Age | Commit message (Collapse) | Author | |
---|---|---|---|
2016-11-03 | making dot to have an accreal return type (consistent with CPU)dotfix | soumith | |
2016-11-02 | TH_INDEX_BASE for nonzero | soumith | |
2016-11-01 | Merge pull request #577 from dbolkensteyn/mathscan | Soumith Chintala | |
CMake: Install generic/THCTensorMathScan.h | |||
2016-11-01 | CMake: Install generic/THCTensorMathScan.h | Dinesh Bolkensteyn | |
2016-11-01 | Merge pull request #576 from torch/distfix | Soumith Chintala | |
dist, cumsum, cumprod for multiple types | |||
2016-11-01 | adding multiple types for distdistfix | soumith | |
2016-11-01 | adding multiple types for cumsum, cumprod | soumith | |
2016-11-01 | Merge pull request #575 from torch/powtrace | Soumith Chintala | |
adding multiple types for pow, trace, diag, tril, triu | |||
2016-11-01 | adding multiple types for pow, trace, diag, tril, triu | soumith | |
2016-11-01 | Merge pull request #574 from torch/nonzero | Soumith Chintala | |
implement torch.nonzero | |||
2016-11-01 | implement torch.nonzero | soumith | |
2016-10-28 | Merge pull request #573 from gchanan/hdotCuda8 | Soumith Chintala | |
Add hdot support for CUDA 8. | |||
2016-10-28 | Add hdot support for CUDA 8. | Gregory Chanan | |
If not compiled with CUDA 8+, an error is raised indicating that CUDA 8.0+ is required. | |||
2016-10-27 | Merge pull request #572 from gchanan/doubleTensorMul | Soumith Chintala | |
Enable '*' for CudaDoubleTensors. | |||
2016-10-27 | Enable '*' for CudaDoubleTensors. | Gregory Chanan | |
2016-10-26 | Merge pull request #569 from colesbury/void | Soumith Chintala | |
Use void for no-arg functions | |||
2016-10-26 | Use 'void' for no-arg functions | Sam Gross | |
2016-10-26 | pushing THCState back to the headerthcstateheader | soumith | |
2016-10-25 | Merge pull request #566 from torch/cachedevice | Soumith Chintala | |
Store the device in THCStorage | |||
2016-10-25 | Store the device in THCStorage | soumith | |
2016-10-25 | Merge pull request #565 from torch/allocatorupdates | Soumith Chintala | |
allow storages to use custom allocators | |||
2016-10-25 | Merge pull request #562 from howard0su/master | Soumith Chintala | |
we can use _NPROCESSOR_CONF instead of a hack for tegra | |||
2016-10-25 | allocator updates | soumith | |
2016-10-25 | some bugfixes for THC | soumith | |
2016-10-25 | Merge branch 'master' into my | Howard Su | |
2016-10-25 | we can use _NPROCESSOR_CONF instead of a hack for tegra | Howard Su | |
2016-10-20 | Merge pull request #558 from gchanan/genericDeviceTensorUtils | Soumith Chintala | |
Add generic type support for toDeviceTensor. | |||
2016-10-20 | Add generic type support for toDeviceTensor. | Gregory Chanan | |
2016-10-18 | correct input types to lua_pushboolean | soumith | |
2016-10-17 | gcc 5 + cuda < 8 workaround improved | Soumith Chintala | |
2016-10-17 | Merge pull request #555 from colesbury/streams | Soumith Chintala | |
Fix cutorch.getStream() | |||
2016-10-17 | Fix cutorch.getStream() | Sam Gross | |
state->numUserStreams does not include the NULL stream, which is stored in res->streams[i] | |||
2016-10-17 | guards for half | Soumith Chintala | |
2016-10-17 | Merge pull request #554 from torch/catmultiple | Soumith Chintala | |
torch.cat for multiple cuda types | |||
2016-10-17 | torch.cat for multiple cuda types | Soumith Chintala | |
2016-10-17 | Update cutorch-1.0-0.rockspec | Soumith Chintala | |
2016-10-17 | Merge pull request #550 from colesbury/streams | Soumith Chintala | |
Add stream API that is not based on indices | |||
2016-10-16 | Merge pull request #548 from BTNC/win-msvc | Soumith Chintala | |
make cunn compile with msvc && fix compilation failure for linux/mac os | |||
2016-10-15 | Add stream API that is not based on indices | Sam Gross | |
This implements the THC code so that we can expose streams as objects instead of simply referring to them by indices. This is not exposed in Lua yet. | |||
2016-10-14 | Merge pull request #549 from colesbury/allocator | Soumith Chintala | |
Fix caching allocator when used from multiple Lua threads | |||
2016-10-14 | Fix caching allocator when used from multiple Lua threads | Sam Gross | |
Use a single, global THCCachingAllocator instance. Previously, each Lua thread had its own THCCachingAllocator instance. However, threads can share storages, which means a segment could be allocated from on THCCachingAllocator and freed on another, which breaks. Fixes #539 | |||
2016-10-14 | make cunn compile with msvc && fix compilation failure for linux/mac os | Rui Guo | |
2016-10-14 | adding hasHalfInstructions and hasFastHalfInstructions exposed to lua | soumith | |
2016-10-14 | Merge pull request #547 from torch/p100fix | Soumith Chintala | |
removing CUDA_HALF_INSTRUCTIONS and enabling hgemm only for P100 | |||
2016-10-14 | removing CUDA_HALF_INSTRUCTIONS and enabling hgemm only for P100p100fix | soumith | |
2016-10-14 | Merge pull request #546 from torch/revert-543-win-msvc | Soumith Chintala | |
Revert "changes to make cunn compile on windows with msvc" | |||
2016-10-14 | Revert "changes to make cunn compile on windows with msvc" | Soumith Chintala | |
2016-10-13 | Merge pull request #543 from BTNC/win-msvc | Soumith Chintala | |
changes to make cunn compile on windows with msvc | |||
2016-10-13 | changes to make cunn compile on windows with msvc | Rui Guo | |
2016-10-13 | Merge pull request #542 from gchanan/atomicAddStaticInline | Soumith Chintala | |
Make atomicAdd functions static inline. |