浏览代码

Preventing the threads from writing into memory in the inner loop yields the expected speedup

Especially bad is writing into memory where another thread is writing
into the same cache line.
Ian Goldberg 1 年之前
父节点
当前提交
0e1836d604
共有 1 个文件被更改,包括 6 次插入3 次删除
  1. 6 3
      online.cpp

+ 6 - 3
online.cpp

@@ -314,13 +314,16 @@ static value_t parallel_streameval_rdpf(MPCIO &mpcio, const RDPF &dpf,
             [&mpcio, &dpf, &scaled_xor, thread_num, threadstart, threadsize] {
                 MPCTIO tio(mpcio, thread_num);
 //printf("Thread %d from %X for %X\n", thread_num, threadstart, threadsize);
-                auto ev = StreamEval(dpf, threadstart, 0, tio.aes_ops());
+                RegXS local_xor;
+                size_t local_aes_ops = 0;
+                auto ev = StreamEval(dpf, threadstart, 0, local_aes_ops);
                 for (address_t x=0;x<threadsize;++x) {
 //if (x%0x10000 == 0) printf("%d", thread_num);
                     DPFnode leaf = ev.next();
-                    RegXS sx = dpf.scaled_xs(leaf);
-                    scaled_xor[thread_num] ^= sx;
+                    local_xor ^= dpf.scaled_xs(leaf);
                 }
+                scaled_xor[thread_num] = local_xor;
+                tio.aes_ops() += local_aes_ops;
 //printf("Thread %d complete\n", thread_num);
             });
         threadstart = (threadstart + threadsize) % totsize;