or qsub <pat script>)


 Generate report and .apa instrumentation file

% pat_report –o my_sampling_report [<sdatafile>.xf |
<sdatadir>]

 Inspect .apa file and sampling report

 Verify if additional instrumentation is needed

June 20, 2011 2011 HPCMP User Group © Cray Inc. Slide 148

# You can edit this file, if desired, and use it # 43.37% 99659 bytes
# to reinstrument the program for tracing like this: -T mlwxyz_
#
# pat_build -O mhd3d.Oapa.x+4125-401sdt.apa # 16.09% 17615 bytes
# -T half_
# These suggested trace options are based on data from:
# # 6.82% 6846 bytes
# /home/crayadm/ldr/mhd3d/run/mhd3d.Oapa.x+4125-401sdt.ap2, -T artv_
/home/crayadm/ldr/mhd3d/run/mhd3d.Oapa.x+4125-401sdt.xf

# 1.29% 5352 bytes
# ----------------------------------------------------------------------
-T currenh_

# HWPC group to collect by default.
# 1.03% 25294 bytes
-T bndbo_
-Drtenv=PAT_RT_HWPC=1 # Summary with instructions metrics.

# Functions below this point account for less than 10% of samples.
# ----------------------------------------------------------------------

# Libraries to trace.
# 1.03% 31240 bytes
# -T bndto_
-g mpi

...
# ----------------------------------------------------------------------

# ----------------------------------------------------------------------
# User-defined functions to trace, sorted by % of samples.
# Limited to top 200. A function is commented out if it has < 1%
-o mhd3d.x+apa # New instrumented program.
# of samples, or if a cumulative threshold of 90% has been reached,
# or if it has size < 200 bytes.
/work/crayadm/ldr/mhd3d/mhd3d.x # Original program.

# Note: -u should NOT be specified as an additional option.

June 20, 2011 149

 biolib Cray Bioinformatics library routines  omp OpenMP API (not supported on
 blacs Basic Linear Algebra communication Catamount)
subprograms  omp-rtl OpenMP runtime library (not
 blas Basic Linear Algebra subprograms supported on Catamount)

 caf Co-Array Fortran (Cray X2 systems only)  portals Lightweight message passing API
 fftw Fast Fourier Transform library (64-bit  pthreads POSIX threads (not supported on
only) Catamount)

 hdf5 manages extremely large and complex  scalapack Scalable LAPACK
data collections  shmem SHMEM
 heap dynamic heap  stdio all library functions that accept or return
 io includes stdio and sysio groups the FILE* construct

 lapack Linear Algebra Package  sysio I/O system calls

 lustre Lustre File System  system system calls

 math ANSI math  upc Unified Parallel C (Cray X2 systems only)
 mpi MPI
 netcdf network common data form (manages
array-oriented scientific data)


0 Summary with instruction 11 Floating point operations
metrics mix (2)
1 Summary with TLB metrics 12 Floating point operations
mix (vectorization)
2 L1 and L2 metrics
13 Floating point operations
3 Bandwidth information mix (SP)
4 Hypertransport information 14 Floating point operations
5 Floating point mix mix (DP)
6 Cycles stalled, resources 15 L3 (socket-level)
idle 16 L3 (core-level reads)
7 Cycles stalled, resources 17 L3 (core-level misses)
full
18 L3 (core-level fills caused
8 Instructions and branches by L2 evictions)
9 Instruction cache 19 Prefetches
10 Cache hierarchy

2011 HPCMP User Group © Cray Inc. June 20, 2011 Slide 151

 Regions, useful to break up long routines
 int PAT_region_begin (int id, const char *label)
 int PAT_region_end (int id)
 Disable/Enable Profiling, useful for excluding initialization
 int PAT_record (int state)
 Flush buffer, useful when program isn’t exiting cleanly
 int PAT_flush_buffer (void)


 Instrument application for further analysis (a.out+apa)

% pat_build –O <apafile>.apa

 Run application

% aprun … a.out+apa (or qsub <apa script>)

 Generate text report and visualization file (.ap2)

% pat_report –o my_text_report.txt [<datafile>.xf |
<datadir>]

 View report in text and/or with Cray Apprentice2

% app2 <datafile>.ap2


 MUST run on Lustre ( /work/… , /lus/…, /scratch/…, etc.)

 Number of files used to store raw data

 1 file created for program with 1 – 256 processes

 √n files created for program with 257 – n processes

 Ability to customize with PAT_RT_EXPFILE_MAX


 Full trace files show transient events but are too large

 Current run-time summarization misses transient events

 Plan to add ability to record:

 Top N peak values (N small)‫‏‬

 Approximate std dev over time

 For time, memory traffic, etc.

 During tracing and sampling


 Call graph profile  Cray Apprentice2
 Communication statistics  is target to help identify and
 Time-line view correct:
 Communication  Load imbalance
 I/O  Excessive communication
 Network contention
 Activity view
 Excessive serialization
 Pair-wise communication statistics  I/O Problems
 Text reports
 Source code mapping

June 20, 2011 157

Switch Overview display


Min, Avg, and Max
Values

-1, +1
Std Dev
marks


HPCMPUG2011 cray tutorial

Recomendados

Recomendados

Más contenido relacionado

La actualidad más candente

La actualidad más candente (20)

Similar a HPCMPUG2011 cray tutorial

Similar a HPCMPUG2011 cray tutorial (20)

Más de Jeff Larkin

Más de Jeff Larkin (16)

Último

Último (20)

HPCMPUG2011 cray tutorial