Jenkinsfile 42 KB
Newer Older
JD's avatar
JD committed
1
def rocmnode(name) {
2
    return '(rocmtest || miopen) && (' + name + ')'
JD's avatar
JD committed
3
4
5
6
7
8
9
}

def show_node_info() {
    sh """
        echo "NODE_NAME = \$NODE_NAME"
        lsb_release -sd
        uname -r
10
        cat /sys/module/amdgpu/version
JD's avatar
JD committed
11
12
13
14
        ls /opt/ -la
    """
}

15
16
17
18
19
20
21
22
23
24
25
26
27
28
def nthreads() {
    def nproc = sh(returnStdout: true, script: 'nproc')
    echo "Number of cores: ${nproc}"
    def n = nproc.toInteger()
    if (n > 32){
        n /= 2
    }
    if (n > 64){
        n = 64
    }
    echo "Number of threads used for building: ${n}"
    return n
}

29
def runShell(String command){
30
    def responseCode = sh returnStatus: true, script: "${command} > tmp.txt"
31
32
33
34
    def output = readFile(file: "tmp.txt")
    return (output != "")
}

35
def getDockerImageName(){
36
    def img
37
38
39
40
    if (params.USE_CUSTOM_DOCKER != ""){
        img = "${params.USE_CUSTOM_DOCKER}"
    }
    else{
41
    if (params.ROCMVERSION != "6.1"){
42
43
       if (params.COMPILER_VERSION == "") {
           img = "${env.CK_DOCKERHUB}:ck_ub20.04_rocm${params.ROCMVERSION}"
44
45
       }
       else{
46
47
48
49
50
51
52
          if (params.COMPILER_COMMIT == ""){
             img = "${env.CK_DOCKERHUB}:ck_ub20.04_rocm${params.ROCMVERSION}_${params.COMPILER_VERSION}"
          }
          else{
             def commit = "${params.COMPILER_COMMIT}"[0..6]
             img = "${env.CK_DOCKERHUB}:ck_ub20.04_rocm${params.ROCMVERSION}_${params.COMPILER_VERSION}_${commit}"
          }
53
       }
54
55
    }
    else{
56
57
       if (params.COMPILER_VERSION == "") {
           img = "${env.CK_DOCKERHUB_PRIVATE}:ck_ub20.04_rocm${params.ROCMVERSION}"
58
59
       }
       else{
60
61
62
63
64
65
66
          if (params.COMPILER_COMMIT == ""){
             img = "${env.CK_DOCKERHUB_PRIVATE}:ck_ub20.04_rocm${params.ROCMVERSION}_${params.COMPILER_VERSION}"
          }
          else{
             def commit = "${params.COMPILER_COMMIT}"[0..6]
             img = "${env.CK_DOCKERHUB_PRIVATE}:ck_ub20.04_rocm${params.ROCMVERSION}_${params.COMPILER_VERSION}_${commit}"
          }
67
       }
68
    }
69
    }
70
71
72
    return img
}

73
def check_host() {
74
75
76
77
    if ("${env.CK_SCCACHE}" != "null"){
        def SCCACHE_SERVER="${env.CK_SCCACHE.split(':')[0]}"
        echo "sccache server: ${SCCACHE_SERVER}"
        sh '''ping -c 1 -p 6379 "${SCCACHE_SERVER}" | echo $? > tmp.txt'''
78
79
80
81
82
83
84
85
86
        def output = readFile(file: "tmp.txt")
        echo "tmp.txt contents: \$output"
        return (output != "0")
    }
    else{
        return 1
    }
}

87
88
89
90
91
92
def build_compiler(){
    def compiler
    if (params.BUILD_COMPILER == "hipcc"){
        compiler = '/opt/rocm/bin/hipcc'
    }
    else{
93
        if (params.COMPILER_VERSION == "amd-staging" || params.COMPILER_VERSION == "amd-mainline-open" || params.COMPILER_COMMIT != ""){
94
            compiler = "/llvm-project/build/bin/clang++"
95
96
        }
        else{
97
            compiler = "/opt/rocm/llvm/bin/clang++"
98
99
100
101
102
        }        
    }
    return compiler
}

103
104
def getDockerImage(Map conf=[:]){
    env.DOCKER_BUILDKIT=1
105
    def prefixpath = conf.get("prefixpath", "/opt/rocm")
106
    def no_cache = conf.get("no_cache", false)
107
    def dockerArgs = "--build-arg BUILDKIT_INLINE_CACHE=1 --build-arg PREFIX=${prefixpath} --build-arg CK_SCCACHE='${env.CK_SCCACHE}' --build-arg compiler_version='${params.COMPILER_VERSION}' --build-arg compiler_commit='${params.COMPILER_COMMIT}' --build-arg ROCMVERSION='${params.ROCMVERSION}' "
108
109
110
111
112
113
114
115
116
117
118
119
    if(no_cache)
    {
        dockerArgs = dockerArgs + " --no-cache "
    }
    echo "Docker Args: ${dockerArgs}"
    def image = getDockerImageName()
    //Check if image exists 
    def retimage
    try 
    {
        echo "Pulling down image: ${image}"
        retimage = docker.image("${image}")
120
121
122
        withDockerRegistry([ credentialsId: "docker_test_cred", url: "" ]) {
            retimage.pull()
        }
123
124
125
126
127
128
129
130
131
132
133
134
135
136
    }
    catch(Exception ex)
    {
        error "Unable to locate image: ${image}"
    }
    return [retimage, image]
}

def buildDocker(install_prefix){
    show_node_info()
    env.DOCKER_BUILDKIT=1
    checkout scm
    def image_name = getDockerImageName()
    echo "Building Docker for ${image_name}"
137
    def dockerArgs = "--build-arg BUILDKIT_INLINE_CACHE=1 --build-arg PREFIX=${install_prefix} --build-arg CK_SCCACHE='${env.CK_SCCACHE}' --build-arg compiler_version='${params.COMPILER_VERSION}' --build-arg compiler_commit='${params.COMPILER_COMMIT}' --build-arg ROCMVERSION='${params.ROCMVERSION}' "
138
139
140

    echo "Build Args: ${dockerArgs}"
    try{
141
142
143
144
        if(params.BUILD_DOCKER){
            //force building the new docker if that parameter is true
            echo "Building image: ${image_name}"
            retimage = docker.build("${image_name}", dockerArgs + ' .')
145
146
147
            withDockerRegistry([ credentialsId: "docker_test_cred", url: "" ]) {
                retimage.push()
            }
148
            sh 'docker images -q -f dangling=true | xargs --no-run-if-empty docker rmi'
149
150
151
152
        }
        else{
            echo "Checking for image: ${image_name}"
            sh "docker manifest inspect --insecure ${image_name}"
153
            echo "Image: ${image_name} found! Skipping building image"
154
        }
155
156
157
158
    }
    catch(Exception ex){
        echo "Unable to locate image: ${image_name}. Building image now"
        retimage = docker.build("${image_name}", dockerArgs + ' .')
159
160
161
        withDockerRegistry([ credentialsId: "docker_test_cred", url: "" ]) {
            retimage.push()
        }
162
163
164
    }
}

JD's avatar
JD committed
165
166
def cmake_build(Map conf=[:]){

167
    def compiler = build_compiler()
JD's avatar
JD committed
168
169
    def config_targets = conf.get("config_targets","check")
    def debug_flags = "-g -fno-omit-frame-pointer -fsanitize=undefined -fno-sanitize-recover=undefined " + conf.get("extradebugflags", "")
170
    def build_envs = "CTEST_PARALLEL_LEVEL=4 " + conf.get("build_env","")
JD's avatar
JD committed
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
    def prefixpath = conf.get("prefixpath","/opt/rocm")
    def setup_args = conf.get("setup_args","")

    if (prefixpath != "/usr/local"){
        setup_args = setup_args + " -DCMAKE_PREFIX_PATH=${prefixpath} "
    }

    def build_type_debug = (conf.get("build_type",'release') == 'debug')

    //cmake_env can overwrite default CXX variables.
    def cmake_envs = "CXX=${compiler} CXXFLAGS='-Werror' " + conf.get("cmake_ex_env","")

    def package_build = (conf.get("package_build","") == "true")

    if (package_build == true) {
        config_targets = "package"
    }

    if(conf.get("build_install","") == "true")
    {
        config_targets = 'install ' + config_targets
192
        setup_args = ' -DBUILD_DEV=On -DCMAKE_INSTALL_PREFIX=../install' + setup_args
JD's avatar
JD committed
193
194
195
    } else{
        setup_args = ' -DBUILD_DEV=On' + setup_args
    }
196
197
198
    if (params.DL_KERNELS){
        setup_args = setup_args + " -DDL_KERNELS=ON "
    }
JD's avatar
JD committed
199
200
201
202
203
204
205
206

    if(build_type_debug){
        setup_args = " -DCMAKE_BUILD_TYPE=debug -DCMAKE_CXX_FLAGS_DEBUG='${debug_flags}'" + setup_args
    }else{
        setup_args = " -DCMAKE_BUILD_TYPE=release" + setup_args
    }

    def pre_setup_cmd = """
207
            #!/bin/bash
JD's avatar
JD committed
208
209
210
211
212
213
214
215
            echo \$HSA_ENABLE_SDMA
            ulimit -c unlimited
            rm -rf build
            mkdir build
            rm -rf install
            mkdir install
            cd build
        """
216
217
218
219
220
221
222
223
224
225
226
227
228
    def invocation_tag=""
    if (setup_args.contains("gfx11")){
        invocation_tag="gfx11"
    }
    if (setup_args.contains("gfx10")){
        invocation_tag="gfx10"
    }
    if (setup_args.contains("gfx90")){
        invocation_tag="gfx90"
    }
    if (setup_args.contains("gfx94")){
        invocation_tag="gfx94"
    }
229
230
    echo "invocation tag: ${invocation_tag}"
    def redis_pre_setup_cmd = pre_setup_cmd
231
    if(check_host() && params.USE_SCCACHE && "${env.CK_SCCACHE}" != "null" && "${invocation_tag}" != "") {
232
        redis_pre_setup_cmd = pre_setup_cmd + """
233
234
235
236
237
238
239
240
241
242
243
244
245
            #!/bin/bash
            export ROCM_PATH=/opt/rocm
            export SCCACHE_ENABLED=true
            export SCCACHE_LOG_LEVEL=debug
            export SCCACHE_IDLE_TIMEOUT=14400
            export COMPILERS_HASH_DIR=/tmp/.sccache
            export SCCACHE_BIN=/usr/local/.cargo/bin/sccache
            export SCCACHE_EXTRAFILES=/tmp/.sccache/rocm_compilers_hash_file
            export SCCACHE_REDIS="redis://${env.CK_SCCACHE}"
            echo "connect = ${env.CK_SCCACHE}" >> ../script/redis-cli.conf
            export SCCACHE_C_CUSTOM_CACHE_BUSTER="${invocation_tag}"
            echo \$SCCACHE_C_CUSTOM_CACHE_BUSTER
            stunnel ../script/redis-cli.conf
246
            ../script/sccache_wrapper.sh --enforce_redis
247
        """
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
        try {
            def cmd1 = conf.get("cmd1", """
                    ${redis_pre_setup_cmd}
                """)
            sh cmd1
            setup_args = " -DCMAKE_CXX_COMPILER_LAUNCHER=sccache -DCMAKE_C_COMPILER_LAUNCHER=sccache " + setup_args
        }
        catch(Exception err){
            echo "could not connect to redis server: ${err.getMessage()}. will not use sccache."
            def cmd2 = conf.get("cmd2", """
                    ${pre_setup_cmd}
                """)
            sh cmd2
        }
    }
    else{
        def cmd3 = conf.get("cmd3",  """
                ${pre_setup_cmd}
            """)
        sh cmd3
268
    }
Chao Liu's avatar
Chao Liu committed
269
    // reduce parallelism when compiling, clang uses too much memory
270
    def nt = nthreads()
Paul Fultz II's avatar
Paul Fultz II committed
271
    def cmd
JD's avatar
JD committed
272
    def execute_cmd = conf.get("execute_cmd", "")
Paul Fultz II's avatar
Paul Fultz II committed
273
274
275
276
    if(!setup_args.contains("NO_CK_BUILD")){
        def setup_cmd = conf.get("setup_cmd", "${cmake_envs} cmake ${setup_args}   .. ")
        def build_cmd = conf.get("build_cmd", "${build_envs} dumb-init make  -j${nt} ${config_targets}")
        cmd = conf.get("cmd", """
JD's avatar
JD committed
277
278
279
280
            ${setup_cmd}
            ${build_cmd}
            ${execute_cmd}
        """)
Paul Fultz II's avatar
Paul Fultz II committed
281
282
283
284
285
286
    }
    else{
        cmd = conf.get("cmd", """
            ${execute_cmd}
        """)
    }
JD's avatar
JD committed
287
288

    echo cmd
289
290
291
292

    dir("build"){
        sh cmd
    }
JD's avatar
JD committed
293
294

    // Only archive from master or develop
295
    if (package_build == true && (env.BRANCH_NAME == "develop" || env.BRANCH_NAME == "amd-master")) {
JD's avatar
JD committed
296
297
298
299
300
301
302
303
304
305
        archiveArtifacts artifacts: "build/*.deb", allowEmptyArchive: true, fingerprint: true
    }
}

def buildHipClangJob(Map conf=[:]){
        show_node_info()

        env.HSA_ENABLE_SDMA=0
        checkout scm

306
        def image = getDockerImageName() 
JD's avatar
JD committed
307
308
309
        def prefixpath = conf.get("prefixpath", "/opt/rocm")

        // Jenkins is complaining about the render group 
310
        def dockerOpts="--device=/dev/kfd --device=/dev/dri --group-add video --group-add render --cap-add=SYS_PTRACE --security-opt seccomp=unconfined"
JD's avatar
JD committed
311
        if (conf.get("enforce_xnack_on", false)) {
312
            dockerOpts = dockerOpts + " --env HSA_XNACK=1 "
JD's avatar
JD committed
313
        }
314
        def dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg CK_SCCACHE='${env.CK_SCCACHE}' --build-arg compiler_version='${params.COMPILER_VERSION}' --build-arg compiler_commit='${params.COMPILER_COMMIT}' --build-arg ROCMVERSION='${params.ROCMVERSION}' "
315
        if (params.COMPILER_VERSION == "amd-staging" || params.COMPILER_VERSION == "amd-mainline-open" || params.COMPILER_COMMIT != ""){
316
317
            dockerOpts = dockerOpts + " --env HIP_CLANG_PATH='/llvm-project/build/bin' "
        }
JD's avatar
JD committed
318
319
320
321

        def variant = env.STAGE_NAME

        def retimage
322
        (retimage, image) = getDockerImage(conf)
323

324
        gitStatusWrapper(credentialsId: "${status_wrapper_creds}", gitHubContext: "Jenkins - ${variant}", account: 'ROCm', repo: 'composable_kernel') {
JD's avatar
JD committed
325
            withDockerContainer(image: image, args: dockerOpts + ' -v=/var/jenkins/:/var/jenkins') {
326
                timeout(time: 48, unit: 'HOURS')
JD's avatar
JD committed
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
                {
                    cmake_build(conf)
                }
            }
        }
        return retimage
}

def reboot(){
    build job: 'reboot-slaves', propagate: false , parameters: [string(name: 'server', value: "${env.NODE_NAME}"),]
}

def buildHipClangJobAndReboot(Map conf=[:]){
    try{
        buildHipClangJob(conf)
    }
    catch(e){
        echo "throwing error exception for the stage"
        echo 'Exception occurred: ' + e.toString()
        throw e
    }
    finally{
        if (!conf.get("no_reboot", false)) {
            reboot()
        }
    }
}

355
356
357
358
359
360
def runCKProfiler(Map conf=[:]){
        show_node_info()

        env.HSA_ENABLE_SDMA=0
        checkout scm

361
        def image = getDockerImageName()
362
363
364
        def prefixpath = conf.get("prefixpath", "/opt/rocm")

        // Jenkins is complaining about the render group 
365
        def dockerOpts="--device=/dev/kfd --device=/dev/dri --group-add video --group-add render --cap-add=SYS_PTRACE --security-opt seccomp=unconfined"
366
        if (conf.get("enforce_xnack_on", false)) {
367
            dockerOpts = dockerOpts + " --env HSA_XNACK=1 "
368
        }
369
        def dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg compiler_version='${params.COMPILER_VERSION}' --build-arg compiler_commit='${params.COMPILER_COMMIT}' --build-arg ROCMVERSION='${params.ROCMVERSION}' "
370
371
372

        def variant = env.STAGE_NAME
        def retimage
373

374
        gitStatusWrapper(credentialsId: "${status_wrapper_creds}", gitHubContext: "Jenkins - ${variant}", account: 'ROCm', repo: 'composable_kernel') {
375
            try {
376
                (retimage, image) = getDockerImage(conf)
377
378
                withDockerContainer(image: image, args: dockerOpts) {
                    timeout(time: 5, unit: 'MINUTES'){
379
380
                        sh 'rocminfo | tee rocminfo.log'
                        if ( !runShell('grep -n "gfx" rocminfo.log') ){
381
                            throw new Exception ("GPU not found")
382
383
384
385
                        }
                        else{
                            echo "GPU is OK"
                        }
386
387
388
                    }
                }
            }
389
390
391
392
            catch (org.jenkinsci.plugins.workflow.steps.FlowInterruptedException e){
                echo "The job was cancelled or aborted"
                throw e
            }
393
394

            withDockerContainer(image: image, args: dockerOpts + ' -v=/var/jenkins/:/var/jenkins') {
395
                timeout(time: 24, unit: 'HOURS')
396
                {
397
398
399
400
401
402
403
404
405
                    sh """
                        rm -rf build
                        mkdir build
                    """
                    dir("build"){
                        unstash 'ckProfiler.tar.gz'
                        sh 'tar -xvf ckProfiler.tar.gz'
                    }

406
					dir("script"){
407
                        if (params.RUN_FULL_QA){
408
                            sh "./run_full_performance_tests.sh 0 QA_${params.COMPILER_VERSION} ${env.BRANCH_NAME} ${NODE_NAME}"
409
410
411
412
413
414
415
416
417
418
419
                            archiveArtifacts "perf_gemm.log"
                            archiveArtifacts "perf_resnet50_N256.log"
                            archiveArtifacts "perf_resnet50_N4.log"
                            archiveArtifacts "perf_batched_gemm.log"
                            archiveArtifacts "perf_grouped_gemm.log"
                            archiveArtifacts "perf_conv_fwd.log"
                            archiveArtifacts "perf_conv_bwd_data.log"
                            archiveArtifacts "perf_gemm_bilinear.log"
                            archiveArtifacts "perf_reduction.log"
                            archiveArtifacts "perf_splitK_gemm.log"
                            archiveArtifacts "perf_onnx_gemm.log"
420
                            archiveArtifacts "perf_mixed_gemm.log"
421
                           // stash perf files to master
422
423
424
425
426
427
428
429
430
431
432
                            stash name: "perf_gemm.log"
                            stash name: "perf_resnet50_N256.log"
                            stash name: "perf_resnet50_N4.log"
                            stash name: "perf_batched_gemm.log"
                            stash name: "perf_grouped_gemm.log"
                            stash name: "perf_conv_fwd.log"
                            stash name: "perf_conv_bwd_data.log"
                            stash name: "perf_gemm_bilinear.log"
                            stash name: "perf_reduction.log"
                            stash name: "perf_splitK_gemm.log"
                            stash name: "perf_onnx_gemm.log"
433
                            stash name: "perf_mixed_gemm.log"
434
                            //we will process results on the master node
435
436
                        }
                        else{
437
438
439
440
                            sh "./run_performance_tests.sh 0 CI_${params.COMPILER_VERSION} ${env.BRANCH_NAME} ${NODE_NAME}"
                            archiveArtifacts "perf_gemm.log"
                            archiveArtifacts "perf_resnet50_N256.log"
                            archiveArtifacts "perf_resnet50_N4.log"
441
                            // stash perf files to master
442
443
444
                            stash name: "perf_gemm.log"
                            stash name: "perf_resnet50_N256.log"
                            stash name: "perf_resnet50_N4.log"
445
                            //we will process the results on the master node
446
                        }
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
					}
                }
            }
        }
        return retimage
}

def runPerfTest(Map conf=[:]){
    try{
        runCKProfiler(conf)
    }
    catch(e){
        echo "throwing error exception in performance tests"
        echo 'Exception occurred: ' + e.toString()
        throw e
    }
    finally{
        if (!conf.get("no_reboot", false)) {
            reboot()
        }
    }
}

470
471
472
473
def Build_CK(Map conf=[:]){
        show_node_info()

        env.HSA_ENABLE_SDMA=0
474
        env.DOCKER_BUILDKIT=1
475
476
477
478
479
480
481
482
483
484
        checkout scm

        def image = getDockerImageName() 
        def prefixpath = conf.get("prefixpath", "/opt/rocm")

        // Jenkins is complaining about the render group 
        def dockerOpts="--device=/dev/kfd --device=/dev/dri --group-add video --group-add render --cap-add=SYS_PTRACE --security-opt seccomp=unconfined"
        if (conf.get("enforce_xnack_on", false)) {
            dockerOpts = dockerOpts + " --env HSA_XNACK=1 "
        }
485
        def dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg compiler_version='${params.COMPILER_VERSION}' --build-arg compiler_commit='${params.COMPILER_COMMIT}' --build-arg ROCMVERSION='${params.ROCMVERSION}' "
486
        if (params.COMPILER_VERSION == "amd-staging" || params.COMPILER_VERSION == "amd-mainline-open" || params.COMPILER_COMMIT != ""){
487
488
            dockerOpts = dockerOpts + " --env HIP_CLANG_PATH='/llvm-project/build/bin' "
        }
489
490
491
492
        def video_id = sh(returnStdout: true, script: 'getent group video | cut -d: -f3')
        def render_id = sh(returnStdout: true, script: 'getent group render | cut -d: -f3')
        dockerOpts = dockerOpts + " --group-add=${video_id} --group-add=${render_id} "
        echo "Docker flags: ${dockerOpts}"
493
494
495

        def variant = env.STAGE_NAME
        def retimage
496
        gitStatusWrapper(credentialsId: "${env.status_wrapper_creds}", gitHubContext: "Jenkins - ${variant}", account: 'ROCm', repo: 'composable_kernel') {
497
498
499
500
            try {
                (retimage, image) = getDockerImage(conf)
                withDockerContainer(image: image, args: dockerOpts) {
                    timeout(time: 5, unit: 'MINUTES'){
501
502
                        sh 'rocminfo | tee rocminfo.log'
                        if ( !runShell('grep -n "gfx" rocminfo.log') ){
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
                            throw new Exception ("GPU not found")
                        }
                        else{
                            echo "GPU is OK"
                        }
                    }
                }
            }
            catch (org.jenkinsci.plugins.workflow.steps.FlowInterruptedException e){
                echo "The job was cancelled or aborted"
                throw e
            }
            withDockerContainer(image: image, args: dockerOpts + ' -v=/var/jenkins/:/var/jenkins') {
                timeout(time: 24, unit: 'HOURS')
                {
518
519
520
521
522
523
524
525
526
527
528
529
                    //check whether running on Navi or MI300 node
                    def navi_node = 0
                    def mi300_node = 0
                    sh 'rocminfo | tee rocminfo.log'
                    if ( runShell('grep -n "gfx1030" rocminfo.log') || runShell('grep -n "gfx1101" rocminfo.log') ){
                        navi_node = 1
                        echo "This is a Navi node"
                    }
                    if ( runShell('grep -n "gfx942" rocminfo.log') ){
                        mi300_node = 1
                        echo "This is MI300 node"
                    }
530
531
                    cmake_build(conf)
                    dir("build"){
532
                        //run tests and examples
533
                        sh 'make -j check'
534
                        if (params.RUN_PERFORMANCE_TESTS && navi_node == 0 && mi300_node == 0 ){
535
                            //we only need the ckProfiler to run the performance tests, so we pack and stash it
536
                            //do not stash profiler on Navi or MI300 nodes
Illia Silin's avatar
Illia Silin committed
537
                           sh 'tar -zcvf ckProfiler.tar.gz bin/ckProfiler'
538
                           stash name: "ckProfiler.tar.gz"
Illia Silin's avatar
Illia Silin committed
539
                        }
540
541
                        if (params.RUN_FULL_QA && mi300_node == 0 ){
                           // build deb packages for all MI100/200/300 targets and prepare to export
542
543
544
                           sh 'make -j package'
                           archiveArtifacts artifacts: 'composablekernel-ckprofiler_*.deb'
                           archiveArtifacts artifacts: 'composablekernel-tests_*.deb'
Illia Silin's avatar
Illia Silin committed
545
                           sh 'mv composablekernel-ckprofiler_*.deb ckprofiler_0.2.0_amd64.deb'
546
                           stash name: "ckprofiler_0.2.0_amd64.deb"
547
                        }
548
                    }
549
550
551
552
553
                    if (params.hipTensor_test && navi_node == 0 ){
                        //build and test hipTensor
                        sh """#!/bin/bash
                            rm -rf "${params.hipTensor_branch}".zip
                            rm -rf hipTensor-"${params.hipTensor_branch}"
554
                            wget https://github.com/ROCm/hipTensor/archive/refs/heads/"${params.hipTensor_branch}".zip
555
556
557
558
559
560
                            unzip -o "${params.hipTensor_branch}".zip
                        """
                        dir("hipTensor-${params.hipTensor_branch}"){
                            sh """#!/bin/bash
                                mkdir -p build
                                ls -ltr
561
                                CC=hipcc CXX=hipcc cmake -Bbuild . -D CMAKE_PREFIX_PATH="${env.WORKSPACE}/install"
562
563
564
565
566
567
568
                                cmake --build build -- -j
                            """
                        }
                        dir("hipTensor-${params.hipTensor_branch}/build"){
                            sh 'ctest'
                        }
                    }
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
                }
            }
        }
        return retimage
}

def Build_CK_and_Reboot(Map conf=[:]){
    try{
        Build_CK(conf)
    }
    catch(e){
        echo "throwing error exception while building CK"
        echo 'Exception occurred: ' + e.toString()
        throw e
    }
    finally{
        if (!conf.get("no_reboot", false)) {
            reboot()
        }
    }
}

591
592
593
def process_results(Map conf=[:]){
    env.HSA_ENABLE_SDMA=0
    checkout scm
594
    def image = getDockerImageName() 
595
596
597
598
599
    def prefixpath = "/opt/rocm"

    // Jenkins is complaining about the render group 
    def dockerOpts="--cap-add=SYS_PTRACE --security-opt seccomp=unconfined"
    if (conf.get("enforce_xnack_on", false)) {
600
        dockerOpts = dockerOpts + " --env HSA_XNACK=1 "
601
602
603
604
605
    }

    def variant = env.STAGE_NAME
    def retimage

606
    gitStatusWrapper(credentialsId: "${env.status_wrapper_creds}", gitHubContext: "Jenkins - ${variant}", account: 'ROCm', repo: 'composable_kernel') {
607
        try {
608
            (retimage, image) = getDockerImage(conf)
609
610
611
612
613
614
615
616
617
618
619
620
621
        }
        catch (org.jenkinsci.plugins.workflow.steps.FlowInterruptedException e){
            echo "The job was cancelled or aborted"
            throw e
        }
    }

    withDockerContainer(image: image, args: dockerOpts + ' -v=/var/jenkins/:/var/jenkins') {
        timeout(time: 1, unit: 'HOURS'){
            try{
                dir("script"){
                    if (params.RUN_FULL_QA){
                        // unstash perf files to master
622
623
624
625
626
627
628
629
630
631
632
                        unstash "perf_gemm.log"
                        unstash "perf_resnet50_N256.log"
                        unstash "perf_resnet50_N4.log"
                        unstash "perf_batched_gemm.log"
                        unstash "perf_grouped_gemm.log"
                        unstash "perf_conv_fwd.log"
                        unstash "perf_conv_bwd_data.log"
                        unstash "perf_gemm_bilinear.log"
                        unstash "perf_reduction.log"
                        unstash "perf_splitK_gemm.log"
                        unstash "perf_onnx_gemm.log"
633
                        unstash "perf_mixed_gemm.log"
634
                        sh "./process_qa_data.sh"
Illia Silin's avatar
Illia Silin committed
635
636
                        unstash "ckprofiler_0.2.0_amd64.deb"
                        sh "sshpass -p ${env.ck_deb_pw} scp -o StrictHostKeyChecking=no ckprofiler_0.2.0_amd64.deb ${env.ck_deb_user}@${env.ck_deb_ip}:/var/www/html/composable_kernel/"
637
638
639
                    }
                    else{
                        // unstash perf files to master
640
641
642
643
                        unstash "perf_gemm.log"
                        unstash "perf_resnet50_N256.log"
                        unstash "perf_resnet50_N4.log"
                        sh "./process_perf_data.sh"
644
645
646
647
648
649
650
651
652
653
654
655
                    }
                }
            }
            catch(e){
                echo "throwing error exception while processing performance test results"
                echo 'Exception occurred: ' + e.toString()
                throw e
            }
        }
    }
}

656
//launch develop branch daily at 23:00 UT in FULL_QA mode and at 19:00 UT with latest staging compiler version
657
658
CRON_SETTINGS = BRANCH_NAME == "develop" ? '''0 23 * * * % RUN_FULL_QA=true;ROCMVERSION=6.0;COMPILER_VERSION=
                                              0 21 * * * % ROCMVERSION=6.0;COMPILER_VERSION=;COMPILER_COMMIT=
659
                                              0 19 * * * % BUILD_DOCKER=true;DL_KERNELS=true;COMPILER_VERSION=amd-staging;COMPILER_COMMIT=;USE_SCCACHE=false
660
                                              0 17 * * * % BUILD_DOCKER=true;DL_KERNELS=true;COMPILER_VERSION=amd-mainline-open;COMPILER_COMMIT=;USE_SCCACHE=false''' : ""
661

JD's avatar
JD committed
662
663
pipeline {
    agent none
664
665
666
    triggers {
        parameterizedCron(CRON_SETTINGS)
    }
JD's avatar
JD committed
667
668
669
    options {
        parallelsAlwaysFailFast()
    }
670
    parameters {
671
672
        booleanParam(
            name: "BUILD_DOCKER",
673
            defaultValue: false,
674
            description: "Force building docker image (default: false), set to true if docker image needs to be updated.")
675
676
677
        string(
            name: 'USE_CUSTOM_DOCKER',
            defaultValue: '',
Paul Fultz II's avatar
Paul Fultz II committed
678
            description: 'If you want to use a custom docker image, please specify it here (default: leave blank).')
679
680
        string(
            name: 'ROCMVERSION', 
681
682
            defaultValue: '6.0', 
            description: 'Specify which ROCM version to use: 6.0 (default).')
683
684
        string(
            name: 'COMPILER_VERSION', 
685
            defaultValue: '', 
686
            description: 'Specify which version of compiler to use: release, amd-staging, amd-mainline-open, or leave blank (default).')
687
688
        string(
            name: 'COMPILER_COMMIT', 
689
            defaultValue: '', 
690
            description: 'Specify which commit of compiler branch to use: leave blank to use the latest commit (default), or use some specific commit of llvm-project branch.')
691
692
        string(
            name: 'BUILD_COMPILER', 
693
694
            defaultValue: 'clang', 
            description: 'Specify whether to build CK with hipcc or with clang (default).')
695
696
697
698
        booleanParam(
            name: "RUN_FULL_QA",
            defaultValue: false,
            description: "Select whether to run small set of performance tests (default) or full QA")
699
700
701
702
        booleanParam(
            name: "DL_KERNELS",
            defaultValue: false,
            description: "Select whether to build DL kernels (default: OFF)")
703
704
705
706
707
708
        booleanParam(
            name: "hipTensor_test",
            defaultValue: true,
            description: "Use the CK build to verify hipTensor build and tests (default: ON)")
        string(
            name: 'hipTensor_branch',
709
710
            defaultValue: 'mainline',
            description: 'Specify which branch of hipTensor to use (default: mainline)')
711
712
713
714
        booleanParam(
            name: "USE_SCCACHE",
            defaultValue: true,
            description: "Use the sccache for building CK (default: ON)")
Illia Silin's avatar
Illia Silin committed
715
716
717
718
        booleanParam(
            name: "RUN_CPPCHECK",
            defaultValue: false,
            description: "Run the cppcheck static analysis (default: OFF)")
719
720
        booleanParam(
            name: "RUN_PERFORMANCE_TESTS",
721
722
            defaultValue: true,
            description: "Run the performance tests (default: ON)")
Paul Fultz II's avatar
Paul Fultz II committed
723
724
725
726
        booleanParam(
            name: "RUN_CODEGEN_TESTS",
            defaultValue: true,
            description: "Run the codegen tests (default: ON)")
727
728
729
730
731
732
733
734
    }
    environment{
        dbuser = "${dbuser}"
        dbpassword = "${dbpassword}"
        dbsship = "${dbsship}"
        dbsshport = "${dbsshport}"
        dbsshuser = "${dbsshuser}"
        dbsshpassword = "${dbsshpassword}"
735
        status_wrapper_creds = "${status_wrapper_creds}"
736
737
        gerrit_cred="${gerrit_cred}"
        DOCKER_BUILDKIT = "1"
738
    }
JD's avatar
JD committed
739
    stages{
740
741
742
743
744
745
        stage("Build Docker"){
            parallel{
                stage('Docker /opt/rocm'){
                    agent{ label rocmnode("nogpu") }
                    steps{
                        buildDocker('/opt/rocm')
746
                        cleanWs()
747
748
749
750
                    }
                }
            }
        }
JD's avatar
JD committed
751
752
        stage("Static checks") {
            parallel{
Illia Silin's avatar
Illia Silin committed
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
                stage('Clang Format and Cppcheck') {
                    when {
                        beforeAgent true
                        expression { params.RUN_CPPCHECK.toBoolean() }
                    }
                    agent{ label rocmnode("nogpu") }
                    environment{
                        execute_cmd = "find .. -not -path \'*.git*\' -iname \'*.h\' \
                                -o -not -path \'*.git*\' -iname \'*.hpp\' \
                                -o -not -path \'*.git*\' -iname \'*.cpp\' \
                                -o -iname \'*.h.in\' \
                                -o -iname \'*.hpp.in\' \
                                -o -iname \'*.cpp.in\' \
                                -o -iname \'*.cl\' \
                                | grep -v 'build/' \
                                | xargs -n 1 -P 1 -I{} -t sh -c \'clang-format-12 -style=file {} | diff - {}\' && \
769
770
771
772
773
                                /cppcheck/build/bin/cppcheck ../* -v -j \$(nproc) -I ../include -I ../profiler/include -I ../library/include \
                                -D CK_ENABLE_FP64 -D CK_ENABLE_FP32 -D CK_ENABLE_FP16 -D CK_ENABLE_FP8 -D CK_ENABLE_BF16 -D CK_ENABLE_BF8 -D CK_ENABLE_INT8 -D DL_KERNELS \
                                -D __gfx908__ -D __gfx90a__ -D __gfx940__ -D __gfx941__ -D __gfx942__ -D __gfx1030__ -D __gfx1100__ -D __gfx1101__ -D __gfx1102__ \
                                -U __gfx803__ -U __gfx900__ -U __gfx906__ -U CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4 \
                                --file-filter=*.cpp --force --enable=all --output-file=ck_cppcheck.log"
Illia Silin's avatar
Illia Silin committed
774
775
776
777
778
779
780
                    }
                    steps{
                        buildHipClangJobAndReboot(setup_cmd: "", build_cmd: "", execute_cmd: execute_cmd, no_reboot:true)
                        archiveArtifacts "build/ck_cppcheck.log"
                        cleanWs()
                    }
                }
JD's avatar
JD committed
781
                stage('Clang Format') {
Illia Silin's avatar
Illia Silin committed
782
783
784
785
                    when {
                        beforeAgent true
                        expression { !params.RUN_CPPCHECK.toBoolean() }
                    }
JD's avatar
JD committed
786
787
                    agent{ label rocmnode("nogpu") }
                    environment{
Illia Silin's avatar
Illia Silin committed
788
789
790
                        execute_cmd = "find .. -not -path \'*.git*\' -iname \'*.h\' \
                                -o -not -path \'*.git*\' -iname \'*.hpp\' \
                                -o -not -path \'*.git*\' -iname \'*.cpp\' \
JD's avatar
JD committed
791
792
793
794
795
                                -o -iname \'*.h.in\' \
                                -o -iname \'*.hpp.in\' \
                                -o -iname \'*.cpp.in\' \
                                -o -iname \'*.cl\' \
                                | grep -v 'build/' \
796
                                | xargs -n 1 -P 1 -I{} -t sh -c \'clang-format-12 -style=file {} | diff - {}\'"
JD's avatar
JD committed
797
798
799
                    }
                    steps{
                        buildHipClangJobAndReboot(setup_cmd: "", build_cmd: "", execute_cmd: execute_cmd, no_reboot:true)
800
                        cleanWs()
JD's avatar
JD committed
801
802
803
804
                    }
                }
            }
        }
Paul Fultz II's avatar
Paul Fultz II committed
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
        stage("Run Codegen Tests")
        {
            parallel
            {
                stage("Run Codegen Tests on MI100/MI200")
                {
                    when {
                        beforeAgent true
                        expression { params.RUN_CODEGEN_TESTS.toBoolean() }
                    }
                    options { retry(2) }
                    agent{ label rocmnode("gfx908 || gfx90a")}
                    environment{
                        setup_args = "NO_CK_BUILD"
                        execute_args = """ cd ../codegen && rm -rf build && mkdir build && cd build && \
                                           cmake -D CMAKE_PREFIX_PATH=/opt/rocm \
                                           -D CMAKE_CXX_COMPILER=/opt/rocm/llvm/bin/clang++ \
                                           -D CMAKE_BUILD_TYPE=Release \
                                           -D GPU_TARGETS="gfx908;gfx90a" \
                                           -DCMAKE_CXX_FLAGS=" -O3 " .. && make -j check"""
                   }
                    steps{
                        buildHipClangJobAndReboot(setup_args:setup_args, no_reboot:true, build_type: 'Release', execute_cmd: execute_args)
                        cleanWs()
                    }
                }
            }
        }
833
		stage("Build CK and run Tests")
834
835
836
        {
            parallel
            {
837
838
839
840
841
842
843
844
                stage("Build CK and run Tests on MI100/MI200/MI300")
                {
                    when {
                        beforeAgent true
                        expression { params.RUN_FULL_QA.toBoolean() }
                    }
                    agent{ label rocmnode("gfx908 || gfx90a") }
                    environment{
845
846
847
848
849
850
851
852
853
                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install \
                                         -DGPU_TARGETS="gfx908;gfx90a;gfx940;gfx941;gfx942" \
                                         -DCMAKE_EXE_LINKER_FLAGS=" -L ${env.WORKSPACE}/script -T hip_fatbin_insert " \
                                         -DCMAKE_CXX_FLAGS=" -O3 " """
                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && \
                                           cmake -DCMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" \
                                           -DGPU_TARGETS="gfx908;gfx90a;gfx940;gfx941;gfx942" \
                                           -DCMAKE_CXX_COMPILER="${build_compiler()}" \
                                           -DCMAKE_CXX_FLAGS=" -O3 " .. && make -j """ 
854
855
                    }
                    steps{
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
                        Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
                        cleanWs()
                    }
                }
                stage("Build CK and run Tests on MI300")
                {
                    when {
                        beforeAgent true
                        expression { params.RUN_FULL_QA.toBoolean() }
                    }
                    agent{ label rocmnode("gfx942") }
                    environment{
                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx942" -DCMAKE_CXX_FLAGS=" -O3 " """
                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && \
                                           cmake -DCMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" \
                                           -DGPU_TARGETS="gfx942" \
                                           -DCMAKE_CXX_COMPILER="${build_compiler()}" \
                                           -DCMAKE_CXX_FLAGS=" -O3 " .. && make -j """
                    }
                    steps{
876
                        Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
877
                        cleanWs()
878
879
                    }
                }
Illia Silin's avatar
Illia Silin committed
880
                stage("Build CK and run Tests on MI100/MI200")
881
                {
882
883
884
885
                    when {
                        beforeAgent true
                        expression { !params.RUN_FULL_QA.toBoolean() }
                    }
886
                    agent{ label rocmnode("gfx908 || gfx90a") }
887
                    environment{
888
889
890
891
892
893
                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx908;gfx90a" -DCMAKE_CXX_FLAGS=" -O3 " """
                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && \
                                           cmake -DCMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" \
                                           -DGPU_TARGETS="gfx908;gfx90a" \
                                           -DCMAKE_CXX_COMPILER="${build_compiler()}" \
                                           -DCMAKE_CXX_FLAGS=" -O3 " .. && make -j """ 
Illia Silin's avatar
Illia Silin committed
894
895
896
                    }
                    steps{
                        Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
897
                        cleanWs()
Illia Silin's avatar
Illia Silin committed
898
899
                    }
                }
900
                stage("Build CK and run Tests on Navi21")
Illia Silin's avatar
Illia Silin committed
901
902
903
904
905
906
907
                {
                    when {
                        beforeAgent true
                        expression { !params.RUN_FULL_QA.toBoolean() }
                    }
                    agent{ label rocmnode("navi21") }
                    environment{
908
909
910
911
912
913
                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx1030" -DDL_KERNELS=ON -DCMAKE_CXX_FLAGS=" -O3 " """ 
                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && \
                                           cmake -DCMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" \
                                           -DGPU_TARGETS="gfx1030" \
                                           -DCMAKE_CXX_COMPILER="${build_compiler()}" \
                                           -DCMAKE_CXX_FLAGS=" -O3 " .. && make -j """
914
                    }
915
916
                    steps{
                        Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
917
                        cleanWs()
918
919
920
921
922
923
924
925
926
927
                    }
                }
                stage("Build CK and run Tests on Navi32")
                {
                    when {
                        beforeAgent true
                        expression { !params.RUN_FULL_QA.toBoolean() }
                    }
                    agent{ label rocmnode("navi32") }
                    environment{
928
929
930
931
932
933
                        setup_args = """ -DCMAKE_INSTALL_PREFIX=../install -DGPU_TARGETS="gfx1101" -DDL_KERNELS=ON -DCMAKE_CXX_FLAGS=" -O3 " """
                        execute_args = """ cd ../client_example && rm -rf build && mkdir build && cd build && \
                                           cmake -DCMAKE_PREFIX_PATH="${env.WORKSPACE}/install;/opt/rocm" \
                                           -DGPU_TARGETS="gfx1101" \
                                           -DCMAKE_CXX_COMPILER="${build_compiler()}" \
                                           -DCMAKE_CXX_FLAGS=" -O3 " .. && make -j """
934
                    }
935
                    steps{
936
                        Build_CK_and_Reboot(setup_args: setup_args, config_targets: "install", no_reboot:true, build_type: 'Release', execute_cmd: execute_args, prefixpath: '/usr/local')
937
                        cleanWs()
JD's avatar
JD committed
938
939
                    }
                }
940
941
            }
        }
942

943
944
945
946
        stage("Performance Tests")
        {
            parallel
            {
Illia Silin's avatar
Illia Silin committed
947
                stage("Run ckProfiler: gfx90*")
948
                {
949
950
                    when {
                        beforeAgent true
951
                        expression { !params.RUN_FULL_QA.toBoolean() && params.RUN_PERFORMANCE_TESTS.toBoolean() }
952
                    }
953
                    options { retry(2) }
954
                    agent{ label rocmnode("gfx908 || gfx90a")}
955
                    environment{
Illia Silin's avatar
Illia Silin committed
956
                        setup_args = """ -DGPU_TARGETS="gfx908;gfx90a" -DBUILD_DEV=On """
957
                   }
958
                    steps{
959
                        runPerfTest(setup_args:setup_args, config_targets: "ckProfiler", no_reboot:true, build_type: 'Release')
960
                        cleanWs()
961
962
963
964
                    }
                }
                stage("Run ckProfiler: gfx90a")
                {
965
966
                    when {
                        beforeAgent true
967
                        expression { params.RUN_FULL_QA.toBoolean() && params.RUN_PERFORMANCE_TESTS.toBoolean() }
968
                    }
969
                    options { retry(2) }
970
971
                    agent{ label rocmnode("gfx90a")}
                    environment{
Illia Silin's avatar
Illia Silin committed
972
                        setup_args = """ -DGPU_TARGETS="gfx90a" -DBUILD_DEV=On """
973
                    }
974
                    steps{
975
                        runPerfTest(setup_args:setup_args, config_targets: "ckProfiler", no_reboot:true, build_type: 'Release')
976
                        cleanWs()
977
978
979
980
                    }
                }
            }
        }
981
982
983
984
        stage("Process Performance Test Results")
        {
            parallel
            {
985
                stage("Process results"){
986
987
988
989
                    when {
                        beforeAgent true
                        expression { params.RUN_PERFORMANCE_TESTS.toBoolean() }
                    }
990
991
                    agent { label 'mici' }
                    steps{
992
                        process_results()
993
                        cleanWs()
994
995
996
997
                    }
                }
            }
        }
JD's avatar
JD committed
998
    }
999
}