use MiB instead of MB for gpu-memory

Signed-off-by: Natasha Romm <nromm@nvidia.com>

Author: natasharomm

docs/gpu-feature-discovery/README.md

@@ -216,7 +216,7 @@ For a similar list of labels generated or used by the device plugin, see [here](
 | nvidia.com/gpu.count           | Integer    | Number of GPUs                                                                                                                                                                         | 2              |
 | nvidia.com/gpu.family          | String     | Architecture family of the GPU                                                                                                                                                         | turing         |
 | nvidia.com/gpu.machine         | String     | Machine type. If in a public cloud provider, value may be set to the instance type.                                                                                                    | DGX-1          |
-| nvidia.com/gpu.memory          | Integer    | Memory of the GPU in megabytes (MB)                                                                                                                                                    | 15360          |
+| nvidia.com/gpu.memory          | Integer    | Memory of the GPU in mebibytes (MiB)                                                                                                                                                    | 15360          |
 | nvidia.com/gpu.product         | String     | Model of the GPU. May be modified by the device plugin if a sharing strategy is employed depending on the config.                                                                      | Tesla-T4       |
 | nvidia.com/gpu.replicas        | String     | Number of GPU replicas available. Will be equal to the number of physical GPUs unless some sharing strategy is employed in which case the GPU count will be multiplied by replicas.    | 4              |
 | nvidia.com/gpu.mode            | String     | Mode of the GPU. Can be either "compute" or "display". Details of the GPU modes can be found [here](https://docs.nvidia.com/grid/13.0/grid-gpumodeswitch-user-guide/index.html#compute-and-graphics-mode) | compute        |

internal/lm/resource.go

@@ -44,7 +44,7 @@ func NewGPUResourceLabeler(config *spec.Config, device resource.Device, count in
 		return nil, fmt.Errorf("failed to get device model: %v", err)
 	}
 
-	totalMemoryMB, err := device.GetTotalMemoryMB()
+	totalMemoryMiB, err := device.GetTotalMemoryMiB()
 	if err != nil {
 		return nil, fmt.Errorf("failed to get memory info for device: %v", err)
 	}
@@ -57,8 +57,8 @@ func NewGPUResourceLabeler(config *spec.Config, device resource.Device, count in
 	}
 
 	memoryLabeler := (Labeler)(&empty{})
-	if totalMemoryMB != 0 {
-		memoryLabeler = resourceLabeler.single("memory", totalMemoryMB)
+	if totalMemoryMiB != 0 {
+		memoryLabeler = resourceLabeler.single("memory", totalMemoryMiB)
 	}
 
 	labelers := Merge(

internal/resource/cuda-device.go

@@ -58,8 +58,8 @@ func (d *cudaDevice) GetDeviceHandleFromMigDeviceHandle() (Device, error) {
 	return nil, fmt.Errorf("GetDeviceHandleFromMigDeviceHandle is unsupported for CUDA devices")
 }
 
-// GetTotalMemoryMB returns the total memory for a device
-func (d *cudaDevice) GetTotalMemoryMB() (uint64, error) {
+// GetTotalMemoryMiB returns the total memory for a device in mebibytes (2^20 bytes)
+func (d *cudaDevice) GetTotalMemoryMiB() (uint64, error) {
 	total, r := cuda.Device(*d).TotalMem()
 	if r != cuda.SUCCESS {
 		return 0, fmt.Errorf("failed to get memory info for device: %v", r)

internal/resource/device_mock.go

@@ -81,8 +81,8 @@ func (d nvmlDevice) GetName() (string, error) {
 	return name, nil
 }
 
-// GetTotalMemoryMB returns the total memory on a device in MB
-func (d nvmlDevice) GetTotalMemoryMB() (uint64, error) {
+// GetTotalMemoryMiB returns the total memory on a device in mebibytes (2^20 bytes)
+func (d nvmlDevice) GetTotalMemoryMiB() (uint64, error) {
 	info, ret := d.Device.GetMemoryInfo()
 	if ret != nvml.SUCCESS {
 		return 0, ret

internal/resource/nvml-device.go

@@ -104,8 +104,8 @@ func (d nvmlMigDevice) GetName() (string, error) {
 	return resourceName, nil
 }
 
-// GetTotalMemoryMB returns the total memory on a device in MB
-func (d nvmlMigDevice) GetTotalMemoryMB() (uint64, error) {
+// GetTotalMemoryMiB returns the total memory on a device in mebibytes (2^20 bytes)
+func (d nvmlMigDevice) GetTotalMemoryMiB() (uint64, error) {
 	attr, err := d.GetAttributes()
 	if err != nil {
 		return 0, err

internal/resource/nvml-mig-device.go

@@ -51,8 +51,8 @@ func (d vfioDevice) GetName() (string, error) {
 	return d.nvidiaPCIDevice.DeviceName, nil
 }
 
-// GetTotalMemoryMB returns the total memory on a device in MB
-func (d vfioDevice) GetTotalMemoryMB() (uint64, error) {
+// GetTotalMemoryMiB returns the total memory on a device in mebibytes (2^20 bytes)
+func (d vfioDevice) GetTotalMemoryMiB() (uint64, error) {
 	_, val := d.nvidiaPCIDevice.Resources.GetTotalAddressableMemory(true)
 	return val, nil
 }

internal/resource/sysfs-device.go

@@ -47,12 +47,12 @@ func NewDeviceMock(migEnabled bool) *DeviceMock {
 			}
 			return 8, 0, nil
 		},
-		GetTotalMemoryMBFunc: func() (uint64, error) { return uint64(300), nil },
-		IsFabricAttachedFunc: func() (bool, error) { return false, nil },
-		IsMigEnabledFunc:     func() (bool, error) { return migEnabled, nil },
-		IsMigCapableFunc:     func() (bool, error) { return migEnabled, nil },
-		GetMigDevicesFunc:    func() ([]resource.Device, error) { return nil, nil },
-		GetPCIClassFunc:      func() (uint32, error) { return 0, nil },
+		GetTotalMemoryMiBFunc: func() (uint64, error) { return uint64(300), nil },
+		IsFabricAttachedFunc:  func() (bool, error) { return false, nil },
+		IsMigEnabledFunc:      func() (bool, error) { return migEnabled, nil },
+		IsMigCapableFunc:      func() (bool, error) { return migEnabled, nil },
+		GetMigDevicesFunc:     func() ([]resource.Device, error) { return nil, nil },
+		GetPCIClassFunc:       func() (uint32, error) { return 0, nil },
 	}}
 	return &d
 }

internal/resource/testing/resource-testing.go

@@ -37,7 +37,7 @@ type Device interface {
 	GetMigDevices() ([]Device, error)
 	GetAttributes() (map[string]interface{}, error)
 	GetName() (string, error)
-	GetTotalMemoryMB() (uint64, error)
+	GetTotalMemoryMiB() (uint64, error)
 	GetDeviceHandleFromMigDeviceHandle() (Device, error)
 	GetCudaComputeCapability() (int, int, error)
 	GetPCIClass() (uint32, error)