Package org.tensorflow.framework
Class GPUOptions.Builder
- java.lang.Object
-
- org.nd4j.shade.protobuf.AbstractMessageLite.Builder
-
- org.nd4j.shade.protobuf.AbstractMessage.Builder<BuilderType>
-
- org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
- org.tensorflow.framework.GPUOptions.Builder
-
- All Implemented Interfaces:
Cloneable
,org.nd4j.shade.protobuf.Message.Builder
,org.nd4j.shade.protobuf.MessageLite.Builder
,org.nd4j.shade.protobuf.MessageLiteOrBuilder
,org.nd4j.shade.protobuf.MessageOrBuilder
,GPUOptionsOrBuilder
- Enclosing class:
- GPUOptions
public static final class GPUOptions.Builder extends org.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder> implements GPUOptionsOrBuilder
Protobuf typetensorflow.GPUOptions
-
-
Method Summary
All Methods Static Methods Instance Methods Concrete Methods Modifier and Type Method Description GPUOptions.Builder
addRepeatedField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field, Object value)
GPUOptions
build()
GPUOptions
buildPartial()
GPUOptions.Builder
clear()
GPUOptions.Builder
clearAllocatorType()
The type of GPU allocation strategy to use.GPUOptions.Builder
clearAllowGrowth()
If true, the allocator does not pre-allocate the entire specified GPU memory region, instead starting small and growing as needed.GPUOptions.Builder
clearDeferredDeletionBytes()
Delay deletion of up to this many bytes to reduce the number of interactions with gpu driver code.GPUOptions.Builder
clearExperimental()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.GPUOptions.Builder
clearField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field)
GPUOptions.Builder
clearForceGpuCompatible()
Force all tensors to be gpu_compatible.GPUOptions.Builder
clearOneof(org.nd4j.shade.protobuf.Descriptors.OneofDescriptor oneof)
GPUOptions.Builder
clearPerProcessGpuMemoryFraction()
Fraction of the available GPU memory to allocate for each process.GPUOptions.Builder
clearPollingActiveDelayUsecs()
In the event polling loop sleep this many microseconds between PollEvents calls, when the queue is not empty.GPUOptions.Builder
clearPollingInactiveDelayMsecs()
This field is deprecated and ignored.GPUOptions.Builder
clearVisibleDeviceList()
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices.GPUOptions.Builder
clone()
String
getAllocatorType()
The type of GPU allocation strategy to use.org.nd4j.shade.protobuf.ByteString
getAllocatorTypeBytes()
The type of GPU allocation strategy to use.boolean
getAllowGrowth()
If true, the allocator does not pre-allocate the entire specified GPU memory region, instead starting small and growing as needed.GPUOptions
getDefaultInstanceForType()
long
getDeferredDeletionBytes()
Delay deletion of up to this many bytes to reduce the number of interactions with gpu driver code.static org.nd4j.shade.protobuf.Descriptors.Descriptor
getDescriptor()
org.nd4j.shade.protobuf.Descriptors.Descriptor
getDescriptorForType()
GPUOptions.Experimental
getExperimental()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.GPUOptions.Experimental.Builder
getExperimentalBuilder()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.GPUOptions.ExperimentalOrBuilder
getExperimentalOrBuilder()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.boolean
getForceGpuCompatible()
Force all tensors to be gpu_compatible.double
getPerProcessGpuMemoryFraction()
Fraction of the available GPU memory to allocate for each process.int
getPollingActiveDelayUsecs()
In the event polling loop sleep this many microseconds between PollEvents calls, when the queue is not empty.int
getPollingInactiveDelayMsecs()
This field is deprecated and ignored.String
getVisibleDeviceList()
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices.org.nd4j.shade.protobuf.ByteString
getVisibleDeviceListBytes()
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices.boolean
hasExperimental()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.protected org.nd4j.shade.protobuf.GeneratedMessageV3.FieldAccessorTable
internalGetFieldAccessorTable()
boolean
isInitialized()
GPUOptions.Builder
mergeExperimental(GPUOptions.Experimental value)
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.GPUOptions.Builder
mergeFrom(org.nd4j.shade.protobuf.CodedInputStream input, org.nd4j.shade.protobuf.ExtensionRegistryLite extensionRegistry)
GPUOptions.Builder
mergeFrom(org.nd4j.shade.protobuf.Message other)
GPUOptions.Builder
mergeFrom(GPUOptions other)
GPUOptions.Builder
mergeUnknownFields(org.nd4j.shade.protobuf.UnknownFieldSet unknownFields)
GPUOptions.Builder
setAllocatorType(String value)
The type of GPU allocation strategy to use.GPUOptions.Builder
setAllocatorTypeBytes(org.nd4j.shade.protobuf.ByteString value)
The type of GPU allocation strategy to use.GPUOptions.Builder
setAllowGrowth(boolean value)
If true, the allocator does not pre-allocate the entire specified GPU memory region, instead starting small and growing as needed.GPUOptions.Builder
setDeferredDeletionBytes(long value)
Delay deletion of up to this many bytes to reduce the number of interactions with gpu driver code.GPUOptions.Builder
setExperimental(GPUOptions.Experimental value)
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.GPUOptions.Builder
setExperimental(GPUOptions.Experimental.Builder builderForValue)
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.GPUOptions.Builder
setField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field, Object value)
GPUOptions.Builder
setForceGpuCompatible(boolean value)
Force all tensors to be gpu_compatible.GPUOptions.Builder
setPerProcessGpuMemoryFraction(double value)
Fraction of the available GPU memory to allocate for each process.GPUOptions.Builder
setPollingActiveDelayUsecs(int value)
In the event polling loop sleep this many microseconds between PollEvents calls, when the queue is not empty.GPUOptions.Builder
setPollingInactiveDelayMsecs(int value)
This field is deprecated and ignored.GPUOptions.Builder
setRepeatedField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field, int index, Object value)
GPUOptions.Builder
setUnknownFields(org.nd4j.shade.protobuf.UnknownFieldSet unknownFields)
GPUOptions.Builder
setVisibleDeviceList(String value)
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices.GPUOptions.Builder
setVisibleDeviceListBytes(org.nd4j.shade.protobuf.ByteString value)
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices.-
Methods inherited from class org.nd4j.shade.protobuf.GeneratedMessageV3.Builder
getAllFields, getField, getFieldBuilder, getOneofFieldDescriptor, getParentForChildren, getRepeatedField, getRepeatedFieldBuilder, getRepeatedFieldCount, getUnknownFields, hasField, hasOneof, internalGetMapField, internalGetMutableMapField, isClean, markClean, newBuilderForField, onBuilt, onChanged, setUnknownFieldsProto3
-
Methods inherited from class org.nd4j.shade.protobuf.AbstractMessage.Builder
findInitializationErrors, getInitializationErrorString, internalMergeFrom, mergeFrom, mergeFrom, mergeFrom, mergeFrom, mergeFrom, mergeFrom, mergeFrom, mergeFrom, mergeFrom, newUninitializedMessageException, toString
-
Methods inherited from class org.nd4j.shade.protobuf.AbstractMessageLite.Builder
addAll, addAll, mergeDelimitedFrom, mergeDelimitedFrom, mergeFrom, newUninitializedMessageException
-
Methods inherited from class java.lang.Object
equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
-
-
-
-
Method Detail
-
getDescriptor
public static final org.nd4j.shade.protobuf.Descriptors.Descriptor getDescriptor()
-
internalGetFieldAccessorTable
protected org.nd4j.shade.protobuf.GeneratedMessageV3.FieldAccessorTable internalGetFieldAccessorTable()
- Specified by:
internalGetFieldAccessorTable
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
clear
public GPUOptions.Builder clear()
- Specified by:
clear
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Specified by:
clear
in interfaceorg.nd4j.shade.protobuf.MessageLite.Builder
- Overrides:
clear
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
getDescriptorForType
public org.nd4j.shade.protobuf.Descriptors.Descriptor getDescriptorForType()
- Specified by:
getDescriptorForType
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Specified by:
getDescriptorForType
in interfaceorg.nd4j.shade.protobuf.MessageOrBuilder
- Overrides:
getDescriptorForType
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
getDefaultInstanceForType
public GPUOptions getDefaultInstanceForType()
- Specified by:
getDefaultInstanceForType
in interfaceorg.nd4j.shade.protobuf.MessageLiteOrBuilder
- Specified by:
getDefaultInstanceForType
in interfaceorg.nd4j.shade.protobuf.MessageOrBuilder
-
build
public GPUOptions build()
- Specified by:
build
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Specified by:
build
in interfaceorg.nd4j.shade.protobuf.MessageLite.Builder
-
buildPartial
public GPUOptions buildPartial()
- Specified by:
buildPartial
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Specified by:
buildPartial
in interfaceorg.nd4j.shade.protobuf.MessageLite.Builder
-
clone
public GPUOptions.Builder clone()
- Specified by:
clone
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Specified by:
clone
in interfaceorg.nd4j.shade.protobuf.MessageLite.Builder
- Overrides:
clone
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
setField
public GPUOptions.Builder setField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field, Object value)
- Specified by:
setField
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
setField
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
clearField
public GPUOptions.Builder clearField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field)
- Specified by:
clearField
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
clearField
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
clearOneof
public GPUOptions.Builder clearOneof(org.nd4j.shade.protobuf.Descriptors.OneofDescriptor oneof)
- Specified by:
clearOneof
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
clearOneof
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
setRepeatedField
public GPUOptions.Builder setRepeatedField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field, int index, Object value)
- Specified by:
setRepeatedField
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
setRepeatedField
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
addRepeatedField
public GPUOptions.Builder addRepeatedField(org.nd4j.shade.protobuf.Descriptors.FieldDescriptor field, Object value)
- Specified by:
addRepeatedField
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
addRepeatedField
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
mergeFrom
public GPUOptions.Builder mergeFrom(org.nd4j.shade.protobuf.Message other)
- Specified by:
mergeFrom
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
mergeFrom
in classorg.nd4j.shade.protobuf.AbstractMessage.Builder<GPUOptions.Builder>
-
mergeFrom
public GPUOptions.Builder mergeFrom(GPUOptions other)
-
isInitialized
public final boolean isInitialized()
- Specified by:
isInitialized
in interfaceorg.nd4j.shade.protobuf.MessageLiteOrBuilder
- Overrides:
isInitialized
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
mergeFrom
public GPUOptions.Builder mergeFrom(org.nd4j.shade.protobuf.CodedInputStream input, org.nd4j.shade.protobuf.ExtensionRegistryLite extensionRegistry) throws IOException
- Specified by:
mergeFrom
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Specified by:
mergeFrom
in interfaceorg.nd4j.shade.protobuf.MessageLite.Builder
- Overrides:
mergeFrom
in classorg.nd4j.shade.protobuf.AbstractMessage.Builder<GPUOptions.Builder>
- Throws:
IOException
-
getPerProcessGpuMemoryFraction
public double getPerProcessGpuMemoryFraction()
Fraction of the available GPU memory to allocate for each process. 1 means to allocate all of the GPU memory, 0.5 means the process allocates up to ~50% of the available GPU memory. GPU memory is pre-allocated unless the allow_growth option is enabled. If greater than 1.0, uses CUDA unified memory to potentially oversubscribe the amount of memory available on the GPU device by using host memory as a swap space. Accessing memory not available on the device will be significantly slower as that would require memory transfer between the host and the device. Options to reduce the memory requirement should be considered before enabling this option as this may come with a negative performance impact. Oversubscription using the unified memory requires Pascal class or newer GPUs and it is currently only supported on the Linux operating system. See https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements for the detailed requirements.
double per_process_gpu_memory_fraction = 1;
- Specified by:
getPerProcessGpuMemoryFraction
in interfaceGPUOptionsOrBuilder
- Returns:
- The perProcessGpuMemoryFraction.
-
setPerProcessGpuMemoryFraction
public GPUOptions.Builder setPerProcessGpuMemoryFraction(double value)
Fraction of the available GPU memory to allocate for each process. 1 means to allocate all of the GPU memory, 0.5 means the process allocates up to ~50% of the available GPU memory. GPU memory is pre-allocated unless the allow_growth option is enabled. If greater than 1.0, uses CUDA unified memory to potentially oversubscribe the amount of memory available on the GPU device by using host memory as a swap space. Accessing memory not available on the device will be significantly slower as that would require memory transfer between the host and the device. Options to reduce the memory requirement should be considered before enabling this option as this may come with a negative performance impact. Oversubscription using the unified memory requires Pascal class or newer GPUs and it is currently only supported on the Linux operating system. See https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements for the detailed requirements.
double per_process_gpu_memory_fraction = 1;
- Parameters:
value
- The perProcessGpuMemoryFraction to set.- Returns:
- This builder for chaining.
-
clearPerProcessGpuMemoryFraction
public GPUOptions.Builder clearPerProcessGpuMemoryFraction()
Fraction of the available GPU memory to allocate for each process. 1 means to allocate all of the GPU memory, 0.5 means the process allocates up to ~50% of the available GPU memory. GPU memory is pre-allocated unless the allow_growth option is enabled. If greater than 1.0, uses CUDA unified memory to potentially oversubscribe the amount of memory available on the GPU device by using host memory as a swap space. Accessing memory not available on the device will be significantly slower as that would require memory transfer between the host and the device. Options to reduce the memory requirement should be considered before enabling this option as this may come with a negative performance impact. Oversubscription using the unified memory requires Pascal class or newer GPUs and it is currently only supported on the Linux operating system. See https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#um-requirements for the detailed requirements.
double per_process_gpu_memory_fraction = 1;
- Returns:
- This builder for chaining.
-
getAllowGrowth
public boolean getAllowGrowth()
If true, the allocator does not pre-allocate the entire specified GPU memory region, instead starting small and growing as needed.
bool allow_growth = 4;
- Specified by:
getAllowGrowth
in interfaceGPUOptionsOrBuilder
- Returns:
- The allowGrowth.
-
setAllowGrowth
public GPUOptions.Builder setAllowGrowth(boolean value)
If true, the allocator does not pre-allocate the entire specified GPU memory region, instead starting small and growing as needed.
bool allow_growth = 4;
- Parameters:
value
- The allowGrowth to set.- Returns:
- This builder for chaining.
-
clearAllowGrowth
public GPUOptions.Builder clearAllowGrowth()
If true, the allocator does not pre-allocate the entire specified GPU memory region, instead starting small and growing as needed.
bool allow_growth = 4;
- Returns:
- This builder for chaining.
-
getAllocatorType
public String getAllocatorType()
The type of GPU allocation strategy to use. Allowed values: "": The empty string (default) uses a system-chosen default which may change over time. "BFC": A "Best-fit with coalescing" algorithm, simplified from a version of dlmalloc.
string allocator_type = 2;
- Specified by:
getAllocatorType
in interfaceGPUOptionsOrBuilder
- Returns:
- The allocatorType.
-
getAllocatorTypeBytes
public org.nd4j.shade.protobuf.ByteString getAllocatorTypeBytes()
The type of GPU allocation strategy to use. Allowed values: "": The empty string (default) uses a system-chosen default which may change over time. "BFC": A "Best-fit with coalescing" algorithm, simplified from a version of dlmalloc.
string allocator_type = 2;
- Specified by:
getAllocatorTypeBytes
in interfaceGPUOptionsOrBuilder
- Returns:
- The bytes for allocatorType.
-
setAllocatorType
public GPUOptions.Builder setAllocatorType(String value)
The type of GPU allocation strategy to use. Allowed values: "": The empty string (default) uses a system-chosen default which may change over time. "BFC": A "Best-fit with coalescing" algorithm, simplified from a version of dlmalloc.
string allocator_type = 2;
- Parameters:
value
- The allocatorType to set.- Returns:
- This builder for chaining.
-
clearAllocatorType
public GPUOptions.Builder clearAllocatorType()
The type of GPU allocation strategy to use. Allowed values: "": The empty string (default) uses a system-chosen default which may change over time. "BFC": A "Best-fit with coalescing" algorithm, simplified from a version of dlmalloc.
string allocator_type = 2;
- Returns:
- This builder for chaining.
-
setAllocatorTypeBytes
public GPUOptions.Builder setAllocatorTypeBytes(org.nd4j.shade.protobuf.ByteString value)
The type of GPU allocation strategy to use. Allowed values: "": The empty string (default) uses a system-chosen default which may change over time. "BFC": A "Best-fit with coalescing" algorithm, simplified from a version of dlmalloc.
string allocator_type = 2;
- Parameters:
value
- The bytes for allocatorType to set.- Returns:
- This builder for chaining.
-
getDeferredDeletionBytes
public long getDeferredDeletionBytes()
Delay deletion of up to this many bytes to reduce the number of interactions with gpu driver code. If 0, the system chooses a reasonable default (several MBs).
int64 deferred_deletion_bytes = 3;
- Specified by:
getDeferredDeletionBytes
in interfaceGPUOptionsOrBuilder
- Returns:
- The deferredDeletionBytes.
-
setDeferredDeletionBytes
public GPUOptions.Builder setDeferredDeletionBytes(long value)
Delay deletion of up to this many bytes to reduce the number of interactions with gpu driver code. If 0, the system chooses a reasonable default (several MBs).
int64 deferred_deletion_bytes = 3;
- Parameters:
value
- The deferredDeletionBytes to set.- Returns:
- This builder for chaining.
-
clearDeferredDeletionBytes
public GPUOptions.Builder clearDeferredDeletionBytes()
Delay deletion of up to this many bytes to reduce the number of interactions with gpu driver code. If 0, the system chooses a reasonable default (several MBs).
int64 deferred_deletion_bytes = 3;
- Returns:
- This builder for chaining.
-
getVisibleDeviceList
public String getVisibleDeviceList()
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices. For example, if TensorFlow can see 8 GPU devices in the process, and one wanted to map visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1", then one would specify this field as "5,3". This field is similar in spirit to the CUDA_VISIBLE_DEVICES environment variable, except it applies to the visible GPU devices in the process. NOTE: 1. The GPU driver provides the process with the visible GPUs in an order which is not guaranteed to have any correlation to the *physical* GPU id in the machine. This field is used for remapping "visible" to "virtual", which means this operates only after the process starts. Users are required to use vendor specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the physical to visible device mapping prior to invoking TensorFlow. 2. In the code, the ids in this list are also called "CUDA GPU id"s, and the 'virtual' ids of GPU devices (i.e. the ids in the device name "/device:GPU:<id>") are also called "TF GPU id"s. Please refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h for more information.
string visible_device_list = 5;
- Specified by:
getVisibleDeviceList
in interfaceGPUOptionsOrBuilder
- Returns:
- The visibleDeviceList.
-
getVisibleDeviceListBytes
public org.nd4j.shade.protobuf.ByteString getVisibleDeviceListBytes()
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices. For example, if TensorFlow can see 8 GPU devices in the process, and one wanted to map visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1", then one would specify this field as "5,3". This field is similar in spirit to the CUDA_VISIBLE_DEVICES environment variable, except it applies to the visible GPU devices in the process. NOTE: 1. The GPU driver provides the process with the visible GPUs in an order which is not guaranteed to have any correlation to the *physical* GPU id in the machine. This field is used for remapping "visible" to "virtual", which means this operates only after the process starts. Users are required to use vendor specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the physical to visible device mapping prior to invoking TensorFlow. 2. In the code, the ids in this list are also called "CUDA GPU id"s, and the 'virtual' ids of GPU devices (i.e. the ids in the device name "/device:GPU:<id>") are also called "TF GPU id"s. Please refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h for more information.
string visible_device_list = 5;
- Specified by:
getVisibleDeviceListBytes
in interfaceGPUOptionsOrBuilder
- Returns:
- The bytes for visibleDeviceList.
-
setVisibleDeviceList
public GPUOptions.Builder setVisibleDeviceList(String value)
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices. For example, if TensorFlow can see 8 GPU devices in the process, and one wanted to map visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1", then one would specify this field as "5,3". This field is similar in spirit to the CUDA_VISIBLE_DEVICES environment variable, except it applies to the visible GPU devices in the process. NOTE: 1. The GPU driver provides the process with the visible GPUs in an order which is not guaranteed to have any correlation to the *physical* GPU id in the machine. This field is used for remapping "visible" to "virtual", which means this operates only after the process starts. Users are required to use vendor specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the physical to visible device mapping prior to invoking TensorFlow. 2. In the code, the ids in this list are also called "CUDA GPU id"s, and the 'virtual' ids of GPU devices (i.e. the ids in the device name "/device:GPU:<id>") are also called "TF GPU id"s. Please refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h for more information.
string visible_device_list = 5;
- Parameters:
value
- The visibleDeviceList to set.- Returns:
- This builder for chaining.
-
clearVisibleDeviceList
public GPUOptions.Builder clearVisibleDeviceList()
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices. For example, if TensorFlow can see 8 GPU devices in the process, and one wanted to map visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1", then one would specify this field as "5,3". This field is similar in spirit to the CUDA_VISIBLE_DEVICES environment variable, except it applies to the visible GPU devices in the process. NOTE: 1. The GPU driver provides the process with the visible GPUs in an order which is not guaranteed to have any correlation to the *physical* GPU id in the machine. This field is used for remapping "visible" to "virtual", which means this operates only after the process starts. Users are required to use vendor specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the physical to visible device mapping prior to invoking TensorFlow. 2. In the code, the ids in this list are also called "CUDA GPU id"s, and the 'virtual' ids of GPU devices (i.e. the ids in the device name "/device:GPU:<id>") are also called "TF GPU id"s. Please refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h for more information.
string visible_device_list = 5;
- Returns:
- This builder for chaining.
-
setVisibleDeviceListBytes
public GPUOptions.Builder setVisibleDeviceListBytes(org.nd4j.shade.protobuf.ByteString value)
A comma-separated list of GPU ids that determines the 'visible' to 'virtual' mapping of GPU devices. For example, if TensorFlow can see 8 GPU devices in the process, and one wanted to map visible GPU devices 5 and 3 as "/device:GPU:0", and "/device:GPU:1", then one would specify this field as "5,3". This field is similar in spirit to the CUDA_VISIBLE_DEVICES environment variable, except it applies to the visible GPU devices in the process. NOTE: 1. The GPU driver provides the process with the visible GPUs in an order which is not guaranteed to have any correlation to the *physical* GPU id in the machine. This field is used for remapping "visible" to "virtual", which means this operates only after the process starts. Users are required to use vendor specific mechanisms (e.g., CUDA_VISIBLE_DEVICES) to control the physical to visible device mapping prior to invoking TensorFlow. 2. In the code, the ids in this list are also called "CUDA GPU id"s, and the 'virtual' ids of GPU devices (i.e. the ids in the device name "/device:GPU:<id>") are also called "TF GPU id"s. Please refer to third_party/tensorflow/core/common_runtime/gpu/gpu_id.h for more information.
string visible_device_list = 5;
- Parameters:
value
- The bytes for visibleDeviceList to set.- Returns:
- This builder for chaining.
-
getPollingActiveDelayUsecs
public int getPollingActiveDelayUsecs()
In the event polling loop sleep this many microseconds between PollEvents calls, when the queue is not empty. If value is not set or set to 0, gets set to a non-zero default.
int32 polling_active_delay_usecs = 6;
- Specified by:
getPollingActiveDelayUsecs
in interfaceGPUOptionsOrBuilder
- Returns:
- The pollingActiveDelayUsecs.
-
setPollingActiveDelayUsecs
public GPUOptions.Builder setPollingActiveDelayUsecs(int value)
In the event polling loop sleep this many microseconds between PollEvents calls, when the queue is not empty. If value is not set or set to 0, gets set to a non-zero default.
int32 polling_active_delay_usecs = 6;
- Parameters:
value
- The pollingActiveDelayUsecs to set.- Returns:
- This builder for chaining.
-
clearPollingActiveDelayUsecs
public GPUOptions.Builder clearPollingActiveDelayUsecs()
In the event polling loop sleep this many microseconds between PollEvents calls, when the queue is not empty. If value is not set or set to 0, gets set to a non-zero default.
int32 polling_active_delay_usecs = 6;
- Returns:
- This builder for chaining.
-
getPollingInactiveDelayMsecs
public int getPollingInactiveDelayMsecs()
This field is deprecated and ignored.
int32 polling_inactive_delay_msecs = 7;
- Specified by:
getPollingInactiveDelayMsecs
in interfaceGPUOptionsOrBuilder
- Returns:
- The pollingInactiveDelayMsecs.
-
setPollingInactiveDelayMsecs
public GPUOptions.Builder setPollingInactiveDelayMsecs(int value)
This field is deprecated and ignored.
int32 polling_inactive_delay_msecs = 7;
- Parameters:
value
- The pollingInactiveDelayMsecs to set.- Returns:
- This builder for chaining.
-
clearPollingInactiveDelayMsecs
public GPUOptions.Builder clearPollingInactiveDelayMsecs()
This field is deprecated and ignored.
int32 polling_inactive_delay_msecs = 7;
- Returns:
- This builder for chaining.
-
getForceGpuCompatible
public boolean getForceGpuCompatible()
Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow, enabling this option forces all CPU tensors to be allocated with Cuda pinned memory. Normally, TensorFlow will infer which tensors should be allocated as the pinned memory. But in case where the inference is incomplete, this option can significantly speed up the cross-device memory copy performance as long as it fits the memory. Note that this option is not something that should be enabled by default for unknown or very large models, since all Cuda pinned memory is unpageable, having too much pinned memory might negatively impact the overall host system performance.
bool force_gpu_compatible = 8;
- Specified by:
getForceGpuCompatible
in interfaceGPUOptionsOrBuilder
- Returns:
- The forceGpuCompatible.
-
setForceGpuCompatible
public GPUOptions.Builder setForceGpuCompatible(boolean value)
Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow, enabling this option forces all CPU tensors to be allocated with Cuda pinned memory. Normally, TensorFlow will infer which tensors should be allocated as the pinned memory. But in case where the inference is incomplete, this option can significantly speed up the cross-device memory copy performance as long as it fits the memory. Note that this option is not something that should be enabled by default for unknown or very large models, since all Cuda pinned memory is unpageable, having too much pinned memory might negatively impact the overall host system performance.
bool force_gpu_compatible = 8;
- Parameters:
value
- The forceGpuCompatible to set.- Returns:
- This builder for chaining.
-
clearForceGpuCompatible
public GPUOptions.Builder clearForceGpuCompatible()
Force all tensors to be gpu_compatible. On a GPU-enabled TensorFlow, enabling this option forces all CPU tensors to be allocated with Cuda pinned memory. Normally, TensorFlow will infer which tensors should be allocated as the pinned memory. But in case where the inference is incomplete, this option can significantly speed up the cross-device memory copy performance as long as it fits the memory. Note that this option is not something that should be enabled by default for unknown or very large models, since all Cuda pinned memory is unpageable, having too much pinned memory might negatively impact the overall host system performance.
bool force_gpu_compatible = 8;
- Returns:
- This builder for chaining.
-
hasExperimental
public boolean hasExperimental()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
- Specified by:
hasExperimental
in interfaceGPUOptionsOrBuilder
- Returns:
- Whether the experimental field is set.
-
getExperimental
public GPUOptions.Experimental getExperimental()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
- Specified by:
getExperimental
in interfaceGPUOptionsOrBuilder
- Returns:
- The experimental.
-
setExperimental
public GPUOptions.Builder setExperimental(GPUOptions.Experimental value)
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
-
setExperimental
public GPUOptions.Builder setExperimental(GPUOptions.Experimental.Builder builderForValue)
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
-
mergeExperimental
public GPUOptions.Builder mergeExperimental(GPUOptions.Experimental value)
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
-
clearExperimental
public GPUOptions.Builder clearExperimental()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
-
getExperimentalBuilder
public GPUOptions.Experimental.Builder getExperimentalBuilder()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
-
getExperimentalOrBuilder
public GPUOptions.ExperimentalOrBuilder getExperimentalOrBuilder()
Everything inside experimental is subject to change and is not subject to API stability guarantees in https://www.tensorflow.org/guide/version_compat.
.tensorflow.GPUOptions.Experimental experimental = 9;
- Specified by:
getExperimentalOrBuilder
in interfaceGPUOptionsOrBuilder
-
setUnknownFields
public final GPUOptions.Builder setUnknownFields(org.nd4j.shade.protobuf.UnknownFieldSet unknownFields)
- Specified by:
setUnknownFields
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
setUnknownFields
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
mergeUnknownFields
public final GPUOptions.Builder mergeUnknownFields(org.nd4j.shade.protobuf.UnknownFieldSet unknownFields)
- Specified by:
mergeUnknownFields
in interfaceorg.nd4j.shade.protobuf.Message.Builder
- Overrides:
mergeUnknownFields
in classorg.nd4j.shade.protobuf.GeneratedMessageV3.Builder<GPUOptions.Builder>
-
-