• Article
[ Printer Friendly ]
Synopsys joins embedded memory IP market
Joint development deal with Novelics pits EDA vendor against former acquisition target MoSys and IP development partner ARM.
Synopsys today uncorked a technology development and distribution agreement with startup Novelics for embedded memory IP, setting the stage for a showdown with ARM, Virage Logic and former Synopsys acquisition target MoSys.
The deal focuses on 1T and 6T SRAM. Synopsys and Novelics claim their technology eliminates the pricing penalty between a single-transistor (1T) version and the six-transistor (6T) version, allowing engineers for the first time to make a tradeoff based solely upon features.
Those tradeoffs include space—a 1T configuration takes up less physical space than a 6T version—as well as energy consumption. While the 6T technology uses less power in active mode, leakage is higher in standby mode. In addition, both companies said their agreement will provide both 6T and 1T versions at roughly the same time for the same node, while in the past 1T modules lagged the 6T versions by at least one process node.
“Before this, the additional cost (for 1T SRAM) made it prohibitive,? said Joachim Kunkel, VP and general manager of Synopsys’ solutions group. “In other cases, chip designers couldn’t wait for it to become available.?
The rough calculation is that for chips containing more than 50 percent embedded memory, a 1T module will save 20 percent in die area. That is particularly important in some portable electronics devices, where savings can reduce the overall size of the device or allow designers to add more features into the same space.
Kunkel said that most companies stick with memory designs for a process node, so if they design in a company’s memory at 130nm, they are unlikely to look for alternatives until they begin using 90nm technology. He said that opens the door for the new joint venture to win new business at 65nm and beyond.
Novelics has been working with Synopsys for about a year, using TSMC to qualify designs for 6T and 8T at 65 nanometers using low-power (LP) processes, and will qualify 1T memory at 130nm next quarter using LPG (triple-gate oxide), according to Eric Collins, Novelics’ VP of business development. Both Novelics and Synopsys are weighing whether to qualify the 1T memory at 90 nanometers, or whether to skip that node and go straight to 65nm, where silicon is currently being qualified. EsinTerzioglu, CTO at Novelics, said there are more requests for the 65nm technology than 90nm versions.
Behind the scenes
Synopsys is no stranger to the embedded memory IP market. The company, which has been pushing standard IP as a way to supplement its EDA tools portfolio, has been eyeing the embedded memory space for at least several years. The company announced plans to acquire MoSys, or Monolithic System Technology, in early 2004, but scrapped those plans in April of that year. MoSys sued Synopsys for breach of contract, but the two settled their differences several months later.
Since then, Synopsys’ embedded memory IP group has been operating in stealth mode, although insiders say the group has always had at least 25 engineers. The company would not comment on what those engineers were working on or why the group’s efforts were kept quiet.
About a year ago, Novelics approached Synopsys and the two began working together on a plan that would go head-to-head with MoSys in the 1T space, and longtime Synopsys development partner ARM and Virage Logic in the 6T space.
BruniBuric, VP of product marketing and strategic found relationships at Virage Logic, said Synopsys continues to be one of its partners and said that Synopsys tried to distribute TSMC’s IP for awhile but was not considered a serious competitor in that space. He said that more competition is good, because it validates a market, but that just because Synopsys has jumped into the embedded memory IP market doesn’t mean it will be successful.
“We have seen it is very difficult to gain the trust of customers,? Buric said. “People are extremely careful about embracing anything new.
MoSys and ARM were unavailable for comment by press time.
Novelics was started three years ago with angel investor funding. The company remains private. Neither Synopsys nor Novelics officials would speculate on how their relationship would evolve. At this point it is confined to the joint technology development based on Synopsys’ IP development environment and Novelics’ MemQuest compiler technology, as well as joint marketing of the technology through Synopsys’ well-developed global distribution channel.
Ed Sperling is a freelance writer and editor based in Silicon Valley. He is a regular contributor to technology publications and a frequent moderator and speaker. He has spent the past two decades immersed in technology and is the recipient of numerous awards for journalistic excellence. The deal focuses on 1T and 6T SRAM. Synopsys and Novelics claim their technology eliminates the pricing penalty between a single-transistor (1T) version and the six-transistor (6T) version, allowing engineers for the first time to make a tradeoff based solely upon features.
Those tradeoffs include space—a 1T configuration takes up less physical space than a 6T version—as well as energy consumption. While the 6T technology uses less power in active mode, leakage is higher in standby mode. In addition, both companies said their agreement will provide both 6T and 1T versions at roughly the same time for the same node, while in the past 1T modules lagged the 6T versions by at least one process node.
“Before this, the additional cost (for 1T SRAM) made it prohibitive,? said Joachim Kunkel, VP and general manager of Synopsys’ solutions group. “In other cases, chip designers couldn’t wait for it to become available.?
The rough calculation is that for chips containing more than 50 percent embedded memory, a 1T module will save 20 percent in die area. That is particularly important in some portable electronics devices, where savings can reduce the overall size of the device or allow designers to add more features into the same space.
Kunkel said that most companies stick with memory designs for a process node, so if they design in a company’s memory at 130nm, they are unlikely to look for alternatives until they begin using 90nm technology. He said that opens the door for the new joint venture to win new business at 65nm and beyond.
Novelics has been working with Synopsys for about a year, using TSMC to qualify designs for 6T and 8T at 65 nanometers using low-power (LP) processes, and will qualify 1T memory at 130nm next quarter using LPG (triple-gate oxide), according to Eric Collins, Novelics’ VP of business development. Both Novelics and Synopsys are weighing whether to qualify the 1T memory at 90 nanometers, or whether to skip that node and go straight to 65nm, where silicon is currently being qualified. EsinTerzioglu, CTO at Novelics, said there are more requests for the 65nm technology than 90nm versions.
Behind the scenes
Synopsys is no stranger to the embedded memory IP market. The company, which has been pushing standard IP as a way to supplement its EDA tools portfolio, has been eyeing the embedded memory space for at least several years. The company announced plans to acquire MoSys, or Monolithic System Technology, in early 2004, but scrapped those plans in April of that year. MoSys sued Synopsys for breach of contract, but the two settled their differences several months later.
Since then, Synopsys’ embedded memory IP group has been operating in stealth mode, although insiders say the group has always had at least 25 engineers. The company would not comment on what those engineers were working on or why the group’s efforts were kept quiet.
About a year ago, Novelics approached Synopsys and the two began working together on a plan that would go head-to-head with MoSys in the 1T space, and longtime Synopsys development partner ARM and Virage Logic in the 6T space.
BruniBuric, VP of product marketing and strategic found relationships at Virage Logic, said Synopsys continues to be one of its partners and said that Synopsys tried to distribute TSMC’s IP for awhile but was not considered a serious competitor in that space. He said that more competition is good, because it validates a market, but that just because Synopsys has jumped into the embedded memory IP market doesn’t mean it will be successful.
“We have seen it is very difficult to gain the trust of customers,? Buric said. “People are extremely careful about embracing anything new.
MoSys and ARM were unavailable for comment by press time.
Novelics was started three years ago with angel investor funding. The company remains private. Neither Synopsys nor Novelics officials would speculate on how their relationship would evolve. At this point it is confined to the joint technology development based on Synopsys’ IP development environment and Novelics’ MemQuest compiler technology, as well as joint marketing of the technology through Synopsys’ well-developed global distribution channel.
......................................................................
