SK Hynix is getting ready to make DDR5-8400 memory

mongeese

Posts: 643   +123
Staff
Oversimplified: Once or twice in a decade, the self-elected council in charge of such things throws new memory technologies into a basket and calls the basket DDR-whatever-the-next-number-is. Then, a few years later, a company like SK Hynix manages to turn that basket into the picnic we’ve all been waiting for.

Unlike the iterative upgrades that define the successive releases of most PC parts, memory upgrades are significant. DDR4 is now six years behind the times and DDR5 is bound to make it utterly obsolete. SK Hynix has thoughtfully warned us ahead of time, with an impressive roster of new specs and a rough release schedule.

The headlining feature is the raw speed: up to 8400 MT/s. At that speed, the per-channel theoretical bandwidth is 67.2 GB/s, which a lot more than DDR4’s rated maximum of 25.6 GB/s. A typical dual-channel configuration will offer 134 GB/s. Quad-channel 267 GB/s; octa-channel 538 GB/s. Real-world speeds tend to be about 30% lower in practice though.

Beyond speed, there’s a whole stack of other technologies.

A fourfold increase in density will make high capacity sticks cheaper, and new volumes of memory possible. Lowered operating voltage and peak-to-peak voltage (VPP) will boost power efficiency. On-die ECC (error correction code) and ECS (error check and scrub) make for much-improved error correction and reliability. Decision feedback equalization circuitry reduces the noise of the electronics which improves the per-pin speeds.

  DDR5 DDR4
Frequency* 3200 → 8400 1600 → 3200
Density 2 Gb → 16 Gb 8 Gb → 64 Gb
Operating Voltage 1.1 V 1.2 V
Peak-to-Peak Voltage 1.8 V 2.5 V
Burst Length 16 8
Bank Groups 8 4
Banks (Total) 32 16
Prefetch Length 16n 8n

The three key numbers when dealing with memory are: the number of banks; they’re what store the data. The prefetch length, which is how much data the memory subsystem can extract in one round. And lastly burst length, which is how long the subsystem has to extract each round of data.

A curious technical limit prevents the memory subsystem from jumping from bank to bank within one round. So it accesses as many banks as it can simultaneously -- but to do this it must reduce the speed it withdraws data from each, thus taking longer overall. In DDR5, the burst length is doubled so the time limit is doubled, and subsequently, the number of banks accessed is doubled. Then you get double the data and you need a doubled prefetch length to carry all that data out of the memory.

That might be an oversimplification but it’s the doubling of these three numbers that make DDR5 twice as fast as DDR4, with a few little enhancements and a node shrink down to 10nm providing the rest. That's also part of why DDR5-8400 isn’t truly operating at 8400 MHz as you’d expect from the name.

SK Hynix is preparing to start mass-production of DDR5 this year. They anticipate it taking 22% of the memory market in 2021 and 43% by 2022, so it won’t be long until all our gaming rigs need an update. AMD’s Zen 3 and Intel’s Sapphire Rapids architectures could both potentially have DDR5 support when they're launched in the coming year.

Image Credit: Liam Briese

Permalink to story.

 
What will be real world impact of this improvement? Do CPUs, GPUs and motherboards need to be updated to match? How will dual channel DDR5 compare to quad channel DDR4?
 
What will be real world impact of this improvement? Do CPUs, GPUs and motherboards need to be updated to match? How will dual channel DDR5 compare to quad channel DDR4?

The entire chipset and usually socket will change to accommodate new memory standards.

Consumer CPU performance won't exploit this bandwidth too much at the moment. Ryzen likes high bandwidth but you only gain so much. It's clearly going to make more of a difference in a few years when we're all on 16 and 32 core processors.

It most definitely would be a real boon for integrated graphics and APU performance though. If you doubled or trebled the bandwidth on machines that have APUs and share memory bandwidth the graphics performance will be much, much faster. Particularly notebooks then.

Quad channel always offers a lot of bandwidth but it means much greater expense and size. Be it CPU controllers, board configuration, power etc. Quad channel will remain a niche feature for consumer grade platforms.
 
Last edited:
This will make AMD's Ryzen G series processors much better at gaming, on new boards that support the fastest DDR5 RAM. Now, if they would only release a Ryzen G, with more than 8 or 11 Vega graphics, to take advantage of the 250% speed improvement for better gaming and faster Photoshop rendering. That would be cool.
 
I think Zen 3 will see a Q4 2020 launch and will be a DDR4 only.

And like every new Memory release over the last 20 years. DDR5 will be slow and expensive at launch and will get faster and cheaper.

Do I expect Zen 4 to be DDR5 and probably PCIe 5.0
 
Theoretically, a fourfold increase in density will make high capacity sticks cheaper, but real-world marketing says it will be 30% more expensive than current iterations - I'm just hoping it isn't going to be the case.
 
Theoretically, a fourfold increase in density will make high capacity sticks cheaper, but real-world marketing says it will be 30% more expensive than current iterations - I'm just hoping it isn't going to be the case.
When they first start trickling out they will be priced much higher than 30% extra compared to DDR4.
 
What will be real world impact of this improvement? Do CPUs, GPUs and motherboards need to be updated to match? How will dual channel DDR5 compare to quad channel DDR4?
Great explanations in this thread, but I'll just chime in with some more comments. Only the CPU and motherboard need to be replaced to support DDR5. But unlike a new GPU release, for example, DDR5 will be 'bundled' as a feature of a new CPU generation, ie when AMD includes it in a new CPU release that release might have a 30% performance boost, but part of that will be DDR5 and part will be improved clock speeds. You don't need to worry about it on an individual level.

As others have said, DDR5 will offer immediate advantages to systems that are already bandwidth starved; so APUs and ultra-high core count systems (24+). But games will be able to make use of the improved bandwidth quite soon I'd imagine, because consoles already use memory that is much faster than PCs.
 
This will not have so much impact especially because of the very limited amount.
So the prices will be for a longer time very high, the gains for high speed memory are for years over rated in general. My experience is that the step from 2133 over to 3200 does make a small impact but the faster the gains become much lower.
We are talking about a few percent in gain not much more, so again I do not see this coming on the market anytime soon at prices normal people can spend.
Maybe some of the super rich will use it when it comes to market, if there are motherboards out which can run them.
Because I am pretty sure they will need yet another memory socket to put them into.
We all know that the next in line has been cooked up already and its just a matter of time that one of these large memory moguls make it a reality.
Which is in my opinion is of course a good thing but I wonder if the rest of the hardware can keep up with the ever faster becoming secundairy hardware.
Videocards truly are still not using any of the real bandwidth the pci-e 4.0 slots can provide they still are only really using pci-e 2.0 x8 data speeds while the next generation already is launched and we still have not seen anything really making use of the full capacity of these slots.
I know they are far ahead in the labs which is not ready for the market for a long time, but it kinda amazes me that nothing really is capable of making use of all the surplus speed.
I admit not being anymore in the server world so I am not sure if the large server builders do make use of the latest improvements. But from what I have seen there is hardly see anything super impressive either.
If we will see videocards released able to make full use of the new standards is again looking at a crystal ball which will not anwer anything either, but this far only minor improvements have been made.
Untill someone is able to cook up some material which can lead us in the next stage of smaller footprint and faster chips. For now we are stuck at the current limits.
@mongeese yes I do not think there will not be much improvement for gamers at all or when it becomes the next level of work memory the impact will be minor as well.

 
Back