LONDON – Reverse engineering and analysis consultancy Chipworks Inc. has torn down a few Samsung GT-E2550 GSM cell phones bought in the United States but intended for different geographic markets in Asia and Europe while on the hunt for phase-change memory ICs.
According to a technology blog by Rajest Krishnamurthy from Chipworks (Ottawa, Ontario) 512-Mbit phase-change memory (PCM) implanted in an approximately 65-nm manufacturing process was found in a multi-chip package in some of the phones.
The Samsung-produced PCM deployed in a Samsung GSM phone is the only known commercial deployment of a memory technology that had shown promise as a potential replacement for flash non-volatile memory technology. The continued scaling of flash memory and lack of uptake has cast doubt on the viability of PCM.
"The main PCB from a few of the GT-E2550 phones had a NOR flash multichip package (MCP) chip, with package marking K5N1229ACD. However, to our surprise, the PCB from some of the phones with specific branding name and with specific marks on the product label, used a PCM MCP memory chip with package marking K571229ACM, whereas others used a K5N122ACD NOR flash part," according to the Chipworks posting.
Numonyx, subsequently acquired by Micron Technologies Inc., released a 90-nm 128-Mbit phase-change memory in 2008, which it formalized as the Omneo range of serial and parallel access memories in April 2010. But the company has not said anything about design wins or volume production. Numonyx was also said to be developing a 1-Gbit phase-change memory in a 45-nm process, but there has been no news of sampling or volume production for a device that was expected to appear in 2010.
The posting includes package shots, x-ray photographs of the multi-chip packages and die shots of the equivalent phase-change memory (identified on the die as KPS1215EZA) and NOR flash (identified on the die as K8S1215EZC).
Chipworks states it is preparing structural analysis reports on both the 65-nm PCM and the 56-nm NOR flash integrated circuits.
The Samsung-produced PRAM deployed in a Samsung GSM phone is NOT A COMMERCIAL DEPLOYMENT of PRAM, and PRAM has shown absolutely no promise as a potential replacement for flash non-volatile memory technology.
As it was explained to you, Samsung's GT-E2550 did in fact originally have K571229ACM in the bill of materials as of Spring 2010. However, the chip was almost immediately replaced with a regular NOR MCP due to power consumption issues. I do possess a GT-E2550 handset, and it has NOR instead of PRAM. You don't believe me? Go ahead and purchase a GT-E2550 unit. You will see!
Slicing dead, non-functioning chips from fake, non-commercial handsets manufactured a year ago only proves Samsung's utter failure to commercialize PRAM.
PRAM is not used in any product on market today. Period.
Numonyx is not developing any 1-Gbit phase-change memory in a 45-nm process.
The PCM/PRAM techno-Ponzi scheme is nearly over.
Is the population of PCM in Samsung phones actually increasing? The original Chipworks report indicated that they had purchased six phones and only one contained PCM. In this latest report they purchased a few and "some of the phones" had PCM. This would appear to mean the PCM population is now at least two in a "few". Now providing the "few" is not greater than 12 it suggests the probability of finding a phone with a PCM is increasing!
It would be nice to know the details of exactly where the phones were purchased so that the experiment could be repeated. The only information in that regard is a source of unlocked phones.
Given the vagaries of the global supply chain, even if we knew when the phones were bought, I don't believe we could be clear as to whether PCM usage is increasing or has been stopped. The only people that know that are in Samsung.
Much time is spent after the first unsuccessful use of PCM, and by that time Flash Memory has traveled a successful trip virtually in most of the portable devices, even Samsung is also try to come with Flash bases HDDs. Looking at this it will be really very hard time to prove PCM.
Peter you are in the Catbird Seat, what about a formal request to Samsung to clarify the situation with respect to these phones and findings.
The pcbs in the tear down appear to have Samsung ID on them and as far as it is possible to determine from the photographs there is no clear sign of rework on the pcb. If so that would suggest to me that Samsung actually fabricated the phones.
The worrying question is if Samsung are seriously trying to get PCM design wins why would Samsung not shout from the rooftops, we have made the commitment to PCM, proved the point, it's now your turn.
I did write to Samsung asking for clarification on the status of their 1G-bit-so far no reply
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.