Commit Graph

  • b938c320a8 Last touchups before ripping out the DSL Steven Robertson 2010-09-13 12:22:08 -0400
  • e4aac6993f A few touchups Steven Robertson 2010-09-13 00:20:15 -0400
  • e0b218feba A new (somewhat experimental) approach to fusing Steven Robertson 2010-09-12 23:45:38 -0400
  • 5a5fcf5bb9 Fix the unbelieveably stupid bug I've been chasing for days. Steven Robertson 2010-09-12 18:42:52 -0400
  • 2f48d01aa9 Fix linear variation typo Steven Robertson 2010-09-12 17:38:51 -0400
  • 5c5122e8c8 Optimization doubles performance... but breaks the output (even more) Steven Robertson 2010-09-12 17:17:08 -0400
  • d01de61952 Simultaneous occupancy microbenchmark Steven Robertson 2010-09-12 16:23:24 -0400
  • 3e4e1d88a2 Allow device call exceptions to propagate after cleanup Steven Robertson 2010-09-12 16:22:56 -0400
  • 70ca6d7729 Fix RNG test Steven Robertson 2010-09-12 16:22:22 -0400
  • a6141f492d A byte is *8* bits Steven Robertson 2010-09-12 15:48:31 -0400
  • 7ef0d334ca ...except I missed the file that actually contained the new method Steven Robertson 2010-09-12 14:06:07 -0400
  • 6ed8907fcb LaunchContext.get_per_thread Steven Robertson 2010-09-12 13:45:55 -0400
  • 3265982fec Change 'ctx.threads' to 'ctx.nthreads', as it should have been from the start Steven Robertson 2010-09-12 11:13:53 -0400
  • a439bf671d Fix occupancy issues (1 block/SM when shuffle was on). Steven Robertson 2010-09-12 11:09:47 -0400
  • ce0571deaf A fake log filter stage while I work on other stuff Steven Robertson 2010-09-12 02:32:03 -0400
  • c13f6a06cf Experiments with larger CTAs for IterThread Steven Robertson 2010-09-12 02:01:03 -0400
  • e2b1c161cf More readable memory allocations Steven Robertson 2010-09-12 01:13:22 -0400
  • 802ca1d585 Allow swapping out store methods for easier testing of performance Steven Robertson 2010-09-12 01:09:04 -0400
  • f368a99a16 Shuffle points between threads of a CTA Steven Robertson 2010-09-12 00:17:18 -0400
  • 40a5ceafde Use a somewhat better writeback mechanism for now Steven Robertson 2010-09-12 00:16:35 -0400
  • aa688564f1 Add Timeouter, for timing out infinite loops so data can be recovered. Steven Robertson 2010-09-11 13:18:40 -0400
  • a5d7c2cc1a Use variations. This works, but is still fragile. Steven Robertson 2010-09-11 13:15:36 -0400
  • 860d7b2fad Add xforms and variations. Steven Robertson 2010-09-11 13:10:41 -0400
  • 383c0f1f9a Fixed bench.py, with the help of Device Assertions™!* Steven Robertson 2010-09-11 00:16:43 -0400
  • 56404b629f Add device assertions to standard library. Steven Robertson 2010-09-11 00:12:02 -0400
  • 3932412539 Test to make sure floating point numbers were in the right range. Steven Robertson 2010-09-10 19:36:39 -0400
  • e71a8422e5 Make store_per_thread reuse gtid in multiple calls when possible Steven Robertson 2010-09-10 18:45:32 -0400
  • 943e92b80c Use pycuda SourceModule to work around crashes, and a few invocation touchups. Steven Robertson 2010-09-10 18:02:37 -0400
  • c3d12d07c2 Fix MWCRNGTest. Steven Robertson 2010-09-10 18:01:50 -0400
  • 36f1c1c056 Rename "cuburnlib" (stupid) to "cuburn" (stupid but shorter) Steven Robertson 2010-09-10 14:48:34 -0400
  • 4552589b35 Refactor call() to be more elegant Steven Robertson 2010-09-10 14:43:20 -0400
  • fb4e5b75e9 Add support for writing float literals in store_per_thread Steven Robertson 2010-09-10 14:33:56 -0400
  • 6eaa80be7a Added property ctx.warps_per_cta Steven Robertson 2010-09-10 12:53:40 -0400
  • 2f3ac42153 Improved DataStream record format Steven Robertson 2010-09-10 12:53:20 -0400
  • 7e0d36af7d Add performance tuning (maxnregs) to entries Steven Robertson 2010-09-10 12:52:47 -0400
  • 086e4e4fb4 Lots-o-stuff. Steven Robertson 2010-09-09 11:36:14 -0400
  • 1f7b00b61e instmethod decorator: another hack (to get around ctx.ptx.instances[]) Steven Robertson 2010-09-08 13:12:46 -0400
  • 094890c324 Use shared memory for iter_count and have each CP processed by only one CTA. Slower, but the code is a bit simpler conceptually, and the difference will be more than accounted for by better scheduling towards the end of the process. Steven Robertson 2010-09-07 14:54:50 -0400
  • aa065dc25d Add the first of many microbenchmarks Steven Robertson 2010-09-07 12:44:12 -0400
  • db72a7d496 Allow register local name rebinding Steven Robertson 2010-09-06 16:50:54 -0400
  • e03f20392d Switch from to_inject() to object insertion. One less kludge to deal with. Steven Robertson 2010-09-06 16:09:37 -0400
  • ada0fe20c7 Random floats (I think) Steven Robertson 2010-09-06 14:19:06 -0400
  • f3298e0bed Finally runs again Steven Robertson 2010-09-06 11:18:20 -0400
  • 27e7fd82a3 Time to go have nightmares about this code again (no really) Steven Robertson 2010-09-03 00:52:27 -0400
  • 2c26ff9ab6 * Fix deptrace typos * Add predicate support to DeviceStream fetches, making them even uglier * Add store_per_thread to PTX stdlib Steven Robertson 2010-09-03 00:51:23 -0400
  • a68fc064a1 Added TODO, because even in my nightmares I underestimate how much is left Steven Robertson 2010-09-03 00:08:58 -0400
  • 7e7fbda2cc Start of render module Steven Robertson 2010-09-02 17:26:16 -0400
  • a23a493d68 Formatter improvements Steven Robertson 2010-09-02 16:12:22 -0400
  • 731c637f80 DataStream. Completely untested. I want to see the bugfixes in the log. Steven Robertson 2010-09-02 16:11:44 -0400
  • bf79dc7fa0 Adjust softjoin and vec to be easier to format Steven Robertson 2010-09-02 15:05:24 -0400
  • 0a3f9551ad get_gtid docstring Steven Robertson 2010-09-01 23:38:29 -0400
  • 4a80d946db Add comment to DSL Steven Robertson 2010-09-01 23:34:24 -0400
  • 32f68ea1d5 Remove some dead code Steven Robertson 2010-09-01 22:46:55 -0400
  • a3660ec6e4 PTX DSL working, at least well enough to pass MWCRNGTest Steven Robertson 2010-09-01 21:09:40 -0400
  • 5f8c2bbf08 Known broken checkin to show algorias Steven Robertson 2010-09-01 13:02:12 -0400
  • cceb75396f Before I rip out tempita and start a DSL Steven Robertson 2010-08-30 14:45:44 -0400
  • 0c78e972b1 Splitting things up a bit Steven Robertson 2010-08-28 16:56:05 -0400
  • 907cbb273f Code builder, RNG working Steven Robertson 2010-08-28 00:28:00 -0400
  • a23ebdcf5f Initial commit. Steven Robertson 2010-08-27 12:28:02 -0400