Intel, Qualcomm win deal to design 7nm silicon for US defense agencies
- Reference: 1629783907
- News link: https://www.theregister.co.uk/2021/08/24/intel_qualcomm_ramp_c/
- Source link:
The two companies were jointly [1]awarded work under the Rapid Assured Microelectronics Prototypes – Commercial (RAMP–C), an effort designed to ensure the Department of Defense (DoD) can source the chips it needs from a US-based fabrication facility, rather than relying on offshore plants.
The scheme also seeks design capabilities for the chips to be produced onshore, which is probably where Qualcomm comes into the picture seeing as it's a fabless company.
[2]Samsung sprints past Intel to become world's semiconductor sales leader
[3]Pick a core, any core, says Intel – we'll magically put the right workload onto one in a hybrid SoC or accelerator
[4]SK hynix to create US-HQ'd NewCo for Intel's outgoing $9bn NAND biz
The DoD prefers its work be done onshore for security reasons, as it enjoys knowing exactly who made its kit and being able to observe chains of custody closely. The Department can also order rather small runs of product, often tweaking commercial-off-the-shelf kit – both to meet its needs and to present a different target to adversaries. Such orders are easier to arrange under arrangements like RAMP–C.
Intel's [5]announcement of its win omits any mention of Qualcomm but does mention its other partners and suppliers. The announcement also focusses on the deal being a sign that Chipzilla's [6]relaunched foundry business is on the right track.
[7]
It's right to do so because RAMP–C specifies access to <7nm fabrication processes, and [8]Intel's record at that scale features missed deadlines and [9]so did its initial 10nm efforts . That the DoD thinks Intel is on track is probably as good an endorsement as Intel can get. ®
Get our [10]Tech Resources
[1] https://nstxl.org/opportunity/rapid-assured-microelectronics-prototypes-commercial-ramp-c/
[2] https://www.theregister.com/2021/08/20/global_semiconductor_sales/
[3] https://www.theregister.com/2021/08/19/intel_architecture_day/
[4] https://www.theregister.com/2021/08/05/sk_hynix_intel_newco/
[5] https://www.intel.com/content/www/us/en/newsroom/news/intel-wins-us-project-develop-foundry-ecosystem.html
[6] https://www.theregister.com/2021/03/24/intel_gelsinger_strategy/
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2YSTDYTgWqjzWbQ3A6hpukAAAAA4&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[8] https://www.theregister.com/2020/07/27/intel_engineering_boss_out/
[9] https://www.theregister.com/2018/12/12/intel_architecture_future/
[10] https://whitepapers.theregister.com/
Re: high end vs bulk
I suppose there are pros and cons.
A lot of automotive would prefer long term reliability over top end speeds.
I'm not quite sure how the mobile arena works these days. Frankly every mid-range phone has a decent 8+ core CPU so then you're into GPU, RAM, screen quality. I find mid-range Asian phones quite adequate for my needs. Although there are some beastly flagship speeds available
Would DoD need something fancy like AI or satellite image processing?
Seems a strange choice.
Intel seems to have lost Process mojo during 22nm to 10nm transition period. They bought Altera but have been failing at design (Cable modem chips and failure of Atom as an alternative to ARM for mobile). The debacle of the non-volatile alternative to Flash.
Qualcomm have bought in a lot of expertise and true in house designs are a narrow field. They prefer income from royalties where they double dip.
I'd have thought that some other US companies such as ADI, Xilinx and TI are better chip designers.
What exactly do they need 7nm for outside of memories, gpus, FPGAs and cpus? Also a custom military memory, gpu, FPGA or cpu is a failed idea. Only a few ASICs would be of value, perhaps prototyped with a Xilinx FPGA with an ARM core.
AI or satellite Image processing is better done with existing commodity HW and put effort into better SW.
high end vs bulk
I wonder if all of this 7nm/10nm stuff is really what is needed. It feels like these modern processes are simply causing problems. Maybe going back to cheaper chips, albeit with less fancy technology, may be better for the bulk of chips we need.
We didn't have chip shortages before every statement became "we must use TSMR 7nm chips"