CUDA ak GPU

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

1. Ci kàddu CUDA, luy 'kernel'?
   • ab fonction buy dox paralel ci ay thread GPU yu bari
   • Xeetu sistem biy doxal
   • Benn xeetu puce mémoire
   • Sistemu seddal GPU bi
2. Luy xeetu jëfandikoo SIMT?
   • Duggal bu yeex, toxal yu bari
   • Benn tegtal, ay thread yu bari dañuy doxal benn tegtal ci done yu wuute
   • Benn nataal, ay mbind yu bari
   • Tegtal yu toppalante, waxtu manuel
3. Lan moo waral PyTorch ak TensorFlow duñu faral di sàkku ci jëfandikukat yi ñu bind CUDA bu ñor?
   • GPU yi mënoo doxal kodu jëfandikukat
   • CUDA jagul yoon ngir jëfandikoo ko ci saasi
   • Dañuy woo bibliotek CUDA yiñ gëna defar lu melni cuDNN ak cuBLAS ci biir
   • CPU yi kese lañuy dox
4. Luy 'warp divergence' te lu tax mu gaañ performance?
   • Jumtukaayu reso buy daaneel ay paket
   • GPU bi dafay bënn ci suufu tàngoor
   • Bug buy efaase mémoire
   • Thread yi ci biir warp yi dañuy jël car yu wuute te dañu wara dox ci serie
5. Lan moo waral 'coalesced' am solo ci CUDA?
   • thread yi ci wetam di jàng adres yi ci wetam moo gëna am njariñ
   • Dafay encrypter ay done ci saasi
   • Dafay dindi palaas ci disk bi
   • Dafay dindi fan GPU yi
6. What is OpenAI Triton in relation to CUDA programming?
   • A CUDA driver that replaces NVIDIA GPUs
   • A Python-like language and compiler for writing efficient GPU kernels
   • A model-serving server that only loads TensorFlow checkpoints
   • A framework that converts GPUs into CPUs
7. What does a CUDA kernel launch configuration specify?
   • A model tokenizer vocabulary
   • The number of training labels
   • The grid and block dimensions used to organize thread execution
   • The host operating system
8. Why might a CUDA kernel use shared memory?
   • To make data available across every cluster without communication
   • To replace all registers
   • To store the model permanently on the GPU
   • To let threads in a block reuse data from a faster on-chip memory space