We Benchmarked H100 vs L40S for ZK Proofs — Here’s What Scales (and What Doesn’t)
Zero-knowledge (ZK) proof systems are becoming a cornerstone technology for privacy-preserving and scalable computation in blockchain and cryptographic applications. As proof complexity and throughput demands grow, optimizing hardware utilization becomes essential to maintain performance and cost-efficiency — particularly in GPU-accelerated proving pipelines.
We at P2P.org have participated in most of the major ZK protocols via different sets of ZK prover hardware. Since Ethereum is moving towards ZK enshrined in the protocol with L2Beat-like “slices” overview projects popping up (https://ethproofs.org/), we wanted to provide the community with an example of one of our researches based on our gathered knowledge on the subject.
This study examines GPU utilization strategies for generating ZK proofs, comparing two leading GPU architectures: the NVIDIA H100 and L40S. The main objective is to evaluate whether allocating multiple GPUs to a single proof improves performance more effectively than generating multiple proofs in parallel, each using a single GPU.
Our benchmark is based on Scroll’s open-source ZK prover implementation, deployed on two high-performance hardware platforms. Below are the technical specifications for each setup:
Hardware Specifications
L40S System:
CPU: 2× AMD EPYC 9354 (3.80 GHz)
RAM: 2 TB
GPU: 8× NVIDIA L40S 48 GB
Storage: 4× 4 TB NVMe SSD
Network: 2× 10 Gbit NICs
H100 System:
CPU: Intel Xeon 8481C (2.7 GHz, 208 cores)
RAM: 1.8 TB
GPU: 8× NVIDIA H100 80 GB
Storage: 12× 400 GB NVMe SSD
Network: 1× 100 Gbit NIC
Using a fixed 8-GPU configuration, we tested two modes: (1) increasing the number of GPUs per proof to measure time reduction, and (2) running multiple proofs concurrently to assess total throughput. This section sets the foundation for analyzing the performance trade-offs, CPU/GPU bottlenecks, and real-world cost-effectiveness of ZK proof generation at scale.
Benchmarking ZK Prover Performance: Parallelization vs Dedicated GPUs
To evaluate GPU utilization efficiency in zero-knowledge proof generation, we conducted a series
of controlled benchmarks on both hardware setups — L40S and H100 — using 8 GPUs in each case.
The goal was to compare two strategies:
Strategy A: Increasing the number of GPUs used for generating a single proof.
Strategy B: Running multiple proofs in parallel, with one GPU assigned per proof.
The Scroll open-source prover was used as the testing framework across both systems. Each configuration
was run with fixed parameters and measured for prover time, proof throughput (proofs per day),
and system resource utilization (CPU, GPU memory, RAM). Below are the summarized results:
L40S Results
Configuration
Prover Time (s)
Proofs per Day
1 GPU on 1 proof
792
109
2 GPUs on 1 proof
705
122
4 GPUs on 1 proof
672
128
8 GPUs on 1 proof
688
125
8 GPUs, 8 proofs in parallel
1420 (total), 60.8 per GPU
486 total
H100 Results
Configuration
Prover Time (s)
Proofs per Day
1 GPU on 1 proof
1047
82
2 GPUs on 1 proof
892
97
4 GPUs on 1 proof
824
105
8 GPUs on 1 proof
803
108
8 GPUs, 8 proofs in parallel
2400 (total), 36 per GPU
288 total
These results demonstrate that assigning a single GPU to each proof and executing them in parallel
yields significantly higher overall throughput, especially on the L40S system. Surprisingly,
H100 performance gains from parallelization were underwhelming, despite its raw power advantage,
suggesting suboptimal software utilization or architectural bottlenecks in the current prover setup.
On the graph we have shown the efficiency we expected to have by adding GPUs with the green line. The red dot on the graph is the generation of 8 ZK proofs simultaneously on the same 8-GPU unit, while the blue line is the result we received by adding GPUs to the proof generation process.
System Resource Utilization During Proof Generation
In addition to measuring prover time and throughput, we monitored system-level resource usage to better understand the efficiency and scaling behavior of each GPU configuration. Metrics recorded include peak CPU utilization, maximum GPU memory usage, and RAM consumption across different levels of parallelism.
The results indicate that running proofs in parallel leads to near full CPU saturation and significantly increased RAM consumption. This suggests that CPU becomes a limiting factor under heavy GPU parallelism unless paired with a properly scaled memory and compute environment.
While GPU memory usage scales linearly with the number of concurrent proofs, the per-proof RAM usage becomes substantial when 8 parallel jobs are running, particularly on H100 hardware.
The RAM usage remains constant at 180 GB across all configurations (1, 2, 4, and 8 GPUs). This suggests that the memory allocation for the proof generation process is not dependent on the number of GPUs involved.
It is likely that the proving software either preallocates the required system memory at the start of the process or that the computational workload is primarily offloaded to the GPU, resulting in negligible variation in RAM consumption.
This behavior indicates that system RAM is not a limiting factor in the scaling of proof generation on the H100 hardware — at least when generating a single proof, regardless of GPU count.
When analyzing GPU memory usage on the H100 for single-proof generation, a clear trend emerges: GPU memory consumption decreases as more GPUs are allocated to the task.
With 1 GPU, the memory usage peaks at 46 GB, but as the workload is distributed across 2, 4, and eventually 8 GPUs, the consumption per GPU drops to 12 GB in the 8-GPU configuration.
This behavior is consistent with the expectation that dividing the computation across more GPUs reduces per-device memory pressure, as intermediate states and computational graphs are split and processed concurrently.
However, despite the lower memory usage, the overall proving time did not improve significantly, suggesting that GPU memory was not the bottleneck. This reinforces the observation that parallel GPU allocation alone is not sufficient to accelerate ZK proof generation without corresponding improvements in software or CPU coordination.
Conclusion
This benchmark study evaluated the performance and hardware efficiency of generating zero-knowledge proofs using two enterprise-grade GPU configurations: the NVIDIA H100 and NVIDIA L40S. The analysis was conducted using Scroll's open-source prover, with a focus on two key strategies: scaling a single proof across multiple GPUs versus running multiple proofs in parallel.
The results demonstrate that parallel generation of proofs using individual GPUs yields significantly better throughput than assigning all GPUs to a single proof process. This effect is especially visible on the L40S platform, where parallel execution nearly quadrupled the number of proofs generated per day compared to the single-proof setup.
Surprisingly, the H100 — despite its superior hardware specs — underperformed in this scenario. Its single-proof generation times were longer than L40S in all configurations, and parallel execution on H100 also delivered lower throughput, indicating that software bottlenecks or suboptimal utilization patterns may limit its current viability for ZK workloads.
Additionally, we found that system RAM and GPU memory were not primary limiting factors in most configurations. RAM usage remained constant during single-proof runs, while GPU memory usage decreased as GPU count increased. Instead, CPU saturation and parallel processing coordination appear to be more critical for maximizing performance in proof generation.
In conclusion, GPU parallelism for a single proof does not scale efficiently beyond a certain point. ZK infrastructure teams aiming to improve throughput should prioritize software optimization, better CPU/GPU coordination, and parallelization across proofs rather than within a single one.
Subscribe to P2P-economy
Get the latest posts delivered right to your inbox
<p>We’re excited to announce a new partnership with Balance, Canada’s oldest digital asset custodian. Through this integration, P2P.org’s industry-leading staking infrastructure is now directly accessible to Balance’s clients — enabling seamless, custody-first staking for top networks like Ethereum, Solana, and more.</p><h2 id="unlocking-staking-in-a-custody-first-environment"><strong>Unlocking Staking in a Custody-First Environment</strong></h2><p>Institutional staking is evolving, and P2P.org is proud to power the next step. With this new integration, Balance clients can now engage with our validator infrastructure without ever relinquishing custody of their assets. By combining Balance’s robust custodial framework with P2P.org’s high-performance validation, institutional participants can stake confidently, securely, and in full alignment with their compliance and risk mandates.</p><h3 id="key-benefits-include">Key benefits include:</h3><p></p><ul><li><strong>Direct access to P2P.org’s staking infrastructure</strong> across major PoS networks like ETH and SOL<br></li><li><strong>Full asset control</strong> throughout the staking lifecycle, preserving both registered and beneficial ownership<br></li><li><strong>Integrated workflows</strong> via Balance’s platform. Stake, unstake, and claim rewards without needing third-party tools<br></li><li><strong>Audit-ready reporting</strong> designed for institutional governance and regulatory standards<br><br></li></ul><p>This partnership is purpose-built for institutions seeking high-assurance staking options that do not compromise on either security or operational efficiency.</p><p></p><h2 id="built-for-institutions-backed-by-performance"><strong>Built for Institutions. Backed by Performance</strong></h2><p>Balance’s infrastructure-first model ensures institutional clients retain oversight and control over their assets at all times. The integration with P2P.org complements this with a trusted staking layer, featuring proven validator performance, multi-chain coverage, and enterprise-grade reliability.</p><p></p><blockquote>“Institutions need secure, scalable ways to participate in staking. By integrating with P2P.org, we’re extending those capabilities inside our custody platform, giving clients the freedom to engage with staking service providers on their own terms.”<br><strong>George Bordianu</strong>, Co-Founder and CEO at Balance.<br></blockquote><blockquote>“Balance’s integration reflects the evolving maturity of institutional staking. Our goal has always been to deliver high-performance validation with rigorous uptime and governance. Now, institutions can access this directly through a trusted custody environment.” <br><strong>Alex Esin</strong>, CEO at P2P.org.</blockquote><p></p><h2 id="accelerating-institutional-adoption"><strong>Accelerating Institutional Adoption</strong></h2><p>This integration marks another milestone in our mission to accelerate secure institutional adoption of staking. As regulatory clarity increases and staking becomes a foundational activity in institutional crypto portfolios, partnerships like this are key to offering compliant, resilient infrastructure at scale.</p><p>To explore staking via P2P.org through Balance’s platform, reach out to [email protected].</p><p>For more on how P2P.org supports institutions with infrastructure-grade validation, <a href="mailto:[email protected]"><u>get in touch directly with us here</u></a>.</p>
<p></p><h2 id="tldr"><strong>TL;DR</strong></h2><ul><li>P2P.org is now available as a Solana validator on Fireblocks, enabling institutional clients to access industry-leading 9.40% total gross rewards without leaving their secure environment.</li><li>This integration delivers a triple-stream reward structure combining standard SOL staking, enhanced MEV, and monthly block rewards—all with zero technical complexity.</li><li>Fireblocks users can start staking within seconds through a streamlined interface, backed by P2P.org's expertise in managing over $10B across 40+ networks.</li></ul><p>We’re thrilled to announce a major achievement for institutional staking: P2P.org is now officially available as a Solana validator on Fireblocks—the industry's leading digital asset infrastructure platform trusted by over 2,000 institutions worldwide. This integration represents months of work and positions P2P.org at the forefront of institutional Solana staking.</p><p>Finally, Fireblocks clients gain access to <a href="http://p2p.org/?ref=p2p.org"><u>P2P.org</u></a>’s market-leading Solana rewards, enabling institutions to receive<strong> Solana Rewards</strong>.</p><h2 id="solana-rewards"><strong>Solana Rewards</strong></h2><p>By staking with <a href="http://p2p.org/?ref=p2p.org"><u>P2P.org</u></a> through Fireblocks, users now finally get direct access to our<strong> reward ecosystem</strong>. This multi-layered reward structure maximizes network reward potential through three distinct revenue streams:</p><ul><li><strong>SOL Staking Rewards</strong> - Automatically compounded every two days to maximize growth</li><li><strong>MEV Rewards</strong> - Enhanced through our proprietary MEV optimization strategies</li><li><strong>Block Rewards</strong> - Transaction fee sharing distributed monthly</li></ul><p>This comprehensive market-leading approach <strong>captures value from every aspect of Solana's reward structure</strong>, delivering a substantial advantage to Fireblocks clients over standard validation services. All of this is achieved while maintaining the institutional level of security and compliance framework that institutions require.</p><figure class="kg-card kg-image-card"><img src="https://p2p.org/economy/content/images/2025/06/image.png" class="kg-image" alt loading="lazy" width="2000" height="1125" srcset="https://p2p.org/economy/content/images/size/w600/2025/06/image.png 600w, https://p2p.org/economy/content/images/size/w1000/2025/06/image.png 1000w, https://p2p.org/economy/content/images/size/w1600/2025/06/image.png 1600w, https://p2p.org/economy/content/images/size/w2400/2025/06/image.png 2400w" sizes="(min-width: 720px) 720px"></figure><figure class="kg-card kg-image-card"><img src="https://lh7-rt.googleusercontent.com/docsz/AD_4nXfjuSzqz8G74OVsGEMd9mvFjbUzlLti71iT0u9aTsF42DPDhqUpoiKqsNs_phg_n3XSV6rwdl_vyVUDb5395U2lokcf8wORzxphjKO09Q2jl-wkTZVJ4t0pmTUa4x9-YkszwFApuQ?key=nGwrnZGkHUhe1fB3t48EAw" class="kg-image" alt loading="lazy" width="1600" height="1084"></figure><p>As part of Fireblocks' strategic expansion of staking capabilities, the platform has integrated P2P.org, giving its clients direct access to one of the highest-performing Solana validators. This integration delivers immediate benefits:</p><p><strong>Superior Rewards, Zero Complexity</strong>: Fireblocks users can now access P2P.org's industry-leading 9.40% total gross rewards, outperforming the network average of 9.16%—without ever leaving their secure Fireblocks environment.</p><p><strong>Institutional-Grade Infrastructure</strong>: Stake with confidence knowing your assets are backed by P2P.org's enterprise validation technology and Fireblocks' unparalleled security architecture.</p><p><strong>Streamlined Experience</strong>: Fireblocks' enhanced validator selection interface makes it simple to compare providers and make informed staking decisions in just a few clicks.</p><h2 id="why-it-matters"><strong>Why It Matters</strong></h2><h3 id="first-mover-advantage"><strong>First-Mover Advantage</strong></h3><p>As one of the first validators available for Solana on Fireblocks, P2P.org gives institutions immediate access to elite staking infrastructure without the traditional technical barriers. Our solution combines years of validator expertise, enterprise-grade performance, and the credibility of managing $10B+ across 40+ networks—delivering a significant competitive edge to early adopters in the institutional Solana ecosystem.</p><h3 id="market-leading-performance"><strong>Market Leading Performance</strong></h3><p>Our advanced validator technology maximizes blockspace efficiency that delivers a measurable financial impact:</p><ul><li><strong>Among the highest Solana stake under management globally</strong></li><li><strong>99.9% block production history</strong></li><li><strong>Built-in MEV rewards </strong> without requiring opt-in</li><li><strong>Higher annualized rewards</strong> (9.40% vs. 9.16% network average)</li><li><strong>Compound growth advantage</strong> that widens over time</li></ul><h3 id="risk-management-through-integration"><strong>Risk Management Through Integration</strong></h3><p>This integration enables institutions to select preferred validators without sacrificing security, compliance, or rewards, allowing for strategic portfolio diversification and reduced exposure to single-provider risks. The addition of further staking providers on Solana also contributes to decentralization and Solana network resilience. </p><h2 id="getting-started-takes-seconds"><strong>Getting Started Takes Seconds</strong></h2><ol><li>Navigate to your Fireblocks staking dashboard and select Solana.</li><li>You’ll now see P2P.org listed as a staking provider.</li><li>Select P2P.org, review and accept the terms, and initiate staking.</li><li>For API-based workflows, use the Fireblocks API with our provider ID<br></li></ol><p>If you’re looking for special offers, please reach out to your personal Fireblocks account manager.</p><p>Comprehensive documentation is available through<a href="https://www.fireblocks.com/partners/?_gl=1*z5h1p8*_up*MQ..*_gs*MQ..&gclid=CjwKCAjw87XBBhBIEiwAxP3_Ay8eSxnWHHj5TA5FdtQJCp6_HtA0Aqqcpt4JqcGkw-2__poTEkPL7BoCrfwQAvD_BwE&gbraid=0AAAAAorYxp8XgphJmjdscfNxOvFOMQF3T&ref=p2p.org"><u> Fireblocks Staking Partners documentation</u></a>.</p><figure class="kg-card kg-video-card kg-width-regular" data-kg-thumbnail="https://p2p.org/economy/content/media/2025/06/Fireblocks-video_thumb.jpg" data-kg-custom-thumbnail>
<div class="kg-video-container">
<video src="https://p2p.org/economy/content/media/2025/06/Fireblocks-video.mp4" poster="https://img.spacergif.org/v1/3040x1650/0a/spacer.png" width="3040" height="1650" loop autoplay muted playsinline preload="metadata" style="background: transparent url('https://p2p.org/economy/content/media/2025/06/Fireblocks-video_thumb.jpg') 50% 50% / cover no-repeat;"></video>
<div class="kg-video-overlay">
<button class="kg-video-large-play-icon" aria-label="Play video">
<svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24">
<path d="M23.14 10.608 2.253.164A1.559 1.559 0 0 0 0 1.557v20.887a1.558 1.558 0 0 0 2.253 1.392L23.14 13.393a1.557 1.557 0 0 0 0-2.785Z"/>
</svg>
</button>
</div>
<div class="kg-video-player-container kg-video-hide">
<div class="kg-video-player">
<button class="kg-video-play-icon" aria-label="Play video">
<svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24">
<path d="M23.14 10.608 2.253.164A1.559 1.559 0 0 0 0 1.557v20.887a1.558 1.558 0 0 0 2.253 1.392L23.14 13.393a1.557 1.557 0 0 0 0-2.785Z"/>
</svg>
</button>
<button class="kg-video-pause-icon kg-video-hide" aria-label="Pause video">
<svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24">
<rect x="3" y="1" width="7" height="22" rx="1.5" ry="1.5"/>
<rect x="14" y="1" width="7" height="22" rx="1.5" ry="1.5"/>
</svg>
</button>
<span class="kg-video-current-time">0:00</span>
<div class="kg-video-time">
/<span class="kg-video-duration">0:25</span>
</div>
<input type="range" class="kg-video-seek-slider" max="100" value="0">
<button class="kg-video-playback-rate" aria-label="Adjust playback speed">1×</button>
<button class="kg-video-unmute-icon" aria-label="Unmute">
<svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24">
<path d="M15.189 2.021a9.728 9.728 0 0 0-7.924 4.85.249.249 0 0 1-.221.133H5.25a3 3 0 0 0-3 3v2a3 3 0 0 0 3 3h1.794a.249.249 0 0 1 .221.133 9.73 9.73 0 0 0 7.924 4.85h.06a1 1 0 0 0 1-1V3.02a1 1 0 0 0-1.06-.998Z"/>
</svg>
</button>
<button class="kg-video-mute-icon kg-video-hide" aria-label="Mute">
<svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 24 24">
<path d="M16.177 4.3a.248.248 0 0 0 .073-.176v-1.1a1 1 0 0 0-1.061-1 9.728 9.728 0 0 0-7.924 4.85.249.249 0 0 1-.221.133H5.25a3 3 0 0 0-3 3v2a3 3 0 0 0 3 3h.114a.251.251 0 0 0 .177-.073ZM23.707 1.706A1 1 0 0 0 22.293.292l-22 22a1 1 0 0 0 0 1.414l.009.009a1 1 0 0 0 1.405-.009l6.63-6.631A.251.251 0 0 1 8.515 17a.245.245 0 0 1 .177.075 10.081 10.081 0 0 0 6.5 2.92 1 1 0 0 0 1.061-1V9.266a.247.247 0 0 1 .073-.176Z"/>
</svg>
</button>
<input type="range" class="kg-video-volume-slider" max="100" value="100">
</div>
</div>
</div>
</figure><h2 id="looking-ahead"><strong>Looking Ahead</strong></h2><p>While Fireblocks plans to expand its validator roster over time, P2P.org is fully integrated and available today. We're committed to building more seamless integration points that make institutional staking accessible, secure, and profitable across all major networks.</p><p>Ready to experience institutional-grade Solana staking? Visit <a href="http://p2p.org/solana?ref=p2p.org"><u>p2p.org/solana</u></a> or <a href="mailto:[email protected]"><u>get in touch directly with us here</u></a>.</p>
ALL
Agoric
Aptos
Auth
Avail
Avalanche
Axelar
Babylon
Bitcoin
BTC
Cardano
Celestia
Chainlink
Cosmos
Crescent
Cyberway
DAOBet
DVT
Dymension
Economy
EigenLayer
Elrond
Ethereum
Evmos
Explain Like I'm Five
Guide
HR
Kava
Kusama
Lido
LSP
Manta
Mantle
Marlin
Matic
Moonbeam
MultiversX
Near
NuCypher
Oasis
Origin
Partnership
Persistence
Pocket
Polkadot
Polygon
Quicksilver
Regen
Renzo
restaking
RPC Node
Serum
Solana
SSV.Network
Staking
Starknet
Sui
symbiotic
Terra
Tezos
TheGraph
Threshold
Update
Vara
Vega
