Hello,
We’ll pretend I’m completely new to computers and workstation GPUs just to make things simple. I wanted to set up my own AI machine to run models locally, and procured a Lenovo P520 workstation with a Xeon processor, 32GB of RAM, 900W power supply, and a 1070 GPU. I found an NVIDIA M60 for dirt cheap and thought the 16GB of VRAM and large number of cores would be useful for my application, albeit slow compared to newer, more expensive cards. That’s okay with me.
I’m running Windows 10, and have installed the latest viable drivers that support both the 1070 and the M60. When I check Device Manager, both cards show up, though the M60 shows “CODE 10: Not enough resources to complete the API”.
Here are some solutions I’ve attempted to no avail:
- Rolled back drivers to 4xx.xx
- Increased/Decreased MMIO bandwidths
- Ensured Above 4G is enabled in BIOS
- Ensured the PCIe slots were capable of 16 lanes for both GPUs
- Added active cooling with two 40mm fans
- Using custom cable adapter to power the M60
- Checked nvidia-smi, and the M60 doesn’t show up
Here are a couple more details that might help:
- The M60’s 16GB of VRAM show up in task manger, but no more than 0.1GB ever gets used, and the card does not appear separately
- The M60 does NOT appear in my NVIDIA control panel, nor does it appear in the NVIDIA app
I understand the M60 is meant to be used for distributed workloads; however, in my mind the card should be usable locally by the host computer without virtualization, but perhaps I just haven’t taken the right steps to do so. Thanks in advance for the help!