| DLRM embedding lookupsMeta Software Defined Memory | 128–256 BQuantized embedding rows in the study | Direct I/O io_uring and an application row cache. | A buffered CPU reader could be tested. The page cache adds another caching layer; the SGL bit-bucket path would not carry over unchanged. | Compare lookup latency and total cache memory using the same row IDs and access distribution. |
| ANN rerankingOriginal FusionANNS paper | 128–384 BRaw vectors discussed in the paper; stored within 4 KiB pages | Direct I/O Page grouping, request deduplication and application buffering. | Possible with a buffered file reader. Preserve layout and scheduling, then compare memory use and performance. | Measure reuse of reranking pages and shared-page reads across concurrent queries. |
| ANN rerankingSamsungDS FusionANN prototype | 512 B by default128 FP32 values; the reader requests 4 KiB pages | Buffered I/O Python file reads and an application page cache. | Already buffered. It is an independent implementation of the paper’s ideas. | Trace actual syscall sizes and page-cache hits. Application read counts alone do not count SSD commands. |
| ANN graph / vector retrievalDiskANN C++ Linux reader | Depends on vector and node layoutSub-4-KiB payloads need to be identified in the selected dataset | Direct I/O
O_DIRECT with asynchronous reads. | Requires a buffered reader and scheduling work. No supported configuration switch is established here. | Measure node payload, full read size, locality and recall before selecting a storage comparison. |
| On-disk vector storageQdrant mmap configuration | Depends on dimensions and encoding | Buffered I/O · mmap Memory-mapped vector files; the OS caches pages. | Already uses the page cache. Buffered pread() would be a separate implementation. | Identify a sub-4-KiB representation and measure faults, residency and query latency. |
| Fraud-graph feature lookupsknlp DGraphFin forced-SSD experiment | 68 B17 FP32 node features | Direct I/O SSD access is forced for the experiment. | A buffered CPU feature reader could be added. Keep the requested nodes and features identical. | Compare page grouping and I/O paths separately. The roughly 250 MB feature store fits in RAM. |
| Relational feature gatheringKumoRFM-2 SSD graph engine | Not reportedThe sub-4-KiB share needs measurement | Buffered I/O · mmap The paper specifies memory-mapped I/O for this engine. | Already uses the page cache. A buffered read() implementation would be a separate path. | Capture sampled row and feature accesses, their useful sizes, page faults and reuse. |
| Out-of-core GNN featuresGIDS | 512 B–4 KiBTypical feature sizes discussed in the paper; the smaller sizes fall within this tracker | GPU direct access GPU-initiated storage, outside CPU buffered reads. | Requires a different architecture. Host buffered reads would add staging, transfers and synchronization. | Compare the full CPU-staged path against the original GPU path on a dataset larger than available memory. |
| Feature / session / agent recordsRocksDB-backed application candidate | Application trace neededValue size and file-block size differ | Buffered by default RocksDB also supports direct I/O. The AI application path is not yet traced. | The engine supports both modes. A service’s configuration and caches determine the actual path. | Obtain record sizes, access patterns and file reads from a specific AI workflow. |