- Status
- Offline
- Joined
- Mar 3, 2026
- Messages
- 546
- Reaction score
- 7
Grabbed one of those budget Stark DMA 75T boards off Amazon recently and I'm hitting a wall with the custom firmware. I've been following the standard guides to roll my own FW for better UD chances, but the results are absolute trash compared to the stock provider files.
The Setup:
The Symptoms:
Whenever I flash my custom bin and reboot the main PC, it either hangs indefinitely on the motherboard splash screen or crawls into Windows with 1 FPS and "robotic" audio. It feels like a massive DPC latency spike or an interrupt storm. The weird part? A speedtest shows the DMA is technically communicating fine, but the system usability is zero—literally like playing on a fridge.
If the stock Capitain FW works, the hardware is obviously fine. This has to be a configuration mismatch in the Vivado project or a bad TLP header setup in the custom source. It’s likely clashing with the high-end Ryzen chipset's PCIe lane management.
Has anyone else dealt with this specific Stark board and managed to get a stable custom build running without the system-wide stutter?
The Setup:
- Main Rig: Ryzen X3D series platform / RTX 4070 Super.
- Hardware: Stark DMA 75T (Artix-7 FPGA).
- Software: PCILeech based custom FW builds.
The Symptoms:
Whenever I flash my custom bin and reboot the main PC, it either hangs indefinitely on the motherboard splash screen or crawls into Windows with 1 FPS and "robotic" audio. It feels like a massive DPC latency spike or an interrupt storm. The weird part? A speedtest shows the DMA is technically communicating fine, but the system usability is zero—literally like playing on a fridge.
- Stock PCILeech Capitain DMA 75T firmware works perfectly. Zero lag, zero boot issues.
- Attempted to expose different USB ports in the config space, no change.
- The custom FW seems to be messing with the PCIe configuration space or BARs in a way that the BIOS/Windows kernel hates.
If the stock Capitain FW works, the hardware is obviously fine. This has to be a configuration mismatch in the Vivado project or a bad TLP header setup in the custom source. It’s likely clashing with the high-end Ryzen chipset's PCIe lane management.
Has anyone else dealt with this specific Stark board and managed to get a stable custom build running without the system-wide stutter?
