Every DMA error message and what it means
An index of the errors a DMA setup prints, with the cause and fix for each. Rows are grouped by layer: card and USB path, initialisation, input device and display path.
Short answer
Match the exact DMA error string to its layer to find the real cause and the cheapest fix. Most have one of four causes: the wrong USB port, a missing driver, a missing memory map, or virtualisation enabled in the game PC's BIOS.
On this page10
- How to find your error
- The four layers of a DMA setup, in order
- Layer 1: card and USB path errors
- Layer 2: DTB, base address, MMAP and process errors
- Layer 3: input device errors
- Layer 4: fuser, capture and overlay problems
- What the numbers in an error tell you
- Before you decide the card is faulty
- Common questions about DMA errors
- Sources
How to find your error
Copy the error text from the console window rather than from a screenshot, then search the rows below for it. The rows are grouped by layer, and the layer matters more than the wording, because two products can print different sentences for the same broken cable. If your exact string is not listed, find the row whose symptom describes the same point in the sequence. That is where the software stopped, so start there.
The four layers of a DMA setup, in order
An error can only come from a layer the software has reached. If initialisation printed anything at all, the card and USB layer already worked, so nothing in the first table applies to you.
- 1. Card and USB
- Is there a card, is it powered, is it enumerated, is the right driver bound
- 2. Initialisation
- Can the reader connect, find the memory layout, locate the DTB, find the game process
- 3. Input device
- Does the box answer on its port, and does the game PC see it as a mouse
- 4. Display path
- Does the fuser or capture path carry a picture, and does the overlay land on it
Layer 1: card and USB path errors
These errors appear before the software knows anything about Windows on the game PC. At this layer the cause is almost never the card itself, but the port, the cable, the driver, or a game PC that is switched off.
| Symptom | Likely cause | Fix |
|---|---|---|
| DEVICE: FPGA: ERROR: Unable to connect to USB/FT601 device [0,v0.0,0000] | There is no working USB link to the data bridge, and the zeros show it: nothing at all was read back from the card. The usual reasons are a cable in the flashing port, a charge-only cable, a game PC that is off, or a missing FTDI driver. | Move the cable to the card's DATA port and use a data-rated USB 3 cable in a rear USB 3 port. Boot the game PC first, then install the FTDI D3XX driver and confirm the device appears in Device Manager. |
| DEVICE: FPGA: ERROR: Unable to load 'FTD3XX.dll' or 'leechcore_driver.dll' [0,v0.0,0000] | The FTDI library is not in the same folder as the executable, which is a missing file rather than a hardware fault. | Place FTD3XX.dll in the same folder as the software, using the version shipped with your order. If your order does not include the file, reinstall the D3XX driver package. |
| DEVICE: FPGA: ERROR: Unable to connect to FPGA device | This is the same failure from a build that prints the short form: either there is no USB connection to the card, or the card was never flashed successfully. | Check the port, cable and driver first. If the card has never worked since it was flashed, go back to the flashing step and re-read the DNA ID before you blame the reader. |
| DEVICE: FPGA: ERROR: Unable to retrieve required Device PCIe ID [4,v4.12,0000] | The USB side is fine, because a version number came back, but the PCIe side never came up. Either the card is not seated, the game PC is off or asleep, the slot is dead, or the firmware is not presenting a device. | Power the game PC down, reseat the card until the retention clip closes, screw the bracket down and try a different slot. Then cold-boot the game PC first and start the second PC after it. |
| Device not found | This is the generic wrapper message for the same problem: the software enumerated no card, and nothing on the second PC is bound to the FT601 bridge. | Open Device Manager and look under Universal Serial Bus controllers for the FT601 bridge device. If it is missing, the cause is the port or the cable, and if it shows a warning triangle, the cause is the driver. |
| Failed to connect to memory acquisition device | A launcher is reporting a layer 1 failure instead of leechcore. The causes are the same, but the wording is one step further away from the real error. | Shut both machines down fully, reseat the card and replug both ends of the USB cable. Boot the game PC first, then check the port, the cable and the driver, in that order. |
| Unknown USB Device (Device Descriptor Request Failed) | Windows on the second PC could not read the bridge's descriptors, which points to a marginal link, a bad cable or a hub in the path. | Remove every hub, dock and extension, and connect directly to a rear USB 3 port with a short data cable. If the message persists on two ports and two cables, suspect the USB socket on the card. |
| Card works for a few minutes, then the device disappears | The usual causes are USB selective suspend, a shared controller, a marginal cable, or the card heating up in a closed case. | Do not re-flash anything. Read the dropout diagnosis guide instead, because the timing of the drop is what separates these four causes. |
Layer 2: DTB, base address, MMAP and process errors
If you have reached this layer, the card is alive and the USB path works. Every error in this table means the reader cannot make sense of memory it can already read. The fix is nearly always a memory map or a BIOS setting on the game PC.
| Symptom | Likely cause | Fix |
|---|---|---|
| vmmWinInit_TryInitialize: Initialization Failed. Unable to locate valid DTB. #2 | The reader read memory but could not find the Windows kernel's Directory Table Base in it. Usually there is no memory map, or VT-d/VT-x or SVM/IOMMU is still enabled on the game PC, so the reader receives a translated view of memory. | Generate a memory map and pass it in, disable virtualisation in the game PC's BIOS, then cold-boot both machines. The full ordered list is in the guide to the DTB error. |
| VmmProc: Unable to auto-identify operating system for PROC file system mount. Specify PageDirectoryBase (DTB/CR3) in -dtb option if value is known. | This is the same failure printed one line later, where the tool says it has stopped guessing and offers you the manual route. | Fix the underlying cause first by checking the memory map, virtualisation and cold boot. Supplying -dtb by hand is a diagnostic rather than a repair, because the value changes on every boot. |
| MOUNT: INFO: PROC file system not mounted. | This is a consequence, not a cause: initialisation failed above this line, so there was nothing to mount. | Scroll up, because the real error is the DTB or device line printed before this one. |
| Failed to initialize DMA | The launcher could not start the reader at all. Either layer 1 never completed, or the memory map it was told to load is missing or malformed. | First confirm the card is visible, then check that the map file exists at the path the software expects. The map must have been generated on this machine, with this firmware, since the last hardware change. |
| Failed to find base address | The reader connected and found the kernel but could not resolve the module base of the target process. Typical causes are a missing or stale memory map, a stale process handle after an alt-tab or a game update, or an unstable read path. | Shut both machines down fully, restart them and regenerate the memory map. If the error only happens after a game update, wait until the software is updated for that build. |
| Process not running / Failed to find process | The game process was not there when the software looked for it. Either the game is not running, is still on its launcher splash screen or is running under a different session, or the reader is reading a stale snapshot. | Start the game and reach the main menu, then start the software in the order your product documents. If the game is clearly running, treat this as a read-path failure and regenerate the map. |
| invalid memory map | The software rejected the map file because it was hand-edited, truncated, produced for different hardware, or generated while the read path was already unstable. | Delete it and generate a new one on the current hardware. Never copy a memory map from another person's machine, because it describes their RAM layout rather than yours. |
| Initialization Failed. Unable to walk EPROCESS | The kernel structures were found but could not be walked, which points to partially readable memory rather than a wrong address. | Generate a memory map so the reader never touches unreadable ranges, and cold-boot both machines. Confirm the read rate is stable before you try again. |
Layer 3: input device errors
The box that moves the cursor is a separate USB device with its own failure modes. Errors at this layer concern a serial port or a network address and have nothing to do with the card.
| Symptom | Likely cause | Fix |
|---|---|---|
| Device not found when the software looks for the input box | The box is not in the second PC's port list, because it is unplugged, plugged into the game PC only, or has no driver bound. | Confirm the box appears in Device Manager on the SECOND PC as a COM port or serial device. The data cable goes to the second PC, and the HID cable goes to the game PC. |
| Access to the port 'COM3' is denied | Another program already holds the port, such as a second copy of the software, a serial terminal, or a flashing tool left open. | Close every other program that could hold the port, then reopen the software. If the error persists, unplug and replug the box so the port re-enumerates. |
| The port 'COM3' does not exist | The configured port number is not the one Windows assigned, and port numbers change when you move to a different USB socket. | Read the current port number in Device Manager and set it in the software. Use the same physical socket every time so the number does not change. |
| The box appears under an unexpected serial device name | These boards use generic USB-to-serial bridges, so Windows names them after the bridge chip rather than the product. | Identify the box by its port number rather than its friendly name. Unplug the box, note which entry disappears, then replug it and use that port. |
| Network box never answers on its IP, or the connection times out | The network-attached box did not get an address, or its address, port or UUID was typed in wrong. If the switch or router has no DHCP, the box comes up unreachable. | Confirm that DHCP is on and that the box's address is on the same subnet as the second PC. Then re-enter the address, port and UUID exactly as printed for your unit. |
| Cursor moves in a test on the second PC but nothing moves in the game | The game PC is not receiving HID input from the box, because the HID cable is in the wrong socket or the box's HID side has not enumerated there. | Check the game PC's Device Manager for a mouse device that appears and disappears as you plug and unplug the box's HID cable. Test it on the desktop before you test it in a game. |
| Movement is doubled, jittery or fights itself | The physical mouse and the box are both sending movement to the game PC at the same time, or two software features are driving the same axis. | Route the physical mouse through the box as your product documents, and enable only one movement source at a time. |
| Unknown USB Device (Device Descriptor Request Failed) for the input box | The box has a marginal USB link, caused by a charge-only cable, a hub or a damaged socket. | Swap the cable and connect directly to a rear port. Then test the box on a second machine, which tells you whether the box or the port is at fault. |
Layer 4: fuser, capture and overlay problems
The display layer only matters once everything above it works. At this point the software is usually fine, and the problem is a video signal, an EDID or a resolution mismatch.
| Symptom | Likely cause | Fix |
|---|---|---|
| No signal on the main monitor after the fuser goes in | There is a break in the video chain, the input on the fuser is not selected, or a cable cannot carry the resolution and refresh rate you were using. | Rebuild the chain one cable at a time, starting with the game PC connected directly to the monitor. Add the fuser only once the direct path works, and use cables rated for your resolution and refresh rate. |
| The game PC drops to a low resolution or 60 Hz after the fuser is inserted | The monitor's EDID is not reaching the game PC, so it falls back to a safe mode. | Use the fuser's EDID pass-through, or copy the monitor's EDID into the fuser as the maker documents. Afterwards, replug the cable at the source end so the game PC reads the EDID again. |
| Black screen or a flicker every few seconds | A failing HDMI handshake, a marginal or overlong cable, or content protection renegotiating on the link. | Shorten the run, use a certified cable, and remove any adapters and splitters. Test at a lower refresh rate to see whether the link simply has no headroom left. |
| The capture app shows a black picture while the monitor shows the game | The source on that input is protected, the wrong input is selected, or the capture device never received a signal format it accepts. | Select the correct input and close anything else that is using the capture device. Check that the source's resolution and refresh rate are on the device's supported list. |
| The overlay draws on the second PC but not through the fuser | The overlay is rendering on the wrong display, or the fuser is receiving a different output from the one the overlay is on. | Feed the fuser from the display output the overlay targets, and confirm the second PC is actually sending a signal on that port. |
| The overlay appears but is offset, scaled or clipped | The overlay resolution does not match the game resolution, or the fuser is scaling one of the two signals. | Set both machines to the same resolution and refresh rate, and turn off scaling in the fuser and in the GPU control panel. |
| Overlay position drifts as soon as the game changes resolution | The overlay was aligned for the previous display mode, and nothing told it that the mode had changed. | Lock the game to one resolution, re-align the overlay once, and avoid switching between borderless and fullscreen during a session. |
Info
What the numbers in an error tell you
The values at the end of these messages carry information. In the FPGA lines, a device ID and version of all zeros means nothing at all was read back over USB. A real version number with a zero PCIe ID means the opposite: USB worked but PCIe did not, which is a different problem with a different fix. The number after a DTB failure identifies which attempt failed. If you shorten the string to "it says DTB error", you lose the part that narrows down the cause.
Before you decide the card is faulty
Cards do fail, but far less often than the causes on this list. Each item below has been the real answer in a ticket that started with "my card is dead".
- The cable is in the card's data port, not the flashing port
- The cable is a data cable, plugged directly into a rear USB 3 port with no hub, dock or extension
- The game PC was booted first and reached the desktop before the second PC was started
- The FTDI D3XX driver is installed on the second PC, and the bridge shows no warning triangle
- Virtualisation is disabled in the game PC's BIOS: VT-d and VT-x on Intel, SVM and IOMMU on AMD
- A memory map exists, was generated on this hardware and is at the path the software expects
- Both machines have had a full power-off cold boot rather than a restart
Common questions about DMA errors
I get two errors at once. Which do I fix first?
Start with the one from the lowest layer. A DTB failure printed after a device error is a consequence of it, and it usually disappears once the device line is fixed.
The error changed after I changed something. Is that progress?
Yes, if the new error comes from a higher layer. Moving from a USB error to a DTB error means the card and the USB path now work and you are one layer further along.
Do I need to reinstall Windows to clear one of these?
Almost never. It is the most expensive step on the list, and every error in the index has a cheaper fix. A clean install that does not address the real cause reproduces the same error on a fresh system.
My exact error is not listed. What should I do?
Match it by layer: work out the last thing that succeeded. Then send the full string in a ticket, with the line before and the line after it, so it can be added.
Does one of these errors mean I have been detected?
No. These are all initialisation or hardware-path failures, and they are the same errors you would see on a machine that has never launched a game.
Sources
The cause and fix next to each string are bench work: they are what the team found when customers sent these errors in.
- 01LeechCore (Ulf Frisk)
the layer that prints the FPGA and FT601 device errors, and the leechcore_driver.dll named in them
github.com
- 02MemProcFS (Ulf Frisk)
the tool that prints the vmmWinInit, VmmProc and MOUNT lines quoted here
github.com
- 03
Read next
Still stuck? Open a ticket on Discord
All guides