1 / 38

CUDA Lecture 7 CUDA Threads and Atomics

CUDA Lecture 7 CUDA Threads and Atomics. Prepared 8/8/2011 by T. O’Neil for 3460:677, Fall 2011, The University of Akron. Topic 1: Atomics. The Problem: how do you do global communication? Finish a kernel and start a new one All writes from all threads complete before a kernel finishes

duy
Download Presentation

CUDA Lecture 7 CUDA Threads and Atomics

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. CUDA Lecture 7CUDA Threads and Atomics Prepared 8/8/2011 by T. O’Neil for 3460:677, Fall 2011, The University of Akron.

  2. Topic 1: Atomics • The Problem: how do you do global communication? • Finish a kernel and start a new one • All writes from all threads complete before a kernel finishes • Would need to decompose kernels into before and after parts CUDA Threads and Atomics – Slide 2

  3. Race Conditions • Alternately write to a predefined memory location • Race condition! Updates can be lost • What is the value of a in thread 0? In thread 1917? CUDA Threads and Atomics – Slide 3

  4. Race Conditions (cont.) • Thread 0 could have finished execution before 1917 started • Or the other way around • Or both are executing at the same time • Answer: not defined by the programming model; can be arbitrary CUDA Threads and Atomics – Slide 4

  5. Atomics • CUDA provides atomic operations to deal with this problem • An atomic operation guarantees that only a single thread has access to a piece of memory while an operation completes • The name atomic comes from the fact that it is uninterruptable • No dropped data, but ordering is still arbitrary CUDA Threads and Atomics – Slide 5

  6. Atomics (cont.) • CUDA provides atomic operations to deal with this problem • Requires hardware with compute capability 1.1 and above • Different types of atomic instructions • Addition/subtraction: atomicAdd, atomicSub • Minimum/maximum: atomicMin, atomicMax • Conditional increment/decrement: atomicInc, atomicDec • Exchange/compare-and-swap: atomicExch, atomicCAS • More types in fermi: atomicAnd, atomicOr, atomicXor CUDA Threads and Atomics – Slide 6

  7. Example: Histogram • atomicAdd returns the previous value at a certain address • Useful for grabbing variable amounts of data from the list CUDA Threads and Atomics – Slide 7

  8. Example: Workqueue CUDA Threads and Atomics – Slide 8

  9. Compare and Swap • if compare equals old value stored at address then val is stored instead • in either case, routine returns the value of old • seems a bizarre routine at first sight, but can be very useful for atomic locks CUDA Threads and Atomics – Slide 9

  10. Compare and Swap (cont.) • Most general type of atomic • Can emulate all others with CAS CUDA Threads and Atomics – Slide 10

  11. Atomics • Atomics are slower than normal load/store • Most of these are associative operations on signed/unsigned integers: • quite fast for data in shared memory • slower for data in device memory • You can have the whole machine queuing on a single location in memory • Atomics unavailable on G80! CUDA Threads and Atomics – Slide 11

  12. Example: Global Min/Max (Naïve) CUDA Threads and Atomics – Slide 12

  13. Example: Global Min/Max (Better) CUDA Threads and Atomics – Slide 13

  14. Global Max/Min • Single value causes serial bottleneck • Create hierarchy of values for more parallelism • Performance will still be slow, so use judiciously CUDA Threads and Atomics – Slide 14

  15. Atomics: Summary • Can’t use normal load/store for inter-thread communication because of race conditions • Use atomic instructions for sparse and/or unpredictable global communication • Decompose data (very limited use of single global sum/max/min/etc.) for more parallelism CUDA Threads and Atomics – Slide 15

  16. Topic 2: Streaming Multiprocessor Execution and Divergence • How a streaming multiprocessor (SM) executes threads • Overview of how a streaming multiprocessor works • SIMT Execution • Divergence CUDA Threads and Atomics – Slide 16

  17. Scheduling Blocks onto SMs • Hardware schedules thread blocks onto available SMs • No guarantee of ordering among thread blocks • Hardware will schedule thread blocks as soon as a previous thread block finished Streaming Multiprocessor Thread Block 5 Thread Block 27 Thread Block 61 Thread Block 2001 CUDA Threads and Atomics – Slide 17

  18. Recall: Warps • A warp = 32 threads launched together • Usually execute together as well Control Control Control Control Control Control ALU ALU ALU ALU ALU ALU Control ALU ALU ALU ALU ALU ALU CUDA Threads and Atomics – Slide 18

  19. Mapping of Thread Blocks • Each thread block is mapped to one or more warps • The hardware schedules each warp independently Thread Block N (128 threads) TB N W1 TB N W2 TB N W3 TB N W4 CUDA Threads and Atomics – Slide 19

  20. Thread Scheduling Example • SM implements zero-overhead warp scheduling • At any time, only one of the warps is executed by SM* • Warps whose next instruction has its inputs ready for consumption are eligible for execution • Eligible warps are selected for execution on a prioritized scheduling policy • All threads in a warp execute the same instruction when selected CUDA Threads and Atomics – Slide 20

  21. Control Flow Divergence • Threads are executed in warps of 32, with all threads in the warp executing the same instruction at the same time • What happens if you have the following code? CUDA Threads and Atomics – Slide 21

  22. Control Flow Divergence (cont.) • This is called warp divergence – CUDA will generate correct code to handle this, but to understand the performance you need to understand what CUDA does with it CUDA Threads and Atomics – Slide 22

  23. Branch Path A Path B Control Flow Divergence (cont.) Branch Path A Path B From Fung et al. MICRO ’07 CUDA Threads and Atomics – Slide 23

  24. Control Flow Divergence (cont.) • Nested branches are handled as well CUDA Threads and Atomics – Slide 24

  25. Control Flow Divergence (cont.) Branch Branch Branch Path A Path B Path C CUDA Threads and Atomics – Slide 25

  26. Control Flow Divergence (cont.) • You don’t have to worry about divergence for correctness • Mostly true, except corner cases (for example intra-warp locks) • You might have to think about it for performance • Depends on your branch conditions CUDA Threads and Atomics – Slide 26

  27. Control Flow Divergence (cont.) • One solution: NVIDIA GPUs have predicated instructions whichare carried out only if a logical flag is true. • In the previous example, all threads compute the logical predicate and two predicated instructions CUDA Threads and Atomics – Slide 27

  28. Control Flow Divergence (cont.) • Performance drops off with the degree of divergence CUDA Threads and Atomics – Slide 28

  29. Control Flow Divergence (cont.) • Performance drops off with the degree of divergence • In worst case, effectively lose a factor of 32 in performance if one thread needs expensive branch, while rest do nothing • Another example: processing a long list of elements where, depending on run-time values, a few require very expensive processing GPU implementation: • first process list to build two sub-lists of “simple” and “expensive” elements • then process two sub-lists separately CUDA Threads and Atomics – Slide 29

  30. Synchronization • Already introduced __syncthreads(); which forms a barrier – all threads wait until every one has reached this point. • When writing conditional code, must be careful to make sure that all threads do reach the __syncthreads(); • Otherwise, can end up in deadlock CUDA Threads and Atomics – Slide 30

  31. Synchronization (cont.) • Fermi supports some new synchronisation instructions which are similar to __syncthreads() but have extra capabilities: • int __syncthreads_count(predicate) • counts how many predicates are true • int __syncthreads_and(predicate) • returns non-zero (true) if all predicates are true • int __syncthreads_or(predicate) • returns non-zero (true) if any predicate is true CUDA Threads and Atomics – Slide 31

  32. Warp voting • There are similar warp voting instructions which operate at the level of a warp: • int __all(predicate) • returns non-zero (true) if all predicates in warp are true • int __any(predicate) • returns non-zero (true) if any predicate is true • unsigned int __ballot(predicate) • sets nth bit based on nth predicate CUDA Threads and Atomics – Slide 32

  33. Topic 3: Locks • Use very judiciously • Always include a max_iter in your spinloop! • Decompose your data and your locks CUDA Threads and Atomics – Slide 33

  34. Example: Global atomic lock • Problem: when a thread writes data to device memory the order of completion is not guaranteed, so global writes may not have completed by the time the lock is unlocked CUDA Threads and Atomics – Slide 34

  35. Example: Global atomic lock (cont.) CUDA Threads and Atomics – Slide 35

  36. Example: Global atomic lock (cont.) • __threadfence_block(); • wait until all global and shared memory writes are visible to all threads in block • __threadfence(); • wait until all global and shared memory writes are visible to all threads in block (or all threads, for global data) CUDA Threads and Atomics – Slide 36

  37. Summary • lots of esoteric capabilities – don’t worry about most of • them • essential to understand warp divergence – can have a • very big impact on performance • __syncthreads(); is vital • the rest can be ignored until you have a critical need – then read the documentation carefully and look for examples in the SDK CUDA Threads and Atomics – Slide 37

  38. End Credits • Based on original material from • Oxford University: Mike Giles • Stanford University • Jared Hoberock, David Tarjan • Revision history: last updated 8/8/2011. CUDA Threads and Atomics – Slide 38

More Related