The tlbtree from ypluo

tlbtree's People

Contributors

Stargazers

Watchers

tlbtree's Issues

Array bound problem in datagen.cc

Hello, I'm trying to run workload generated by /Single/test/gen.sh, but when runing workload1.txt (Random RO) I got output:

somthing not found

With some debugging codes added, I found an impossible line in workload1.txt:

0 0

After some more digging, I believe codes below caused the bug

TLBtree/Single/test/datagen.cc

Lines 115 to 129 in 8f95ad7

 void gen_workload(int64_t *arr, int64_t scale, QueryType * querys, WorkloadType w) { 

 std::mt19937 gen(getRandom()); 

 std::uniform_int_distribution<int64_t> idx1_dist(0, scale); 

 zipfian_int_distribution<int64_t> idx2_dist(0, scale, w.skewness); 

 OperationGenerator op_gen(w); 

 for(int i = 0; i < w.operations; i++) { 

 OperationType op = op_gen.next(); 

 int idx = (w.dist == RAND ? idx1_dist(gen) : idx2_dist(gen)); 

 // for insert operations, we should make sure the key does not exist in the dataset 

 int64_t key = (op == OperationType::INSERT ? arr[idx] + getRandom(): arr[idx]); 

 querys[i] = {op, key}; 

 } 

 }

According to cppreference for uniform_int_distribution (https://en.cppreference.com/w/cpp/numeric/random/uniform_int_distribution) and some simple test for zipfian_int_distribution, I found both distributions parameters a, b means [a, b] instead of [a, b), for which idx1_dist(0, scale); may lead to a return of scale and the bug appear for index out of bound.
It seems that modifying scale to scale - 1 can fix it.

ypluo / tlbtree Goto Github PK

tlbtree's People

Contributors

Stargazers

Watchers

Forkers

tlbtree's Issues

Array bound problem in datagen.cc

rebuilding issue

Concurrent preload

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent

	void gen_workload(int64_t arr, int64_t scale, QueryType querys, WorkloadType w) {
	std::mt19937 gen(getRandom());
	std::uniform_int_distribution<int64_t> idx1_dist(0, scale);
	zipfian_int_distribution<int64_t> idx2_dist(0, scale, w.skewness);
	OperationGenerator op_gen(w);

	for(int i = 0; i < w.operations; i++) {
	OperationType op = op_gen.next();
	int idx = (w.dist == RAND ? idx1_dist(gen) : idx2_dist(gen));
	// for insert operations, we should make sure the key does not exist in the dataset
	int64_t key = (op == OperationType::INSERT ? arr[idx] + getRandom(): arr[idx]);

	querys[i] = {op, key};
	}
	}