News & Analysis
Flash will ride DRAM bus in 2014, says Micron
Rick Merritt
2/7/2013 3:10 PM EST
Hybrid DIMMs leapfrog SSD
Click on image to enlarge.
Click on image to enlarge.
Hybrid DIMMs fill a gap between on-chip cache and PCI Express SSDs, Micron said.
|
|
|
|
|
Navigate to related information
rick.merritt
2/7/2013 6:09 PM EST
Will flash hop on DDR4? What impact will that have for SSDs?
Sign in to Reply
bberg
2/8/2013 1:39 PM EST
Viking Technology's ArxCis-NV has this same basic architecture for DDR3, and it is on the market now.
Sign in to Reply
resistion
2/7/2013 8:49 PM EST
There are serious errors in the chart on p. 2 like mlc and cell size for stt.
Sign in to Reply
resistion
2/8/2013 7:18 AM EST
"..post-DRAM world" Did Micron really use that expression?
Sign in to Reply
rick.merritt
2/12/2013 1:39 AM EST
It's a Micron foil
Sign in to Reply
resistion
2/12/2013 9:26 AM EST
It's rife with errors, like where did they get 10 nJ per bit for NAND? And STT needs transistor isolation, that's at least 6F^2. MLC is not even an option with low MR ratio.
Sign in to Reply
resistion
2/7/2013 8:54 PM EST
Mcp seems a commonly proposed solution for companies which make both dram and flash. Is it possible they're not making enough margin selling them separately? Their financials may indicate so.
Sign in to Reply
eewiz
2/7/2013 11:30 PM EST
Interesting concept. Helps for an Instant boot in PCs I guess.
Sign in to Reply
greenpattern
2/8/2013 1:53 AM EST
Looks incompatible with Intel CPUs with integrated memory controllers.
Sign in to Reply
rick.merritt
2/8/2013 12:36 PM EST
Samsung and Microsoft officially responded: No comment
Sign in to Reply
resistion
2/8/2013 12:46 PM EST
Micron also working with AgigA on NVDIMM
http://www.xbitlabs.com/news/memory/display/20121114132456_Micron_and_AgigA_to_Develop_Non_Volatile_DIMM_Technology.html
Sign in to Reply
vshadan
2/8/2013 12:54 PM EST
FLASH requires ERASE and WRITE, Just the WRITE is much longer time than DRAM's WRITE. How is that taken care?
Sign in to Reply
iniewski
2/9/2013 10:30 AM EST
Memory world is getting to be interesting...
Sign in to Reply
EFosters
2/9/2013 2:55 PM EST
Sounds similar to what Spansion and Virident tried to do a few years back, but with NOR. EcoRam DIMMs.
Sign in to Reply
resistion
2/10/2013 11:41 AM EST
I've always felt DRAM and NVM (i.e., Flash) had different applications, almost exclusive of each other. Generally, I don't expect what is buffered in DRAM to be stored directly, nor what has been stored to be directly read into a DRAM buffer.
Sign in to Reply
dnh
2/11/2013 3:17 PM EST
The placing of Flash on the DIMM is an exciting development, with the potential of greatly improving PC performance.
Unfortunately the figures in this article are so bad that it lacks credibility: What is 25M ns? Is that 25ms?
Also the figures appear to be scanned in at a low resolution and are very hard to read, especially in the General/PrintView.
While this is unfortunately typical of EETimes articles, the author of a post is responsible for his/her own postings.
Rick Merritt - please fix your own post, and try to exert your influence to fix this generic problem.
Sign in to Reply
rick.merritt
2/12/2013 1:37 AM EST
@dnh I used foils provided by Micron at the largest size we can accommodate (did you note you can "click to enlarge" them?)
Can you provide more specifics on what you would like to see improved?
Also, don't hesitate to email me at rick.merritt@ubm.com
Sign in to Reply