Samsung Mass Producing 6Gb LPDDR3 Mobile DRAM

News
By Kevin Parrish
published

South Korea-based Samsung Electronics announced on Wednesday that it is now using 20 nm process technology to mass produce 6 Gb LPDDR3 mobile DRAM. The announcement follows Samsung's move to use 20 nm processing technology on 4 Gb DDR3 for PCs back in March, which was an industry first.

"Our new 20 nm 6 Gb LPDDR3 DRAM provides the most advanced mobile memory solution for the rapidly expanding high-performance mobile DRAM market," said Jeeho Baek, vice president, memory marketing, Samsung Electronics.

According to the company, four of these 6 Gb LPDDR3 dies can be crammed into a single 3 GB package (24 Gb) that's now 20 percent smaller and uses 10 percent less energy than current 3 GB memory packed with 25 nm 6 Gb LPDDR3 chips.

Samsung states that the new 20 nm 6 Gb chip has a per-pin data transfer rate of up to 2,133 Mbps. The company also reports that its own productivity has improved more than 30 percent when compared to using the older 25 nm processing technology.

"We are working closely with our global customers to offer next-generation mobile memory solutions that can be applied to a more extensive range of markets ranging from the premium to standard segments," Baek added.

Samsung indicated that it plans to use the 20 nm process technology in the future to strengthen its product portfolio and stay ahead in the high-density mobile DRAM market.

Follow Kevin Parrish @exfileme. Follow us @tomshardware, on Facebook and on Google+.

Kevin Parrish
11 Comments Comment from the forums
  • firefoxx04
    Should the title have 6GB (capital B) instead of 6Gb?

    6Gb (lower case b) makes me think about a data rate like 6Gb/s transfer speed, not a capacity. Not to mention, Gigabyte is GB not Gb (Gigabit).

    Not trying to be a nitpick, this actually confused me when I started reading.
    Reply
  • IInuyasha74
    No its Gb. Gigabit instead of Gigabyte. A gigabyte is eight times larger than a gigabit, hence why 4 of these chips which have 6Gb each, adds up to 24Gb, or in other words 3GB of storage.

    Its the same Gb used in speed, just one is larger than the other. In terms of RAM chips Gb is the commonly used term
    Reply
  • erendofe
    little Gb is GIGA-bits and GB is gigabytes a byte is what ever the bus with is the CPU is IE> a 32 bit CPU uses 32 bit bytes, 64 bit cut is 64 bit bytes etc.
    Reply
  • IInuyasha74
    14204029 said:
    little Gb is GIGA-bits and GB is gigabytes a byte is what ever the bus with is the CPU is IE> a 32 bit CPU uses 32 bit bytes, 64 bit cut is 64 bit bytes etc.

    Not quite. There isn't always an "s" on the end. It can be read either way, sort of like mm and cm, or written either way but the "s" is not a constant. It is not so uncommon to see "1 GB" and "1 Gb" also which would indicate single measurements and therefore not be read or written out with the "s" on the end.

    A byte is not the bus inside of the CPU. A byte is 8-bits. It goes back to binary coding. Binary is written in three groups of eight digits being either 0 or 1. Each digit either 0 or 1 consumes 1 bit to store the information.So a full line in binary, 00000000.00000000.00000000 would require 24 bits to store. Or just 3 bytes.

    In computers connections, typically called a "bus" the "bit" used is different. For example, each memory channel to the CPU has a 64-bit connection to the CPU. What this is directly referring to is the existence of 64 wire connections connecting them. The reason for these connections often being called "bits" is owing to the ability of them each to be transferring a single bit of data at the same time. Of course that is increased by clock speed, but in basic that is why they are called bits.
    http://www.hardwaresecrets.com/article/Everything-You-Need-to-Know-About-the-Dual-Triple-and-Quad-Channel-Memory-Architectures/133/1
    Reply
  • erendofe
    14205103 said:
    14204029 said:
    little Gb is GIGA-bits and GB is gigabytes a byte is what ever the bus with is the CPU is IE> a 32 bit CPU uses 32 bit bytes, 64 bit cut is 64 bit bytes etc.

    Not quite. There isn't always an "s" on the end. It can be read either way, sort of like mm and cm, or written either way but the "s" is not a constant. It is not so uncommon to see "1 GB" and "1 Gb" also which would indicate single measurements and therefore not be read or written out with the "s" on the end.

    A byte is not the bus inside of the CPU. A byte is 8-bits. It goes back to binary coding. Binary is written in three groups of eight digits being either 0 or 1. Each digit either 0 or 1 consumes 1 bit to store the information.So a full line in binary, 00000000.00000000.00000000 would require 24 bits to store. Or just 3 bytes.

    In computers connections, typically called a "bus" the "bit" used is different. For example, each memory channel to the CPU has a 64-bit connection to the CPU. What this is directly referring to is the existence of 64 wire connections connecting them. The reason for these connections often being called "bits" is owing to the ability of them each to be transferring a single bit of data at the same time. Of course that is increased by clock speed, but in basic that is why they are called bits.
    http://www.hardwaresecrets.com/article/Everything-You-Need-to-Know-About-the-Dual-Triple-and-Quad-Channel-Memory-Architectures/133/1

    Reply
  • erendofe
    actually a byte at the hardware level as I understood it was determined by a cpu's physical data bus width. the address bus dictates the amount of memory the CPU can actually access. the byte grouping you illustrated as I understand is represented at the software level and is hardware androgynous. itel its self requires a 64 bit cpu bus to enable some of its coding features we not take for granted because the instructions are 64 bits long internally
    Reply
  • IInuyasha74
    14209867 said:
    actually a byte at the hardware level as I understood it was determined by a cpu's physical data bus width. the address bus dictates the amount of memory the CPU can actually access. the byte grouping you illustrated as I understand is represented at the software level and is hardware androgynous. itel its self requires a 64 bit cpu bus to enable some of its coding features we not take for granted because the instructions are 64 bits long internally

    Your understanding is flawed.

    A byte is eight bits. It is in hardware in RAM and RAM like storage such as cache. It is also present in hard drives.
    http://en.wikipedia.org/wiki/Byte
    http://searchstorage.techtarget.com/definition/byte
    http://programmers.stackexchange.com/questions/120126/what-is-the-history-of-why-bytes-are-eight-bits
    https://web.stanford.edu/class/cs101/hardware-1.html
    http://en.wikipedia.org/wiki/Hard_disk_drive#Calculation

    In a CPU there is the address register that relates directly to how much RAM the CPU is able to see, however this is also measured to be 64-bits, not bytes. This also is not a type of bus, this is an address register which means its more of a listing of the information present or a mapping and has no affect on transfer speed, all computer buses have a direct impact on transfer speed.
    http://en.wikipedia.org/wiki/64-bit_computing

    As for the last bit again assuming you are referring to bytes being androgynous to storage devices, the amount of storage on a HDD or how much the PC can see has no connection with the CPU being 32-bit or 64-bit. If the motherboard supports a 4TB hard disk, it will see 4TB with a 32-bit or 64-bit CPU or OS installed.
    Reply
  • erendofe
    14210093 said:
    14209867 said:
    actually a byte at the hardware level as I understood it was determined by a cpu's physical data bus width. the address bus dictates the amount of memory the CPU can actually access. the byte grouping you illustrated as I understand is represented at the software level and is hardware androgynous. itel its self requires a 64 bit cpu bus to enable some of its coding features we not take for granted because the instructions are 64 bits long internally

    Your understanding is flawed.

    A byte is eight bits. It is in hardware in RAM and RAM like storage such as cache. It is also present in hard drives.
    http://en.wikipedia.org/wiki/Byte
    http://searchstorage.techtarget.com/definition/byte
    http://programmers.stackexchange.com/questions/120126/what-is-the-history-of-why-bytes-are-eight-bits
    https://web.stanford.edu/class/cs101/hardware-1.html
    http://en.wikipedia.org/wiki/Hard_disk_drive#Calculation

    In a CPU there is the address register that relates directly to how much RAM the CPU is able to see, however this is also measured to be 64-bits, not bytes. This also is not a type of bus, this is an address register which means its more of a listing of the information present or a mapping and has no affect on transfer speed, all computer buses have a direct impact on transfer speed.
    http://en.wikipedia.org/wiki/64-bit_computing

    As for the last bit again assuming you are referring to bytes being androgynous to storage devices, the amount of storage on a HDD or how much the PC can see has no connection with the CPU being 32-bit or 64-bit. If the motherboard supports a 4TB hard disk, it will see 4TB with a 32-bit or 64-bit CPU or OS installed.

    Reply
  • erendofe
    no a byte is not necessarily 8 bits it is dictated by the cpu's data bus x86 cpu's or example are 64 bits wide the can toggle between 64 bit and dual 32 bit code execution. thus each instruction is either written in 64 or 32 bits. each instruction MUST at least 1 byte. and breaking the instruction into groups of 8 bits would cause cohesion loss because the CPU operates on all 32 bits simultaneously. reaking the instruction into 8 bit groups could change it from a load command for example to a multiply command, or even cause a execution halt. also the data buss IS NOT the address bus they are separate. data registers do not dictate maximum RAM the track memory addressed being accessed the physical number of address lines (wires) dictate mx ram. registers in them selves are actually memory cells inside the cpu. for those interested I recommend reading "upgrading and repairing PC's by Scott Muller. he;s an engineer actually and his work is used for foundation of many college level computes technician courses. in fach it's the meat of the A+ exam
    Reply
  • IInuyasha74
    14210395 said:
    no a byte is not necessarily 8 bits it is dictated by the cpu's data bus x86 cpu's or example are 64 bits wide the can toggle between 64 bit and dual 32 bit code execution. thus each instruction is either written in 64 or 32 bits. each instruction MUST at least 1 byte. and breaking the instruction into groups of 8 bits would cause cohesion loss because the CPU operates on all 32 bits simultaneously. reaking the instruction into 8 bit groups could change it from a load command for example to a multiply command, or even cause a execution halt. also the data buss IS NOT the address bus they are separate. data registers do not dictate maximum RAM the track memory addressed being accessed the physical number of address lines (wires) dictate mx ram. registers in them selves are actually memory cells inside the cpu. for those interested I recommend reading "upgrading and repairing PC's by Scott Muller. he;s an engineer actually and his work is used for foundation of many college level computes technician courses. in fach it's the meat of the A+ exam

    Again, no it isn't. All the information is there. Your understanding is flawed, you don't know what a byte is. Nor does it seem you really understand what an address bus is, why it exists, and are confused on the working of the CPU in several ways. I would highly suggest you read more before attempting to help anyone or answer any questions, otherwise you will give people wrong information and negatively impact other people trying to understand these systems.

    I suggest you start with the first link I posted above as it goes over multi-channel configurations and specifically will tell you that the number of channels does not determine the max memory size the CPU can recognize.
    Reply