LONDON – Ceva, a DSP IP licensor, has offered $75 million in cash for processor IP licensor MIPS Technologies in an attempt to outbid GPU licensing company Imagination Technologies.
MIPS (Sunnyvale, Calif.) announced on Nov. 5 that Imagination would acquire outstanding MIPS shares for $60 million in cash following the completion of the proposed patent sale transaction with Bridge Crossing.
The competing bid by Ceva (Mountain View, Calif.) is reportedly structured along the same lines as Imagination's (Kings Langley, England). That bid calls for a large proportion of the MIPS patent portfolio to be sold to Bridge Crossing for about $350 million. Ceva has valued the remainder of MIPS at a 25 percent higher value than Imagination did in its bid. The Ceva proposal is still subject to the approval of its board of directors and due diligence, MIPS said.
Ceva has emerged as one of the beneficiaries of the changing mobile handset IC market. With more than $150 million cash on hand earlier this year, the company was talking about expanding its customer base and diversifying its revenue sources.
MIPS said its board is evaluating Ceva's proposal before deciding which offer to accept. It said it is now in discussions with Ceva.
Ceva's DSP cores complement MIPS' architecture processors, but in ways that differ from Imagination's graphics cores that are used alongside MIPS processors. For example, Ceva and MIPS cores are often used together in home entertainment systems, although Ceva has also been successful in mobile communications applications where MIPS cores are rarely found.
MIPS architecture is seldom been used in Embedded system as I know in China. Comparing with ARM, it lag behind much.
It is not a bad thing to be purchased. I plan to obtain some information about MIPS to learn, now I change my mind.
If I read this right they won't "kill" MIPS outright but the IP licensing biz will surely be dead? Who is going to license MIPS cores from a major chip competitor? They may continue to issue "MIPS architecture licenses" for those who design their own MIPS cores.
That will lead to dilution of the MIPS software ecosystem with multiple proprietary cores which each require different tweaks for software to run on them. This is exactly why Linaro was created for ARM to reduce that problem. For example they are working on a single linux kernel that runs on multiple vendors' ARM-based cpus.
Broadcom and others could have bought MIPS back in early 2000's when the stock price was as low as $1.25 per share but they wanted MIPS to remain an independent IP company not controlled by one vendor. The reason companies license a core instead of designing their own proprietary core is to exploit a proven architecture and its software ecosystem. Also reduces training cost because you can readily find engineers who already know that architecture. Same reasons you license a RTOS instead of creating your own. But if its controlled by your competitor you won't use it. And if only one company uses it the software ecosystem won't build up critical mass.
Like I said before MIPS won't die but it won't be a major widely known cpu architecture anymore. It will become lesser known like PowerPC. See my previous comments about why MIPS lost to ARM at end of this article:
It's all about the software people, not the hardware.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.