publish from a016ccad (gopub)

intel · Dec 25, 2023 · dcc3345 · dcc3345
1 parent e97df1b
commit dcc3345
Show file tree

Hide file tree

Showing 46 changed files with 846 additions and 552 deletions.
diff --git a/README.md b/README.md
@@ -3,7 +3,7 @@ Intel(R) XPU Manager is a free and open-source tool for monitoring and managing
 
 It is designed to simplify administration, maximize reliability and uptime, and improve utilization.
 
-XPU Manager can be used standalone through its command line interface (CLI) to manage GPUs locally, or through its RESTful APIs to manage GPUs remotely. Intel(R) XPU System Management Interface (XPU-SMI) is the daemon-less version of XPU Manager and it only provides the local interface. XPU-SMI feature scope is the subset of XPU Manager. Their features are listed in the table below. Please note that XPU-SMI has been included in the GPU driver repository. If you want to use XPU Manager, please uninstall XPU-SMI and install XPU Manager. 
+XPU Manager can be used standalone through its command line interface (CLI) to manage GPUs locally, or through its RESTful APIs to manage GPUs remotely. Intel(R) XPU System Management Interface (XPU-SMI) is the daemon-less version of XPU Manager and it only provides the local interface. XPU-SMI feature scope is the subset of XPU Manager. Their features are listed in the table below. Please note that XPU-SMI and XPU Manager can't be installed or executed on the same system due to some resource conflict. XPU-SMI has been included in the GPU driver repository. If you want to use XPU Manager, please uninstall XPU-SMI and install XPU Manager. 
 
 amcmcli is a portable CLI tool to manage GPU AMC firmware on Linux OS. It is independent of GPU driver. 
 
@@ -74,8 +74,8 @@ Update firmware successfully.
 ```
 
 
-## Feature set of XPU Manager, XPU-SMI and Windows CLI tool
-|                           | XPU Manager            | XPU-SMI              | Windows CLI tool             | amcmcli         |
+## Feature set of XPU Manager, XPU-SMI and XPU-SMI Windows CLI tool
+|                           | XPU Manager            | XPU-SMI              | XPU-SMI Windows CLI          | amcmcli         |
 | :------------------------ | :--------------------: | :------------------: | :--------------------------: | :-------------: |
 | Device Info and Topology  | Yes                    | Yes                  | Yes                          | No              |
 | GPU Telemetries           | Yes (aggregated data)  | Yes (real-time data) | Yes (real-time data)         | No              |
@@ -98,16 +98,16 @@ You may get the latest installers or binaries in [Releases](https://github.com/i
 ## Supported OSes
 * XPU Manager
 	* Ubuntu 20.04.3/22.04
-	* RHEL 8.5/8.6/8.8/9.2
+	* RHEL 8.8/9.2
 	* CentOS 8/9 Stream
 	* CentOS 7.4/7.9
-	* SLES 15 SP3/SP4
+	* SLES 15 SP4/SP5
 * XPU-SMI
 	* Ubuntu 20.04.3/22.04
-	* RHEL 8.5/8.6/8.8/9.2
+	* RHEL 8.8/9.2
 	* CentOS 8/9 Stream
 	* CentOS 7.4/7.9
-	* SLES 15 SP3/SP4
+	* SLES 15 SP4/SP5
 	* Debian 10.13
 	* Windows Server 2019/2022 (limited features including: GPU device info, GPU telemetry, GPU firmware update and GPU configuration)
 
@@ -121,6 +121,7 @@ You may get the latest installers or binaries in [Releases](https://github.com/i
 * Refer to [DockerHub](https://hub.docker.com/r/intel/xpumanager) for a Docker container image that can be used as a Prometheus exporter in a Kubernetes environment.
 * Refer to [Building XPU Manager Installer](BUILDING.md) to build XPU Manager installer packages. 
 * Refer to [XPU Manager/XPU-SMI API documents](https://intel.github.io/xpumanager/smi_index.html) to integrate the library or RESTFul interface. 
+* A simple introduction video on [Youtube](https://www.youtube.com/watch?v=1bKeqlriDX0).
 
 ## Architecture
 ![XPU Manager Architecture](doc/img/architecture.PNG)

diff --git a/VERSION b/VERSION
@@ -1 +1 @@
-1.2.25
+1.2.26
diff --git a/cli/src/comlet_diagnostic.cpp b/cli/src/comlet_diagnostic.cpp
@@ -187,8 +187,7 @@ Alternatively the strings \"yesterday\", \"today\" are also understood.\n\
 Relative times also may be specified, prefixed with \"-\" referring to times before the current time.\n\
 Scanning would start from the latest boot if it is not specified.");
 
-    auto singleTestIdList = addOption("--singletest", this->opts->singleTestIdList,
-              "Selectively run some particular tests. Separated by the comma.\n\
+std::string singleTestIdListDesc = "Selectively run some particular tests. Separated by the comma.\n\
       1. Computation\n\
       2. Memory Error\n\
       3. Memory Bandwidth\n\
@@ -197,7 +196,13 @@ Scanning would start from the latest boot if it is not specified.");
       6. Power\n\
       7. Computation functional test\n\
       8. Media Codec functional test\n\
-      9. Xe Link Throughput");
+      9. Xe Link Throughput";
+#ifdef DAEMONLESS  
+    singleTestIdListDesc += "\nNote that in a multi NUMA node server, it may need to use numactl to specify which node the PCIe bandwidth test runs on.\n\
+Usage: numactl [ --membind nodes ] [ --cpunodebind nodes ] xpu-smi diag -d [deviceId] --singletest 5\n\
+It also applies to diag level tests.";
+#endif
+    auto singleTestIdList = addOption("--singletest", this->opts->singleTestIdList, singleTestIdListDesc);  
     singleTestIdList->delimiter(',');
     singleTestIdList->check(CLI::Range(1, (int)testIdToType.size()));
 

diff --git a/cli/src/comlet_dump.cpp b/cli/src/comlet_dump.cpp
@@ -173,10 +173,10 @@ bool ComletDump::dumpIdlePowerOnly() {
                 auto bdf_pos = uevent.find(bdf_key); 
                 if (bdf_pos != std::string::npos) {
                     auto device_id = uevent.substr(pos + key.length(), 4);
-                    if (device_id.compare(0, 3, "0BD") == 0 || device_id.compare(0, 3, "0BE") == 0) {
+                    if (device_id.compare(0, 3, "0BD") == 0 || device_id.compare(0, 3, "0BE") == 0 || device_id.compare(0, 3, "0B6") == 0) {
                         auto bdf = uevent.substr(bdf_pos + bdf_key.length(), 12);
                         gpu_bdfs.insert(bdf);
-                        if (device_id.compare("0BD9") == 0 || device_id.compare("0BDA") == 0 || device_id.compare("0BDB") == 0) {
+                        if (device_id.compare("0BD9") == 0 || device_id.compare("0BDA") == 0 || device_id.compare("0BDB") == 0 || device_id.compare("0B6E") == 0) {
                             gpu_bdf_to_tile_num[bdf] = 1;
                         } else {
                             gpu_bdf_to_tile_num[bdf] = 2;

diff --git a/cli/src/grpc_stub/grpc_core_stub.cpp b/cli/src/grpc_stub/grpc_core_stub.cpp
@@ -7,6 +7,7 @@
 #include "core_stub.h"
 
 #include <grpc++/grpc++.h>
+#include <grpc/impl/codegen/grpc_types.h>
 
 #include <cassert>
 #include <chrono>
@@ -35,7 +36,9 @@ GrpcCoreStub::GrpcCoreStub(bool priv) {
     }
     std::string unixSockName{unixSockDir + (priv ? "xpum_p.sock" : "xpum_up.sock")};
     std::string serverAddr{"unix://" + unixSockName};
-    this->channel = grpc::CreateChannel(serverAddr, grpc::InsecureChannelCredentials());
+    grpc::ChannelArguments args;
+    args.SetInt(GRPC_ARG_ENABLE_HTTP_PROXY, 0);
+    this->channel = grpc::CreateCustomChannel(serverAddr, grpc::InsecureChannelCredentials(), args);
     this->stub = XpumCoreService::NewStub(this->channel);
 }
 

diff --git a/cli/src/local_functions.cpp b/cli/src/local_functions.cpp
@@ -523,7 +523,9 @@ std::unique_ptr<nlohmann::json> addKernelParam() {
          *  Refer: https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/system_administrators_guide/ch-working_with_the_grub_2_boot_loader
          */
         cmdStr = "grub2-mkconfig -o /boot/efi/EFI/rhel/grub.cfg";
-    }
+    } else if (osRelease == LINUX_OS_RELEASE_OPEN_EULER && isFileExists("/boot/efi/EFI/openEuler/grub.cfg")) {
+        cmdStr = "grub2-mkconfig -o /boot/efi/EFI/openEuler/grub.cfg";
+    } 
     if (execCommand(cmdStr, cmdRes) != 0) {
         (*json)["error"] = "Fail to update grub.";
         (*json)["errno"] = XPUM_CLI_ERROR_VGPU_ADD_KERNEL_PARAM_FAILED;

diff --git a/cli/src/utility.cpp b/cli/src/utility.cpp
@@ -139,6 +139,8 @@ linux_os_release_t getOsRelease() {
                 return LINUX_OS_RELEASE_RHEL;
             } else if (value.find("debian") != std::string::npos) {
                 return LINUX_OS_RELEASE_DEBIAN;
+            } else if (value.find("openEuler") != std::string::npos) {
+                return LINUX_OS_RELEASE_OPEN_EULER;
             } else {
                 return LINUX_OS_RELEASE_UNKNOWN;
             }

diff --git a/cli/src/utility.h b/cli/src/utility.h
@@ -17,6 +17,7 @@ typedef enum linux_os_release_t {
     LINUX_OS_RELEASE_SLES,
     LINUX_OS_RELEASE_RHEL,
     LINUX_OS_RELEASE_DEBIAN,
+    LINUX_OS_RELEASE_OPEN_EULER,
     LINUX_OS_RELEASE_UNKNOWN,
 } linux_os_release_t;
 

diff --git a/core/resources/config/diagnostics.conf b/core/resources/config/diagnostics.conf
@@ -57,20 +57,41 @@ SINGLE_PRECISION_MIN_GFLOPS = 8000       # GFLOPS
 POWER_MIN_STRESS_WATT = 80               # W
 MEMORY_BANDWIDTH_MIN_GBPS = 320          # GBPS
 
-# PVC 2T 2048EU 128GB
+# PVC 2T 1024EU 128GB
+NAME = Intel(R) Graphics [0x0bd4]
+PCIE_BANDWIDTH_MIN_GBPS = 22             # GBPS
+SINGLE_PRECISION_MIN_GFLOPS = 27000      # GFLOPS
+POWER_MIN_STRESS_WATT = 240              # W
+MEMORY_BANDWIDTH_MIN_GBPS = 1000         # GBPS
+
+# PVC 2T 1024EU 128GB
 NAME = Intel(R) Graphics [0x0bd5]
 PCIE_BANDWIDTH_MIN_GBPS = 22             # GBPS
 SINGLE_PRECISION_MIN_GFLOPS = 27000      # GFLOPS
 POWER_MIN_STRESS_WATT = 240              # W
 MEMORY_BANDWIDTH_MIN_GBPS = 1000         # GBPS
 
-# PVC 2T 2048EU 128GB
+# PVC 2T 1024EU 128GB
 NAME = Intel(R) Graphics [0x0bd6]
 PCIE_BANDWIDTH_MIN_GBPS = 32             # GBPS
 SINGLE_PRECISION_MIN_GFLOPS = 35000      # GFLOPS
 POWER_MIN_STRESS_WATT = 320              # W
 MEMORY_BANDWIDTH_MIN_GBPS = 1600         # GBPS
 
+# PVC 2T 896EU 96GB
+NAME = Intel(R) Graphics [0x0bd7]
+PCIE_BANDWIDTH_MIN_GBPS = 30             # GBPS
+SINGLE_PRECISION_MIN_GFLOPS = 35000      # GFLOPS
+POWER_MIN_STRESS_WATT = 360              # W
+MEMORY_BANDWIDTH_MIN_GBPS = 1100         # GBPS
+
+# PVC 2T 768EU 96GB
+NAME = Intel(R) Graphics [0x0bd8]
+PCIE_BANDWIDTH_MIN_GBPS = 30             # GBPS
+SINGLE_PRECISION_MIN_GFLOPS = 30000      # GFLOPS
+POWER_MIN_STRESS_WATT = 360              # W
+MEMORY_BANDWIDTH_MIN_GBPS = 1100         # GBPS
+
 # PVC 1T 448EU 48GB
 NAME = Intel(R) Graphics [0x0bdb]
 PCIE_BANDWIDTH_MIN_GBPS = 32             # GBPS
@@ -85,7 +106,14 @@ SINGLE_PRECISION_MIN_GFLOPS = 16000      # GFLOPS
 POWER_MIN_STRESS_WATT = 240              # W
 MEMORY_BANDWIDTH_MIN_GBPS = 560          # GBPS
 
-# PVC 2T 2048EU 128GB
+# PVC 1T 448EU 48GB
+NAME = Intel(R) Graphics [0x0b6e]
+PCIE_BANDWIDTH_MIN_GBPS = 28             # GBPS
+SINGLE_PRECISION_MIN_GFLOPS = 15000      # GFLOPS
+POWER_MIN_STRESS_WATT = 220              # W
+MEMORY_BANDWIDTH_MIN_GBPS = 520          # GBPS
+
+# PVC 2T 1024EU 128GB
 NAME = Intel(R) Graphics [0x0b69]
 PCIE_BANDWIDTH_MIN_GBPS = 36             # GBPS
 SINGLE_PRECISION_MIN_GFLOPS = 42000      # GFLOPS

diff --git a/core/resources/config/vgpu.conf b/core/resources/config/vgpu.conf
@@ -161,7 +161,7 @@ SCHED_IF_IDLE=0 #Same as default configuration
 DRIVERS_AUTOPROBE=0 #Same as default configuration
 
 
-NAME=0bd5N1,0bd6N1
+NAME=0bd5N1,0bd6N1,0bd4N1
 VF_LMEM=128849018880
 VF_LMEM_ECC=128849018880
 VF_CONTEXTS=2048
@@ -174,7 +174,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bd5N2,0bd6N2
+NAME=0bd5N2,0bd6N2,0bd4N2
 VF_LMEM=64424509440
 VF_LMEM_ECC=64424509440
 VF_CONTEXTS=1024
@@ -187,7 +187,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bd5N4,0bd6N4
+NAME=0bd5N4,0bd6N4,0bd4N4
 VF_LMEM=32212254720
 VF_LMEM_ECC=32212254720
 VF_CONTEXTS=1024
@@ -200,7 +200,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bd5N6,0bd6N6
+NAME=0bd5N6,0bd6N6,0bd4N6
 VF_LMEM=21474836480
 VF_LMEM_ECC=21474836480
 VF_CONTEXTS=1024
@@ -213,7 +213,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bd5N8,0bd6N8
+NAME=0bd5N8,0bd6N8,0bd4N8
 VF_LMEM=16106127360
 VF_LMEM_ECC=16106127360
 VF_CONTEXTS=1024
@@ -226,7 +226,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bd5N16,0bd6N16
+NAME=0bd5N16,0bd6N16,0bd4N16
 VF_LMEM=8053063680
 VF_LMEM_ECC=8053063680
 VF_CONTEXTS=1024
@@ -239,7 +239,7 @@ PF_PREEMPT_TIMEOUT=16000
 SCHED_IF_IDLE=1
 DRIVERS_AUTOPROBE=0
 
-NAME=0bd5N32,0bd6N32
+NAME=0bd5N32,0bd6N32,0bd4N32
 VF_LMEM=4026531840
 VF_LMEM_ECC=4026531840
 VF_CONTEXTS=1024
@@ -252,7 +252,7 @@ PF_PREEMPT_TIMEOUT=8000
 SCHED_IF_IDLE=1
 DRIVERS_AUTOPROBE=0
 
-NAME=0bd5N62,0bd6N62
+NAME=0bd5N62,0bd6N62,0bd4N62
 VF_LMEM=2013265920
 VF_LMEM_ECC=2013265920
 VF_CONTEXTS=1024
@@ -266,7 +266,7 @@ SCHED_IF_IDLE=1
 DRIVERS_AUTOPROBE=0
 
 # If a specified number of VGPU is not in this file, the configuration below would be used.
-NAME=0bd5DEF,0bd6DEF
+NAME=0bd5DEF,0bd6DEF,0bd4DEF
 VF_LMEM=128849018880 #Equally divided to each vgpu
 VF_LMEM_ECC=128849018880 #Equally divided to each vgpu
 VF_CONTEXTS=1024 #Same as default configuration
@@ -280,7 +280,7 @@ SCHED_IF_IDLE=1 #Same as default configuration
 DRIVERS_AUTOPROBE=0 #Same as default configuration
 
 
-NAME=0bdaN1,0bdbN1
+NAME=0bdaN1,0bdbN1,0b6eN1
 VF_LMEM=47244640256
 VF_LMEM_ECC=47244640256
 VF_CONTEXTS=1024
@@ -293,7 +293,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bdaN2,0bdbN2
+NAME=0bdaN2,0bdbN2,0b6eN2
 VF_LMEM=23622320128
 VF_LMEM_ECC=23622320128
 VF_CONTEXTS=1024
@@ -306,7 +306,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bdaN3,0bdbN3
+NAME=0bdaN3,0bdbN3,0b6eN3
 VF_LMEM=15748213418
 VF_LMEM_ECC=15748213418
 VF_CONTEXTS=1024
@@ -319,7 +319,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bdaN4,0bdbN4
+NAME=0bdaN4,0bdbN4,0b6eN4
 VF_LMEM=11811160064
 VF_LMEM_ECC=11811160064
 VF_CONTEXTS=1024
@@ -332,7 +332,7 @@ PF_PREEMPT_TIMEOUT=128000
 SCHED_IF_IDLE=0
 DRIVERS_AUTOPROBE=0
 
-NAME=0bdaN8,0bdbN8
+NAME=0bdaN8,0bdbN8,0b6eN8
 VF_LMEM=5905580032
 VF_LMEM_ECC=5905580032
 VF_CONTEXTS=1024
@@ -345,7 +345,7 @@ PF_PREEMPT_TIMEOUT=16000
 SCHED_IF_IDLE=1
 DRIVERS_AUTOPROBE=0
 
-NAME=0bdaN16,0bdbN16
+NAME=0bdaN16,0bdbN16,0b6eN16
 VF_LMEM=2952790016
 VF_LMEM_ECC=2952790016
 VF_CONTEXTS=1024
@@ -358,7 +358,7 @@ PF_PREEMPT_TIMEOUT=8000
 SCHED_IF_IDLE=1
 DRIVERS_AUTOPROBE=0
 
-NAME=0bdaN32,0bdbN32
+NAME=0bdaN32,0bdbN32,0b6eN32
 VF_LMEM=1476395008
 VF_LMEM_ECC=1476395008
 VF_CONTEXTS=1024
@@ -371,7 +371,7 @@ PF_PREEMPT_TIMEOUT=4000
 SCHED_IF_IDLE=1
 DRIVERS_AUTOPROBE=0
 
-NAME=0bdaN63,0bdbN63
+NAME=0bdaN63,0bdbN63,0b6eN63
 VF_LMEM=738197504
 VF_LMEM_ECC=738197504
 VF_CONTEXTS=1024
@@ -385,7 +385,7 @@ SCHED_IF_IDLE=1
 DRIVERS_AUTOPROBE=0
 
 # If a specified number of VGPU is not in this file, the configuration below would be used.
-NAME=0bdaDEF,0bdbDEF
+NAME=0bdaDEF,0bdbDEF,0b6eDEF
 VF_LMEM=47244640256 #Equally divided to each vgpu
 VF_LMEM_ECC=47244640256 #Equally divided to each vgpu
 VF_CONTEXTS=1024 #Same as default configuration

diff --git a/core/src/api/device_model.cpp b/core/src/api/device_model.cpp
@@ -17,6 +17,7 @@ int getDeviceModelByPciDeviceId(int deviceId) {
             return XPUM_DEVICE_MODEL_ATS_M_3;
         case 0x0b69:
         case 0x0bd0:
+        case 0x0bd4:
         case 0x0bd5:
         case 0x0bd6:
         case 0x0bd7:
@@ -25,6 +26,7 @@ int getDeviceModelByPciDeviceId(int deviceId) {
         case 0x0bda:
         case 0x0bdb:
         case 0x0be5:
+        case 0x0b6e:
             return XPUM_DEVICE_MODEL_PVC;
         case 0x4907:
             return XPUM_DEVICE_MODEL_SG1;

diff --git a/core/src/data_logic/engine_group_utilization_data_handler.cpp b/core/src/data_logic/engine_group_utilization_data_handler.cpp
@@ -50,6 +50,11 @@ void EngineGroupUtilizationDataHandler::calculateData(std::shared_ptr<SharedData
                 auto pre_extended = pre_data->second->getExtendedDatas()->find(extended_data->first);
                 if (pre_extended != pre_data->second->getExtendedDatas()->end()) {
                     if (extended_data->second.type == ZES_ENGINE_GROUP_COMPUTE_ALL || extended_data->second.type == ZES_ENGINE_GROUP_RENDER_ALL || extended_data->second.type == ZES_ENGINE_GROUP_MEDIA_ALL || extended_data->second.type == ZES_ENGINE_GROUP_COPY_ALL || extended_data->second.type == ZES_ENGINE_GROUP_3D_ALL) {
+                        if (extended_data->second.timestamp == 
+                                pre_extended->second.timestamp) {
+                            ++extended_data;
+                            continue;
+                        }
                         uint64_t val = Configuration::DEFAULT_MEASUREMENT_DATA_SCALE * 100 * (extended_data->second.active_time - pre_extended->second.active_time) / (extended_data->second.timestamp - pre_extended->second.timestamp);
                         if (val > Configuration::DEFAULT_MEASUREMENT_DATA_SCALE * 100) {
                             val = Configuration::DEFAULT_MEASUREMENT_DATA_SCALE * 100;