Guides

Custom DMA firmware and RDMA setups, explained

4 min read, updated 1 November 2026

Two DMA cards with the same chip can behave very differently depending on their firmware. Here is what the firmware is, what custom firmware changes, and what RDMA setups are.

What the firmware is

The FPGA on a DMA card is programmed with a firmware that decides how the card presents itself on the PCIe bus. Default firmware identifies the card as what it is: a development board. Anything that lists PCIe devices can see that.

Custom firmware makes the card present itself as an ordinary device, for example a network or sound card, with matching IDs and configuration. That is why sellers charge for it.

RDMA and vendor runtimes

Some setups do not use the standard USB connection. RDMA and similar setups move the memory reads over the network or through a different driver, and they ship their own versions of the runtime files (leechcore and its plugins).

The QOWED loader checks the runtime files next to it. Files it does not recognise are treated as your vendor's build and left in place; standard files that are missing or outdated are downloaded and verified by checksum. Any .ini your vendor ships is copied along. You do not have to swap DLLs by hand.

If your card is not found

  • Check that the vendor's files and their .ini sit next to the QOWED loader.
  • The client log names the runtime it loaded. Send it with the "Send log to support" button; support can see whether your vendor build or the standard one was used.