Accelerating Systems with Real-time AI Solutions
Built for leaders needing the fastest time-to-value
Groq offers comprehensive end-to-end acceleration solutions from scalable, ultra low latency systems to generalized software. We improve results by orders of magnitude for our customers needing to modernize underperforming systems. Groq increases performance and enables innovation unlike any other technology provider.
Our First Generation Solutions
A single GroqChip in a PCIe Gen4 x16 interface with 230 MB of on-die memory delivers up to 750 TOPs, 188 TFLOPs (INT8, FP16 @900 MHz).
Eight GroqCard accelerators in a rack-ready 4U server chassis supports millions of parameters and is a scalable building block for a single global hop network, delivering up to 6 POPs, 1.5 PFLOPs (INT8, FP16 @900MHz).
Eight plus one interconnected GroqNode servers delivers low system latency of ~1.6 µsec and up to 48 POPs, 12 PFLOPs (INT8, FP16 @900MHz).
GroqCloud delivers 216 POPs, 54 PFLOPs (INT8, FP16) and is growing.
Redefining the Developer Experience
- Groq™ Compiler: Out-of-the-box applicability, serving the majority of industry standard models
- Groq API: Meets tailored solution needs with fine-grained control
- Productivity Tools: GroqView™ Visualization and Profiler, Performance Estimator, and GroqFlow™ Tool Chain
- Try Out Groq: Contact us for access to the GroqWare™ Developer Tools Package
Providing Value to Customers Today
Groq Enables Drug Discovery in Minutes at Argonne National Laboratory
“Using the Groq platform at Argonne, we were able to accelerate our efforts to identify promising COVID-19 drug candidates from a vast number of small molecules. The system’s AI capabilities enabled us to achieve significantly more inferences a second, reducing the time needed for each search from days to minutes.”
Key Advantages and Differentiators
We look forward to working with Groq to help our government partners address their enduring need for higher performance, lower latency compute solutions to process large volumes of data faster and use less power.
Groq validates that it is developing disruptive technology that can address rapidly expanding AI and ML opportunities with the ability to drive powerful use cases...
The TSP is designed to extend single core performance across multiple chips, can perform inference tasks quickly, and is deterministic—it delivers consistent, predictable, and repeatable performance with zero overhead for context switching.
Groq’s TSP stands out in peak performance and its [deep-learning] accelerator is the fastest available on the merchant market.
Groq is one of the leaders, if not the leader, in machine learning...