Loading…
Topic
812 posts about Performance. Every summary links to the original.
See the idea
Bare metal gives engineers direct control over GPU, network, and virtualization layers, but it shifts elasticity and lifecycle work onto the team. This trade-off map shows when that bargain pays off.
Bun packs a runtime, package manager, bundler, and test runner into one binary. Here is how those four parts fit together and where the seams still show.
A five-step how-to for deciding when an AMD Instinct GPU should stay dedicated and when SR-IOV virtual functions, VMs, or sidecar isolation can safely share it.
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
Newest first
Bullet operates a DeFi exchange for perpetuals, spot trading, and lending, with trades settled on a blockchain while users retain custody of their assets. Its sub-millisecond trading system produces more than 150 million event rows per hour, creating a need for historical indexing, real-time analytics, and operational monitoring beyond the blockchain’s state-machine model. ClickHouse Cloud consumes events from Kafka, stores them in dedicated tables, and supports user features, leaderboard and reward calculations, P&L charts, audit logs, and Grafana-backed alerts. Replacing Databricks reduced query latency from 10–15 seconds to milliseconds, improved data freshness from 1–2 hours to under 5 seconds, and removed a separate DynamoDB serving layer at comparable cost while handling 1,000 times more data across staging, testnet, and mainnet.
ClickHouse's @clickhouse/rowbinary is a Node.js reader and writer for the RowBinary, RowBinaryWithNames, and RowBinaryWithNamesAndTypes formats, offering both a generic parser and an accompanying Agent Skill. RowBinary avoids per-row overhead but is difficult to consume in JavaScript because nested types, LEB128 varints, precision-sensitive DateTime64 values, and recursive self-describing types require extensive runtime dispatch; JSON alternatives can also round large UInt64 values. The package provides small, composable, monomorphizable reading primitives, while SKILL.md instructs a coding agent to assemble query-specific parsers with inlined operations and selected policies such as BigInt handling and array materialization. Generated parsers run 1.5–3.4x faster than composing library functions in loops, with a fixed-width benchmark parser reaching 3.41x faster, and produce reviewable source that can be tested and benchmarked like ordinary application code.
Adapting vision reasoning models for specialized physical AI video tasks often demands days of manual container configuration, training script authoring, and hyperparameter tuning. NVIDIA Cosmos 3 Nano, an omnimodal mixture-of-transformers model, pairs an autoregressive Reasoner tower with an iterative diffusion Generator tower for video question answering. By integrating NVIDIA TAO agent skills with a coding agent like Codex, developers can automate data patching, evaluation, and Low-Rank Adaptation (LoRA) post-training. In experiments using the Woven Traffic Safety dataset, LoRA fine-tuning boosted zero-shot baseline accuracy from 54.41% to 87.14% while using approximately seven times fewer GPU hours than full-parameter supervised fine-tuning. Combining this agent workflow with TAO AutoML further elevated model accuracy to 93.35% before automated deployment via NVIDIA NIM.
ClickHouse Managed Postgres configures Linux huge pages to reduce the per-connection page-table cost and translation overhead of a large PostgreSQL shared_buffers cache. The configuration reserves a quarter of machine memory before PostgreSQL starts, verifies HugePages_Total, sets huge_pages = 'on' to prevent silent fallback, and sizes shared_buffers against the complete shared-memory segment, including buffer descriptors, WAL buffers, and lock tables. On an r7i.4xlarge with 32GB shared_buffers, 4KB pages added 31.1MB of page tables per connection and reached 6.12GB at 200 connections, while 2MB huge pages reached 111MB. A select-only pgbench test at 100 clients also rose from 373,083 TPS to 418,087 TPS, a 12% increase, while startup failure without a sufficient pool made misconfiguration explicit.
Serving Qwen3.5-397B, a 397-billion-parameter sparse Mixture-of-Experts model, on Ironwood (TPU v7x) requires fitting roughly 400 GB of weights into HBM while sustaining utilization across contrasting workloads. The report applies a modular, model-agnostic playbook built from reusable JAX/Pallas kernels and hardware-aware cost models, focusing new work on Gated DeltaNet linear attention and Attention Data Parallelism while integrating changes into vLLM and SGLang. The architecture activates exactly 17 billion parameters per token and combines 15 repeating blocks of three GDN layers and one GQA layer. Benchmarks use a single host with four Ironwood chips, eight logical devices, 8K/1K prefill-heavy and 1K/8K decode-heavy workloads, and concurrency tiers from 64 to 512. Between April and June 2026, stated optimizations improved inference performance by approximately 4.7x for 512-concurrency prefill-heavy workloads and 3.1x for decode-heavy workloads.
Fault-tolerant quantum computing requires effective quantum error correction, but color codes have historically been constrained by difficult decoding demands. NVIDIA introduced the Ising Decoder ColorCode 1 Fast, a pre-decoder designed to accelerate and improve color code logical error rates. The framework uses a 3D Convolutional Neural Network trained via cuStabilizer in NVIDIA cuQuantum alongside PyTorch to handle localized error syndromes. In benchmark testing at code distance d=31 with a 0.3% physical error rate, the decoder achieved over 347.7x lower logical error rates and 7.3x faster runtime compared to the Chromobius decoder. These local, space-time predictions scale to arbitrary distances, supporting parallel blockwise decoding for real-time quantum algorithm execution.
The post presents a reusable real-time guardrail pattern for flagging suspicious Ethereum blockchain transactions and routing them for downstream action. Its rules identify impossible blocks where gas_used exceeds gas_limit and scan extra_data for email addresses, JWT tokens, or AWS access-key patterns, producing ALLOW or QUARANTINE decisions with reasons. The implementation uses Apache Spark Structured Streaming Real-Time Mode, whose continuous data flow, pipeline scheduling, streaming shuffle, pre-allocated execution pipelines, and asynchronous checkpointing target millisecond latency without a separate streaming engine. On a four-worker DBR 16.4 LTS cluster, the stateless Kafka-to-Kafka test processed about 23 million records at 69,713 rows per second, with P95 below 0.5 milliseconds and P99 at 1 millisecond. The post notes that RTM with a Kafka sink provides at-least-once delivery and that more complex stateful workloads may incur higher latency.
Meta’s ads serving fleet processes more than 5 million requests per second, so millisecond-scale latency changes can affect ads relevance and advertiser ROI. During a move from kernel 6.4 with CFS to kernel 6.9, the team observed a latency regression associated with EEVDF and used sched_ext, an upstream BPF-based framework, to encode ads-specific scheduling priorities. The policy soft-partitions CPUs between latency-critical request-path threads and less-sensitive work, adjusts pool sizes with load-based heuristics, and runs as a user-space binary that loads BPF, enabling policy changes without rebuilding or reinstalling the kernel. Initial results were a 28% reduction in ads retrieval p99 latency, 3.28 megawatts of fleet power savings, and a 1.1% increase in weighted ads ranked; two later user-space updates delivered an additional 60% p99 reduction and an 18% decrease in critical-path timeout errors, with iterations shipping in days rather than months.
Insider One’s Unified Customer Database (UCD) unifies event data, user attributes, identity resolution, and profiles across channels to support real-time segmentation, personalization, and analytics. The platform initially ran ClickHouse in a self-managed AWS deployment, but growth brought larger datasets, more concurrent queries, and increasing work tuning clusters and balancing performance against cost. In a proof of concept, the team migrated an anonymized dataset from one of its largest customers to ClickHouse Cloud and replayed UI segmentation, audience-count, and scheduled-campaign workloads. Average query latency fell 50–60% without significant changes, reaching 70–75% for the largest customers, while latency percentiles became more predictable; the managed service also reduced operational burden and gave the team more capacity for product development and its AI-oriented roadmap.
Large language model training workloads frequently face GPU high-bandwidth memory (HBM) capacity constraints before compute limits are reached. To alleviate this memory pressure in JAX training workloads, host offloading transfers selected forward-pass activations to pinned host memory over high-speed NVLink-C2C interconnects and reloads them during the backward pass instead of recomputing them. Benchmarks executed on 128 GPUs within NVIDIA GB200 NVL72 systems using MaxText demonstrated significant improvements. For DeepSeek-V3 671B, host offloading combined with the Latency Hiding Scheduler and pipelined offloading reached 908.2 TFLOPs/s/device, surpassing activation rematerialization throughput by 57% and unlocking larger batch configurations. Similarly, QKV activation offloading on Llama 3.1 405B achieved a 2.9% throughput increase when transfer latency was hidden behind computation.
amazon.science
Robotic manipulation in contact-rich settings requires accurate tactile force perception, but existing physics simulators either run too slowly for reinforcement learning or oversimplify shear forces. HydroShear addresses this tactile reality gap by incorporating path-dependent force tracking into hydroelastic contact models. By remembering object motion history across soft elastomer sensors, the GPU-parallelizable simulator computes full 3-D contact deformations, friction, and slipping dynamics. Reinforcement learning policies trained entirely in HydroShear transferred directly to a real Franka robot equipped with GelSight Mini sensors without fine-tuning. Across four challenging manipulation tasks, HydroShear achieved a 93 percent average success rate, outperforming baseline models TacSL and FOTS.
GPU compute often outpaces memory bandwidth, causing separate kernels to suffer from intermediate global memory round-trips and launch overhead. Kernel fusion resolves this by executing multiple operations within a single kernel to keep intermediate data inside registers. Using the sum(abs(x)) operation as a benchmark on an NVIDIA GeForce RTX 4090, a naive two-kernel setup moved 3 GB of global memory over 3.51 ms. In contrast, manual CUDA C++ fusion, PyTorch compiler fusion via torch.compile, and explicit Python fusion using cuda.compute all achieved a 3x speedup. Across all three approaches, global memory transfers dropped to 1 GB by computing absolute values inline during reduction.
Efficient large language model serving requires co-designing architectures to balance datacenter token throughput and user interactivity within hardware constraints. Roofline analysis shows that linear layers become memory-bound when hidden or intermediate projection dimensions are small, leaving arithmetic hardware underutilized even at large token counts. For prefill workloads, Chunked Pipeline Parallelism reduces first-token latency on long contexts when models use regular, repeatable layer patterns that split into balanced stages. For latency-critical decoding, decoupling attention and feed-forward network parallelization enables independent optimization, applying expert parallelism for feed-forward networks and Helix parallelism to shard the key-value cache across sequences. Following architectural design rules—such as dimension alignment, width prioritization, and low-precision NVFP4 compatibility—allows models to achieve higher utilization on modern accelerator hardware.
When migrating GitHub Copilot code review from custom navigation tools to shared Unix-style CLI utilities—grep, glob, and view—benchmarks revealed higher review costs and fewer caught issues. Rather than an issue with the underlying tools, trace analysis showed the agent used general-purpose coding assistant instructions that triggered expansive repository browsing loops. In response, the team rewrote the tool guidance to enforce a review-specific workflow anchored to pull request diffs. The updated instructions direct the agent to narrow candidate call sites and files using batched discovery before reading minimal line ranges with view. This workflow tuning reduced average review costs by approximately 20% in production while maintaining review quality.
ClickHouse announces an official ADBC driver, a modern Arrow-native alternative to ODBC and JDBC for analytics and AI applications. It provides end-to-end columnar movement: query results leave ClickHouse as Apache Arrow and arrive in applications without a row-oriented conversion or staging step. A single Rust-built implementation connects through ClickHouse's HTTP interface, supports TLS, and serves languages and tools with ADBC bindings, including Ruby, R, and C. Users can install precompiled, signed binaries through the ADBC Driver Foundry and connect with a clickhouse:// URI; dbt Fusion distributes the driver directly. The driver also supports streaming Arrow inserts, while schema discovery, improved bulk ingest, usage examples, and JSON support remain on the roadmap.
Biomolecular structure prediction and co-folding pipelines encounter latency and memory ceilings during sequence alignment and multi-residue inference. NVIDIA addresses these constraints through the BioNeMo Agent Toolkit, which connects AI agents to accelerated biological computing components. The pipeline offloads multiple sequence alignment to MMseqs2-GPU, while cuEquivariance and the OpenFold3 NIM accelerate core geometric attention kernels to cut inference latency up to fourfold. To handle large molecular assemblies beyond single-GPU memory limits, the Fold-CP framework distributes context across multiple GPUs with O(N²/P) memory scaling. Together, these tools enable the prediction of complexes up to 32,000 tokens on 64 NVIDIA B300 GPUs, making large-scale virtual screening and ribosome-scale modeling computationally feasible.
An Android APK investigation found high-resolution PNG variants beside vector drawable XML files, although those raster assets had not been added directly to the codebase. The cause was AAPT2 rasterizing vectors that used features unavailable below API 24, including gradients in android:fillColor, despite the app’s minSdkVersion being 23; the generated files could be much larger than their XML sources, and were produced even for unused resources. The described remedy was to remove unused raster-prone assets, use VectorDrawableCompat with app:srcCompat for affected vectors, and tell Gradle and AAPT to use support-library rendering rather than density-specific PNGs. Removing unused assets cut download size by more than 1 MB, while raising minSdkVersion to 24 produced a further 2.7 MB reduction, for 3.7 MB total; similar gains were observed in Toing and Instamart.
An Android monorepo’s CI pipeline averaged about 35 minutes in its build phase, with a P90 of roughly 44 minutes, as repeated dependency downloads, environment-driven invalidations, variant bloat, and machine limits increased work. The team reworked caching with branch- and architecture-aware keys, release/develop fallbacks, cached SDK and dependency data, and a Gradle Remote Cache, while filtering repositories and restricting CI to required variants. It also tuned 32GB runners to a 16GB heap and 12GB Metaspace, disabled configuration-cache saves in ephemeral CI, upgraded Kotlin, AGP 8.10.1, and Gradle 8.11.1, and moved critical workflows to Apple Silicon. Gradle tasks fell from 3399 to 2745, while remote-cache hit rates reached P50 44% and P90 56%. After the Apple Silicon refresh, total CI time reached P50 10 and P90 11 minutes, average credits per build dropped from 55 to 43, and a comparable rollout window saved about 5800 credits.
Molecular dynamics packages like GROMACS face severe scaling bottlenecks when CPU-orchestrated MPI communication introduces blocking synchronizations during halo exchanges. To resolve this, researchers implemented GPU-initiated communication using NVIDIA NVSHMEM to eliminate CPU-GPU handoffs. The redesigned approach fuses packing, remote memory access, and completion waiting into device-side kernels using fine-grained signaling. Benchmarks on NVIDIA DGX H100 systems demonstrated up to 1.5x performance gains intra-node and up to 2x on GB200 NVL72 multi-node clusters, particularly on latency-bound systems. While device-initiated communication introduces a slight 1–3% overhead on very large compute-bound configurations at low node counts, it consistently improves strong scaling for production workloads.
Trigger.dev, an open-source platform for serverless asynchronous tasks, needed to scale observability for increasingly complex, long-running AI workflows spanning thousands of tenants. Its initial Postgres design stored logs, traces, and events in a growing table, making indexes, analytics, and operational reliability increasingly difficult to manage. The team moved observability workloads to ClickHouse after evaluating performance, scalability, open-source compatibility, deployment flexibility, and support for dynamic telemetry, including its native JSON data type. Migration occurred gradually at the task-execution level, routing each run’s logs to either database without downtime or customer intervention. ClickHouse now supports real-time telemetry analysis, approximately 200-millisecond P95 latency for queries returning tens of thousands of rows, and a dramatically smaller telemetry storage footprint.
The best engineering reads, once a week.
