Commit eb8baaba authored by Junchao Zhang's avatar Junchao Zhang Committed by Pavan Balaji
Browse files

Set almost all cvar scope as MPI_T_SCOPE_ALL_EQ



This is a conservative setting, but should make MPICH "MPI_T-correct".
Once we know more about cvars, we can incrementally relax their scopes.

Fixes #1959
Signed-off-by: Pavan Balaji's avatarPavan Balaji <balaji@mcs.anl.gov>
parent b7186c70
......@@ -61,7 +61,7 @@ cvars:
default : 256
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the short message algorithm will be used if the per-destination
message size (sendcount*size(sendtype)) is <= this value
......@@ -72,7 +72,7 @@ cvars:
default : 32768
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the medium message algorithm will be used if the per-destination
message size (sendcount*size(sendtype)) is <= this value and
......@@ -84,7 +84,7 @@ cvars:
default : 32
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
max no. of irecvs/isends posted at a time in some alltoall
algorithms. Setting it to 0 causes all irecvs/isends to be
......@@ -96,7 +96,7 @@ cvars:
default : 524288
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the long message algorithm will be used if the operation is commutative
and the send buffer size is >= this value (in bytes)
......@@ -107,7 +107,7 @@ cvars:
default : 8
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the minimum number of processes in a communicator to use a non-binomial
broadcast algorithm
......@@ -118,7 +118,7 @@ cvars:
default : 12288
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the short message algorithm will be used if the send buffer size is <
this value (in bytes)
......@@ -129,7 +129,7 @@ cvars:
default : 524288
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the long message algorithm will be used if the send buffer size is >=
this value (in bytes)
......@@ -140,7 +140,7 @@ cvars:
default : 81920
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
For MPI_Allgather and MPI_Allgatherv, the short message algorithm will
be used if the send buffer size is < this value (in bytes).
......@@ -151,7 +151,7 @@ cvars:
default : 524288
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
For MPI_Allgather and MPI_Allgatherv, the long message algorithm will be
used if the send buffer size is >= this value (in bytes)
......@@ -162,7 +162,7 @@ cvars:
default : 2048
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the short message algorithm will be used if the send buffer size is <=
this value (in bytes)
......@@ -173,7 +173,7 @@ cvars:
default : 2048
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
the short message algorithm will be used if the send buffer size is <=
this value (in bytes)
......@@ -184,7 +184,7 @@ cvars:
default : 1024
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
use a temporary buffer for intracommunicator MPI_Gather if the send
buffer size is < this value (in bytes)
......@@ -195,7 +195,7 @@ cvars:
default : 2048
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
use the short message algorithm for intercommunicator MPI_Gather if the
send buffer size is < this value (in bytes)
......@@ -206,7 +206,7 @@ cvars:
default : 32
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Use Ssend (synchronous send) for intercommunicator MPI_Gatherv if the
"group B" size is >= this value. Specifying "-1" always avoids using
......@@ -219,7 +219,7 @@ cvars:
default : 2048
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
use the short message algorithm for intercommunicator MPI_Scatter if the
send buffer size is < this value (in bytes)
......@@ -230,7 +230,7 @@ cvars:
default : 32768
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
The smallest message size that will be used for the pipelined, large-message,
ring algorithm in the MPI_Allgatherv implementation.
......@@ -241,7 +241,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Enable SMP aware collective communication.
......@@ -251,7 +251,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Enable SMP aware barrier.
......@@ -261,7 +261,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Enable SMP aware broadcast.
......@@ -271,7 +271,7 @@ cvars:
default : 0
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Maximum message size for which SMP-aware broadcast is used. A
value of '0' uses SMP-aware broadcast for all message sizes.
......@@ -282,7 +282,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Enable SMP aware reduce.
......@@ -292,7 +292,7 @@ cvars:
default : 0
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Maximum message size for which SMP-aware reduce is used. A
value of '0' uses SMP-aware reduce for all message sizes.
......@@ -303,7 +303,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Enable SMP aware allreduce.
......@@ -313,7 +313,7 @@ cvars:
default : 0
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Maximum message size for which SMP-aware allreduce is used. A
value of '0' uses SMP-aware allreduce for all message sizes.
......@@ -326,7 +326,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Use qsort(3) in the implementation of MPI_Comm_split instead of bubble sort.
......@@ -338,7 +338,7 @@ cvars:
default : false
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, causes processes to wait in MPI_Init and
MPI_Initthread for a debugger to be attached. Once the
......@@ -352,7 +352,7 @@ cvars:
default : 64
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Size of the "MPIR" debugger interface proctable (process table).
......@@ -362,7 +362,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, dump the proctable entries at MPIR_WaitForDebugger-time.
(currently compile-time disabled by "#if 0")
......@@ -375,7 +375,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
DEPRECATED! Will be removed in MPICH-3.2
Collectives called on a communicator with a failed process
......@@ -392,7 +392,7 @@ cvars:
default : false
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, MPI will call MPI_Abort at MPI_Finalize if any MPI object
handles have been leaked. For example, if MPI_Comm_dup is called
......@@ -410,7 +410,7 @@ cvars:
default : 2
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
The MPIR_CVAR_CTXID_EAGER_SIZE environment variable allows you to
specify how many words in the context ID mask will be set aside
......@@ -423,7 +423,7 @@ cvars:
default : false
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If set to true, MPICH will initiate an additional thread to
make asynchronous progress on all communication operations
......@@ -444,7 +444,7 @@ cvars:
default : "MPI_THREAD_SINGLE"
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Sets the default thread level to use when using MPI_INIT.
......@@ -455,7 +455,7 @@ cvars:
type : boolean
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
verbosity : MPI_T_VERBOSITY_MPIDEV_DETAIL
scope : MPI_T_SCOPE_LOCAL
description : >-
If true, list any memory that was allocated by MPICH and that
......@@ -467,7 +467,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, perform checks for errors, typically to verify valid inputs
to MPI routines. Only effective when MPICH is configured with
......@@ -479,7 +479,7 @@ cvars:
default : true
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, print an error stack trace at error handling time.
......@@ -489,7 +489,7 @@ cvars:
default : 0
class : device
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If >0, truncate error stack output lines this many characters
wide. If 0, do not truncate, and if <0 use a sensible default.
......@@ -502,7 +502,7 @@ cvars:
default : true
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Use the immediate accumulate optimization
......@@ -512,7 +512,7 @@ cvars:
default : 4000
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Threshold at which the RMA implementation attempts to complete requests
while completing RMA operations and while using the lazy synchonization
......@@ -525,7 +525,7 @@ cvars:
default : 128
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Threshold for the number of new requests since the last attempt to
complete pending requests. Higher values can increase performance,
......@@ -538,7 +538,7 @@ cvars:
default : false
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Issue a request for the passive target RMA lock immediately. Default
behavior is to defer the lock request until the call to MPI_Win_unlock.
......@@ -549,7 +549,7 @@ cvars:
default : true
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Enable/disable an optimization that merges lock, op, and unlock
messages, for single-operation passive target epochs.
......@@ -562,7 +562,7 @@ cvars:
default : false
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, force all processes to operate as though all processes
are located on another node. For example, this disables shared
......@@ -576,7 +576,7 @@ cvars:
default : false
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, odd procs on a node are seen as local to each other, and even
procs on a node are seen as local to each other. Used for debugging on
......@@ -590,7 +590,7 @@ cvars:
defaultliteral : NULL
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If non-NULL, this cvar specifies the IP address that
other processes should use when connecting to this process.
......@@ -604,7 +604,7 @@ cvars:
default : 131072
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
This cvar controls the message size at which CH3 switches
from eager to rendezvous mode.
......@@ -618,7 +618,7 @@ cvars:
default : "0:0"
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
The MPIR_CVAR_CH3_PORT_RANGE environment variable allows you to
specify the range of TCP ports to be used by the process
......@@ -633,7 +633,7 @@ cvars:
default : 1000
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
When MPICH is in a busy waiting loop, it will periodically
call a function to yield the processor. This cvar sets
......@@ -646,7 +646,7 @@ cvars:
default : 2097152
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Messages larger than this size will use the "dma" (knem)
intranode LMT implementation, if it is enabled and available.
......@@ -657,7 +657,7 @@ cvars:
default : ""
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If non-empty, this cvar specifies which network module
should be used for communication.
......@@ -668,7 +668,7 @@ cvars:
default : -1
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
This cvar controls the message size at which Nemesis
switches from eager to rendezvous mode for shared memory.
......@@ -681,7 +681,7 @@ cvars:
default : -2
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
This cvar controls the message size at which Nemesis
switches from eager to rendezvous mode for ready-send
......@@ -695,7 +695,7 @@ cvars:
default : false
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If true, enables checkpointing support and returns an error if
checkpointing library cannot be initialized.
......@@ -710,7 +710,7 @@ cvars:
defaultliteral : NULL
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If non-NULL, this cvar specifies which pseudo-ethernet
interface the tcp netmod should use (e.g., "eth1", "ib0").
......@@ -725,7 +725,7 @@ cvars:
default : 10
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
This cvar controls the number of times to retry the
gethostbyname() function before giving up.
......@@ -738,7 +738,7 @@ cvars:
default : true
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
If set to false, communication override functionality will be
disabled for netmods that provide the override feature.
......@@ -755,7 +755,7 @@ cvars:
defaultliteral : NULL
class : none
verbosity : MPI_T_VERBOSITY_USER_BASIC
scope : MPI_T_SCOPE_LOCAL
scope : MPI_T_SCOPE_ALL_EQ
description : >-
Sets the directory to use for MPI service publishing in the
file nameserv implementation. Allows the user to override
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment