Max Krasnyansky
PulseAugur coverage of Max Krasnyansky — every cluster mentioning Max Krasnyansky across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
llama.cpp optimizes hexagon architecture with fused RMS_NORM operations
The llama.cpp project has released version b12181, which includes optimizations for the hexagon architecture. These changes focus on fusing post-norm residual additions into the RMS_NORM_MUL operation and refactoring RM…
-
llama.cpp adds DSV4_HC_PRE and DSV4_HC_POST operations
The llama.cpp project has released version b12092, introducing new operations for DSV4_HC_PRE and DSV4_HC_POST. These updates include optimizations for double-buffered DMA, precomputation of kernel parameters, and suppo…
-
llama.cpp b12030 fixes hexagon DMA transfer issues across platforms
The llama.cpp project has released version b12030, which includes a fix for DMA transfers on the hexagon architecture when strides exceed 24 bits. This issue could lead to truncated strides and incorrect addresses for l…
-
llama.cpp b11970 adds Hexagon conv2d support across platforms
The llama.cpp project has released version b11970, which includes significant enhancements to its Hexagon backend, specifically adding support for the conv2d operation. This update refactors the conv2d kernel parameters…
-
llama.cpp adds Q1_0 and Q2_0 quantization support for Hexagon backend
The llama.cpp project has released an update, b11716, which introduces support for Q1_0 and Q2_0 quantization formats. This update includes optimizations for the hexagon backend, specifically enhancing the HMX MUL_MAT p…
-
llama.cpp fixes Hexagon IM2COL DMA overflow bug
The llama.cpp project released version b11516, which addresses an Integer-to-Integer (IM2COL) patch-embed Direct Memory Access (DMA) ring overflow issue. This overflow occurred when the system attempted to queue more DM…
-
llama.cpp project releases b11434 with hexagon and ssm-conv updates
The llama.cpp project has released version b11434, which includes updates to its hexagon and ssm-conv components. These updates focus on improving the efficiency and structure of DMA operations for prefill and decode pr…
-
llama.cpp adds multi-device model splitting in b10920 release
The llama.cpp project has released version b10920, introducing support for multi-device model splitting, also known as row-splitting. This enhancement allows for more efficient distribution of model computations across …