Welcome to QuEST Discussions!

[TysonRayJones]

👋 Welcome!

We’re testing using Discussions to keep track of what the community is working on, and what new features are coming to QuEST.

Feel free to create discussions which:

• Describe new facilities you are working on which you hope to integrate into the QuEST repo.
• Ask about new features, or propose new facilities.

This will, at least initially, be a very relaxed forum so feel free also to create discussions about anything you wish! You can of course also create an Issue.

[fieldofnodes]

Hi, Great to see.

[ketan-more-github]

when i am using quest as backend GPU i am not able to saw tracess of usage of GPU using nvidia-smi cmd

[TysonRayJones]

Hi Ketan,

That likely indicates QuEST is not running on the GPU. Have you compiled QuEST with GPU-acceleration enabled, as outlined here?

The upcoming v4 (only days away! 🎉) conveniently provides a reportQuESTEnv() function to make ascertaining the utilised deployments much easier. You'd simply run:

reportQuESTEnv();

to see output like:

QuEST execution environment:
...
  [compilation]
    isMpiCompiled...........1
    isGpuCompiled...........1
    isOmpCompiled...........1
    isCuQuantumCompiled.....0
  [deployment]
    isMpiEnabled.....1
    isGpuEnabled.....1
    isOmpEnabled.....1
  [cpu]
...
  [gpu]
    numGpus...........4
    gpuDirect.........1
    gpuMemPools.......1
    gpuMemory.........79.3 GiB per gpu
    gpuMemoryFree.....78.8 GiB per gpu
    gpuCache..........0 bytes per gpu
  [distribution]
    isMpiGpuAware.....1
    numMpiNodes.......4
...
  [statevector autodeployment]
    8 qubits......[omp]
    12 qubits.....[omp] [gpu]
    28 qubits.....[omp] [gpu] [mpi]
  [density matrix autodeployment]
    4 qubits......[omp]
    6 qubits......[omp] [gpu]
    14 qubits.....[omp] [gpu] [mpi]

Or you can see the parallelisation automatically chosen for a particular Qureg with

reportQuregParams(qureg);

which might show

Qureg:
  [deployment]
    isMpiEnabled.....1
    isGpuEnabled.....1
    isOmpEnabled.....1
  [dimension]
    isDensMatr.....0
    numQubits......34
    numCols........N/A
    numAmps........2^34 = 17179869184
  [distribution]
    numNodes.....2^2 = 4
    numCols......N/A
    numAmps......2^32 = 4294967296 per node
  [memory]
    cpuAmps...........32 GiB per node
    gpuAmps...........32 GiB per node
    cpuCommBuffer.....32 GiB per node
    gpuCommBuffer.....32 GiB per node
    globalTotal.......512 GiB

If you're just getting started with a new project, I would recommend you trying the new v4 branch. Setup is a little different - I'm happy to guide you through it while we polish the docs!

[ketan-more-github]

Yes i have compiled source with gpu and cu quantum also and i get effective results. But didn't get insight on gpu card..for running process No issue i will love to try V4 branch i will check it

…

On Tue, Mar 4, 2025, 1:26 AM Tyson Jones ***@***.***> wrote: Hi Ketan, That likely indicates QuEST is *not* running on the GPU. Have you compiled QuEST with GPU-acceleration enabled, as outlined here <https://github.com/QuEST-Kit/QuEST/blob/master/examples/README.md#compiling> ? The upcoming v4 <https://github.com/QuEST-Kit/QuEST/tree/v4> (only days away! 🎉) conveniently provides a reportQuESTEnv() function to make ascertaining the utilised deployments much easier. You'd simply run: reportQuESTEnv(); to see output like: QuEST execution environment: ... [compilation] isMpiCompiled...........1 isGpuCompiled...........1 isOmpCompiled...........1 isCuQuantumCompiled.....0 [deployment] isMpiEnabled.....1 isGpuEnabled.....1 isOmpEnabled.....1 [cpu] ... [gpu] numGpus...........4 gpuDirect.........1 gpuMemPools.......1 gpuMemory.........79.3 GiB per gpu gpuMemoryFree.....78.8 GiB per gpu gpuCache..........0 bytes per gpu [distribution] isMpiGpuAware.....1 numMpiNodes.......4 ... [statevector autodeployment] 8 qubits......[omp] 12 qubits.....[omp] [gpu] 28 qubits.....[omp] [gpu] [mpi] [density matrix autodeployment] 4 qubits......[omp] 6 qubits......[omp] [gpu] 14 qubits.....[omp] [gpu] [mpi] Or you can see the parallelisation automatically chosen for a particular Qureg with reportQuregParams(qureg); which might show Qureg: [deployment] isMpiEnabled.....1 isGpuEnabled.....1 isOmpEnabled.....1 [dimension] isDensMatr.....0 numQubits......34 numCols........N/A numAmps........2^34 = 17179869184 [distribution] numNodes.....2^2 = 4 numCols......N/A numAmps......2^32 = 4294967296 per node [memory] cpuAmps...........32 GiB per node gpuAmps...........32 GiB per node cpuCommBuffer.....32 GiB per node gpuCommBuffer.....32 GiB per node globalTotal.......512 GiB If you're just getting started with a new project, I would recommend you trying the new v4 <https://github.com/QuEST-Kit/QuEST/tree/v4> branch. Setup is a little different - I'm happy to guide you through it while we polish the docs! — Reply to this email directly, view it on GitHub <#393 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AXZ3MBZARPH3VQ2IQUNCOTD2SSXWRAVCNFSM6AAAAABYG3SAH2VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTEMZYGAYTQNY> . You are receiving this because you commented.Message ID: ***@***.***>