Re: [HTCondor-users] GPUs not detected in 9.0.6 version

[John M Knoeller <johnkn@xxxxxxxxxxx>]

Yep.  I'm pretty sure that the crash is caused by an nvml library that does not have the

nvmlDeviceGetMaxMigDeviceCount function.  we are missing a test for this to be null in 

one of the code paths

thanks again,

-tj

---

From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> on behalf of John M Knoeller <johnkn@xxxxxxxxxxx>
Sent: Thursday, September 30, 2021 11:14 AM
To: HTCondor-Users Mail List <htcondor-users@xxxxxxxxxxx>
Subject: Re: [HTCondor-users] GPUs not detected in 9.0.6 version

 

Thank you.  this output will be helpful

-tj

---

From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> on behalf of Anderson, Stuart B. <sba@xxxxxxxxxxx>
Sent: Thursday, September 30, 2021 10:58 AM
To: HTCondor-Users Mail List <htcondor-users@xxxxxxxxxxx>
Subject: Re: [HTCondor-users] GPUs not detected in 9.0.6 version

 

Carles,
        In case it helps, here is what I see from "condor_gpu_discovery -verbose -diag" on an SL7 system with a single GTX 1050 Ti running condor 9.0.6, CUDA 11.2 and NVIDIA driver 460.32.03.


root@node1 config.d]# /usr/libexec/condor/condor_gpu_discovery -verbose -diag
diag: clearing environment before device enumeration
diag: using nvcuda for gpu discovery
# querying ordinal:0, dev:0x7ffc00000000 using cuDevice* API
# cuDeviceTotalMem(0) returns 0, value = 4236312576
# cuDeviceTotalMem(0x7ffc00000000) returns 0, value = 4236312576
# nvml_getBasicProps() for GPU-23b6505e-b534-990a-9ec9-f4dca5662ab0 returns 0
diag: skipping uuid=GPU-23b6505e-b534-990a-9ec9-f4dca5662ab0 during nvml enumeration because it matches CUDA0
DetectedGPUs="GPU-23b6505e"


Looks like the call to nvml_getBasicProps() is new in 9.0.6. I don't know if will help, but it might also be worth comparing the output of compiling and running /usr/local/cuda/samples/1_Utilities/deviceQuery to see if there are different BasicProps being returned that Condor is chocking on. Here is what I see,


[root@node1 deviceQuery]# ./deviceQuery
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "GeForce GTX 1050 Ti"
  CUDA Driver Version / Runtime Version          11.2 / 11.2
  CUDA Capability Major/Minor version number:    6.1
  Total amount of global memory:                 4040 MBytes (4236312576 bytes)
  ( 6) Multiprocessors, (128) CUDA Cores/MP:     768 CUDA Cores
  GPU Max Clock rate:                            1392 MHz (1.39 GHz)
  Memory Clock rate:                             3504 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 1048576 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        98304 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Managed Memory:                Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 7 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 11.2, CUDA Runtime Version = 11.2, NumDevs = 1
Result = PASS


Thanks.


> On Sep 29, 2021, at 10:07 PM, Carles Acosta <cacosta@xxxxxx> wrote:
>
> Hi TJ,
>
> Here you have the results:
>
> # /usr/libexec/condor/condor_gpu_discovery-9.0.6 -verbose -diag
> diag: clearing environment before device enumeration
> diag: using nvcuda for gpu discovery
> # querying ordinal:0, dev:0x833117100000000 using cuDevice* API
> # cuDeviceTotalMem(0) returns 0, value = 4236312576
> # cuDeviceTotalMem(0x833117100000000) returns 0, value = 4236312576
> # nvml_getBasicProps() for GPU-c659279d-ce12-c3b9-f9c4-05a68df7c711 returns 0
> Segmentation fault
>
> On the other hand, using the 9.0.5 version:
>
> # /usr/libexec/condor/condor_gpu_discovery-9.0.5 -verbose -diag
> diag: using nvcuda for gpu discovery
> # querying ordinal:0, dev:0xa2f81b3c00000000 using cuDevice* API
> # cuDeviceTotalMem(0) returns 0, value = 4236312576
> # cuDeviceTotalMem(0xa2f81b3c00000000) returns 0, value = 4236312576
> DetectedGPUs="GPU-c659279d"
>

--
Stuart Anderson
sba@xxxxxxxxxxx




_______________________________________________
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users

The archives can be found at:
https://lists.cs.wisc.edu/archive/htcondor-users/