Skip to content

issue/979 optimize paged attention #980

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
PanZezhong1725 wants to merge 3 commits into
base: main
Choose a base branch
from
Draft

issue/979 optimize paged attention #980

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
cdf8fa2
issue/979 optimize paged attention
PanZezhong1725 Jan 23, 2026
35d1524
issue/979 - removed commented paged attn codes
wooway777 Jan 26, 2026
57d6e8d
issue/983 - adapted the optimized paged attention to metax
wooway777 Jan 26, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions src/infiniop/devices/metax/metax_ht2mc.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -85,4 +85,7 @@
#define hcclSuccess mcclSuccess
#define hcclCommDestroy mcclCommDestroy
#define hcclAllReduce mcclAllReduce
#define hcGetDevice mcGetDevice
#define hcDeviceAttributeMultiProcessorCount mcDeviceAttributeMultiProcessorCount
#define hcDeviceGetAttribute mcDeviceGetAttribute
#endif
6 changes: 6 additions & 0 deletions src/infiniop/devices/metax/metax_kernel_common.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,12 @@ using cuda_bfloat16 = hpcc_bfloat16;
using cuda_bfloat162 = hpcc_bfloat162;
using cuda_fp8_e4m3 = __hpcc_fp8_e4m3;

#ifdef ENABLE_METAX_MC_API
using __nv_bfloat16 = __maca_bfloat16;
#else
using __nv_bfloat16 = __hpcc_bfloat16;
#endif

namespace device::metax {

// get the memory offset of the given element in a tensor given its flat index
Expand Down
Loading