Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms

Behnam Khaleghi; Mohsen Imani; Sahand Salamat; Tajana Rosing

REVIEW 3 major objections 4 minor 47 references

Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper claims that scaling the core and block-RAM voltage rails together, guided by a pre-characterized library and workload prediction, cuts power by 4.0x on average across five neural-network accelerators while preserving quality of…

desk verdict A genuinely useful dual-rail DVFS framework, but the headline 4.0X power saving depends on a static critical-path model that the paper itself admits can break under voltage scaling. read the letter →

arxiv 1908.06519 v2 pith:43MDPFCT submitted 2019-08-18 cs.AR cs.PF

Sahand Salamat , Behnam Khaleghi , Mohsen Imani , Tajana Rosing This is my paper · ORCID

classification cs.ARcs.PF

keywords FPGADVFSmulti-FPGAdatacentersvoltagescalingworkloadpredictionblockRAMneuralnetworkacceleratorspowerefficiency

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

Multi-FPGA platforms in data centers often run far below peak capacity, yet they burn close to peak power. This paper claims that the wasted slack can be converted into large energy savings by scaling two voltage rails—the core rail and the block-RAM rail—in tandem, rather than power-gating nodes or scaling a single rail. The key is a pre-characterized library of how each FPGA resource's delay and power respond to voltage, which the authors show is necessary because logic, routing, memory, and DSP blocks have very different sensitivities. On five neural-network accelerators, the framework reports an average 4.0x power reduction while maintaining quality of service, beating core-only DVFS by 33.6% and BRAM-only scaling by up to 83%.

What carries the argument

The central machinery is a two-rail voltage optimization driven by resource-level characterization curves. For a critical path with logic/routing delay $d_{l0}$ and memory delay $d_{m0}$, the paper uses $d_{cp}=d_{l0}D_l(V_{core})+d_{m0}D_m(V_{bram})$ and the constraint $D_l(V_{core})+\alpha D_m(V_{bram})\leq(1+\alpha)S_w$, where $\alpha=d_{m0}/d_{l0}$ is the memory-to-logic delay ratio and $S_w$ is the workload stretch factor; a companion power model $p_{cir}\propto P_l(V_{core},d_{cp})+\beta P_m(V_{bram},d_{cp})$ ranks the timing-feasible pairs. The optimal pair is precomputed at design time and looked up at runtime, so the runtime controller only predicts the workload, sets frequency, and fetches the stored voltages.

What would settle it

Measure actual timing margins with on-chip or external monitors while a neural-network accelerator runs under a 30% workload at the framework's selected $(V_{core}, V_{bram})$. If any path fails to meet the stretched clock period, or the accelerator's output error rate exceeds the QoS bound, the fixed-critical-path model is false. A cheaper check is to run static timing analysis at the selected low-voltage pair and see whether a path not included in the characterization becomes critical.

Watch

Extended reading notes

Core claim

The paper argues that for any target frequency there are many $(V_{core}, V_{bram})$ pairs that satisfy timing, but exactly one pair minimizes power, and finding that pair requires knowing the voltage-delay and voltage-power behavior of the resources on the critical path. It models critical-path delay as the sum of a logic/routing term scaled by $D_l(V_{core})$ and a memory term scaled by $D_m(V_{bram})$, and uses this constraint together with a power model to select voltages from a pre-characterized library. A lightweight workload predictor sets the frequency for the next time step, and the voltage selector then picks the lowest-power pair that still meets the stretched clock period. In experiments with five neural-network accelerators under a bursty, self-similar workload, the method achieves 4.0x average power reduction versus nominal operation and outperforms single-rail core scaling by 33.6% and memory-only scaling by up to 83%, while preserving throughput QoS.

Load-bearing premise

The load-bearing premise is that the critical path characterized at design time stays critical when both $V_{core}$ and $V_{bram}$ are lowered; if a different path takes over at low voltage, the framework's voltage pair can violate timing and QoS.

Editorial extensions

If this is right

At the 10-50% utilizations typical of FPGA clouds, the approach converts underutilization into roughly proportional power savings without waking or power-gating compute nodes.
Dual-rail scaling is consistently better than single-rail DVFS: 33.6% average over core-only, and 83% over BRAM-only in the best benchmark, because it exploits the different voltage sensitivities of logic and memory.
The same accelerator can run at different optimal voltages under different workload bins, so a static voltage setting leaves energy on the table and runtime adaptation is required.
The two-PLL clocking scheme removes the PLL relock stall, making frequent voltage and frequency updates cheap enough for time steps in the millisecond-to-second range.
Because voltage selection is based on pre-characterized resource curves rather than per-application re-characterization, the framework is portable across accelerator designs on the same FPGA architecture.

Reading between the lines

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

A natural generalization: whenever two voltage domains have different delay-voltage slopes, the energy-optimal point is a joint solution; the paper's single-rail intuition of scaling only the rail with headroom is a special case.
The same characterization library could support thermal or aging-aware scaling: if the library included temperature as a parameter, the controller could back off voltages as leakage rises, a direction the paper addresses only indirectly.
The fixed-critical-path assumption is testable: run static timing analysis at the chosen low-voltage pairs and check whether a non-characterized path becomes critical; the paper itself cites related work noting this risk, so a direct measurement would either confirm or bound the framework's safe operating range.
For latency-sensitive workloads, the throughput-based QoS model would need extension to per-request deadlines; the frequency-setting rule would then be driven by worst-case latency rather than average arrival rate.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Read this as a dual-rail DVFS paper, not a measurement paper. The core idea is sound and worth engaging: instead of scaling Vcore or Vbram alone, the framework uses a pre-characterized resource delay/power library to pick the pair that meets throughput at minimum power. The motivation in Section III is clear and the figures make the case that resource sensitivities differ. The dependency on COFFE/PTM for characterization is appropriate and reproducible, and running five DNN accelerators through the same flow is honest work.

The soft spot is the critical-path model. Equation (1) assumes one fixed critical path with known logic/routing and memory delays, and the voltage selector uses that path to choose operating points. When you scale two rails independently, a different path with a different resource mix can become critical. The paper itself cites Zhao et al. acknowledging exactly this risk, but the proposed framework never re-evaluates path criticality, and Section V says voltages are computed at synthesis time and stored. If the true critical path after scaling is not the one in Eq. (1), the chosen pair can violate timing and break QoS. That is a direct threat to the headline 4.0X number. I don't think the idea is wrong, but the timing-safety claim needs either all-path static timing analysis across the voltage range or a runtime guard.

Two smaller issues. The workload predictor is never validated; Figure 10 assumes a workload but shows no prediction accuracy or how often the 5% margin covers mispredictions. And the evaluation is entirely simulated — COFFE, VTR, SPICE-derived libraries — so the 4.0X is an estimate, not a measured result. The paper should say that plainly in the abstract. The DVS overhead dismissal via a fast on-chip converter is also a bit quick, since board-level PMBUS rail changes are usually slower; minor.

Who this is for: FPGA architects and data-center power management researchers. It is an incremental but real step beyond single-rail DVFS. I would send it to peer review with a request for a revised version that addresses the critical-path reordering issue, either by proof that Eq. (1) is conservative or by adding runtime slack monitoring.

Referee Report

3 major / 4 minor

Summary. The paper proposes a framework for reducing power consumption of multi-FPGA data-center platforms by dynamically scaling both the core voltage (Vcore) and block-RAM voltage (Vbram) in response to predicted workload levels. The approach uses a pre-characterized delay-voltage and power-voltage library for FPGA resources, models the critical path delay as a weighted sum of logic/routing and memory delays (Eq. 1), and selects voltage/frequency points at design synthesis time. Workload prediction is done with a lightweight Markov-chain model with a 5% throughput margin. The authors evaluate the method on five neural-network accelerators (Tabla, DnnWeaver, DianNao, Stripes, Proteus) using COFFE/PTM characterization, Quartus synthesis, and VTR place-and-route, reporting an average 4.0× power reduction, 33.6% better than core-only voltage scaling and up to 83% better than BRAM-only scaling.

Significance. If the claimed 4.0× power reduction is reliable, the paper addresses an important problem: multi-FPGA data centers are often underutilized, and existing DVFS approaches typically scale only the core rail or only the memory rail. The idea of jointly and independently scaling Vcore and Vbram using a pre-characterized resource library is novel and practically motivated, and the resource library is based on external tools (COFFE/PTM, NanGate) rather than fitted constants, which is a strength. The approach also makes falsifiable predictions about which voltage pairs are optimal for given resource mixes. However, the evaluation is simulation-only, uses a single synthetic workload trace, never evaluates the workload predictor's accuracy or the resulting QoS, and rests on a critical-path model that the paper itself acknowledges can fail under voltage scaling. With additional verification, the result would be significant for FPGA-based cloud acceleration.

major comments (3)

[Section III, Eq. (1); Section V] The timing model in Eq. (1) assumes a single fixed critical path whose memory-to-logic delay ratio α remains constant after voltage scaling. The paper itself notes in Section II, in the discussion of Zhao et al. [25], that 'originally non-critical paths might become critical when the voltage changes.' Since Section V states that the optimal voltage pairs are calculated during design synthesis from this model, the selected (Vcore, Vbram) points may violate timing if a different path becomes critical after scaling. Figure 1 shows that logic, routing, memory, and DSP have materially different delay-voltage sensitivities, making path reordering plausible. The reported 4.0× power reduction and the QoS guarantee therefore depend on an unverified assumption. Please re-evaluate critical paths at the selected voltage pairs (e.g., via path enumeration or static timing analysis at each voltage point) and report the actual critical paths and slack for the five benchmarks; if path reordering occurs, the voltage selection must consider multiple paths.
[Section IV-A; Section VI-B] The central QoS claim is never directly evaluated. Section IV-A describes a Markov-chain workload predictor with a t=5% throughput margin, but no experiment reports prediction accuracy, misprediction rate, or whether workload under-estimation actually causes throughput or timing violations. Figures 10 and 12 show normalized power under an assumed workload trace, but not the predicted versus actual workload, and no QoS metric is defined or measured. The power-saving numbers assume the frequency/voltage pair chosen for each step is always safe. The 5% margin is asserted by citing [37] and not validated for this workload or this DVFS scheme. Please report prediction accuracy, end-to-end throughput relative to the QoS target, and the power impact of mispredictions.
[Section VI-A; Section VI-B; Section III Eq. (3)] The 4.0× average power reduction is obtained entirely from an analytical model, not from measurements on a physical FPGA. The model is never calibrated against silicon, and the per-benchmark values of α and β in Eqs. (1)–(3) are not reported, even though Section VI-B asserts that α parameters are 'close' across accelerators. Without these values and without any validation of the delay/power model at actual voltage points, the reader cannot assess how much of the reported gain is due to the voltage-selection algorithm versus the model's assumptions. Please report the extracted α and β values for each accelerator and, if possible, validate the delay and power predictions on at least one commercial FPGA across a range of voltage pairs.

minor comments (4)

[Section IV-C] The text says 'This adopter is used as a proof of concept'; 'adopter' should be 'adapter'.
[Section IV-A] The sentence 'Considering the The total probability of the outgoing edges of state Si has to be 1' contains a grammar error ('The The') and should be reworded.
[Section VI-A] The paper does not report the time-step duration τ used in the experiments, although the PLL overhead analysis in Section V explicitly depends on τ. Please state the value used.
[Section III, Fig. 4] In Figure 4, the power-gating (PG) curve is described as working better at very low workloads, but the text does not specify the crossover point or how PG is modeled; a sentence clarifying the PG assumption (e.g., number of nodes powered off, wake-up overhead) would help reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the dual-rail voltage-scaling optimization is driven by external characterization libraries and workload prediction, not by the reported 4.0x power result.

full rationale

The paper's derivation chain is self-contained rather than circular. The reported average power reduction is obtained by synthesizing five accelerators with Quartus/VTR, characterizing FPGA resource delay and power with COFFE/PTM and external process/architecture data, and then selecting Vcore/Vbram pairs using Eq. 2 and Eq. 3. The key parameters (alpha, beta, dl0, dm0) are either derived from post-place-and-route critical paths or taken from external sources, not fitted to the 4.0x result or to the comparison numbers in Table II. The comparison against core-only and bram-only schemes evaluates the same physical model with one rail fixed, so the proposed two-rail optimizer's advantage follows from having more degrees of freedom; that is a modeling implication, not circularity. Self-citations such as [29] appear in related work and in the general idea of resource pre-characterization, but the actual characterization in Section VI is independently generated with COFFE/PTM and VTR, and the workload predictor is adapted from the external PRESS work [37], so no load-bearing self-citation is present. The fixed-critical-path assumption in Eq. 1 is a genuine correctness and fidelity risk under dual-rail voltage scaling, since a different path may become critical after scaling; however, this is a modeling-validity concern, not circularity, because the delay model's inputs are external characterizations rather than derivatives of the claimed power savings.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The quantitative headline is built on a small number of hand-set parameters (alpha, beta, crash voltage) and on external SPICE-based characterization. The prediction and transition-overhead assumptions are asserted without experimental support.

free parameters (7)

alpha (memory-to-logic/routing critical-path delay ratio) = 0.2 (motivational analysis); asserted close across benchmarks but not reported
Appears in Eq. (2); the optimal voltage point depends on alpha. The experiments do not list per-benchmark alpha values.
beta (BRAM power contribution ratio) = 0.4 (motivational analysis)
Appears in Eq. (3); determines the weight of BRAM power in the objective. Per-benchmark beta values are not reported.
t (throughput margin for misprediction) = 5%
Used to offset workload underestimation; adopted from [37] and not tuned or evaluated here.
Crash voltage = ~0.50V
Set as the lower bound of voltage scaling in the optimization; taken from the characterization, not from hardware measurement.
PLL power (P_PLL) = 0.1W
Hand-assumed value used in the two-PLL overhead argument (Eq. 5).
Design power (P_design) = ~20W
Hand-assumed fully utilized FPGA power used in Eq. (5).
PLL lock time (t_lock) = ~10 microseconds
Hand-assumed PLL lock time used in Eq. (4)-(5).

assumptions (5)

domain assumption The critical path delay is a weighted sum of logic/routing delay and memory delay, with each part scaling independently with its rail voltage (Eq. 1).
Used throughout Section III and V to select voltages; ignores other paths that could become critical and assumes memory/core delays are independent.
domain assumption Total circuit power is a weighted sum of core-rail power and memory-rail power with a fixed activity factor beta (Eq. 3).
Used to choose the minimum-power voltage pair; per-benchmark beta is not measured.
domain assumption COFFE/PTM 22nm characterization of logic, routing, memory, and DSP delay/power versus voltage (Figs. 1-3) represents the target FPGA platform.
All quantitative results depend on this external simulation model; no silicon validation is provided.
ad hoc to paper The voltage change does not make a different path critical; a single static critical path with fixed alpha is sufficient.
Eq. (1) uses one alpha; the paper cites the path-reordering risk in prior work [25] but does not address it in the timing model.
ad hoc to paper Workload prediction is accurate enough with a 5% margin, and DVS transition latency is negligible.
Section IV-A and IV-C; the experiments do not model misprediction, and the DVS latency claim relies on an on-chip converter [39] that is not representative of board-level regulators.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms." pith.science (2026). https://pith.science/paper/43MDPFCT

@misc{pith2026190806519,
  author       = {Pith},
  title        = {Pith review of: Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/43MDPFCT}},
  note         = {Machine review of arXiv:1908.06519}
}

read the original abstract

The continuous growth of big data applications with high computational and scalability demands has resulted in increasing popularity of cloud computing. Optimizing the performance and power consumption of cloud resources is therefore crucial to relieve the costs of data centers. In recent years, multi-FPGA platforms have gained traction in data centers as low-cost yet high-performance solutions particularly as acceleration engines, thanks to the high degree of parallelism they provide. Nonetheless, the size of data centers workloads varies during service time, leading to significant underutilization of computing resources while consuming a large amount of power, which turns out as a key factor of data center inefficiency, regardless of the underlying hardware structure. In this paper, we propose an efficient framework to throttle the power consumption of multi-FPGA platforms by dynamically scaling the voltage and hereby frequency during runtime according to prediction of, and adjustment to the workload level, while maintaining the desired Quality of Service (QoS). This is in contrast to, and more efficient than, conventional approaches that merely scale (i.e., power-gate) the computing nodes or frequency. The proposed framework carefully exploits a pre-characterized library of delay-voltage, and power-voltage information of FPGA resources, which we show is indispensable to obtain the efficient operating point due to the different sensitivity of resources w.r.t. voltage scaling, particularly considering multiple power rails residing in these devices. Our evaluations by implementing state-of-the-art deep neural network accelerators revealed that, providing an average power reduction of 4.0X, the proposed framework surpasses the previous works by 33.6% (up to 83%).

Figures

Figures reproduced from arXiv: 1908.06519 by the authors.

**Figure 2.** Dynamic power of FPGA resources versus voltage. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 6.** Comparing DVFS techniques in different BRAM power [PITH_FULL_IMAGE:figures/full_fig_p004_6.png] view at source ↗

Figures from the paper (4 more)

**Figure 5.** Figure 5: Comparing DVFS techniques in different critical paths. [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]

**Figure 8.** Figure 8: Example of Markov chain for workload prediction. [PITH_FULL_IMAGE:figures/full_fig_p005_8.png]

**Figure 9.** Figure 9: (a) the architecture of the proposed energy-efficient multi-FPGA platform. The details of the (b) central controller, and [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]

**Figure 11.** Figure 11: Voltage adjustment in different voltage scaling tech [PITH_FULL_IMAGE:figures/full_fig_p007_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 47 canonical work pages

[25]

A universal self-calibrating dynamic voltage and frequency scaling (dvfs) scheme with thermal compensation for energy savings in fpgas,

S. Zhao et al., “A universal self-calibrating dynamic voltage and frequency scaling (dvfs) scheme with thermal compensation for energy savings in fpgas,” in IEEE Applied Power Electronics Conference and Exposition (APEC) , pp. 1882–1887, IEEE, 2016

work page 2016
[37]

Press: Predictive elastic resource scaling for cloud systems,

Z. Gong, X. Gu, and J. Wilkes, “Press: Predictive elastic resource scaling for cloud systems,” in 2010 International Conference on Network and Service Management , pp. 9–16, Ieee, 2010

work page 2010
[1]

Why end-users move to the cloud: a migration- theoretic analysis,

A. Bhattacherjee and S. C. Park, “Why end-users move to the cloud: a migration- theoretic analysis,” European Journal of Information Systems , vol. 23, no. 3, pp. 357–372, 2014

work page 2014
[2]

Future views on waste heat utilization–case of data centers in northern europe,

M. Wahlroos et al., “Future views on waste heat utilization–case of data centers in northern europe,” Renewable and Sustainable Energy Reviews , vol. 82, pp. 1749– 1764, 2018

work page 2018
[3]

United states data center energy usage report,

A. Shehabi et al., “United states data center energy usage report,” 2016

work page 2016
[4]

An energy efﬁcient vm management scheme with power-law characteristic in video streaming data centers,

H.-W. Tseng et al., “An energy efﬁcient vm management scheme with power-law characteristic in video streaming data centers,” IEEE Transactions on Parallel and Distributed Systems, vol. 29, no. 2, pp. 297–311, 2018

work page 2018
[5]

Data analytics for energy-efﬁcient clouds: design, implementation and evaluation,

A. Altomare, E. Cesario, and A. Vinci, “Data analytics for energy-efﬁcient clouds: design, implementation and evaluation,” International Journal of Parallel, Emergent and Distributed Systems , pp. 1–16, 2018

work page 2018
[6]

Asic clouds: special- izing the datacenter,

I. Magaki, M. Khazraee, L. V . Gutierrez, and M. B. Taylor, “Asic clouds: special- izing the datacenter,” in 2016 ACM/IEEE 43rd Annual International Symposium on Computer Architecture (ISCA) , pp. 178–190, IEEE, 2016

work page 2016

Show all 47 references

[7]

Rnsnet: In-memory neural network acceleration using residue number system,

S. Salamat, M. Imani, S. Gupta, and T. Rosing, “Rnsnet: In-memory neural network acceleration using residue number system,” in2018 IEEE International Conference on Rebooting Computing (ICRC) , pp. 1–12, IEEE, 2018

2018
[8]

In-datacenter performance analysis of a tensor processing unit,

N. P. Jouppi et al., “In-datacenter performance analysis of a tensor processing unit,” in ACM/IEEE 44th Annual International Symposium on Computer Architecture (ISCA), pp. 1–12, IEEE, 2017

2017
[9]

Diannao: A small-footprint high-throughput accelerator for ubiquitous machine-learning,

T. Chen et al. , “Diannao: A small-footprint high-throughput accelerator for ubiquitous machine-learning,” in ACM Sigplan Notices , vol. 49, pp. 269–284, ACM, 2014

2014
[10]

F5-hd: Fast ﬂexible fpga-based framework for refreshing hyperdimensional computing,

S. Salamat, M. Imani, B. Khaleghi, and T. Rosing, “F5-hd: Fast ﬂexible fpga-based framework for refreshing hyperdimensional computing,” in ACM International Symposium on Field-Programmable Gate Arrays , pp. 53–62, ACM, 2019

2019
[11]

Fach: Fpga-based ac- celeration of hyperdimensional computing by reducing computational complexity,

M. Imani, S. Salamat, S. Gupta, J. Huang, and T. Rosing, “Fach: Fpga-based ac- celeration of hyperdimensional computing by reducing computational complexity,” in Proceedings of the 24th Asia and South Paciﬁc Design Automation Conference , pp. 493–498, ACM, 2019

2019
[12]

Sparsehd: Algorithm-hardware co-optimization for efﬁcient high-dimensional computing,

M. Imani, S. Salamat, B. Khaleghi, M. Samragh, F. Koushanfar, and T. Rosing, “Sparsehd: Algorithm-hardware co-optimization for efﬁcient high-dimensional computing,” in IEEE International Symposium on Field-Programmable Custom Computing Machines (FCCM) , pp. 190–198, IEEE, 2019

2019
[13]

Tabla: A uniﬁed template-based framework for accelerating statistical machine learning,

D. Mahajan et al. , “Tabla: A uniﬁed template-based framework for accelerating statistical machine learning,” in IEEE International Symposium on High Perfor- mance Computer Architecture (HPCA) , pp. 14–26, IEEE, 2016

2016
[14]

From high-level deep neural models to fpgas,

H. Sharma et al. , “From high-level deep neural models to fpgas,” in The 49th Annual IEEE/ACM International Symposium on Microarchitecture , p. 17, IEEE Press, 2016

2016
[15]

Accelerating deep convolutional neural networks using specialized hardware,

K. Ovtcharov et al. , “Accelerating deep convolutional neural networks using specialized hardware,” Microsoft Research Whitepaper , vol. 2, no. 11, pp. 1–4, 2015

2015
[16]

A reconﬁgurable fabric for accelerating large-scale datacenter services,

A. Putnam et al. , “A reconﬁgurable fabric for accelerating large-scale datacenter services,” ACM SIGARCH Computer Architecture News, vol. 42, no. 3, pp. 13–24, 2014

2014
[17]

Network-attached fp- gas for data center applications,

J. Weerasinghe, R. Polig, F. Abel, and C. Hagleitner, “Network-attached fp- gas for data center applications,” in 2016 International Conference on Field- Programmable Technology (FPT), pp. 36–43, IEEE, 2016

2016
[18]

A survey on virtual machine migra- tion: Challenges, techniques, and open issues,

F. Zhang, G. Liu, X. Fu, and R. Yahyapour, “A survey on virtual machine migra- tion: Challenges, techniques, and open issues,” IEEE Communications Surveys & Tutorials, vol. 20, no. 2, pp. 1206–1243, 2018

2018
[19]

Quantifying and mitigating the costs of fpga virtualization,

S. Yazdanshenas and V . Betz, “Quantifying and mitigating the costs of fpga virtualization,” in 2017 27th International Conference on Field Programmable Logic and Applications (FPL) , pp. 1–7, IEEE, 2017

2017
[20]

Interconnect solutions for virtualized ﬁeld- programmable gate arrays,

S. Yazdanshenas and V . Betz, “Interconnect solutions for virtualized ﬁeld- programmable gate arrays,” IEEE Access, vol. 6, pp. 10497–10507, 2018

2018
[21]

Dynamic voltage scaling for commercial fpgas,

C. T. Chow, L. S. M. Tsui, P. H. W. Leong, W. Luk, and S. J. Wilton, “Dynamic voltage scaling for commercial fpgas,” in Proceedings. 2005 IEEE International Conference on Field-Programmable Technology, 2005., pp. 173–180, IEEE, 2005

2005
[22]

A distributed critical-path timing monitor for a 65nm high- performance microprocessor,

A. Drake et al. , “A distributed critical-path timing monitor for a 65nm high- performance microprocessor,” in IEEE International Solid-State Circuits Confer- ence. Digest of Technical Papers , pp. 398–399, IEEE, 2007

2007
[23]

Enabling fpgas in hyperscale data centers,

J. Weerasinghe, F. Abel, C. Hagleitner, and A. Herkersdorf, “Enabling fpgas in hyperscale data centers,” in IEEE Intl Conf on Ubiquitous Intelligence and Computing (UIC-ATC-ScalCom), pp. 1078–1086, IEEE, 2015

2015
[24]

Dynamic voltage & frequency scaling with online slack measurement,

J. M. Levine, E. Stott, and P. Y . Cheung, “Dynamic voltage & frequency scaling with online slack measurement,” in Proceedings of the 2014 ACM/SIGDA international symposium on Field-programmable gate arrays , pp. 65–74, ACM, 2014

2014
[26]

Reliability-aware design to suppress aging,

H. Amrouch, B. Khaleghi, A. Gerstlauer, and J. Henkel, “Reliability-aware design to suppress aging,” in2016 53nd ACM/EDAC/IEEE Design Automation Conference (DAC), pp. 1–6, IEEE, 2016

2016
[27]

A dynamic timing error avoidance technique using prediction logic in high-performance designs,

M. Ahmadi, S. Salamat, and B. Alizadeh, “A dynamic timing error avoidance technique using prediction logic in high-performance designs,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 27, no. 3, pp. 734–737, 2018

2018
[28]

Comprehensive evaluation of supply voltage underscaling in fpga on-chip memories,

B. Salami, O. S. Unsal, and A. C. Kestelman, “Comprehensive evaluation of supply voltage underscaling in fpga on-chip memories,” in 2018 51st Annual IEEE/ACM International Symposium on Microarchitecture (MICRO), pp. 724–736, IEEE, 2018

2018
[29]

Thermal-aware design and ﬂow for fpga perfor- mance improvement,

B. Khaleghi and T. ˇS. Rosing, “Thermal-aware design and ﬂow for fpga perfor- mance improvement,” in 2019 Design, Automation & Test in Europe Conference & Exhibition (DATE) , pp. 342–347, IEEE, 2019

2019
[30]

Dynamically optimizing fpga ap- plications by monitoring temperature and workloads,

P. H. Jones, Y . H. Cho, and J. W. Lockwood, “Dynamically optimizing fpga ap- plications by monitoring temperature and workloads,” in International Conference on VLSI Design (VLSID’07) , pp. 391–400, IEEE, 2007

2007
[31]

Don’t forget the memory: Automatic block ram modelling, optimization, and architecture exploration,

S. Yazdanshenas, K. Tatsumura, and V . Betz, “Don’t forget the memory: Automatic block ram modelling, optimization, and architecture exploration,” in Proceedings of the 2017 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, pp. 115–124, ACM, 2017

2017
[32]

Coffe: Fully-automated transistor sizing for fpgas,

C. Chiasson and V . Betz, “Coffe: Fully-automated transistor sizing for fpgas,” in 2013 International Conference on Field-Programmable Technology (FPT), pp. 34– 41, IEEE, 2013

2013
[33]

Autonomic sla-driven provisioning for cloud applications,

N. Bonvin, T. G. Papaioannou, and K. Aberer, “Autonomic sla-driven provisioning for cloud applications,” in IEEE/ACM international symposium on cluster, cloud and grid computing , pp. 434–443, IEEE Computer Society, 2011

2011
[34]

Resource provisioning with budget constraints for adaptive applications in cloud environments,

Q. Zhu and G. Agrawal, “Resource provisioning with budget constraints for adaptive applications in cloud environments,” in ACM International Symposium on High Performance Distributed Computing , pp. 304–307, ACM, 2010

2010
[35]

Empirical prediction models for adaptive resource provisioning in the cloud,

S. Islam, J. Keung, K. Lee, and A. Liu, “Empirical prediction models for adaptive resource provisioning in the cloud,” Future Generation Computer Systems, vol. 28, no. 1, pp. 155–162, 2012

2012
[36]

Workload prediction using arima model and its impact on cloud applications qos,

R. N. Calheiros, E. Masoumi, R. Ranjan, and R. Buyya, “Workload prediction using arima model and its impact on cloud applications qos,” IEEE Transactions on Cloud Computing , vol. 3, no. 4, pp. 449–458, 2015

2015
[38]

Texas instruments (ti),

“Texas instruments (ti), ”fusion digital power designer”.” http://www.ti.com/tool/ FUSION DIGITAL POWER DESIGNER
[39]

A 0.45–1 v fully-integrated distributed switched capacitor dc-dc converter with high density mim capacitor in 22 nm tri-gate cmos,

R. Jain et al. , “A 0.45–1 v fully-integrated distributed switched capacitor dc-dc converter with high density mim capacitor in 22 nm tri-gate cmos,” IEEE Journal of Solid-State Circuits , vol. 49, no. 4, pp. 917–927, 2014

2014
[40]

Coffe 2: Automatic modelling and optimization of complex and heterogeneous fpga architectures,

S. Yazdanshenas and V . Betz, “Coffe 2: Automatic modelling and optimization of complex and heterogeneous fpga architectures,” ACM Transactions on Reconﬁg- urable Technology and Systems (TRETS) , vol. 12, no. 1, p. 3, 2019

2019
[41]

Predictive technology model

“Predictive technology model.”
[42]

Vtr 7.0: Next generation architecture and cad system for fpgas,

J. Luu et al. , “Vtr 7.0: Next generation architecture and cad system for fpgas,” ACM Transactions on Reconﬁgurable Technology and Systems (TRETS) , vol. 7, no. 2, p. 6, 2014

2014
[43]

Stratix iv device handbook

“Stratix iv device handbook.” Datasheet, September 2014

2014
[44]

Nangate open cell library

“Nangate open cell library.”
[45]

Stripes: Bit-serial deep neural network computing,

P. Judd et al. , “Stripes: Bit-serial deep neural network computing,” in 2016 49th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO) , pp. 1–12, IEEE, 2016

2016
[46]

Proteus: Exploiting numerical precision variability in deep neural networks,

P. Judd et al., “Proteus: Exploiting numerical precision variability in deep neural networks,” in International Conference on Supercomputing , p. 23, ACM, 2016

2016
[47]

Burse: A bursty and self-similar workload generator for cloud computing,

J. Yin et al. , “Burse: A bursty and self-similar workload generator for cloud computing,” IEEE Transactions on Parallel and Distributed Systems, vol. 26, no. 3, pp. 668–680, 2015

2015

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [25]

A universal self-calibrating dynamic voltage and frequency scaling (dvfs) scheme with thermal compensation for energy savings in fpgas,

S. Zhao et al., “A universal self-calibrating dynamic voltage and frequency scaling (dvfs) scheme with thermal compensation for energy savings in fpgas,” in IEEE Applied Power Electronics Conference and Exposition (APEC) , pp. 1882–1887, IEEE, 2016

work page 2016

[2] [37]

Press: Predictive elastic resource scaling for cloud systems,

Z. Gong, X. Gu, and J. Wilkes, “Press: Predictive elastic resource scaling for cloud systems,” in 2010 International Conference on Network and Service Management , pp. 9–16, Ieee, 2010

work page 2010

[3] [1]

Why end-users move to the cloud: a migration- theoretic analysis,

A. Bhattacherjee and S. C. Park, “Why end-users move to the cloud: a migration- theoretic analysis,” European Journal of Information Systems , vol. 23, no. 3, pp. 357–372, 2014

work page 2014

[4] [2]

Future views on waste heat utilization–case of data centers in northern europe,

M. Wahlroos et al., “Future views on waste heat utilization–case of data centers in northern europe,” Renewable and Sustainable Energy Reviews , vol. 82, pp. 1749– 1764, 2018

work page 2018

[5] [3]

United states data center energy usage report,

A. Shehabi et al., “United states data center energy usage report,” 2016

work page 2016

[6] [4]

An energy efﬁcient vm management scheme with power-law characteristic in video streaming data centers,

H.-W. Tseng et al., “An energy efﬁcient vm management scheme with power-law characteristic in video streaming data centers,” IEEE Transactions on Parallel and Distributed Systems, vol. 29, no. 2, pp. 297–311, 2018

work page 2018

[7] [5]

Data analytics for energy-efﬁcient clouds: design, implementation and evaluation,

A. Altomare, E. Cesario, and A. Vinci, “Data analytics for energy-efﬁcient clouds: design, implementation and evaluation,” International Journal of Parallel, Emergent and Distributed Systems , pp. 1–16, 2018

work page 2018

[8] [6]

Asic clouds: special- izing the datacenter,

I. Magaki, M. Khazraee, L. V . Gutierrez, and M. B. Taylor, “Asic clouds: special- izing the datacenter,” in 2016 ACM/IEEE 43rd Annual International Symposium on Computer Architecture (ISCA) , pp. 178–190, IEEE, 2016

work page 2016