How do I practice coding and algorithm questions?

Use PracHub's coding console to write, test, and debug your solutions in Python or JavaScript. View hints, test against sample inputs, and compare with official solutions.

What difficulty level is this coding question?

This is a medium difficulty Coding & Algorithms question, commonly asked during Onsite rounds at Anthropic.

What role is this question designed for?

This question is commonly asked for Software Engineer candidates at Anthropic during technical interviews.

Design high-throughput hashing for kernels

Quick Overview

This question evaluates understanding of cache-efficient, high-throughput hash lookup design and related competencies such as hashing strategies, probe behavior, memory layout, vectorization, bitwise optimizations, prefetching, alignment, and microarchitectural performance trade-offs.

Company: Anthropic

Role: Software Engineer

Category: Coding & Algorithms

Difficulty: medium

Interview Round: Onsite

Design a high-throughput hash-based lookup to be called inside a tight kernel. Choose between open addressing and chaining, specify the load factor, probe sequence, and table layout to favor vectorization and contiguous accesses. Show how to use bitwise operations, prefetching, and alignment to reduce collisions and cache misses. Explain trade-offs and how you would profile and validate the design.

Quick Answer: This question evaluates understanding of cache-efficient, high-throughput hash lookup design and related competencies such as hashing strategies, probe behavior, memory layout, vectorization, bitwise optimizations, prefetching, alignment, and microarchitectural performance trade-offs.

You are modeling the core of a hash-based lookup that will be called inside a tight kernel. To favor contiguous memory accesses, the design uses open addressing with linear probing instead of chaining. Implement the exact hash table behavior below. 1. Let n be the number of input pairs. 2. Choose the table capacity as the smallest power of two greater than or equal to ceil(10 * n / 7), with a minimum capacity of 1. 3. Store entries in a contiguous table using open addressing. 4. Use this hash for the initial slot: initial_index = ((key * 2654435761) & 0xFFFFFFFF) & (capacity - 1) 5. On collision, probe linearly: next_index = (index + 1) & (capacity - 1) 6. Insert pairs in order. If a key appears more than once, overwrite the previous value in place. 7. For lookups, return the stored value, or -1 if the key is not found. A probe is counted every time you examine one table slot during insertion or lookup. Return a tuple: (capacity, answers, insert_probes, lookup_probes) where answers is the list of lookup results for the query keys. This problem focuses on simulating a kernel-friendly table layout. In a lower-level language, the same design would naturally support alignment, prefetching, and contiguous accesses.

Constraints

0 <= len(pairs), len(queries) <= 2 * 10^5
Keys are 32-bit signed integers
Values are integers in the range [0, 10^9]
Capacity is the smallest power of two >= ceil(10 * len(pairs) / 7), with minimum 1

Examples

Input: ([(1, 10), (9, 20), (17, 30)], [1, 17, 2])

Expected Output: (8, [10, 30, -1], 6, 7)

Explanation: Capacity is 8. Keys 1, 9, and 17 all start at slot 1 and are placed at slots 1, 2, and 3. Insert probes = 1 + 2 + 3 = 6. Lookups use the same probe sequence, giving results [10, 30, -1] and lookup probes 1 + 3 + 3 = 7.

Input: ([(5, 1), (13, 2), (5, 7)], [5, 13, 21])

Expected Output: (8, [7, 2, -1], 4, 6)

Explanation: Capacity is 8. Key 5 is inserted, 13 collides and moves to the next slot, then key 5 appears again and overwrites its existing value. Insert probes = 1 + 2 + 1 = 4. Query results are [7, 2, -1] with lookup probes 1 + 2 + 3 = 6.

Input: ([(-1, 4), (7, 8), (-9, 6)], [-9, -1, 15])

Expected Output: (8, [6, 4, -1], 6, 8)

Explanation: With capacity 8, -1, 7, and -9 all start at slot 7 and wrap around to slots 7, 0, and 1. Insert probes = 1 + 2 + 3 = 6. Lookups return [6, 4, -1] with lookup probes 3 + 1 + 4 = 8.

Input: ([], [1, -1])

Expected Output: (1, [-1, -1], 0, 2)

Explanation: An empty table still has minimum capacity 1. No inserts are performed. Each lookup examines the only empty slot once, so lookup probes = 2.

Hints

Because the capacity is a power of two, you can wrap around with & (capacity - 1) instead of using modulo.
With open addressing, a lookup can stop as soon as it reaches an empty slot, because the key could not appear later in that probe sequence.

Quick Overview