Low Latency Interrupt Handling with Selective Moderation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interrupt moderation techniques increase latency in network interface controllers, making them unsuitable for latency-sensitive applications, as they delay interrupt requests even for sparse events, affecting the accuracy of round-trip time measurements and overall system performance.
Innovation Solution
A system and method that selectively delay interrupt requests, allowing immediate generation of the first interrupt request and coalescing subsequent requests within a predefined time window, thereby reducing unnecessary delays and maintaining low latency for sparse events while coalescing multiple interrupts for high-rate events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If interrupt moderation is enabled to reduce the number of interrupt requests, then productivity is improved, but latency increases making it unsuitable for latency-sensitive applications
Solution Approach 1:
The interrupt moderation system dynamically adjusts its behavior based on the type of interrupt event. For packet reception events, it applies coalescing to reduce interrupt frequency and improve productivity. For transmission completion events, it disables coalescing to maintain low latency and enable accurate round-trip time measurement. This dynamic configuration resolves the contradiction by optimizing for different operational requirements.
Solution Approach 2:
The patent applies different interrupt moderation strategies to different types of interrupt events locally. Packet reception interrupts use coalescing with configurable thresholds to reduce overall interrupt load, while transmission completion interrupts use immediate notification to preserve timing accuracy. This localized differentiation allows the system to achieve both productivity improvement and latency preservation for critical operations.
2Productivity
If interrupt coalescing is applied to all interrupt requests, then the number of interrupt requests is reduced, but measurement precision of round-trip time deteriorates
Solution Approach 1:
The interrupt handling system is segmented into different paths based on event type. Transmission completion events are routed to an immediate interrupt path that bypasses the coalescing mechanism, while packet reception events are routed to a coalescing path. This segmentation ensures that timing-critical measurements are not affected by coalescing delays while still benefiting from interrupt reduction in other areas.
Solution Approach 2:
Instead of applying coalescing universally and then making exceptions, the system inverts the approach by defaulting to immediate interrupt generation for transmission events and selectively enabling coalescing only for packet reception events. This inversion prioritizes measurement precision for critical operations while still achieving productivity gains elsewhere.
Data Source
AI summary
An interrupt request generating process can include determining a first interrupt triggering event has occurred after a predetermined period of time in which no interrupt triggering event occurred. In response to determining that the first interrupt triggering event has occurred, the interrupt request generating process may generate a first interrupt request without adding an intentional delay, and initiating a timer configured to expire after a predetermined time interval. When a second interrupt triggering event is determined to have occurred before the timer expires, a second interrupt request is delayed from being generated until the timer expires.