@charles_irl: Added a fun lil widget to the LLM Engineer's Almanac -- a "Token Timing Simulator" so you can get a visceral feel for w…
Summary
A token timing simulator widget was added to the LLM Engineer's Almanac, demonstrating the DFlash technique achieving ~1k TPS, to help users viscerally understand benchmark performance numbers.
View Cached Full Text
Cached at: 05/15/26, 07:07 PM
Added a fun lil widget to the LLM Engineer’s Almanac – a “Token Timing Simulator” so you can get a visceral feel for what a benchmark perf number means. Here’s @dcw02’s latest work with @zhijianliu’s DFlash technique in @sgl_project – ~1k TPS! https://modal.com/llm-almanac/token-timing-simulator…
LLM Engineer’s Almanac - Token Timing Simulator
Source: https://modal.com/llm-almanac/token-timing-simulator TextInput tokensOutput tokens
Similar Articles
Getting a feel for how fast X tokens/second really is.
The author introduces a web-based script designed to help users intuitively understand token-per-second speeds in local LLM setups by simulating text, code, and reasoning generation rates.
@charles_irl: This block quant visualizer is another page in our LLM Engineer's Almanac -- a one-stop shop for engineers looking to o…
A new page in the LLM Engineer's Almanac provides a block quant visualizer to help engineers understand quantization formats for owning their LLM inference.
How fast is N tokens per second really?
A web tool that lets users visually experience different LLM token generation rates (e.g., 5–800 tok/s) across code, text, reasoning, and agent modes, helping internalize performance numbers from benchmarks.
[Benchmark] DFlash2 vs MTP comparison. 5090RTX, Qwen 3.8 27B, Dynamic v3 GGUF, llama.cpp. Token generation, latency and available context.
The benchmark compares DFlash2 and MTP techniques in llama.cpp, showing that DFlash2 offers around 20% faster token generation but reduces available context by 38%.
@LottoLabs: This is awesome work Dflash for qwen 3.5/6 series
Charles Frye announces the co-release with Z Lab of six new DFlash speculators for Alibaba Qwen 3.x models, achieving over 1k output tokens per second for Qwen 3.5 122B-A10B on a B200.