This time at MLPerf, which is considered the Olympics of Machine Learning where chips are pitted against each other, a four-year-old start-up beats the world’s most celebrated Chip company Nvidia, which has been winning the competition for decades.
ML System on Chip (SOC) platform on SIMa beat Nvidia’s Jetson AGX Orin in Closed Edge Power category. The CEO Krishna Rangasayee says, ‘SiMa earned top inference achievements in all aspects of the benchmark.’ The best part was, they did not even tweak their chip for the competition.
This news got me curious about what went into creating such a groundbreaking company, with meager resources and the number of people that can be counted on fingers.
And here are the answers:
1. Thinking from Scratch: Instead of thinking of becoming another Nvidia, they tried understanding customer needs and followed software-centric approach to solve the problem unlike silicon-centric approach followed by Nvidia.
2. Partnering with the Best: While SiMa’s people focus on design, it largely depends on partners like TSMC, Synopsys, Arm, Allegro, GUC and Arteris for other aspects of the business from manufacturing to software interface.
3. Redefining the Limits: SiMa ML System on Chip (SOC) delivers 10x performance while solving computer vision applications at a fraction of power consumption.
4. Rate of Acceleration Led by Disregard for Bureaucracy: A clear advantage SiMa had was no bureaucracy to slow down the decision-making and experimentation that start ups thrive on.
5. Courage: Krishna Rangasayee’s courage to venture into highly complex and technical business that needs heavy investments and resources deserves a salute as the nearest competitor has 20000 employees.
What are your thoughts? Would love to hear.
Change your game!
Vivek
Picture Courtesy: ET Auto
#executivecoach #executivecoaching #leadershipcoach #leadershipcoaching #bestcoachinindia