## Research Papers

All papers and articles from Code Metal Research.

### Paper List

1. **[The Trust Problem Has Shifted: What Formal Verification Can and Cannot Guarantee About AI-Generated Code](/content/research/the-trust-problem-has-shifted-what-formal-verification-can-and-cannot-guarantee-about-ai-generated-code/index.html)**  
     
   *Code Metal Article*  
   *Formal Verification*  
   *July 8, 2026*  
   A clear-eyed technical assessment of formal verification for AI-generated code: which approaches are credible, what barriers remain, and where the market will emerge first.

2. **[The Real Cost of Leaving NVIDIA](/content/research/the-real-cost-of-leaving-nvidia/index.html)**  
     
   *Code Metal Article*  
   *GPU/CUDA*  
   *June 4, 2026*  
   What Automated Transpilation Actually Costs, and What It Doesn't.

3. **[AI-generated code that works — and proves it](/content/research/ai-generated-code-that-works-and-proves-it/index.html)**  
     
   *Code Metal Article*  
   *Formal Verification*  
   *May 18, 2026*  
   How Code Metal combines AI with formal methods to build trusted code translation systems, and welcoming Prof. Loris D'Antoni as our first Code Metal Scholar.

4. **[Counting Without Running: Evaluating LLMs' Reasoning About Code Complexity](/content/research/counting-without-running-evaluating-llms-reasoning-about-code-complexity/index.html)**  
     
   *Research Paper*  
   *GPU/CUDA*  
   *December 4, 2025*  
   Introduces gpuFLOPBench, a benchmark containing 577 CUDA kernels to evaluate whether language models can predict floating-point operation counts without execution, revealing limitations in understanding hardware-specific performance details.

5. **[Verified Code Transpilation with LLMs](/content/research/verified-code-transpilation-with-llms/index.html)**  
     
   *Research Paper*  
   *Formal Verification*  
   *October 15, 2025*  
   LLMLift uses large language models (LLMs) to automate code translation between domain-specific languages (DSLs), ensuring functional correctness and outperforming traditional tools in speed and efficiency.

6. **[Workflows vs Agents for Code Translation](/content/research/workflows-vs-agents-for-code-translation/index.html)**  
     
   *Research Paper*  
   *LLMs*  
   *September 22, 2025*  
   Compares structured workflows versus agentic approaches for MATLAB-to-HDL translation, showing that agentic methods with the Model Context Protocol increase simulation reach rates by over 20 percentage points on mid-sized models.

7. **[UniPar: A Unified LLM-Based Framework for Parallel and Accelerated Code Translation in HPC](/content/research/unipar-unified-framework-for-parallel-code-translation/index.html)**  
     
   *Research Paper*  
   *HPC*  
   *September 15, 2025*  
   Introduces UniPar, an evaluation framework for assessing how large language models translate between parallel programming languages, achieving 69% compilation success and 33% functional correctness through fine-tuning and compiler-guided repair.

8. **[MONOCODER: Domain-Specific Code Language Model for HPC Codes and Tasks](/content/research/monocoder-domain-specific-code-language-model-for-hpc-codes-and-tasks/index.html)**  
     
   *Research Paper*  
   *HPC*  
   *December 18, 2024*  
   MONOCODER is a smaller, HPC-specific language model that outperforms larger, general-purpose LLMs on HPC code tasks, offering better performance in code generation and comprehension.

9. **[Combining AI with formal verification for efficient migration of legacy code](/content/research/combining-ai-with-formal-verification-for-efficient-migration-of-legacy-code/index.html)**  
     
   *Code Metal Article*  
   *Formal Verification*  
   *December 18, 2024*  
   Advanced techniques for deploying and optimizing large language models on resource-constrained edge devices without compromising performance.

10. **[Code migration with formal verification for performance improvement of legacy code](/content/research/code-migration-with-formal-verification-for-performance-improvement-of-legacy-code/index.html)**  
      
   *Code Metal Article*  
   *Formal Verification*  
   *December 20, 2023*  
   Domain-specific languages (DSL) have been a common alternative to general-purpose programming languages to implement domain-specific optimizations and simplify programming of domain-specific hardware.