Samsung Unveils World's First 10nm-class 8 Gb LPDDR5 DRAM

BrianFagioli writes: Today, Samsung announces yet another milestone, this time with its low-powered memory. You see, Samsung has created what it calls the "industry's first 10-nanometer (nm) class 8-gigabit (Gb) LPDDR5 DRAM." The company promises significant power reduction -- up to 30 percent over LPDDR4X DRAM. This should be important for the upcoming 5G explosion. "The 8Gb LPDDR5 boasts a data rate of up to 6,400 megabits per second (Mb/s), which is 1.5 times as fast as the mobile DRAM chips used in current flagship mobile devices (LPDDR4X, 4266Mb/s). With the increased transfer rate, the new LPDDR5 can send 51.2 gigabytes (GB) of data, or approximately 14 full-HD video files (3.7GB each), in a second," says Samsung.

The Galaxy-maker further says, "The 10nm-class LPDDR5 DRAM will be available in two bandwidths -- 6,400Mb/s at a 1.1 operating voltage (V) and 5,500Mb/s at 1.05V -- making it the most versatile mobile memory solution for next-generation smartphones and automotive systems."

Re:

[Anonymous Coward]

14 full-HD video files (3.7GB each), in a second

That's over 10 Trump pee-pee tapes.

Surely that depends on how many Russian hookers he pays to pee on him and how full their bladders are.

Re:Bad math?

[Desler]

6400 Mbps is the data rate per pin. The higher data rate would be for a 64-bit memory bus.

Re:

[Desler]

6400 Mbps would be a 1-bit bus.

Re:

[Anonymous Coward]

maybe you should take this nerd shit elsewhere
this is slashdot we are a elon musk / facebook / apple fan blog

Re:

[hcs_$reboot]

you should check other pants than yours, you'll be surprised.

Re:

[ourlovecanlastforeve]

Samsung loves "class."

They use it on their TVs to make them look larger, too.

"50 inch class" usually means it's a TV that's 45 inches wide and 2 inches tall.

Re:Very Impressive. 10nm of what?

[ChrisMaple]

From https://en.wikichip.org/wiki/10_nm_lithography_process [wikichip.org] gate length 20 nm, 51 nm metal pitch. Similar to the practices of other semiconductor manufacturers, no honest person could call this a 10 nm process.

Re:

[religionofpeas]

It's the 10 nm/half a gate length process.

6400 Mbps != 51.2 GB/s

[sanosuke001]

Am I missing something or are the transfer amounts being interpolated over a number of 8gb chips on a single DRAM module? It says 6400 Mbps transfer which is 51.2 GB/s which doesn't seem like the same thing at all.

51.2GB = 419430.4 Mb = 65.536 times the stated transfer speed of 6400 mbps. It seems odd that the transfer speed is exactly a power of two...

Re:

[Desler]

Yes, you're missing that 6,400 Mbps is the data rate per pin. There's more than one pin on the memory bus.

Re:

[Desler]

Your math is off by a factor of 8. 51.2 GB/s is for a 64-bit bus.

Re:6400 Mbps != 51.2 GB/s

[iCEBaLM]

51.2GByte * 8 = 409.6Gbit
409.6Gbit / 6.4Gbit = 64

So it's 6.4Gbps (6400Mbps) per pin on a 64bit bus.

Why would you need 10 gigabit on mobile?

[Anonymous Coward]

This should be important for the upcoming 5G explosion. "The 8Gb LPDDR5 boasts a data rate of up to 6,400 megabits per second (Mb/s), which is 1.5 times as fast as the mobile DRAM chips used in current flagship mobile devices (LPDDR4X, 4266Mb/s).

Full motion compressed HD video is something like 25-40 megabits/second. What could you possibly do with gigabit speeds on a mobile phone? Then, even that's not fast enough and we need it 1.5 times faster?

Whoever wrote this article doesn't understand computing. T

Re:

[Desler]

It's a Brian Fagioli article. Of course the person is clueless.

Re:

[AHuxley]

8K display?

Re:

[Desler]

Huh? You're statement makes no sense. What does system memory bandwidth have to do with the video signal bandwidth of the display interface?

Re:

[AHuxley]

To make a $1100 phone stand out from a $250 phone. Why stop at 4K and HDR tech? Get more pixels per inch. That needs more hardware. Then the new device can have a 8K sticker.

so you can hit your cap and rack up $10/gig overag

[Joe_Dragon]

so you can hit your cap and rack up $10/gig overages so we can pay off the new 5G network in 6mo

Re:

[shmlco]

I don't want to ask why people are so clueless... but why are people so clueless?

Faster chips mean that processors can do more work... or do the same amount of work in less time. Accomplishing X in half the time means the processor is running at speed half the time, which dramatically reduces the amount of power needed to accomplish the task, which in turn improves battery life.

Which, in your case, means you can spend more time trolling Facebook and news sites.

Re:

[darkain]

First, the connection isnt one-to-one, it is one-to-many. Using time-division-multiplexing (TDM), this bandwidth is shared between several client devices (cell phones, tables, etc) per radio channel. Next, as you increase bandwidth, you can decrease the length of real-time used for each time slice within a TDM system, and with a lower time slice, latency is reduced.

So, while YOU as an INDIVIDUAL may not need 10gbps on your personal cell phone, when you're in a building with 50,000k other people (sporting ev

Re:

[religionofpeas]

Full motion compressed HD video is something like 25-40 megabits/second.

That's compressed, like you said. But before it goes to the display, it needs to be uncompressed, which involves a number of operations for each bit, involving multiple memory accesses.

Re:

[Z80a]

Running 3D games do require every bit you can give.
A regular PS4 for example have 176GB/s of memory speed access.

Re:

[ausekilis]

But what if I want to watch 300 HD streams at once?

Latency?

[lobiusmoop]

Article doesn't mention if the DRAM latency is any better or not... (DRAM latency hasn't changed significantly in the last 15 years)

Re:

[hcs_$reboot]

As long as we have the speed, does the latency really matter?

Re:

[Kokuyo]

Uh Hell Yeah?
That's what RAM is used for after all.

Re:

[alvinrod]

If you're worried about latency, get a bigger cache. Maybe we could drop latency on RAM, but it would probably be at the expense of other important aspects such as capacity. Most people would probably prefer the added capacity since having to go to disk (even if it's an SSD) is much worse in terms of extra latency than having to go from cache (especially once you get out to L3) to RAM.

Perhaps you could argue that now we have more than enough RAM (64GB ought to be enough for anybody) but for most of histo

Re:

[TeknoHog]

It depends on what you mean by "speed". Usually, perceived computing speed involves both throughput and latency in some way. IMHO, interactive and realtime operations are much more about latency, while throughput only helps you process more stuff per frame. I'm not into gaming, but I'd expect that bad latency can easily make things completely unplayable. Bad throughput, OTOH, will only mean things like lower graphics quality. For a non-realtime look at latency, one of my favourite quotes:

"Why people think "performace" means "throughput" is something I'll never understand. Throughput is _always_ secondary to latency, and really only becomes interesting when it becomes a latency number (ie "I need higher throughput in order to process these jobs in 4 hours instead of 8" - notice how the real issue was again about _latency_)." -- Linus Torvalds

Re:

[GuB-42]

Latency matters a lot. In computing, that's an aspect that is too often overlooked.

Increasing speed is easy, just add more of the same, but decreasing latency is impossible. The common analogy is: no matter what you do, you won't get a baby in less than 9 months.

Have you already wondered why oldshool computers sometimes feel faster than top of the line modern machines. You have your answer. Modern machines are extremely fast, but the countless layers of abstraction between you and the moving electrons creat

Online

[Wijayanti]

apotik jual obat perangsang [aldyperkasa.com]

Hammer me

[HalWasRight]

I wonder how much design or testing they've done to prevent Rowhammer flaws. The smaller the geometry the less charge is held.