1. 1 Utility-Based Partitioning of Shared Caches Moinuddin K. Qureshi Yale N. Patt International Symposium on Microarchitecture (MICRO) 2006

2. 2 Introduction CMP and shared caches are common Applications compete for the shared cache Partitioning policies critical for high performance Traditional policies: oEqual (half-and-half) Performance isolation. No adaptation oLRU Demand based. Demand benefit (e.g. streaming)

3. 3 Background Utility U a b = Misses with a ways – Misses with b ways Low Utility High Utility Saturating Utility Num ways from 16-way 1MB L2 Misses per 1000 instructions

4. 4 Motivation Num ways from 16-way 1MB L2 Misses per 1000 instructions (MPKI) equake vpr LRU UTIL Improve performance by giving more cache to the application that benefits more from cache

5. 5 Outline Introduction and Motivation Utility-Based Cache Partitioning Evaluation Scalable Partitioning Algorithm Related Work and Summary

6. 6 Framework for UCP Three components: Utility Monitors (UMON) per core Partitioning Algorithm (PA) Replacement support to enforce partitions I$ D$ Core1 I$ D$ Core2 Shared L2 cache Main Memory UMON1 UMON2 PA

7. 7 Utility Monitors (UMON) For each core, simulate LRU policy using ATD Hit counters in ATD to count hits per recency position LRU is a stack algorithm: hit counts utility E.g. hits(2 ways) = H0+H1 MTD Set B Set E Set G Set A Set C Set D Set F Set H ATD Set B Set E Set G Set A Set C Set D Set F Set H +++ + (MRU)H0 H1 H2…H15(LRU)

8. 8 Dynamic Set Sampling (DSS) Extra tags incur hardware and power overhead DSS reduces overhead [Qureshi+ ISCA06] 32 sets sufficient (analytical bounds)analytical bounds Storage < 2kB/UMON MTD ATD Set B Set E Set G Set A Set C Set D Set F Set H +++ + (MRU)H0 H1 H2…H15(LRU) Set B Set E Set G Set A Set C Set D Set F Set H Set B Set E Set G Set A Set C Set D Set F Set H Set B Set E Set G UMON (DSS)

9. 9 Partitioning algorithm Evaluate all possible partitions and select the best With a ways to core1 and (16-a) ways to core2: Hits core1 = (H 0 + H 1 + … + H a-1 ) ---- from UMON1 Hits core2 = (H 0 + H 1 + … + H 16-a-1 ) ---- from UMON2 Select a that maximizes (Hits core1 + Hits core2 ) Partitioning done once every 5 million cycles

10. 10 Way Partitioning Way partitioning support: [ Suh+ HPCA02, Iyer ICS04 ] 1.Each line has core-id bits 2.On a miss, count ways_occupied in set by miss-causing app ways_occupied < ways_given Yes No Victim is the LRU line from other app Victim is the LRU line from miss-causing app

11. 11 Outline Introduction and Motivation Utility-Based Cache Partitioning Evaluation Scalable Partitioning Algorithm Related Work and Summary

12. 12 Methodology Configuration: Two cores: 8-wide, 128-entry window, private L1s L2: Shared, unified, 1MB, 16-way, LRU-based Memory: 400 cycles, 32 banks Used 20 workloads (four from each type) Benchmarks: Two-threaded workloads divided into 5 categories 1.01.2 1.4 1.6 1.82.0 Weighted speedup for the baseline

13. 13 Metrics Three metrics for performance: 1.Weighted Speedup (default metric) perf = IPC 1 /SingleIPC 1 + IPC 2 /SingleIPC 2 correlates with reduction in execution time 2.Throughput perf = IPC 1 + IPC 2 can be unfair to low-IPC application 3.Hmean-fairness perf = hmean(IPC 1 /SingleIPC 1, IPC 2 /SingleIPC 2 ) balances fairness and performance

14. 14 Results for weighted speedup UCP improves average weighted speedup by 11%

15. 15 Results for throughput UCP improves average throughput by 17%

16. 16 Results for hmean-fairness UCP improves average hmean-fairness by 11%

17. 17 Effect of Number of Sampled Sets Dynamic Set Sampling (DSS) reduces overhead, not benefits 8 sets 16 sets 32 sets All sets

18. 18 Outline Introduction and Motivation Utility-Based Cache Partitioning Evaluation Scalable Partitioning Algorithm Related Work and Summary

19. 19 Scalability issues Time complexity of partitioning low for two cores (number of possible partitions number of ways) Possible partitions increase exponentially with cores For a 32-way cache, possible partitions: 4 cores 6545 8 cores 15.4 million Problem NP hard need scalable partitioning algorithm

20. 20 Greedy Algorithm [Stone+ ToC 92] GA allocates 1 block to the app that has the max utility for one block. Repeat till all blocks allocated Optimal partitioning when utility curves are convex Pathological behavior for non-convex curves Num ways from a 32-way 2MB L2 Misses per 100 instructions

21. 21 Problem with Greedy Algorithm In each iteration, the utility for 1 block: U(A) = 10 misses U(B) = 0 misses Problem: GA considers benefit only from the immediate block. Hence it fails to exploit huge gains from ahead Blocks assigned Misses All blocks assigned to A, even if B has same miss reduction with fewer blocks

22. 22 Lookahead Algorithm Marginal Utility (MU) = Utility per cache resource MU a b = U a b /(b-a) GA considers MU for 1 block. LA considers MU for all possible allocations Select the app that has the max value for MU. Allocate it as many blocks required to get max MU Repeat till all blocks assigned

23. 23 Lookahead Algorithm (example) Time complexity ways 2 /2 (512 ops for 32-ways) Iteration 1: MU(A) = 10/1 block MU(B) = 80/3 blocks B gets 3 blocks Result: A gets 5 blocks and B gets 3 blocks (Optimal) Next five iterations: MU(A) = 10/1 block MU(B) = 0 A gets 1 block Blocks assigned Misses

24. 24 Results for partitioning algorithms Four cores sharing a 2MB 32-way L2 Mix2 (swm-glg-mesa-prl) Mix3 (mcf-applu-art-vrtx) Mix4 (mcf-art-eqk-wupw) Mix1 (gap-applu-apsi-gzp) LA performs similar to EvalAll, with low time-complexity LRU UCP(Greedy) UCP(Lookahead) UCP(EvalAll)

25. 25 Outline Introduction and Motivation Utility-Based Cache Partitioning Evaluation Scalable Partitioning Algorithm Related Work and Summary

26. 26 Related work Zhou+ [ASPLOS04] Perf += 11% Storage += 64kB/core X UCP Perf += 11% Storage += 2kB/core Suh+ [HPCA02] Perf += 4% Storage += 32B/core Performance Low High Overhead LowHigh UCP is both high-performance and low-overhead

27. 27 Summary CMP and shared caches are common Partition shared caches based on utility, not demand UMON estimates utility at runtime with low overhead UCP improves performance : oWeighted speedup by 11% oThroughput by 17% oHmean-fairness by 11% Lookahead algorithm is scalable to many cores sharing a highly associative cache

28. 28 Questions

29. 29 DSS Bounds with Analytical Model Us = Sampled mean (Num ways allocated by DSS) Ug = Global mean (Num ways allocated by Global) P = P(Us within 1 way of Ug) By Cheb. inequality: P 1 – variance/n n = number of sampled sets In general, variance 3 back

30. 30 Phase-Based Adapt of UCP

31. 31 Galgel – concave utility galgel twolf parser

32. 32 LRU as a stack algorithm