Searched refs:benefits (Results 1 – 25 of 167) sorted by relevance
1234567
61 llvm::SmallDenseMap<const Pattern *, PatternBenefit> benefits; in applyCostModel() local62 auto cmp = [&benefits](const Pattern *lhs, const Pattern *rhs) { in applyCostModel()63 return benefits[lhs] > benefits[rhs]; in applyCostModel()80 benefits.clear(); in applyCostModel()82 benefits.try_emplace(pat, model(*pat)); in applyCostModel()87 while (!list.empty() && benefits[list.back()].isImpossibleToMatch()) { in applyCostModel()
16 name: "benefits"60 name: "benefits"111 name: "benefits"
16 name: "benefits"60 name: "benefits"
23 name: "benefits"25 Parameter benefits updated by the MDL Adagrad Light optimization algorithm.
23 name: "benefits"25 Value of benefits used in the MDL Adagrad Light optimization algorithm.
46 However, `broadcast_to` does not carry with it any such benefits.
3 ; Make sure that vectors get the same benefits as scalars when using unsafe-fp-math.
8 **Here's why we have this policy:** We want to focus on the work that benefits
13 … TensorFlow developers respond to issues. We want to focus on work that benefits the whole communi…
4 ; Make sure that vectors get the same benefits as scalars when using unsafe-fp-math.
23 All options are key-values. There are both drawbacks and benefits to this.25 information might seem redundant. The main benefits is scalability and code
24 * that everyone benefits.
24 which reduces the benefits of laziness).**
10 the benefits of resolving the issues above. We expect to make CFStream the default networking
36 For the benefits of packaging tool authors and users seeking a deeper
38 For the benefits of packaging tool authors and users seeking a deeper
28 You can leverage all benefits the Task Library infrastructure provides and
11 ; only duplicate code with no benefits.
31 The on-device model comes with several benefits. It is:
51 benefits, I think this is a very reasonable tradeoff.
7 ; registers in unwind info. Since there are benefits to aligned stores, it's