GPU L7: Computation

Piyush Avinash Chincholikar cs20m045: Good morning sir KARTHIK SURESH ee16b140: Has sir KARTHIK SURESH ee16b140: sir's voice cut off? KASHYAPI SHUBHAM BHALCHANDRA mm16b027: Yes Nishant Prabhu me17b084: His audio has stopped Sumit Negi cs20m067: yes NIRANJAN KAMATH ph16b006: Yeah Jaladi Sashedher cs20m027: yes Nishitha R ee20s064: Is the video stuck for other too? Shibobrota Das cs20m059: alright i'm not the only one Rajesh Pandian M: yes Anuj Kumar Singh cs20m013: Yes Nishitha R ee20s064: others* Mr Rahul Mastram Verma CS20S038: Sir you your voice is not coming Shouvick Mondal: yes KASHYAPI SHUBHAM BHALCHANDRA mm16b027: Yes Nishitha R ee20s064: Yes, sir Nishant Prabhu me17b084: yes Piyush Avinash Chincholikar cs20m045: yes Jeswant Krishna ae17b030: yes Mr Rahul Mastram Verma CS20S038: No Nishitha R ee20s064: Operation will be column-wis? Banavath Tarun cs17b102: no change Aditya Agrawal CS20S026: will it be column major now? Nishant Prabhu me17b084: It will be column msjor Aditya Agrawal CS20S026: where does grid come here? Shagnik Pal ee17b147: Sir you said warps can change the SMs. Since warp is a part of the block, if block doesn't change the SM, how can warps change? Rigved Sah cs20m053: Does all threads within a warp run on same core? Shagnik Pal ee17b147: Ok sir Parth S. Shah ee17b059: Sir, if a warp has a stall, does a different warp get executed on the cores during the stall? Shouvick Mondal: FAQs? Parth S. Shah ee17b059: Like for example a stall due to a memory operation maybe? Vishnu Kotte: Sir can we assume a multi-processor has tens of thousands of cores because each thread goes to a different core?? M Venkat Rahul cs18b026: sir can you explain the point about warp moving in the same SM? i didnt get it ee20s136 Rahual G S: Sir, at a given point of time how many thread blocks can reside in a Streaming Multiprocessor? Shouvick Mondal: All 32 warp threads share a single program counter, right? Rigved Sah cs20m053: Sir can you give an example regarding the previous programs Parth S. Shah ee17b059: So the time-multiplexing in each SM is done only at a warp level but not a thread level right? K Sampreeth Prem cs17b013: Does all threads in a warp execute same instruction? ee20s136 Rahual G S: Sir when the data to be processed is large it's better to use the less number of threads (as a multiple of number of cores in a SM) and increase the number of blocks, since this will increase the number of SMs utilized simultaneously right?