Non-Speculative Store Coalescing in Total Store Order

Abstract

We present a non-speculative solution for a coalescing store buffer in total store order (TSO) consistency. Coalescing violates TSO with respect to both conflicting loads and conflicting stores, if partial state is exposed to the memory system. Proposed solutions for coalescing in TSO resort to speculation-and-rollback or centralized arbitration to guarantee atomicity for the set of stores whose order is affected by coalescing. These solutions can suffer from scalability, complexity, resource-conflict deadlock, and livelock problems. A non-speculative solution that writes out coalesced cachelines, one at a time, over a typical directory-based MESI coherence layer, has the potential to transcend these problems if it can guarantee absence of deadlock in a practical way. There are two major problems for a non-speculative coalescing store buffer: i) how to present to the memory system a group of coalesced writes as atomic, and ii) how to not deadlock while attempting to do so. For this, we introduce a new lexicographical order. Relying on this order, conflicting atomic groups of coalesced writes can be individually performed per cache block, without speculation, rollback, or replay, and without deadlock or livelock, yet appear atomic to conflicting parties and preserve TSO. One of our major contributions is to show that lexicographical orders based on a small part of the physical address (sub-address order) are deadlock-free throughout the system when taking into account resourceconflict deadlocks. Our approach exceeds the performance and energy benefits of two baseline TSO store buffers and matches the coalescing (and energy savings) of a release-consistency store buffer, at comparable cost.

Free atomics

“…Deadlock avoidance has been also studied in proposals allow loads [43,45] and stores [44] to be non-speculatively reordered while offering an ordered behaviour to the programmer. Writers-Block [43] resolves the deadlocks by guaranteeing the loads always to progress through tear-off copies.…”

“…Writers-Block [43] resolves the deadlocks by guaranteeing the loads always to progress through tear-off copies. Non-speculative store coalescing [44] relies on a predetermined order of writes to solve deadlocks. Applying a predetermined order to Free atomics could lead to a reduction in timeouts, when a predetermined order in performing the stores older than the Free atomic can be guaranteed.…”

Regional Out-of-Order Writes in Total Store Order

“…Ros and Kaxiras [28] propose coalescing stores in the store buffer and avoid breaking the store order by performing stores in atomic groups. In contrast to our approach where DRF stores can perform completely out-of-order, stores in an atomic group perform following a globally defined order.…”

“…Previous proposals alleviate the store buffer bottleneck by performing the stores out-of-order while still complying with TSO [8,15,28,33,34]. However, they either require complex hardware structures and speculative writes [34], or they only enable limited reordering [28,33].…”

“…Previous proposals alleviate the store buffer bottleneck by performing the stores out-of-order while still complying with TSO [8,15,28,33,34]. However, they either require complex hardware structures and speculative writes [34], or they only enable limited reordering [28,33]. For example, the compile-time analysis by Singh et al [33] only enables reordering of ≈75% of the stores, on average, while Ros and Kaxiras [28] simply force a different (lexicographical) order for all stores.…”