The interview experience was decent. They conducted one round of virtual interview. After ten days, they scheduled another four rounds of virtual interviews. The team I interviewed for was the GPU firmware team.

Round-1: (Technical)

1. Questions on serial port and GPIO. How do you debug if a display didn't come up at boot?
2. Tell me about a challenging issue you debugged.
3. Rotate nibbles in a 32-bit integer. E.g., Input: 0xABCD1234, Output: 0x4321DCBA.
4. Given the address of a node in a singly linked list, but the head pointer is not provided. How do you delete this node?

Round-2: (Technical Design)

1. The interviewer showed a board block diagram. Suppose there are two GPUs on a single graphics card board. Design a way such that one of the GPUs enumerates as a VGA device, then detects and enumerates the monitor to SBIOS.
2. Follow-up questions on previous questions about GPIO pins, firmware, communication between two GPIOs, and SBIOS.

Round-3: (Technical + DSA)

1. How does a GPU communicate with SBIOS saying it's ready? Follow-up questions on ucode implementation high-level design.
2. How to automate driver code implementation of register definitions, which are provided in the HW spec?
3. For multiple SKUs in the same generation, how do you implement code? (Answer: Use device-id. Define bits of device-id to differentiate features and SKUs).
4. Implement code for checking if a matrix is a chessboard or not.
5. Given a function SetBits(value, endBit, startBit). Write the function implementation and handle corner cases.

Round-4: (DSA)

1. Implement integer division without the division operator.
2. Reverse a singly linked list in groups of k.
3. Rotate a 2D array by 90 degrees clockwise.

Round-5: (Display Ecosystem)

1. Questions on the MIPI DSI protocol: about video mode, command mode, BTA, and ESD.
2. Questions on the Display Port protocol: about SST, MST, and link training.
3. Questions on USB-C, PD, Alt-mode, and tunneling.
4. Questions on the eDP protocol.