RapidlyA Tel Aviv -based starting start -up that is being developed to accelerate the Big Data Analytic and the AI workload increased a round of funding of the B $ 44 million, bringing its overall capital to $ 114 million.
The B series was led by its existing investors, such as Walden Catalyst Ventures, 83north, Koch Disruptive Technologies, Pitango First and Viola Ventures, as well as strategic investors, including Lip-Bu Tan, Managing Director of Intel and Managing Director Cataly Ventures and Eyal Waldman, co -founder and former CEO of Mellanox Technologies.
The APU architecture focuses on dealing with specific points of congestion of analytical components at computer level, as opposed to graphics processing units (GPU), which were originally designed for graphics and later amended for AI work and data related, according to the start.
“For decades, data detailed data have been based on standard processing units and, more recently, companies such as Nvidia have invested GPU boost for Analytics’ workload,” said Adi Gelvan, CEO of Speedata, in an interview with TechCrunch. “But these are either general purpose processors or processors designed for other workloads, not ground -made brands for data analysis. Our APU is designed for data processing and a single APU can replace servers’ shelves, providing dramatically better performance.”
Speedata was founded in 2019 by six founders, some of whom were the first researchers to develop Rebelled Architecture Technology (CGRA). The founders collaborated with ASIC design experts to tackle a fundamental problem: Data analyzes were carried out by general purpose processors. If the workloads became very complicated, they could need to use hundreds of servers. The founders believed they could develop a single special processor to complete the job faster using less energy.
“We have seen this as an opportunity to put decades of silicon research to convert how the industry processes the data,” Gelvan said.
His APU is targeting today Apache SPARK workloadBut its course map includes the support of every important data analysis platform, according to the company’s chief executive.
“Our goal is to become the standard data processor – just as GPUs have become the default for AI training, we want the APU to be the default for data analyzes on each database and detailed platform,” Gelvan told Techcrunch.
The start says it has several large companies to try its APU, though it refused to name them. The official product launch is set for Data & AI’s Databricks Summit in June second week. Gelvan said he would publicly present APU for the first time at the event.
Speedata claims a specific case where the APU completed a 19 -minute pharmaceutical workload, which was significantly faster than the 90 hours it took when a non -specialized processing unit was used, resulting in 280x speed improvement.
The start has stated that it has achieved many milestones from the latest concentration of funds, including finalizing the design and construction of the first APU in late 2024.
‘We have moved from the idea to testing A Field View Portal Array (FPGA)And now we are proud to say that we have a work material that we are starting today. We already have a growing business customer pipeline eagerly awaiting this technology and we are ready to escalate our market businesses, “Gelvan said.
