--- title: ASPLOS 2025: Paper Summaries and Insights | daily.dev description: A comprehensive summary of research papers presented at ASPLOS 2025, covering advances in AI/ML systems, memory and storage, cloud infrastructure, security,... canonical: https://daily.dev/posts/asplos-2025-paper-summaries-and-insights-2fyyreega twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: ASPLOS 2025: Paper Summaries and Insights | daily.dev og:description: A comprehensive summary of research papers presented at ASPLOS 2025, covering advances in AI/ML systems, memory and storage, cloud infrastructure, security,... og:url: https://daily.dev/posts/asplos-2025-paper-summaries-and-insights-2fyyreega og:image: https://api.daily.dev/og/posts/2FYYREEGA.png og:image:alt: ASPLOS 2025: Paper Summaries and Insights og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # ASPLOS 2025: Paper Summaries and Insights **[eunomia eBPF Blog](https://daily.dev/sources/eunomia)** · 40 min read · 0 upvotes · 0 comments ## Summary A comprehensive summary of research papers presented at ASPLOS 2025, covering advances in AI/ML systems, memory and storage, cloud infrastructure, security, programming languages, and hardware acceleration. Key themes include efficient LLM serving and training (quantization, MoE, pipeline parallelism), CXL-based memory disaggregation, RowHammer defenses, privacy-preserving computation (FHE, ZK proofs), serverless GPU scaling, and observability in microservices. The conference highlights a strong trend toward domain-specific hardware/software co-design, adaptive systems, and formal verification entering traditionally pragmatic engineering fields. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the COMET approach to LLM quantization and what bit-widths does it use? COMET uses 4-bit weights, activations, keys, and values (W4A4KV4) to shrink LLM size while maintaining model accuracy. This enables faster inference with limited precision, reducing memory and compute requirements so large models become more practical to deploy without sacrificing meaningful quality. _Teams shipping LLM inference infrastructure track quantization trade-offs like these on daily.dev._ ### What is the Marionette RowHammer attack and how does it bypass existing mitigations? Marionette is a RowHammer variant that exploits electromagnetic coupling between DRAM rows to induce bit flips in a victim row by hammering neighbor-of-neighbor rows rather than immediate neighbors. This indirect hammering pattern bypasses mitigations that only protect against direct adjacent-row attacks, requiring defenses to account for more complex coupling patterns. _Security engineers hardening memory systems against RowHammer variants follow developments like this on daily.dev._ ### How does the Helix LLM serving system schedule inference across heterogeneous GPUs? Helix models LLM inference scheduling across heterogeneous GPUs as a max-flow problem, partitioning model execution and routing data through GPUs of different capabilities connected by network. This approach efficiently stretches model execution across dissimilar hardware, achieving low latency while maximizing utilization of existing multi-GPU servers in data centers. _Engineers building multi-GPU inference pipelines keep up with scheduling research like Helix on daily.dev._ ## Similar posts on daily.dev - [Best of metadata in 2025](https://daily.dev/posts/best-of-metadata-in-2025-gs9xv613f) · Metadata · 1 upvotes · 0 comments - [AI Papers to Read in 2025](https://daily.dev/posts/ai-papers-to-read-in-2025-kzq7ahqog) · Towards Data Science · 6 upvotes · 0 comments - [Machine Learning Blog \| ML@CMU \| Carnegie Mellon University](https://daily.dev/posts/machine-learning-blog-ml-cmu-carnegie-mellon-university-ynzk2dhmg) · ML CMU · 2 upvotes · 0 comments - [DevDays 2026 Conf From a Speaker’s View](https://daily.dev/posts/devdays-2026-conf-from-a-speaker-s-view-uawono0aq) · We Are .NET · 1 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#architecture](https://daily.dev/tags/architecture), [#distributed-systems](https://daily.dev/tags/distributed-systems), [#ai-inference](https://daily.dev/tags/ai-inference) [View this post on daily.dev](https://daily.dev/posts/asplos-2025-paper-summaries-and-insights-2fyyreega) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"ASPLOS 2025: Paper Summaries and Insights","url":"https://daily.dev/posts/asplos-2025-paper-summaries-and-insights-2fyyreega","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/asplos-2025-paper-summaries-and-insights-2fyyreega"},"datePublished":"2026-08-11T08:54:30.694Z","dateModified":"2026-08-11T09:24:36.232Z","description":"A comprehensive summary of research papers presented at ASPLOS 2025, covering advances in AI/ML systems, memory and storage, cloud infrastructure, security,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b59e04eefe0ea85df0b7e50151b0b12e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b59e04eefe0ea85df0b7e50151b0b12e?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"eunomia eBPF Blog","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"eunomia eBPF Blog","logo":"https://media.daily.dev/image/upload/s--z-nTl7TK--/f_auto,q_auto/v1786438304/logos/eunomia?_a=BAMAMicg0","url":"https://daily.dev/sources/eunomia"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/asplos-2025-paper-summaries-and-insights-2fyyreega","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,architecture,distributed-systems,ai-inference","timeRequired":"PT40M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"eunomia eBPF Blog","item":"https://daily.dev/sources/eunomia"},{"@type":"ListItem","position":3,"name":"ASPLOS 2025: Paper Summaries and Insights"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/asplos-2025-paper-summaries-and-insights-2fyyreega#faq","mainEntity":[{"@type":"Question","name":"What is the COMET approach to LLM quantization and what bit-widths does it use?","acceptedAnswer":{"@type":"Answer","text":"COMET uses 4-bit weights, activations, keys, and values (W4A4KV4) to shrink LLM size while maintaining model accuracy. This enables faster inference with limited precision, reducing memory and compute requirements so large models become more practical to deploy without sacrificing meaningful quality. Teams shipping LLM inference infrastructure track quantization trade-offs like these on daily.dev."}},{"@type":"Question","name":"What is the Marionette RowHammer attack and how does it bypass existing mitigations?","acceptedAnswer":{"@type":"Answer","text":"Marionette is a RowHammer variant that exploits electromagnetic coupling between DRAM rows to induce bit flips in a victim row by hammering neighbor-of-neighbor rows rather than immediate neighbors. This indirect hammering pattern bypasses mitigations that only protect against direct adjacent-row attacks, requiring defenses to account for more complex coupling patterns. Security engineers hardening memory systems against RowHammer variants follow developments like this on daily.dev."}},{"@type":"Question","name":"How does the Helix LLM serving system schedule inference across heterogeneous GPUs?","acceptedAnswer":{"@type":"Answer","text":"Helix models LLM inference scheduling across heterogeneous GPUs as a max-flow problem, partitioning model execution and routing data through GPUs of different capabilities connected by network. This approach efficiently stretches model execution across dissimilar hardware, achieving low latency while maximizing utilization of existing multi-GPU servers in data centers. Engineers building multi-GPU inference pipelines keep up with scheduling research like Helix on daily.dev."}}]} ```