I recently put together a new build which has a i9-9900K and a GTX1660TI as the GPU. Under my previous build, which was running a GTX970 I didn't have any problems with work units failing. After reinstalling Win 10, the World Community Grid BOINC client (7.14.2) and latest Nvidia drivers (440.97), I downloaded 3 work units which all immediately failed. Here is the information for one of the 3. They all seem to be missing a module [.nonbonded.cu.] for device version 750.

I had done a search for other work units having failed but came up with nothing. I saw that some of the long run work units were being tweeted, but wasn't sure if this was the case or because of my 1660ti not being compatible with current work units. Not had any issues with Seti@home or Einstein@home GPU work units, though that is sort of an apples to oranges comparision!

Name e81s55_e70s250p0f20-PABLO_V4_UCB_p27_sj403_short_005_salt_IDP-0-2-RND3069_0
Workunit 16828457
Created 25 Oct 2019 | 5:07:28 UTC
Sent 25 Oct 2019 | 5:07:36 UTC
Received 25 Oct 2019 | 9:23:23 UTC
Server state Over
Outcome Computation error
Client state Compute error
Exit status -59 (0xffffffffffffffc5) Unknown error number
Computer ID 515441
Report deadline 30 Oct 2019 | 5:07:36 UTC
Run time 1.04
CPU time 0.00
Validate state Invalid
Credit 0.00
Application version Long runs (8-12 hours on fastest card) v9.23 (cuda80)

Stderr output

<core_client_version>7.14.2</core_client_version>
<![CDATA[
<message>
(unknown error) - exit code -59 (0xffffffc5)</message>
<stderr_txt>
# GPU [GeForce GTX 1660 Ti] Platform [Windows] Rev [3212] VERSION [80]
# SWAN Device 0 :
# Name : GeForce GTX 1660 Ti
# ECC : Disabled
# Global mem : 6144MB
# Capability : 7.5
# PCI ID : 0000:01:00.0
# Device clock : 1860MHz
# Memory clock : 6001MHz
# Memory width : 192bit
# Driver version : r440_92 : 44097
#SWAN: FATAL: cannot find image for module [.nonbonded.cu.] for device version 750

</stderr_txt>

Any help would be appreciated!

Please, take a look to this Retvari Zoltan's post, regarding incompatibility of ACEMD tasks with Turing cards:
https://www.gpugrid.net/forum_thread.php?id=4954&nowrap=true#52822

I just did a search for the error code 59 and found it. Perhaps, I'll put the old 970 card in the build to do work units in the mean time.

Thank you for the reply!