Pith. sign in

REVIEW 4 major objections 7 minor 69 references

DICE: Detailed Inter-Chiplet End-to-End PHY Modeling for Accurate Chiplet Simulation

T0 review · 4 major / 7 minor · reviewed 2026-07-31 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Fixed-latency chiplet-link models miss runtime PHY behavior and can shift simulated IPC by 6.8% on average and up to 27.6%.

desk verdict Real gem5 PHY integration that moves chiplet sim beyond fixed-delay links; the headline IPC numbers sit on a jitter/rate calibration slip that needs fixing before you trust the percentages. read the letter →

arxiv 2607.24221 v2 pith:SQFSMDCE submitted 2026-07-27 cs.AR

Rashid Aligholipour , Stefanos kaxiras , Yuan Yao This is my paper · ORCID
classification cs.AR
keywords chipletPHYmodelinggem5inter-chipletinterconnectQC-LDPCPAM4FECsimulationfidelity
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Chiplet systems move data across short-reach physical links that are noisy, modulated, and protected by iterative error correction. Most architecture simulators still treat those links as fixed delays, which erases channel noise, decoder convergence, retransmissions, and application-driven traffic that only appear at runtime. This paper shows that those omissions change packet-latency composition and high-level metrics such as IPC, sometimes optimistically and sometimes pessimistically, so fixed-delay chiplet studies can give off-trend answers. DICE embeds an end-to-end PHY path—QC-LDPC encode/decode, PAM4, lossy channel, soft demodulation, adaptive resend, and PHY flow control—inside gem5 and calibrates component latencies from synthesis and public link specs. With that model, packet time moves into the PHY boundary, long-tail latencies grow, and system IPC diverges from both monolithic and fixed-latency chiplet baselines, closer to measured core-to-core behavior on real multi-CCD processors.

What carries the argument

DICE: an in-simulation, runtime end-to-end PHY model in gem5 that chains QC-LDPC flit encoding, PAM4 modulation, AWGN channel noise (base SNR, jitter, crosstalk), LLR demodulation, bounded-iteration layered min-sum decoding with NACK resend, and PHY-level cut-through flow control at chiplet-boundary routers.

What would settle it

On a production multi-CCD processor, if measured core-to-core latency distributions and application IPC under the same workloads matched a carefully throttled fixed-latency model as well as or better than DICE, or if sweeping the paper’s SNR/parity/iteration knobs erased the reported IPC gap, the central claim would fail.

Watch

Extended reading notes

Core claim

Neglecting dynamic inter-chiplet PHY effects—SNR, jitter, crosstalk, iterative FEC convergence, and flit retransmissions—distorts packet-level timing and system IPC. Modeling the full end-to-end PHY datapath in simulation reshapes latency breakdown and shifts IPC by 6.8% on average (up to 27.6%) versus fixed-latency chiplet links, revealing variability that constant-delay abstractions cannot capture or correct by simple throttling.

Load-bearing premise

The chosen default channel and coding stack—about 35 dB base SNR, 1 ps jitter, 20 dB crosstalk, two parity bytes per flit, and a four-iteration decode budget—is representative enough of real and near-future chiplet PHYs that the IPC and latency gaps generalize beyond this calibration.

Editorial extensions

If this is right

  • Chiplet design-space studies that use constant link delay will mis-rank global vs local LLC, IOD speed, and SerDes rate choices because they miss PHY-induced tails.
  • Out-of-order cores and coherence/synchronization paths are especially sensitive: long-tail cross-chiplet flits, not mean latency, drive stalls and multi-threaded slowdown.
  • Architects can co-evaluate reliability knobs (parity bytes, decode budget, symbol rate) against IPC inside the same full-system run instead of offline BER tables.
  • Validation against real C2C measurements becomes a first-class check for any chiplet interconnect model claiming fidelity.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • As UCIe and similar standards push higher GT/s, the gap between fixed-delay and PHY-accurate models should widen, making constant-latency chiplet NoCs increasingly misleading for server-class DSE.
  • Memoizing common LLR/decode patterns, as the authors sketch, could make detailed PHY modeling cheap enough for routine gem5 sweeps rather than special studies.
  • The same variability argument that motivated detailed DRAM models now applies to die-to-die fabrics; interconnect abstraction level may need to track memory-model rigor in chiplet-era papers.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The manuscript presents DICE, an extension of gem5/Garnet that models the inter-chiplet physical layer end to end at runtime: QC-LDPC encoding/decoding with a bounded layered min-sum iteration budget, PAM4 modulation, an AWGN channel that folds baseline SNR, jitter, and crosstalk into an effective SNR (Eq. 5), LLR-based soft demodulation, and a flit-level ACK/NACK flow-control scheme with packet-granularity buffer reservation. The authors calibrate FEC encoder/decoder latencies via hardware synthesis (TSMC 40nm), pick defaults (35 dB base SNR, 2 parity bytes per 128-bit flit, N=4 iterations, 32 GT/s) from UCIe/IEEE HIR sources, and validate against measured core-to-core latencies on three production AMD processors, reporting lower RMSE than HeteroGarnet (HG). The central claim is that fixed-latency chiplet abstractions such as HG distort packet-latency composition and IPC — DICE shifts IPC by 6.8% on average and up to 27.6% vs HG — because performance is driven by latency variability and tails (Fig. 22) and by coherence/synchronization traffic (Fig. 23), not by means. The paper is generally well organized, the sensitivity studies are extensive, and the matched-average-latency HG+ control is a genuinely informative experiment. However, the default channel calibration contains an arithmetic inconsistency that propagates into the effective SNR and therefore into every FER-, retransmission-, and IPC-dependent result.

Significance. If the calibration issues are resolved, this is a useful contribution to the architecture community. Strengths that deserve explicit credit: (i) a complete, openly described PHY pipeline integrated into a standard full-system simulator; (ii) hardware-synthesis calibration of the FEC encoder (Yosys/OpenSTA, TSMC 40nm, Fig. 7); (iii) validation against three production AMD processors (EPYC 9454P, EPYC 7R13, ThreadRipper 3960X) on both max and average C2C latency with quantitative RMSE; (iv) a matched-average-latency control (HG+, Fig. 22) that isolates variability from mean latency — a good falsifiable experiment; and (v) broad sensitivity studies (SNR, parity, symbol rate, IOD latency, GS/LS LLC, multi-threaded synchronization). The qualitative thesis — that fixed-latency abstractions erase tail behavior that matters for OoO cores and coherence — is well argued and likely robust. The quantitative headline numbers, however, currently rest on an unresolved rate/noise calibration, which caps the significance until corrected.

major comments (4)
  1. [§III-E, Eq. (4)-(5); Table II/III] The default jitter SNR is inconsistent with the stated symbol rate. With T_sym = 1/32 GT/s = 31.25 ps and sigma_t = 1 ps, Eq. (4) gives 20 log10(31.25/pi) = 20.0 dB, not the reported 26.0 dB; 26 dB corresponds to T_sym = 62.5 ps, i.e. 16 GT/s (32 Gb/s PAM4). The text ('T_sym according to the network clock rate (32 Gb/s)') suggests bit rate was used as symbol rate. Recomputing Eq. (5) with jitter=20 dB, base=35 dB, XT=20 dB gives SNR_eff = 16.9 dB, not 19.0 dB. Since Figs 9/13/14 show FER strongly nonlinear in this regime, a ~2 dB shift can materially change post-FEC FER, retransmission rates, latency tails, and hence the headline 6.8%/27.6% IPC shifts and the C2C RMSE validation. Please fix the calibration, state whether the default link is 16 or 32 GT/s, and re-run or bound all affected results.
  2. [§IV-A, Fig. 3, Fig. 15-16, Table II] The iso-bandwidth basis of the central DICE-vs-HG comparison is not documented. Table II lists a 32 GT/s symbol rate, but Figs 15-16 sweep 'symbols/cycle' (2-32) without stating the network clock that maps this to GT/s; the on-die links are 128-bit at 2.0/1.0 GHz; and neither the SerDes lane count nor HeteroGarnet's throttled-channel configuration (used to match DICE's effective bandwidth, including the R=0.88 FEC overhead) is ever given. If HG is not bandwidth-matched, part of the IPC gap in Fig. 3(b) could be a bandwidth artifact rather than PHY dynamics. The HG+ control in Fig. 22 matches only average latency, not bandwidth. Please state HG's throttle settings, lane count, and the symbols/cycle-to-GT/s conversion.
  3. [§III-G, 'FEC-decoder latency'] The abstract claims decoder iteration timing is calibrated 'through hardware synthesis', but only the FEC *encoder* synthesis is reported (Fig. 7). The decoder assumptions - 1-cycle syndrome, 1 cycle per layered min-sum iteration at 2.0 GHz - are asserted without synthesis results or citations to decoder ASICs. A full layered iteration (all check-node and variable-node LLR updates across m layers) in one 500 ps cycle is a strong claim, and decode latency feeds directly into packet latency, tail behavior, and the IPC results. Please provide decoder synthesis data (cell count, critical path) or justify from prior art, and report the sensitivity of the headline IPC numbers to L_iter = 2-3 cycles.
  4. [§IV-D, Figs. 22-23] Two load-bearing observations lack root-cause analysis. (i) Fig. 23: multi-threaded XSBench under DICE slows 9.53x vs monolithic, vs 1.74x for HG - a 5.5x gap between models. Is this FEC retransmission/backpressure, globally-shared-LLC contention, or an artifact (e.g., NACK storms at the default SNR_eff)? A breakdown (retransmission rate, decoder-iteration distribution, queue occupancy) is needed before this can be read as realism rather than pathology. (ii) Fig. 22(c): HG+ on bc 'induces long backlogs that ultimately lead to simulation failure' - an unexplained simulator failure inside a central experiment. Please explain its cause and why it does not indicate a flow-control deadlock that could also affect DICE.
minor comments (7)
  1. [Listing 1, §III-E] Line 3 assigns Es/SNR_eff (a variance, per the text's sigma^2 = Es/SNR_eff) to 'sigma', which is then passed as the stddev of the normal distribution. Align the pseudocode with the equation.
  2. [Table II] '32GB DDR5, 4400 GHz' should read 4400 MT/s (or MHz). Also 'L2 Cache (LLC)' conflicts with Fig. 1, where the LLC is L3.
  3. [§III-C vs Table II] Table II lists Z=8, but the worked example in §III-C uses Z=16 with 16-bit chunks. Please reconcile the expansion factor.
  4. [§I, citation [18]] 'QC-LDPC decoding is NP-hard [18]' - Gallager's thesis does not establish this; the standard reference is Berlekamp, McEliece, and van Tilborg (1978) on ML decoding of linear codes.
  5. [Fig. 5] The annotations '= 0.23', '= 0.27' in the three panels are never defined (presumably parity overhead or code rate). Please label them.
  6. [§III-B, §I, §III-H] Typos: 'to capture actuate cross-die packet transmission' (-> accurate); 'inherentlydynamic'; §III-H list markers '1, 2, 3,'. Fig. 3(a) and Fig. 22 legends are very small in print.
  7. [§IV-B2, Figs. 13-14] It would help to state the random seeds / number of noise realizations per FER point and the statistical error on post-FEC FER, since several conclusions (e.g., the 97.8% correction figure) rest on rare-event counts at high SNR.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: IPC and latency shifts are simulation outputs under externally sourced PHY parameters, not identities forced by construction or self-citation.

full rationale

DICE’s load-bearing chain is (1) PHY component models (QC-LDPC, PAM4, AWGN+jitter+XT, layered min-sum, flit-level flow control) with defaults taken from IEEE HIR, UCIe, PCI-SIG, and public datasheets (Table III; §III-C–G); (2) runtime gem5 execution that produces packet-latency distributions and IPC; (3) comparison to HeteroGarnet fixed-latency links and to external AMD EPYC/ThreadRipper C2C measurements (§IV-B1, Fig. 12, Table V). Parity bytes, iteration budget N=4, and SNR_base≈35 dB are chosen from sensitivity/FER plots and roadmaps (Figs. 5, 9, 10), not by fitting so that IPC equals a target. Hardware synthesis calibrates encoder/decoder cycle costs independently of application IPC. Validation RMSE improvements versus real processors are post-hoc checks, not inputs that define the reported 6.8%/27.6% IPC shifts. Any dependence of results on the chosen default stack is ordinary model sensitivity, not circular reduction of a claimed prediction to its fitted inputs. The GT/s vs jitter-dB calibration tension raised by the skeptic is a correctness/consistency issue, not circularity under this pass’s definitions. No self-definitional loop, fitted-input-as-prediction, load-bearing self-citation uniqueness claim, or renamed known identity appears in the derivation.

Assumptions & free parameters 7 free parameters · 5 assumptions · 2 invented entities

The central IPC/latency claims rest on a stack of standard communications models plus several hand-chosen operating points and a specific FEC/flow-control microarchitecture. No new physical particle or force is invented; the ‘new entity’ is the DICE tool and boundary router. Free parameters are the channel and code knobs that set error and iteration behavior; axioms are AWGN/SNR composition, QC-LDPC suitability, and that C2C latency is a meaningful fidelity proxy for full-system PHY effects.

free parameters (7)
  • SNR_base = ≈35 dB
    Default intrinsic channel quality set to ~35 dB from IEEE HIR; dominates pre-FEC FER and whether 2-byte parity suffices.
  • jitter_RMS_sigma_t = ≈1 ps (→ ~26 dB SNR_jitter)
    Clock-edge jitter used in SNR_jitter formula; set to ~1 ps from PCI-SIG-order guidance.
  • SNR_crosstalk = ≈20 dB
    Fixed crosstalk contribution in harmonic SNR sum; taken from UCIe SI guidance.
  • parity_bytes_per_flit / code_rate_R = 2 bytes / R≈0.88
    Chosen via FER sensitivity sweet-spot study (Fig. 5), not derived from a uniqueness theorem; sets bandwidth overhead and post-FEC residual errors.
  • decoder_iteration_budget_N = N=4 (often converges in ≤2 at 35 dB)
    Cap on layered min-sum iterations; set empirically from convergence violins at baseline SNR (Fig. 10).
  • PAM4_swing_d_and_symbol_rate = d=50 mV; up to 32 GT/s
    Voltage levels and 32 GT/s rate set serialization time and noise variance via Es and sigma.
  • per_iteration_and_encode_latencies_in_cycles = encode 1 cy; syndrome 1 cy; iter 1 cy
    Mapped from synthesis/timing targets into gem5 cycle counts (1-cycle encode/modulate pieces; Latency(N)=2N+1).
assumptions (5)
  • domain assumption Inter-chiplet impairments can be aggregated as independent AWGN sources via harmonic sum of linear SNRs (base, jitter, crosstalk).
    §III-E Eq. (5); standard comms approximation, not re-derived from package EM.
  • domain assumption QC-LDPC with flit-granularity encoding and layered min-sum decoding is an appropriate model of near-future chiplet FEC behavior.
    Motivated by HIR/UCIe trends and SSD/high-speed FEC practice (§I, §III-C, §III-G); production Infinity Fabric today is described as CRC-oriented in the paper’s own comparison.
  • ad hoc to paper Cut-through flit-level ACK/NACK PHY flow control with packet-granularity buffer reservation correctly captures backpressure and retransmission effects on NoC timing.
    §III-H microarchitecture choice specific to DICE’s integration.
  • domain assumption Maximum and average core-to-core latency on Linux full-system runs is a valid external benchmark for PHY-model fidelity relevant to application IPC.
    §IV-B1 validation design; authors argue tails matter more than means for OoO performance (§IV-D).
  • standard math Standard digital communications identities for PAM4 Gray mapping, LLR soft demodulation, and QC-LDPC parity-check constraints.
    §III-D–G; textbook constructions used as implementation specs.
invented entities (2)
  • DICE end-to-end PHY pipeline in gem5 (encode→PAM4→AWGN link→LLR→iterative decode→PHY flow control) independent evidence
    purpose: Replace fixed-latency inter-chiplet links with runtime-variable timing and errors inside full-system simulation.
    Primary contribution; a software model, not a claimed new physical mediator. Independent evidence partial via C2C hardware comparison, not via open silicon PHY traces of the same FEC stack.
  • Chiplet-boundary PHY router output/input units with send/receive buffers and modulation arbiter
    purpose: Implement serialization, FEC, and flit ACK/NACK at CCD/IOD edges.
    Microarchitectural construct introduced to host the PHY stack (§III-H, Fig. 11).

how reviews work

0 comments
Cite this review

Pith. "Pith review of DICE: Detailed Inter-Chiplet End-to-End PHY Modeling for Accurate Chiplet Simulation." pith.science (2026). https://pith.science/paper/SQFSMDCE

@misc{pith2026260724221,
  author       = {Pith},
  title        = {Pith review of: DICE: Detailed Inter-Chiplet End-to-End PHY Modeling for Accurate Chiplet Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SQFSMDCE}},
  note         = {Machine review of arXiv:2607.24221}
}
read the original abstract

Scaling monolithic multicores is increasingly constrained by power/thermal limits, yield, and rising manufacturing and testing costs. Chiplet designs address these challenges by partitioning large dies into smaller parts (typically multiple core-complex dies and an I/O die) linked via high-bandwidth physical fabrics (PHY). As bandwidth and wiring density scale, however, these short-reach links are pushed closer to their signal-integrity limits, increasing susceptibility to noise, crosstalk, and channel loss, motivating stronger link-level reliability mechanisms such as forward error correction (FEC). Despite this trend, state-of-the-art simulation infrastructures often approximate inter-chiplet links using oversimplified, fixed-latency models. Such abstractions overlook the inherently dynamic, runtime-dependent behavior of the PHY -- including channel conditions (e.g., signal-to-noise ratio shifts, signal crosstalk, clock jitter), iterative decoder convergence and packet retransmissions, and application dynamics (e.g., LLC-misses that travel across chiplet boundaries) -- all of which are hard to determine offline. We show that neglecting these effects distorts inter-chiplet packet-level timing and high-level performance metrics such as IPC, leading to off-trend simulation results. We present DICE, an in-simulation, runtime PHY modeling in gem5 that captures the end-to-end inter-chiplet datapath, including QC-LDPC encoding/decoding, PAM4 modulation, lossy-channel transmission, LLR-based demodulation, adaptive packet re-sending, and PHY-level flow control between chiplets.

Figures

Figures reproduced from arXiv: 2607.24221 by the authors.

Figure 1
Figure 1. An example chiplet architecture, where inter-chiplet [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. End-to-end inter-chiplet data communication in [PITH_FULL_IMAGE:figures/full_fig_p001_2.png] view at source ↗
Figure 3
Figure 3. Impact of PHY-realistic modeling in DICE on overall packet latency and system IPC. The problems. Despite the wide adoption of chiplet￾based design, most architectural simulators still “wire up” chiplets using interconnect models originally developed for monolithic dies, resulting in two problems. First, these models typically assume fixed link latencies, abstracting away the details of the inter-chiplet PHY, where p… view at source ↗
Figures from the paper (15 more)
Figure 4
Figure 4. Figure 4: Detailed end-to-end inter-chiplet communication mod [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 6
Figure 6. Figure 6: FEC-encoding for 128-bit flit with 2-byte parity bits. [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Logic synthesis results for flit- and packet-level FEC [PITH_FULL_IMAGE:figures/full_fig_p005_7.png]
Figure 8
Figure 8. Figure 8: Error distribution under two example base [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 9
Figure 9. Figure 9: Pre- and post-FEC FER sensitivity across varying [PITH_FULL_IMAGE:figures/full_fig_p006_9.png]
Figure 11
Figure 11. Figure 11: Microarchitecture for sender (Router A) and receiver [PITH_FULL_IMAGE:figures/full_fig_p008_11.png]
Figure 12
Figure 12. Figure 12: Validation of DICE against AMD EPYC 9454P (8 CCDs × 6 cores): C2C max latency for all 6 cores within one representative CCD, measuring communication from each of these cores to every other of the 47 cores in the system. As shown, by modeling PHY dynamics, DICE (with d…
Figure 13
Figure 13. Figure 13: Pre-FEC FER based on varied SNRbase. bc bfs cc leelamcf omnet. lu ocean radix radio.volr. kmns. sc XSBe. GMEAN 10 8 10 6 10 4 10 2 Post FEC FER 20dB 25dB 30dB 35dB 40dB 45dB [PITH_FULL_IMAGE:figures/full_fig_p010_13.png]
Figure 14
Figure 14. Figure 14: Post-FEC FER based on varied SNRbase. (SNRbase) with jitter and crosstalk via Equation 5. While jitter and crosstalk are primarily determined by channel characteristics (e.g., link frequency and distance between wires), SNRbase drifts with runtime operating conditions…
Figure 15
Figure 15. Figure 15: Normalized avg packet latency across symbol rates. [PITH_FULL_IMAGE:figures/full_fig_p011_15.png]
Figure 16
Figure 16. Figure 16: Normalized execution time across symbol rates. [PITH_FULL_IMAGE:figures/full_fig_p011_16.png]
Figure 19
Figure 19. Figure 19: Normalized APL in global- vs. local-shared LLC. bc bfs cc leelamcf omnet. lu ocean radix radio.volr. kmns. sc XSBe. GMEAN 0 1 2 3 4 Execution time (norm.) Monolithic Garnet GS Garnet LS DICE GS DICE LS [PITH_FULL_IMAGE:figures/full_fig_p012_19.png]
Figure 20
Figure 20. Figure 20: Normalized exe. time in global- vs. local-shared LLC. work [63], [64]. As is evident, a GS LLC introduces higher packet latency for both DICE and HG. The discrepancy be￾tween DICE and HG varies significantly between benchmarks with the resulting DICE GS performance be…
Figure 22
Figure 22. Figure 22: HG, HG+, and DICE: Latency variability and resulting IPC for 3 high-MPKI workloads. Note: In bc, HG+ induces long backlogs that ultimately lead to simulation failure. radix radiosity XSBench GMEAN 0 1 2 3 4 Norm. avg load latency 1.00 1.06 1.36 1.13 1.00 1.13 1.70 1.2…
Figure 23
Figure 23. Figure 23: Single- vs. multi-threaded: Load latency (left) and application execution time (right), normalized to Monolithic. DICE vs. 39.26 for HG), but rather is a result of the large difference of the tail latencies (61 cycles for HG vs. 104 cycles for DICE). Further, statical…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

69 extracted references · 2 linked inside Pith

  1. [1]

    Interconnect Design for Heterogeneous Integration of Chiplets in the AMD Instinct MI300X Accelerator,

    A. Smith, G. H. Loh, S. Naffziger, J. Wuu, N. Kalyanasundharam, E. Chapman, R. Swaminathan, T. Huang, W. Jung, A. Kaganov, H. McIntyre, and R. Mangaser, “Interconnect Design for Heterogeneous Integration of Chiplets in the AMD Instinct MI300X Accelerator, ”IEEE Micro, vol. 45, no. 1, pp. 57–66, 2025

    2025

  2. [2]

    Enabling Scalable Chiplet-Based Uniform Memory Architectures with Silicon Photonics,

    P. Fotouhi, S. Werner, J. Lowe-Power, and S. J. B. Yoo, “Enabling Scalable Chiplet-Based Uniform Memory Architectures with Silicon Photonics, ” inInternational Symposium on Memory Systems (MemSys), (New York, NY, USA), p. 222–334, Association for Computing Machinery, 2019

    2019

  3. [3]

    Pioneering Chiplet Technology and Design for the AMD EPYC and Ryzen Processor Families: Industrial Product,

    S. Naffziger, N. Beck, T. Burd, K. Lepak, G. H. Loh, M. Subramony, and S. White, “Pioneering Chiplet Technology and Design for the AMD EPYC and Ryzen Processor Families: Industrial Product, ” in International Symposium on Computer Architecture (ISCA), pp. 57–70, 2021

    2021

  4. [4]

    Active Interposer Technology for Chiplet- Based Advanced 3D System Architectures,

    P. Coudrain, J. Charbonnier, A. Garnier, P. Vivet, R. Vélard, A. Vinci, F. Ponthenier, A. Farcy, R. Segaud, P. Chausse, L. Arnaud, D. Lat- tard, E. Guthmuller, G. Romano, A. Gueugnot, F. Berger, J. Beltritti, T. Mourier, M. Gottardi, S. Minoret, C. Ribière, G. Romero, P.-E. Philip, Y. Exbrayat, D. Scevola, D. Campos, M. Argoud, N. Allouti, R. Eleouet, C. ...

    2019

  5. [5]

    Coherence Attacks and Countermeasures in Interposer- Based Chiplet Systems,

    G. A. Chacon, C. Williams, J. Knechtel, O. Sinanoglu, P. V. Gratz, and V. Soteriou, “Coherence Attacks and Countermeasures in Interposer- Based Chiplet Systems, ”ACM Transactions on Architecture and Code Optimization (TACO), vol. 21, no. 2, 2024

    2024

  6. [6]

    AMD Next-Generation “Zen 4

    R. Bhargava and K. Troester, “AMD Next-Generation “Zen 4” Core and 4th Gen AMD EPYC Server CPUs, ”IEEE Micro, vol. 44, no. 3, pp. 8–17, 2024

    2024

  7. [7]

    Accelerating Innovation Through a Standard Chiplet Inter- face: The Advanced Interface Bus (AIB),

    D. Kehlet, “Accelerating Innovation Through a Standard Chiplet Inter- face: The Advanced Interface Bus (AIB), ” white paper, Intel Corpora- tion, 2017

    2017

  8. [8]

    Universal Chiplet Interconnect Express (UCIe): An Open Industry Standard for Innovations With Chiplets at Package Level,

    D. Das Sharma, G. Pasdast, Z. Qian, and K. Aygun, “Universal Chiplet Interconnect Express (UCIe): An Open Industry Standard for Innovations With Chiplets at Package Level, ”IEEE Transactions on Components, Packaging and Manufacturing Technology (TCPMT), vol. 12, no. 9, pp. 1423–1431, 2022

    2022

Show all 69 references
  1. [9]

    Analysis of UCIe 48/64 GT/s Electrical Links,

    Z. Wu and J.-R. Guo, “Analysis of UCIe 48/64 GT/s Electrical Links, ” IEEE Open Journal of the Solid-State Circuits Society, pp. 1–1, 2025

    2025

  2. [10]

    Chapter 2: High Performance Computing (HPC),

    IEEE Heterogeneous Integration Roadmap Technical Working Group, “Chapter 2: High Performance Computing (HPC), ” inIEEE Heteroge- neous Integration Roadmap (HIR) 2024 Edition, IEEE Electronics Pack- aging Society, 2021

    2024

  3. [11]

    The Road to 64G UCIe IP: What Designers Need to Know,

    A. Tarde, “The Road to 64G UCIe IP: What Designers Need to Know, ” Oct. 2025. Synopsys. https://www.synopsys.com/articles/ucie- 3-0-64gbps-challenges.html

    2025

  4. [12]

    Kite: A Family of Heterogeneous Interposer Topologies Enabled via Accurate Inter- connect Modeling,

    S. Bharadwaj, J. Yin, B. Beckmann, and T. Krishna, “Kite: A Family of Heterogeneous Interposer Topologies Enabled via Accurate Inter- connect Modeling, ” inDesign Automation Conference (DAC), pp. 1–6, 2020

    2020

  5. [13]

    The gem5 Simulator: Version 20.0+,

    J. Lowe-Power, A. M. Ahmad, A. Akram, M. Alian, R. Amslinger, M. An- dreozzi, A. Armejach, N. Asmussen, S. Bharadwaj, G. Black, G. Bloom, B. R. Bruce, D. R. Carvalho, J. Castrillón, L. Chen, N. Derumigny, S. Diestelhorst, W. Elsasser, M. Fariborz, A. F. Farahani, P. Fotouhi, R...

    2007 arXiv

  6. [14]

    J. G. Proakis and M. Salehi,Digital Communications. McGraw-Hill, 5 ed., 2007

    2007

  7. [15]

    Quasi-Cyclic LDPC Codes for Fast Encoding,

    S. Myung, K. Yang, and J. Kim, “Quasi-Cyclic LDPC Codes for Fast Encoding, ”IEEE Transactions on Information Theory, vol. 51, no. 8, pp. 2894–2901, 2005

    2005

  8. [16]

    LDPC- in-SSD: Making Advanced Error Correction Codes Work Effectively in Solid State Drives,

    K. Zhao, W. Zhao, H. Sun, X. Zhang, N. Zheng, and T. Zhang, “LDPC- in-SSD: Making Advanced Error Correction Codes Work Effectively in Solid State Drives, ” inUSENIX Conference on File and Storage Technologies (FAST), pp. 243–256, 2013

    2013

  9. [17]

    The PCIe 6.0 Specification Webinar Q&A: Error Detection and Correction (FEC),

    PCI-SIG, “The PCIe 6.0 Specification Webinar Q&A: Error Detection and Correction (FEC), ” 2021

    2021

  10. [18]

    Low-Density Parity-Check Codes,

    R. Gallager, “Low-Density Parity-Check Codes, ”IRE Transactions on information theory, vol. 8, no. 1, pp. 21–28, 2003

    2003

  11. [19]

    GARNET: A Detailed On-Chip Network Model Inside a Full-System Simulator,

    N. Agarwal, T. Krishna, L.-S. Peh, and N. K. Jha, “GARNET: A Detailed On-Chip Network Model Inside a Full-System Simulator, ” in IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), pp. 33–42, IEEE, 2009

    2009

  12. [20]

    Sniper: Exploring the Level of Abstraction for Scalable and Accurate Parallel Multi-Core Simulation (SC),

    T. E. Carlson, W. Heirman, and L. Eeckhout, “Sniper: Exploring the Level of Abstraction for Scalable and Accurate Parallel Multi-Core Simulation (SC), ” inInternational Conference for High Performance Computing, Networking, Storage and Analysis (SC), pp. 1–12, 2011

    2011

  13. [21]

    DARSIM: A Parallel Cycle-Level NoC Simulator,

    M. Lis, K. S. Shim, M. H. Cho, P. Ren, O. Khan, and S. Devadas, “DARSIM: A Parallel Cycle-Level NoC Simulator, ” inAnnual Workshop on Modeling, Benchmarking and Simulation, 2010

    2010

  14. [22]

    A Detailed and Flexible Cycle- Accurate Network-on-Chip Simulator,

    N. Jiang, D. U. Becker, G. Michelogiannakis, J. Balfour, B. Towles, D. E. Shaw, J. Kim, and W. J. Dally, “A Detailed and Flexible Cycle- Accurate Network-on-Chip Simulator, ” inIEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), pp. 86–96, 2013

    2013

  15. [23]

    Noxim: An Open, Extensible and Cycle-Accurate Network on Chip Simulator,

    V. Catania, A. Mineo, S. Monteleone, M. Palesi, and D. Patti, “Noxim: An Open, Extensible and Cycle-Accurate Network on Chip Simulator, ” inIEEE International Conference on Application-Specific Systems, Archi- tectures and Processors (ASAP), pp. 162–163, IEEE, 2015

    2015

  16. [24]

    Muchisim: A Simulation Framework for Design Exploration of Multi-Chip Many- core Systems,

    M. Orenes-Vera, E. Tureci, M. Martonosi, and D. Wentzlaff, “Muchisim: A Simulation Framework for Design Exploration of Multi-Chip Many- core Systems, ” inInternational Symposium on Performance Analysis of Systems and Software (ISPASS), 2024

    2024

  17. [25]

    BZSim: Fast, Large-Scale Mi- croarchitectural Simulation with Detailed Interconnect Modeling,

    P. Strikos, A. Ejaz, and I. Sourdis, “BZSim: Fast, Large-Scale Mi- croarchitectural Simulation with Detailed Interconnect Modeling, ” in IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), pp. 167–178, IEEE, 2024

    2024

  18. [26]

    Evaluating Chiplet-based Large- Scale Interconnection Networks via Cycle-Accurate Packet-Parallel Simulation,

    Y. Feng, Y. Wei, D. Xiang, and K. Ma, “Evaluating Chiplet-based Large- Scale Interconnection Networks via Cycle-Accurate Packet-Parallel Simulation, ” inUSENIX Annual Technical Conference (ATC), (Santa Clara, CA), pp. 731–747, USENIX Association, July 2024

    2024

  19. [27]

    RapidChiplet: A Toolchain for Rapid Design Space Exploration of Inter-Chiplet Interconnects,

    P. Iff, B. Bruggmann, B. Morel, M. Besta, L. Benini, and T. Hoe- fler, “RapidChiplet: A Toolchain for Rapid Design Space Exploration of Inter-Chiplet Interconnects, ” inACM International Conference on Computing Frontiers (CF), CF ’25, (New York, NY, USA), p. 168–171, Associati...

    2025

  20. [28]

    32-bit Cyclic Redundancy Codes for Internet Applica- tions,

    P. Koopman, “32-bit Cyclic Redundancy Codes for Internet Applica- tions, ” inInternational Conference on Dependable Systems and Networks (DSN), pp. 459–468, 2002

    2002

  21. [29]

    Richardson and R

    T. Richardson and R. Urbanke,Modern Coding Theory. Cambridge University Press, 2008

    2008

  22. [30]

    Characterizing and Optimizing LDPC Performance on 3D NAND Flash Memories,

    Q. Li, Y. Chen, G. Wu, Y. Du, M. Ye, X. Gan, J. Zhang, Z. Shen, J. Shu, and C. Xue, “Characterizing and Optimizing LDPC Performance on 3D NAND Flash Memories, ”ACM Transactions on Architecture and Code Optimization (TACO), vol. 21, no. 3, 2024

    2024

  23. [31]

    D. R. Stauffer, S. Mirabbasi, and M. Zargari,High-Speed SerDes Devices and Applications. Cham, Switzerland: Springer, 2018

    2018

  24. [32]

    Clocking for PCIe Applications

    Texas Instruments, “Clocking for PCIe Applications. ” https://www.ti. com/lit/an/snaa386/snaa386.pdf, 2023. Application Report SNAA386

    2023

  25. [33]

    Per- formance Analysis of Bump in Tapered TSV: Impact on Crosstalk and Power Loss,

    S. Chandrakar, D. Gupta, M. K. Majumder, and B. K. Kaushik, “Per- formance Analysis of Bump in Tapered TSV: Impact on Crosstalk and Power Loss, ”IEEE Open Journal of Nanotechnology, vol. 3, pp. 227–235, 2022

    2022

  26. [34]

    A 32 Gb/s 0.36 pJ/bit 3 nm Chiplet IO Using 2.5- D CoWoS Package With Real-Time and Per-Lane CDR and Bathtub Monitoring,

    J. Gu, J. Ma, A. Ahmed Chowdhury, J. Guo, X. Zhang, J. Ding, H. Wang, and K. Chang, “A 32 Gb/s 0.36 pJ/bit 3 nm Chiplet IO Using 2.5- D CoWoS Package With Real-Time and Per-Lane CDR and Bathtub Monitoring, ”IEEE Journal of Solid-State Circuits (JSSC), vol. 60, no. 4, pp. 1289–...

    2025

  27. [35]

    Beyond CPO: A Motivation and Approach for Bringing Optics Onto the Silicon Interposer,

    B. G. Lee, N. Nedovic, T. H. Greer, and C. T. Gray, “Beyond CPO: A Motivation and Approach for Bringing Optics Onto the Silicon Interposer, ”Journal of Lightwave Technology, vol. 41, no. 4, pp. 1152– 1162, 2023

    2023

  28. [36]

    Digital Clock and Data Recovery Circuit Design: Challenges and Tradeoffs,

    M. Talegaonkar, R. Inti, and P. K. Hanumolu, “Digital Clock and Data Recovery Circuit Design: Challenges and Tradeoffs, ” inIEEE Custom Integrated Circuits Conference (CICC), pp. 1–8, 2011

    2011

  29. [37]

    An Equalizer Adaptation Algorithm to Reduce Jitter in Binary Receivers,

    A. Carusone, “An Equalizer Adaptation Algorithm to Reduce Jitter in Binary Receivers, ”IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 53, no. 9, pp. 807–811, 2006

    2006

  30. [38]

    A Study of BER and EVM Degradation in Digital Modulation Schemes Due to PLL Jitter and Communication- Link Noise,

    M. Oveisi and P. Heydari, “A Study of BER and EVM Degradation in Digital Modulation Schemes Due to PLL Jitter and Communication- Link Noise, ”IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 69, no. 8, pp. 3402–3415, 2022

    2022

  31. [39]

    A Novel Interposer Channel Structure with Vertical Tabbed Vias to Reduce Far-End Crosstalk for Next-Generation High-Bandwidth Memory,

    H. Kim, S. Lee, K. Song, Y. Shin, D. Park, J. Park, J. Cho, and S. Ahn, “A Novel Interposer Channel Structure with Vertical Tabbed Vias to Reduce Far-End Crosstalk for Next-Generation High-Bandwidth Memory, ”Micromachines, vol. 13, no. 7, p. 1070, 2022

    2022

  32. [40]

    Seamless Transition to PCIe 5.0 Technology in System Implementations Webinar Q&A,

    PCI-SIG, “Seamless Transition to PCIe 5.0 Technology in System Implementations Webinar Q&A, ” 2021

    2021

  33. [41]

    Optimized Signal and Power Integrity of Silicon Interposer for HBM2E in CoWoS Packaging,

    K.-J. Lin, C.-M. Lin, and R.-B. Wu, “Optimized Signal and Power Integrity of Silicon Interposer for HBM2E in CoWoS Packaging, ”IEEE Transactions on Signal and Power Integrity, vol. 3, pp. 159–168, 2024

    2024

  34. [42]

    High-Bandwidth Chiplet Interconnects for Advanced Packaging Technologies in AI/ML Appli- cations: Challenges and Solutions,

    S. Li, M.-S. Lin, W.-C. Chen, and C.-C. Tsai, “High-Bandwidth Chiplet Interconnects for Advanced Packaging Technologies in AI/ML Appli- cations: Challenges and Solutions, ”IEEE Open Journal of the Solid-State Circuits Society, vol. 4, pp. 351–364, 2024

    2024

  35. [43]

    Achieving Better Chiplet Design Signal Integrity with UCIe,

    T. Wang-Lee, “Achieving Better Chiplet Design Signal Integrity with UCIe, ” 2024. Post on the UCIe orgnization’s official web- site. https://www.uciexpress.org/post/achieving-better-chiplet-design- signal-integrity-with-ucie

    2024

  36. [44]

    UCIe Electrical, Form-Factor, and Compliance,

    UCIe Consortium, “UCIe Electrical, Form-Factor, and Compliance, ” in Proceedings of the Hot Chips 2023 Tutorial Sessions, Aug. 2023

    2023

  37. [45]

    Signal Integrity Design and Analysis of Universal Chiplet Interconnect Express (UCIe) Channel in Silicon Interposer for Advanced Package,

    T. Shin, K. Kim, H. Park, B. Sim, S. Kim, J. Kim, S. Choi, J. Park, J. Song, J. Kim, J. W. Park, D. Kang, and J. Kim, “Signal Integrity Design and Analysis of Universal Chiplet Interconnect Express (UCIe) Channel in Silicon Interposer for Advanced Package, ” inIEEE Electrical ...

    2023

  38. [46]

    Bit-Interleaved Coded Modula- tion,

    G. Caire, G. Taricco, and E. Biglieri, “Bit-Interleaved Coded Modula- tion, ”IEEE Transactions on Information Theory, vol. 44, no. 3, pp. 927– 946, 1998

    1998

  39. [47]

    A Reduced Complexity Decoder Architecture via Layered Decoding of LDPC Codes,

    D. Hocevar, “A Reduced Complexity Decoder Architecture via Layered Decoding of LDPC Codes, ” inIEEE Workshop on Signal Processing Systems, pp. 107–112, 2004

    2004

  40. [48]

    UCIe Specification Revision 2.0,

    Universal Chiplet Interconnect Express (UCIe) Consortium, “UCIe Specification Revision 2.0, ” 2024

    2024

  41. [49]

    Pre-FEC and Post-FEC BER as Criteria for Optimizing Wireline Transceivers,

    M. Yang, S. Shahramian, H. Wong, P. Krotnev, and A. C. Carusone, “Pre-FEC and Post-FEC BER as Criteria for Optimizing Wireline Transceivers, ” inInternational Symposium on Circuits and Systems (ISCAS), pp. 1–5, 2021

    2021

  42. [50]

    The GAP Benchmark Suite,

    S. Beamer, K. Asanovic, and D. A. Patterson, “The GAP Benchmark Suite, ”CoRR, vol. abs/1508.03619, 2015

    2015 arXiv

  43. [51]

    SPEC Bench- marks and Tools

    Standard Performance Evaluation Corporation (SPEC), “SPEC Bench- marks and Tools. ” https://www.spec.org/, 2017

    2017

  44. [52]

    Splash- 4: A Modern Benchmark Suite with Lock-Free Constructs,

    E. J. Gómez-Hernández, J. M. Cebrian, S. Kaxiras, and A. Ros, “Splash- 4: A Modern Benchmark Suite with Lock-Free Constructs, ” inIEEE International Symposium on Workload Characterization (IISWC), pp. 51– 64, 2022

    2022

  45. [53]

    Rodinia: A Benchmark Suite for Heterogeneous Com- puting,

    S. Che, M. Boyer, J. Meng, D. Tarjan, J. W. Sheaffer, S.-H. Lee, and K. Skadron, “Rodinia: A Benchmark Suite for Heterogeneous Com- puting, ” inIEEE International Symposium on Workload Characterization (IISWC), pp. 44–54, 2009

    2009

  46. [54]

    XSBench: The Development and Verification of a Performance Abstraction for Monte Carlo Reactor Analysis,

    C. R. Wilson, A. T. Evans, A. R. Siegel, and T. Islam, “XSBench: The Development and Verification of a Performance Abstraction for Monte Carlo Reactor Analysis, ” inPHYSOR: The Role of Reactor Physics Toward a Sustainable Future, 2014

    2014

  47. [55]

    Core-to-Core Latency Measurement Tool

    N. Viennot, “Core-to-Core Latency Measurement Tool. ” https://github. com/nviennot/core-to-core-latency, 2026

    2026

  48. [56]

    Addressing End-to-End Memory Access Latency in NoC-Based Multicores,

    A. Sharifi, E. Kultursay, M. Kandemir, and C. R. Das, “Addressing End-to-End Memory Access Latency in NoC-Based Multicores, ” in International Symposium on Microarchitecture (MICRO), pp. 294–304, 2012

    2012

  49. [57]

    Pitstop: Enabling a Virtual Net- work Free Network-on-Chip,

    H. Farrokhbakht, H. Kao, K. Hasan, P. V. Gratz, T. Krishna, J. San Miguel, and N. E. Jerger, “Pitstop: Enabling a Virtual Net- work Free Network-on-Chip, ” inInternational Symposium on High- Performance Computer Architecture (HPCA), pp. 682–695, 2021

    2021

  50. [58]

    A Quad-Chiplet AI SoC with Full-Chip Scalable Mesh Over 16Gb/s UCIe-Advanced Die-to-Die Interface for Large-Scale AI Inferencing,

    C.-H. Yu, J. Bae, J. Kim, H. Kim, W. Shin, J.-S. Yoon, Y.-J. Jin, J. Oh, J. Lee, E. Kim,et al., “A Quad-Chiplet AI SoC with Full-Chip Scalable Mesh Over 16Gb/s UCIe-Advanced Die-to-Die Interface for Large-Scale AI Inferencing, ” inIEEE International Solid-State Circuits Confer...

    2026

  51. [59]

    Zeppelin: An SoC for Multichip Architectures,

    N. Beck, S. White, M. Paraschou, and S. Naffziger, “Zeppelin: An SoC for Multichip Architectures, ” inIEEE International Solid-State Circuits Conference (ISSCC), pp. 40–42, IEEE, 2018

    2018

  52. [60]

    Compute Express Link (CXL) 3.1 Specification

    CXL Consortium, “Compute Express Link (CXL) 3.1 Specification. ” https://computeexpresslink.org/cxl-specification/, 2023

    2023

  53. [61]

    Scaling Out Chip Interconnect Networks with Implicit Sequence Numbers,

    G. Jung, S. Gorgin, J. Kim, and J. Kim, “Scaling Out Chip Interconnect Networks with Implicit Sequence Numbers, ” inInternational Conference for High Performance Computing, Networking, Storage and Analysis (SC), pp. 1240–1251, 2025

    2025

  54. [62]

    Sapphire Rapids: The Next-Generation Intel Xeon Scalable Processor,

    N. Nassif, A. O. Munch, C. L. Molnar, G. Pasdast, S. V. Lyer, Z. Yang, O. Mendoza, M. Huddart, S. Venkataraman, S. Kandula, R. Marom, A. M. Kern, W. J. Bowhill, D. R. Mulvihill, S. Nimmagadda, V. Kalidindi, J. Krause, M. M. Haq, R. Sharma, and K. Duda, “Sapphire Rapids: The Ne...

    2022

  55. [63]

    Invalidate or Update? Revisiting Coherence for Tomorrow’s Cache Hierarchies,

    M. Zhu, A. Shahab, A. Katsarakis, and B. Grot, “Invalidate or Update? Revisiting Coherence for Tomorrow’s Cache Hierarchies, ” inInterna- tional Conference on Parallel Architectures and Compilation Techniques (PACT), pp. 226–241, 2021

    2021

  56. [64]

    An Adaptive Cache Coherence Pro- tocol Optimized for Producer-Consumer Sharing,

    L. Cheng, J. B. Carter, and D. Dai, “An Adaptive Cache Coherence Pro- tocol Optimized for Producer-Consumer Sharing, ” inInternational Sym- posium on High Performance Computer Architecture (HPCA), pp. 328– 339, 2007

    2007

  57. [65]

    Anatomy of the gem5 Simulator

    J. Söderström, R. Aligholipour, and Y. Yao, “Anatomy of the gem5 Simulator. ” Presented at the gem5 Workshop at ISCA’25, 2025

    2025

  58. [66]

    Demystify- ing Complex Workload-DRAM Interactions: An Experimental Study,

    S. Ghose, T. Li, N. Hajinazar, D. S. Cali, and O. Mutlu, “Demystify- ing Complex Workload-DRAM Interactions: An Experimental Study, ” Proceedings of the ACM on Measurement and Analysis of Computing Systems (SIGMETRICS), vol. 3, no. 3, pp. 1–50, 2019

    2019

  59. [67]

    A First-Order Superscalar Processor Model,

    T. S. Karkhanis and J. E. Smith, “A First-Order Superscalar Processor Model, ” inInternational Symposium on Computer Architecture (ISCA), pp. 338–349, 2004

    2004

  60. [68]

    A Mechanistic Performance Model for Superscalar Out-of-Order Processors,

    S. Eyerman, L. Eeckhout, T. Karkhanis, and J. E. Smith, “A Mechanistic Performance Model for Superscalar Out-of-Order Processors, ”ACM Transactions on Computer Systems (TOCS), vol. 27, no. 2, pp. 1–37, 2009

    2009

  61. [69]

    Callback: Efficient Synchronization without Invalidation with a Directory Just for Spin-Waiting,

    A. Ros and S. Kaxiras, “Callback: Efficient Synchronization without Invalidation with a Directory Just for Spin-Waiting, ” inInternational Symposium on Computer Architecture (ISCA), pp. 427–438, 2015

    2015

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.