HazyResearch/ThunderKittens issues and pull requests

#58 - Refactor complex code into separate files + add old kernel versions

Pull Request - State: open - Opened by ArjunParthasarathy 11 days ago

#57 - Could you provide a valid mirror?

Issue - State: open - Opened by ziyuhuang123 12 days ago - 1 comment

#56 - cannot find -lcuda: No such file or directory

Issue - State: open - Opened by ziyuhuang123 12 days ago

#55 - Could you provide a gemm kernel?

Issue - State: open - Opened by ziyuhuang123 12 days ago

#54 - h100.cu(97): error: "wait" is ambiguous

Issue - State: open - Opened by ziyuhuang123 13 days ago - 1 comment

#53 - Fix typos.

Pull Request - State: closed - Opened by jasondavies 22 days ago

#52 - Add complex tile functionality to test TK FFT kernels

Pull Request - State: closed - Opened by ArjunParthasarathy 23 days ago

#51 - Merge 32x32 FFTConv Kernel into TK

Pull Request - State: closed - Opened by ArjunParthasarathy 25 days ago

#50 - When will ThunderKittens support AMD GPUs, specifically the W7900?

Issue - State: open - Opened by lahmuller about 1 month ago

#49 - Build infra

Pull Request - State: closed - Opened by benjaminfspector about 1 month ago

#48 - Confusing Comment in rt.cuh

Issue - State: open - Opened by KAOZUOI about 1 month ago

#47 - Redo mbar

Pull Request - State: closed - Opened by benjaminfspector 2 months ago - 1 comment

#46 - Swizzy

Pull Request - State: closed - Opened by benjaminfspector 2 months ago - 1 comment

#45 - c++20 does not work?

Issue - State: open - Opened by ziyuhuang123 2 months ago - 1 comment

#44 - Support for global load/store padding

Issue - State: open - Opened by Hprairie 3 months ago

#43 - Template error

Issue - State: open - Opened by Hprairie 3 months ago - 1 comment

#42 - Cross-GPU portability

Issue - State: open - Opened by janEbert 3 months ago

#41 - Is it possible to support non-contiguous input tensor ?

Issue - State: open - Opened by ProHuper 3 months ago

#40 - Error running make

Issue - State: closed - Opened by BurhanUlTayyab 3 months ago

#39 - Support `softmax_scale` and `dropout` options for fwd_attend_ker_dim ?

Issue - State: closed - Opened by ProHuper 3 months ago

#38 - Support `softmax_scale` and `dropout` options for fwd_attend_ker_dim ?

Issue - State: closed - Opened by ProHuper 3 months ago

#37 - [bug report][4090 attn] cudaCheckError(): too many resources requested for launch

Issue - State: open - Opened by kexve 4 months ago - 1 comment

#36 - TK kernelize M1 for loop body

Pull Request - State: closed - Opened by LeoXinhaoLee 4 months ago

#35 - Support for TPUs?

Issue - State: open - Opened by jaanli 4 months ago

#34 - fix async bug

Pull Request - State: closed - Opened by xiayuqing0622 4 months ago

#33 - [bug report] h100 attn_causal kernel

Issue - State: open - Opened by xiayuqing0622 4 months ago - 3 comments

#32 - attn_bias rel-pos support to the FAv2 example

Issue - State: closed - Opened by vadimkantorov 4 months ago - 1 comment

#31 - add suport for a100 atten

Issue - State: closed - Opened by MichoChan 4 months ago

#30 - why there is no zero(attn) before compute [email protected] in h100 example?

Issue - State: closed - Opened by xiayuqing0622 4 months ago - 2 comments

#29 - added tma reductions

Pull Request - State: closed - Opened by Aaryan0404 4 months ago - 1 comment

#29 - added tma reductions

Pull Request - State: closed - Opened by Aaryan0404 4 months ago - 1 comment

#28 - [feat] add simple half gemm example

Pull Request - State: closed - Opened by luliyucoordinate 4 months ago

#27 - Load with ldmatrix

Issue - State: open - Opened by liyanc 4 months ago - 2 comments

#26 - Add support for head dimension 128

Issue - State: open - Opened by perklet 4 months ago - 4 comments

#25 - Two questions

Issue - State: closed - Opened by dongrixinyu 4 months ago - 1 comment

#24 - fix a small typo

Pull Request - State: closed - Opened by lancerts 4 months ago - 1 comment

#23 - [Feature Request] GEMM benchmarks and FP8 Support

Issue - State: open - Opened by jwfromm 4 months ago - 7 comments

#22 - unable to reproduce attn_causal speeds

Issue - State: closed - Opened by 152334H 4 months ago - 3 comments

#21 - [Question] Supported compute capabilities?

Issue - State: open - Opened by bayley 4 months ago - 3 comments

#20 - Based gen

Pull Request - State: closed - Opened by simran-arora 4 months ago

#19 - Causal

Pull Request - State: closed - Opened by Aaryan0404 4 months ago

#18 - Cutlass parity

Pull Request - State: closed - Opened by benjaminfspector 5 months ago

#17 - Kvstates

Pull Request - State: closed - Opened by simran-arora 5 months ago

#16 - Descriptor refactor

Pull Request - State: closed - Opened by benjaminfspector 5 months ago - 1 comment

#15 - Binding

Pull Request - State: closed - Opened by simran-arora 5 months ago

#14 - conda setup docs

Pull Request - State: closed - Opened by jordan-benjamin 5 months ago

#13 - add based upate step

Pull Request - State: closed - Opened by simran-arora 5 months ago

#12 - Cooperative refactor

Pull Request - State: closed - Opened by benjaminfspector 5 months ago - 1 comment