All major releases, enhancements, and what they mean for your simulations.
Webinar Events – Eastern Hemisphere: 24 Feb | 09:00 UTC | Western Hemisphere: 26 Feb | 16:00 UTC
We’ve created a wealth of educational assets to advance your industry and technical capabilities.
View Available Training Courses
Speed is only valuable when it honors the physics that matter.
By now, the trade-off between simulation speed and accuracy should feel familiar. Faster results enable better decisions, but only if the underlying physics is sound. Cutting corners may deliver answers sooner, but it also introduces blind spots that surface later, when decisions are harder or impossible to undo.
The real question, then, is not whether speed matters - it does. The question is how speed is achieved. In this final article of the Accuracy at Speed series, I want to focus on what that balance looks like in practice.
Achieving faster simulations without compromising accuracy often comes down to a few controllable choices teams can make today.
These best practices only work when the underlying software continues to evolve, and that’s where sustained engineering investment matters.
Our teams made 2025 a year of continuous performance advancements. Early in the year, we delivered a commercial release of our black-oil simulator (IMEX) on GPU, demonstrating a ~2.4x speedup over the previous version on real field models.
Building on that momentum, we applied multiple optimization strategies to our compositional simulator (GEM). Now, in Q4 2025, we have rolled out the latest version of GEM, delivering up to 1.6x faster compositional simulation performance.
We have also delivered a pre-commercial release of GEM on GPU, which our customers are testing across various models.
CMG’s iterative approach – develop, test with users, refine, then release – highlights a best practice in the software industry. By continuously working with real-world models and incorporating customer feedback, we ensure that speedups are achieved without sacrificing accuracy or stability.
What’s important here is not just the individual speedups, but the pattern behind them. Each release reflects the same principle: performance gains should come from better math, better algorithms, better use of modern hardware, and not from cutting physics. This is how we continue to expand what “Accuracy at Speed” means in practice, giving users the confidence to run bigger models, explore more scenarios, and make decisions faster without compromising the science underneath.
In the dynamic world of oil and gas, the ability to rapidly simulate and forecast reservoir behavior has shifted from a technical convenience to a strategic necessity. Faster simulation speeds enable teams to evaluate more scenarios, reduce uncertainty, and optimize outcomes in a timely manner. From avoiding costly drilling mistakes to maximizing recovery through agile field management, the impacts of speed are felt across the value chain.
Speed, however, must come with accuracy. The industry rightly insists that any accelerated simulation still honors the physics and field data. Because the barrels that the simulation misses won’t magically appear in production. They vanish quietly into poor decisions, never to be recovered. And that’s the hidden cost of choosing 'good enough'.
Written by Rahul Jain
December 2025
