deepseek-v4.1POST

Porting TensorFold to run V4.1 Flash on dual A6000 TP=2 with a DGX Spark

The author is porting TensorFold to dual Ampere A6000s in TP=2, keeping experts on the GPUs and other weights on a DGX Spark, with 384GB RAM and 96GB VRAM and about 75% of the code done.

Jeremy Tregunna

@noozieArchived 6 days agoOriginally posted 2026.10.24

Jeremy Tregunna @noozieOriginally posted 2026.10.24

Open original on X

Spent my evening and now part of my morning porting TensorFold to run on Ampere generation A6000 GPUs in TP=2, while keeping all experts on the GPUs, all other weights on a dgx spark, and yet remaining things in memory on the GPU node. Yeah I'm trying to see how far I can push the hardware I have, and DeepSeek V4.1 Flash 1017 seems like it was built to help me with that. 384GB ram, 48GB x 2 VRAM, 128GB Unified memory and an rdma link between gpu node and spark... About 75% done the code prepping for first test today.

x.com/noozie/status/2114020044443165011Archived 2026.10.25 12:03