GPU: Simplify memory alignment for host allocations and support stacked memory for direct allocations

[davidrohr]

No description provided.

[github-actions]

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2023-pbpb-apass4
async-2023-pp-apass4
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0
async-2024-PbPb-apass1
async-2024-ppRef-apass1
async-2024-PbPb-apass2
async-2023-PbPb-apass5

[davidrohr]

@mconcas : could you check if this causes any trouble with ITS GPU tracking?
Or could you send me a command line to run dpl-workflow.sh, so I can check myself?

[alibuild]

Error while checking build/O2/fullCI_slc9 for 99853a8 at 2025-04-21 23:47:

## sw/BUILD/O2Physics-latest/log
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:


## sw/BUILD/O2-sim-challenge-test-latest/log
./sim-challenge.logtask timeout reached .. killing all processes


## sw/BUILD/o2checkcode-latest/log
--
========== List of errors found ==========
++ GRERR=0
++ grep -v clang-diagnostic-error error-log.txt
++ grep ' error:'
grep: error-log.txt: binary file matches
++ GRERR=1
++ [[ 1 == 0 ]]
++ mkdir -p /sw/INSTALLROOT/56e69c041f8dab1d10b6be08d129ea13bc2e0ffa/slc9_x86-64/o2checkcode/1.0-local251/etc/modulefiles
++ cat
--

Full log here.

[mconcas]

@mconcas : could you check if this causes any trouble with ITS GPU tracking? Or could you send me a command line to run dpl-workflow.sh, so I can check myself?

I can test it later today, if you want to try: this is what I used to validate fst:

#! /bin/bash -x

rm -f /dev/shm/*fmq*;
# export WORKFLOWMODE=print
# export DISABLE_ROOT_OUTPUT=0
export ROCR_VISIBLE_DEVICES="0"
export GPUTYPE=HIP
export WORKFLOW_DETECTORS=TPC,ITS
export WORKFLOW_DETECTORS_GPU=ITS,TPC
export GPUMEMSIZE=10000000000
export SYNCMODE=0
export CTFINPUT=1
# export GLOBALDPLOPT="-b"
export NTIMEFRAMES=1
export INPUT_FILE_LIST=data/o2_ctf_run00559456_orbit0139181824_tf0003905868_epn294.root
export CONFIG_EXTRA_PROCESS_o2_gpu_reco_workflow="GPU_proc.debugLevel=2;GPU_global.deviceType=$GPUTYPE;GPU_proc.forceMemoryPoolSize=8000000000;GPU_proc.forceHostMemoryPoolSize=1073741824;"
~/alice/latest/O2/prodtests/full-system-test/dpl-workflow.sh

It is based on: /alice/data/2024/LHC24aq/559456/raw/0700/o2_ctf_run00559456_orbit0139181824_tf0003905868_epn294.root .

[alibuild]

Error while checking build/O2/fullCI_slc9 for c789667 at 2025-04-22 11:06:

## sw/BUILD/O2Physics-latest/log
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:
Error in cling::AutoLoadingVisitor::InsertIntoAutoLoadingState:


## sw/BUILD/O2-sim-challenge-test-latest/log
./sim-challenge.logtask timeout reached .. killing all processes
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/37}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/38}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/40}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/42}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/43}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/45}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/46}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/47}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/48}
./sim-challenge.log[ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/49}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/37}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/38}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/40}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/42}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/43}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/45}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/46}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/47}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/48}
./sim-challenge.log[7740:TRDDigitizer]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/49}
./sim-challenge.log[7739:internal-dpl-ccdb-backend]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/37}
./sim-challenge.log[7739:internal-dpl-ccdb-backend]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/38}
./sim-challenge.log[7739:internal-dpl-ccdb-backend]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/40}
./sim-challenge.log[7739:internal-dpl-ccdb-backend]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/42}
./sim-challenge.log[7739:internal-dpl-ccdb-backend]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/43}
./sim-challenge.log[7739:internal-dpl-ccdb-backend]: [ERROR] Found duplicate input binding with different spec.:collisioncontext {SIM/COLLISIONCONTEXT/45}
[0 more errors; see full log]

Full log here.

[alibuild]

Error while checking build/O2/fullCI_slc9 for 03c1b38 at 2025-04-22 17:48:

## sw/BUILD/o2checkcode-latest/log
--
========== List of errors found ==========
++ GRERR=0
++ grep -v clang-diagnostic-error error-log.txt
++ grep ' error:'
/sw/SOURCES/O2/14208-slc9_x86-64/0/GPU/GPUTracking/TPCClusterFinder/GPUTPCNNClusterizerHost.cxx:154:21: error: use '= default' to define a trivial destructor [modernize-use-equals-default]
/sw/SOURCES/O2/14208-slc9_x86-64/0/GPU/GPUTracking/TPCClusterFinder/GPUTPCNNClusterizerHost.cxx:194:27: error: statement should be inside braces [readability-braces-around-statements]
++ [[ 0 == 0 ]]
++ exit 1
--

Full log here.

[mconcas]

FST tested locally, it works.