Modal Technical Screen

Given a series of questions based on the following:

• Think out loud — interviewers care more about your reasoning than the perfect answer.
  
• Reason from first principles — you won't recognize everything, and that's expected.
  
• Be ready to discuss tradeoffs — correctness, performance, edge cases, and what inputs could break the code.
  
• Brush up on async/concurrency — systems and concurrency concepts may come up.
  
• Get comfortable with order-of-magnitude thinking — memory, compute, and performance at scale, including CPU & memory dynamics.
  
• Please refrain from using an AI tools
  

-----
Asked to determine what a mystery_fn does. It converted integers into human readabl

您好！
本帖隐藏的内容需要积分高于 188 才可浏览
您当前积分为 0。
使用VIP即刻解锁阅读权限或查看其他获取积分的方式

游客，您好！
本帖隐藏的内容需要积分高于 188 才可浏览
您当前积分为 0。
VIP即刻解锁阅读权限 或 查看其他获取积分的方式
Unlock interview details and practice with AI
Curated Interview Questions from Top Companies

n a fair random way).

Use a fixed size array with ten buckets, iterate through and collect the buckets. Then iterate through the buckets and generate the output.

With 100M digits in a compiled language like rust or c++

How long would this take?

The work:

1. 100M reads        // count buckets
2. 100M bucket bumps // buckets stay hot in cache
3. 100M writes       // generate sorted output

复制代码

Memory:

1. 100M u8 digits     = ~100 MB input
2. 100M usize digits  = ~800 MB input on 64-bit
3. output Vec<usize>  = another ~800 MB if using usize

复制代码

for u8

1. CPU = ~3 GHz
2. memory bandwidth = ~20-80 GB/s
4. 100M [i] 1 byte = 100 MB
6. Counting pass
7. 100M [/i] 2 cycles / 3GHz  ~= 0.07s
8. 100M * 8 cycles / 3GHz  ~= 0.27s
10. Output pass
11. 100 MB written
12. 100 MB / 20 GB/s ~= 0.005s
13. 100 MB / 80 GB/s ~= 0.001s
15. accounting for loop overhead
16. counting: ~0.07s-0.27s
17. output:   ~0.02s-0.15s
18. total:    ~0.1s-0.5s

复制代码

Also depends on input type. Parsing from strings could take seconds.