DeepSeek Open-Sources TileKernels, GPU Kernel Library for Large Model Training and Inference

Gate News message, April 23 — DeepSeek has open-sourced TileKernels under the MIT license, a GPU kernel library written in TileLang for large language model training and inference. TileLang is a domain-specific language developed by the tile-ai team for expressing high-performance GPU kernels in Python. DeepSeek stated that most kernels in the library have approached hardware performance limits in compute density and memory bandwidth, with portions already deployed in internal training and inference operations.

The library comprises six categories of kernels: MoE (mixture of experts) gating and routing, including Top-k expert selection, token-to-expert mapping, and fused expand/shrink with weight normalization; quantization supporting FP8, FP4, and E5M6 formats with per-token, per-block, and per-channel quantization, including fused SwiGLU+quantization operations; batch transpose; Engram gating with fused RMSNorm forward/backward propagation and weight gradient reduction; Manifold HyperConnection with Sinkhorn normalization and mixed split/apply; and high-level autograd interfaces that wrap low-level kernels into trainable layers.

Engram and Manifold HyperConnection are proprietary components of DeepSeek’s model architecture, with implementation details disclosed publicly for the first time. The library requires NVIDIA SM90 or SM100 architecture GPUs (H100/H200 or Blackwell series), CUDA Toolkit 13.1 or higher, and PyTorch 2.10 or higher.

Disclaimer: The information on this page may come from third parties and does not represent the views or opinions of Gate. The content displayed on this page is for reference only and does not constitute any financial, investment, or legal advice. Gate does not guarantee the accuracy or completeness of the information and shall not be liable for any losses arising from the use of this information. Virtual asset investments carry high risks and are subject to significant price volatility. You may lose all of your invested principal. Please fully understand the relevant risks and make prudent decisions based on your own financial situation and risk tolerance. For details, please refer to Disclaimer.

Related Articles

XRP Breakout Holds as XRPL Lending Vote Gains Momentum

Key Insights XRP maintains weekly strength above major cryptocurrencies as price holds above key EMAs, reflecting sustained momentum despite minor daily declines in trading sessions. XRPL validators advance lending upgrades through XLS-65 and XLS-66, introducing pooled liquidity vaults and f

CryptoNewsLand1h ago

XRP Breakout Holds as XRPL Lending Vote Gains Momentum

XRP shows weekly strength, trading above EMAs after breaking from a descending wedge; XRPL advances XLS-65/66 lending upgrades with pooled vaults and fixed-term loans; derivatives rise in volume, open interest, and options activity. Abstract: This report notes XRP's persistent weekly momentum and price strength above key moving averages following a breakout from a descending wedge. It covers XRPL validators voting on XLS-65 and XLS-66, enabling native lending, pooled liquidity vaults, and fixed-term loans to expand on-chain financial activity. It also reports rising derivatives participation, with higher trading volume, open interest, and a surge in options activity, suggesting increasing trader positioning for a continued breakout.

CryptoNewsLand1h ago

MagicBlock Launches Mirage, Command-Line Privacy Payment Tool for Solana

Gate News message, April 23 — MagicBlock has released Mirage, a command-line privacy payment tool designed for the Solana network. The tool enables users to create wallets, deposit funds, and send private transactions through terminal commands, bots, or AI agents. Mirage is built on Private

GateNews2h ago

Aave Suspends rsETH Reserve Operations Across Ethereum, Arbitrum, and Other Networks to Mitigate Systemic Risk

Gate News message, April 23 — Aave announced it has suspended rsETH reserve-related operations across Ethereum mainnet, Arbitrum, Base, Mantle, and Linea networks to mitigate systemic risk during the asset recovery process. The measure aims to preserve additional funds while the recovery plan is

GateNews2h ago

Morpho Association to Execute Series of Wallet Transfers Over Coming Weeks for Fund Management

Gate News message, April 23 — Morpho Association will conduct a series of transfers from its association-owned wallet over the coming weeks as part of routine fund management operations.

GateNews3h ago

Circle Chief Economist Proposes Raising USDC Rates on Aave Amid KelpDAO Fallout

Gate News message, April 23 — Gordon Liao, Circle's chief economist, has proposed raising USDC lending parameters on Aave v3 Ethereum Core this week following a $292 million KelpDAO rsETH exploit that triggered a liquidity crisis across the protocol. Liao's Request for Comment suggests increasing "S

GateNews3h ago
Comment
0/400
No comments