Diff
Modified: trunk/PerformanceTests/ChangeLog (173528 => 173529)
--- trunk/PerformanceTests/ChangeLog 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/ChangeLog 2014-09-11 19:33:47 UTC (rev 173529)
@@ -1,3 +1,38 @@
+2014-09-11 Geoffrey Garen <[email protected]>
+
+ Some MallocBench refinements
+ https://bugs.webkit.org/show_bug.cgi?id=136750
+
+ Reviewed by Sam Weinig.
+
+ * MallocBench/MallocBench/Interpreter.cpp:
+ (Interpreter::run): Allow for null entries in the object list so that
+ we can test in modes that exclude large or small allocations.
+
+ * MallocBench/MallocBench/churn.cpp:
+ (benchmark_churn):
+ * MallocBench/MallocBench/flickr.cpp:
+ (benchmark_flickr):
+ * MallocBench/MallocBench/fragment.cpp:
+ (benchmark_fragment_iterate):
+ * MallocBench/MallocBench/list.cpp:
+ (benchmark_list_allocate):
+ * MallocBench/MallocBench/reddit.cpp:
+ (benchmark_reddit): Updated test runtimes to weight them more equally,
+ for the sake of arithmetic mean.
+
+ * MallocBench/MallocBench/stress.cpp:
+ (Object::Object):
+ (allocate):
+ (deallocate):
+ (benchmark_stress): Verify the contents of memory as we go. Also,
+ force scavenging each time through the loop to test the scavenging path.
+
+ * MallocBench/MallocBench/theverge.cpp:
+ (benchmark_theverge):
+ * MallocBench/MallocBench/tree.cpp:
+ (benchmark_tree_churn): Re-weighted, as above.
+
2014-09-08 Myles C. Maxfield <[email protected]>
PerformanceTests/SVG/SVG-Text.html has unparsable output
Modified: trunk/PerformanceTests/MallocBench/MallocBench/Interpreter.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/Interpreter.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/Interpreter.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -101,13 +101,15 @@
break;
}
case op_free: {
- assert(m_objects[op.slot].object);
+ if (!m_objects[op.slot].object)
+ continue;
mbfree(m_objects[op.slot].object, m_objects[op.slot].size);
m_objects[op.slot] = { 0, 0 };
break;
}
case op_realloc: {
- assert(m_objects[op.slot].object);
+ if (!m_objects[op.slot].object)
+ continue;
m_objects[op.slot] = { mbrealloc(m_objects[op.slot].object, m_objects[op.slot].size, op.size), op.size };
break;
}
Modified: trunk/PerformanceTests/MallocBench/MallocBench/churn.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/churn.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/churn.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -41,7 +41,7 @@
void benchmark_churn(bool isParallel)
{
- size_t times = 10000000;
+ size_t times = 7000000;
if (isParallel)
times /= cpuCount();
Modified: trunk/PerformanceTests/MallocBench/MallocBench/flickr.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/flickr.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/flickr.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -44,7 +44,7 @@
void benchmark_flickr(bool isParallel)
{
- size_t times = 1;
+ size_t times = 3;
Interpreter interpreter("flickr.ops");
for (size_t i = 0; i < times; ++i)
Modified: trunk/PerformanceTests/MallocBench/MallocBench/fragment.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/fragment.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/fragment.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -110,7 +110,7 @@
void benchmark_fragment_iterate(bool isParallel)
{
size_t nodeCount = 512 * 1024;
- size_t times = 32;
+ size_t times = 20;
if (isParallel)
nodeCount /= cpuCount();
size_t replaceCount = nodeCount / 4;
Modified: trunk/PerformanceTests/MallocBench/MallocBench/list.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/list.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/list.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -99,7 +99,7 @@
void benchmark_list_allocate(bool isParallel)
{
Node* head = 0;
- size_t times = 96;
+ size_t times = 70;
size_t nodes = 32 * 1024;
if (isParallel) {
nodes /= cpuCount();
Modified: trunk/PerformanceTests/MallocBench/MallocBench/reddit.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/reddit.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/reddit.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -44,7 +44,7 @@
void benchmark_reddit(bool isParallel)
{
- size_t times = 1;
+ size_t times = 6;
Interpreter interpreter("reddit.ops");
for (size_t i = 0; i < times; ++i)
Modified: trunk/PerformanceTests/MallocBench/MallocBench/stress.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/stress.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/stress.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -39,14 +39,16 @@
static const size_t MB = kB * kB;
struct Object {
- Object(void* pointer, size_t size)
+ Object(void* pointer, size_t size, long uuid)
: pointer(pointer)
, size(size)
+ , uuid(uuid)
{
}
void* pointer;
size_t size;
+ long uuid;
};
class SizeStream {
@@ -100,6 +102,24 @@
size_t m_count;
};
+Object allocate(size_t size)
+{
+ Object object(mbmalloc(size), size, random());
+ for (size_t i = 0; i < size / sizeof(long); ++i)
+ (static_cast<long*>(object.pointer))[i] = object.uuid;
+ return object;
+}
+
+void deallocate(const Object& object)
+{
+ for (size_t i = 0; i < object.size / sizeof(long); ++i) {
+ if ((static_cast<long*>(object.pointer))[i] != object.uuid)
+ abort();
+ }
+
+ mbfree(object.pointer, object.size);
+}
+
void benchmark_stress(bool isParallel)
{
const size_t heapSize = 100 * MB;
@@ -112,18 +132,17 @@
SizeStream sizeStream;
- size_t lastSize = 0;
- for (size_t remaining = heapSize; remaining; remaining -= std::min(remaining, lastSize)) {
- lastSize = sizeStream.next();
- Object object(mbmalloc(lastSize), lastSize);
- objects.push_back(object);
+ size_t size = 0;
+ for (size_t remaining = heapSize; remaining; remaining -= std::min(remaining, size)) {
+ size = sizeStream.next();
+ objects.push_back(allocate(size));
}
for (size_t i = 0; i < churnCount; ++i) {
std::vector<Object> objectsToFree;
- for (size_t remaining = churnSize; remaining; remaining -= std::min(remaining, lastSize)) {
- lastSize = sizeStream.next();
- Object object(mbmalloc(lastSize), lastSize);
+ for (size_t remaining = churnSize; remaining; remaining -= std::min(remaining, size)) {
+ size = sizeStream.next();
+ Object object = allocate(size);
size_t index = random() % objects.size();
objectsToFree.push_back(objects[index]);
@@ -131,7 +150,9 @@
}
for (auto& object : objectsToFree)
- mbfree(object.pointer, object.size);
+ deallocate(object);
+
+ mbscavenge();
}
for (auto& object : objects)
Modified: trunk/PerformanceTests/MallocBench/MallocBench/theverge.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/theverge.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/theverge.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -44,7 +44,7 @@
void benchmark_theverge(bool isParallel)
{
- size_t times = 1;
+ size_t times = 3;
Interpreter interpreter("theverge.ops");
for (size_t i = 0; i < times; ++i)
Modified: trunk/PerformanceTests/MallocBench/MallocBench/tree.cpp (173528 => 173529)
--- trunk/PerformanceTests/MallocBench/MallocBench/tree.cpp 2014-09-11 19:26:16 UTC (rev 173528)
+++ trunk/PerformanceTests/MallocBench/MallocBench/tree.cpp 2014-09-11 19:33:47 UTC (rev 173529)
@@ -207,7 +207,7 @@
void benchmark_tree_churn(bool isParallel)
{
- size_t times = 160;
+ size_t times = 130;
size_t depth = 15;
if (isParallel) {
times *= 4;
