Repository Issue Activity (beta)

#138 Question About AMD Infinity Fabric P2P Bandwidth and Its Impact on Training

#133 [mamf-finder] Adding FP4 support

#130 Question

#129 Typo： ROMA -> RDMA

#127 Additional tools for storage: dust

#122 wrong content in model-parallelism?

#120 Inter-node vs Intra-node

#121 Link fail

#115 benchmark v2 gaudi2/gaudi3 results

#113 SkyPilot examples

#109 some thing about the formula

#72 GPU utilization monitoring

#100 About the Maximum Achievable Matmul FLOPS comparison table

#91 RuntimeError: "_tunable_scaled_gemm" not implemented for 'Float8_e4m3fn'

#77 Empirical measurements of sustained bandwidth and flops

#52 Adding another logbook (kinda)

#76 AMD MI250X MAMF efficiency is wrong

#75 How can operations with the INT8 data type be performed using a GPU accelerator card? Is dequantization required?

#74 MAMF Finder usage of MxNxK differs from BLAS gemm MxNxK

#73 PDF link in readme doesn't work

#71 Performance Profiling

#66 [Question] `FSDP` vs `Deepspeed ZeRO3 / ZeRO++`

#67 bb

#68 f