TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Regent: A Language for Implicit Dataflow Parallelism

33 点作者 federicoponzi将近 5 年前

5 条评论

mratsim将近 5 年前
Hey, I&#x27;ve been following up on Legion and Regent quite a bit, excellent work there.<p>Do you have a set of benchmarks that others can reimplement to compare the approaches?<p>I&#x27;ve added Dataflow Parallelism to my own multithreading runtime[1] but I didn&#x27;t had dataflow focused benchmarks yet, well I could add Cholesky Decomposition but it&#x27;s quite involved.<p>I expect the people from TaskFlow[2] and Habanero[3] (via Data-Driven Futures) would be quite interested as well in a common set of dataflow parallelism benchmarks.<p>By the way if you didn&#x27;t read the DaCe paper[4] you absolutely should, seems like the age of Dataflow parallelism and properly optimizing for data is coming.<p>[1]: <a href="https:&#x2F;&#x2F;github.com&#x2F;mratsim&#x2F;weave#dataflow-parallelism" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;mratsim&#x2F;weave#dataflow-parallelism</a><p>[2]: <a href="https:&#x2F;&#x2F;github.com&#x2F;taskflow&#x2F;taskflow" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;taskflow&#x2F;taskflow</a><p>[3]: <a href="https:&#x2F;&#x2F;github.com&#x2F;habanero-rice&#x2F;hclib" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;habanero-rice&#x2F;hclib</a><p>[4]: <a href="https:&#x2F;&#x2F;github.com&#x2F;spcl&#x2F;dace" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;spcl&#x2F;dace</a>, <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1902.10345" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1902.10345</a>
评论 #23612102 未加载
brudgers将近 5 年前
some past comments, <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=10764268" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=10764268</a>
smabie将近 5 年前
I&#x27;ve always wondered why there isn&#x27;t a general purpose purely functional language that computes a graph of deps and implicitly parallelizes all operations. For some things, like a map over a list, I understand that the overhead of distributing the work is greater than just using one thread, but for things known at compile time (like deps between variables), the cost should be zero to distribute.
评论 #23598303 未加载
chenzhekl将近 5 年前
This idea seems to be pretty similar to TensorFlow. Could anybody familiar to this elaborate the difference?
评论 #23597563 未加载
评论 #23596902 未加载
BubRoss将近 5 年前
I don&#x27;t understand why something like this would need a separate language. Switching languages means starting over in many ways with regards to tools libraries and semantics. A graph of tasks can be made with a cdecl library.
评论 #23596258 未加载
评论 #23596808 未加载
评论 #23597524 未加载