We Broke Big Data Economics
3TB enterprise workload for less than a cup of coffee
Not a demo.
A production-grade result.
Yeedu executed the full 99-query TPC-DS benchmark suite across 1TB, 3TB, and 10TB datasets with zero failures and 0 line-item rewrites.
The benchmark proved that Yeedu's Turbo Engine delivers predictable performance and cloud cost savings at scale.
Below is a breakdown of execution costs, node configurations, and performance metrics across scale factors.
100% Query Pass Rate
Every single query in the 99-query TPC-DS benchmark suite executed to completion with zero failures or code modifications.
Zero Line-Item Query Modification
Standard TPC-DS SQL queries were executed as-is, proving true compatibility without tuning query syntax.
Predictable Performance
Consistent linear scaling from 1TB to 10TB datasets with no performance degradation or cluster thrashing.
Enterprise Scale Ready for AI
Heavy analytical workloads finish up to 10x faster, leaving data ready for downstream AI/ML consumption.
Benchmark Results
Execution Time: 11 mins 42 secs
Zero failures across 99 queries
Execution Time: 16 mins 18 secs
Predictable cost-per-terabyte
Execution Time: 31 mins 12 secs
Linear scalability for giant datasets
Infrastructure Details
Cloud hardware specs and configuration details for performance verification.
| Configuration Metric | 1TB Scale | 3TB Scale | 10TB Scale |
|---|---|---|---|
| Instance Type | r7g.4xlarge | r7g.8xlarge | r7g.16xlarge |
| vCPUs | 16 | 32 | 64 |
| RAM (GB) | 128 GB | 256 GB | 512 GB |
| Instance Storage | EBS gp3 (10,000 IOPS) | EBS gp3 (10,000 IOPS) | EBS gp3 (10,000 IOPS) |
| Executors | 2 nodes | 4 nodes | 8 nodes |
| Total Run Cost | $0.52 Total | $2.33 Total | $13.47 Total |
How is Turbo the Fastest Analytical Engine?
SIMD Vectorized Execution
Vectorized query engine processes multiple data rows per CPU instruction, maximizing modern processor hardware registers.
Predictive Memory Access
Pre-fetches memory and reduces cache misses, allowing heavy queries to execute with minimal CPU wait states.
Smart Scheduling
Eliminates I/O bottlenecks and optimizes task multiplexing across cluster nodes.
Architecture Flow Diagram
Every dollar saved on data infrastructure is a dollar spent on AI
Reallocating compute budget from legacy data infrastructure to high-impact AI/ML projects.
Compute spend for traditional Spark platforms consumes up to 80% of data budgets. Switching to Yeedu shifts budget from infrastructure overhead to strategic AI projects.
Shifting Cloud Spend From Legacy Data Engines to AI/ML
* All benchmark metrics were executed on AWS Graviton4 instances in US-East region.
Bring your workload.
We'll show you the gap.
Benchmark your real-world Spark workloads in your own cloud environment with a 30-day proof of concept.
Access Turbo Engine
Schedule a call with our technical team to see how Yeedu can accelerate your workloads and lower your cloud bill.
We respect your privacy. No spam.