Dan Luu

According to the Travis docs, they'll give us 3GB. In addition to reducing the total amount of memory allocated, I also removed the `-j4` flag on tests so that tests...

Now that I think about it, I don't think this should be that hard. `gtest` lets you filter tests by name with `--gtest_filter=POSTIVE_PATTERNS[-NEGATIVE_PATTERNS]` We can have a convention that puts...

In cmake, we can give it `-E` to exclude some tests. See https://cmake.org/Wiki/CMake/Testing_With_CTest for more information.

I think this means that, until we fix the issue where we can't handle rows that aren't as wide as a qword, we have to pass either a "max rank"...

Oh, this looks like it's probably trivial. I think it's that we exceed the maximum number of allowed rows. The current value seems like an arbitrary number, but because of...

The core of this is fixed in 5be82ec450e12bdd7467468f3883a8c7d24b2956, but we then crash on TermTableBuild with a density of 0.35 because we exceed: ~~~ static const size_t c_log2MaxRowCount = 4; static...

Each `RowConfiguration::Entry` has 8 bits, which are used as follows: ~~~ m_data = static_cast(rowCount | (rank

With the current test setup (8-core single socket Xeon-D), we see a ~20% speedup from HT. However, in this setup we also see that compilation consumes 3/4 of execution time....

Additionally, `drd` (data race detector) fails with: ~~~ [ RUN ] TokenManager.Basic ==31961== not a condition variable: cond 0xffefffe10 ==31961== at 0x4C38D24: pthread_cond_destroy@* (in /usr/lib/valgrind/vgpreload_drd-amd64-linux.so) ==31961== by 0x4639A8: BitFunnel::TokenManager::~TokenManager() (in...

The non-drd valgrind failures listed above were fixed after bf9b06ff3d4d7205ba8d5042367bf358e1c24308 and d2db7088aabdec7b788fda72e560e581603a7b64. However, we still have drd failures such as: ~~~ [ RUN ] TokenManager.Basic ==26202== not a condition variable:...