News
 

Bookmark and Share

(15) 

The next-generation DDR4 SDRAM memory will bring rather ultimate performance improvements to both desktops and laptops as well as servers and workstations. But the new performance heights will demand a rather radical change to topology of memory sub-system.

At a recent MemCon conference in Tokyo, Japan, Bill Gervasi, vice president of engineering at US Modular and a member of the JEDEC board of directors, revealed that the target effective clock-speeds for DDR4 memory would be 2133MHz - 4266MHz, an increase from previously discussed frequencies.  Apparently, JEDEC and memory manufacturers decided that the progress of DDR3 leaves no space for DDR4 data rates below 2133Mb/s.

The designers of DDR4 memory are looking forward 1.2V and 1.1V voltage settings for the new memory type and are even considering 1.05V option to greatly reduce power consumption of the forthcoming systems. It is expected that manufacturers of dynamic random access memory (DRAM) will have to use advanced fabrication technology to make the DDR4 chips. The first chips are likely to be made using 32nm or 36nm process technologies.

At present JEDEC expects to finalize the DDR4 specification in 2011 and start commercial production in 2012. Actual mass transition to the next-generation memory is projected to occur towards 2015.

But extreme performance will require a tradeoff. In DDR4 memory sub-systems every memory channel will only support one memory module, reports PC Watch web-site, since the developers substituted current multi-drop bus in facour of point-to-point topology. In order to overcome potential inability to install appropriate amount of memory into high-end clients as well as servers, the developers have reportedly presented two approaches:

  • DRAM manufacturers will need to dramatically increase capacities of memory chips by using multi-layer technique with through silicon via (TSV) technology. As a result, DDR4 memory chips with very high density will become relatively inexpensive. Obviously, this will naturally make memory upgrades slightly more complicated as in order to sustain multi-channel memory performance, all memory modules will have to replaced with more advanced DIMMs.
  • In case of server multi-layer DRAM IC approach only will not be viable for high-end machines. As a result, it is proposed that special switches are installed onto mainboards to allow multiple memory modules to work on a single memory channel.

The transition to DDR3 memory has taken a long time already and will take a couple more years to complete. But the transition to DDR4 memory will take even longer since it will be much more complicated for all the participants of the ecosystem: the DRAM chip makers, memory module manufacturers, mainboard makers, microprocessor producers, system builders and end-users.

Tags: DDR4, DRAM, JEDEC, , Intel, AMD

Discussion

Comments currently: 15
Discussion started: 08/16/10 03:23:48 PM
Latest comment: 09/02/10 06:55:59 AM
Expand all threads | Collapse all threads

[1-6]

1. 
"the target effective clock-speeds for DDR4 memory would be 2133MHz - 4266MHz."
Wow up to 4266Mhz! Im looking forward to that. If DDR4 can reach those speeds why was GDDR4 never adopted? Why did they skip straight to GDDR5?
0 0 [Posted by: dizzystuff  | Date: 08/16/10 03:23:48 PM]
Reply
- collapse thread

 
GDDR4 is based on DDR3 with some tweaks for graphics/point-to-point applications.

GDDR5 features some of DDR4 elements.
0 0 [Posted by:  | Date: 08/16/10 04:47:39 PM]
Reply
 
"GDDR4 is based on DDR3"

errrr ... wrong GDDR3/GDDR4 are based on old DDR2 in fact GDDR3 was first to sport on-die termination (at least you should remember it), and large block refresh/erase

GDDR4 is only lower voltage requirement of GDDR3 something more like of DDR3 LP-DDR3 transition today. It was good way of further OCing GDDR3 to 3600MHz x32 data rates

GDDR5 is more like VIA's QDR, except that external arbiter IC is integrated directly onto memory chip. And GDDR5 is direct advancement onto GDDR3, and instead multiplexed 2bit (GDDR3/DDR2 alike) it had 4bit multiplex (DDR3 alike) of course all that doubled working as DDR SDRAM.

Far better question would be how would really stack memory chip transistor density if NIp Bstds didnt effectively kill extremely advanced Qimonda (year ahead anyone else they had first GDDR5). Maybe we would see inexpensive 8MB DDR3 workstation modules based on 4Mb today not next year.

Anyway could you further explain cross-links between GDD5 and DDR4
0 0 [Posted by: OmegaHuman  | Date: 08/17/10 05:05:09 PM]
Reply
 
Graphics memory doesn't parallel desktop memory like that. Both GDDR3 and GDDR4 are based on DDR2, GDDR5 is based on DDR3.
0 0 [Posted by: Alereon  | Date: 08/17/10 08:42:13 AM]
Reply
 
simple... GDDR =! DDR.
GDDR5 is actually very similar to high clocked DDR3... with a few specialized attributes meant for video cards... GDDR3 and GDDR4 are both based on DDR2 though.

I recommend you read about what those terms mean in more detail here:
http://en.wikipedia.org/wiki/Double_data_rate
http://en.wikipedia.org/wiki/GDDR

EDIT: Wow, I got ninjad to hell and back
0 0 [Posted by: taltamir  | Date: 08/18/10 01:39:41 AM]
Reply

2. 
I'm highly skeptical of the value proposition of DDR4 within the next few years. Most applications (with the notable exception of data compression) still don't show a real performance improvement with DDR3 even over dual-channel DDR2-1066, and with the official DDR3 specs scaling beyond 2Ghz right now there just doesn't seem to be much of a need for a new standard. This is doubly true when you consider that upcoming high-end and server platforms are supporting quad-channel DDR3, and that Low Voltage (1.35v) and Ultra Low Voltage (1.25v) DDR3 products are available in the channel right now.
0 0 [Posted by: Alereon  | Date: 08/17/10 08:58:33 AM]
Reply
- collapse thread

 
Well there's definitely need for new standard now when even Sandy Bridge will sport quad channel DDR3 IMC Memory needs might not be such an issue on desktop market for quite some time (with those stunning speeds >4300MHz all promised up) but how would Intel/AMD make desktop/server CPU distinction if not by number of IMC

Thing it's kind of interesting (unanswered) question. Would still in the future CPU sport 64bit IMC or enormous number of 32bit ones? So in the end we could have 6 32bit IMCs sport six memory slots on desktop/workstation and 16 32bit IMCs in server segment? Is this really delusion or just ugly fact how cpu manufacturers could improve market segmentation?

DDR4 should certainly support enormous BW needs of future GP-GPU alike processor that will serve as CPU and could have humongous RAM (few TB) build with cheaper per GB DDR4 over some GDDR5/6 faster derivatives that waste 3x-6x times more power over cheaper DDR4 per GB, and being less durable at the same time. GDDR5 support ECC if i'm not mistaken
0 0 [Posted by: OmegaHuman  | Date: 08/17/10 05:28:45 PM]
Reply
 
"Well there's definitely need for new standard now when even Sandy Bridge will sport quad channel DDR3 IMC"
I hear AMD Bulldozer is dual channel and will support DDR3 speeds up to about 2133. If this is true, shouldn't quad channel sandy bridge easily beat bulldozer? Or does the amount of channels not really matter if higher speeds (and lower timings) are supported?
0 0 [Posted by: dizzystuff  | Date: 08/17/10 05:53:52 PM]
Reply
 
the point of a new standard, is to render all currently manufactured components "obsolete"... that using the "new standard" results in a mere 1% improvement to performance is irrelevant, they want people buying new parts and tossing old ones in the trash (because they are "too old to bother selling, nobody will pay anything for it"
0 0 [Posted by: taltamir  | Date: 08/18/10 01:45:35 AM]
Reply

3. 
This won't happen too soon ...
0 0 [Posted by: East17  | Date: 08/17/10 12:06:36 PM]
Reply

4. 
Yeah and I'm still waiting for my holographic memory cube that will hold exabytes of data on a desktop.In reality some DDR2 configured properly is faster that DDR3 so DDR 4 does not excite me in the least.On top of that RAMBUS will be waiting until it's popular (if ever) THEN file law suits against the manufacturers.FDunn
0 0 [Posted by: fdunn  | Date: 08/17/10 07:14:32 PM]
Reply
- collapse thread

 
RAMBUS has been filing lawsuits for YEARS... and winning!
RAMBUS was a new start up company that made some revolutionary advancements, the major manufacturers conspired to sell at a LOSS to drive them out of the market, RAMBUS went belly up, its talented staff fired, and its tattered remains bought by patent trolls who have been suing and harassing companies for years. (although, some of the "victims" of said patent trolls are those companies who originally ruined rambus by price fixing... so its hard to feel sorry for them)

Conspiring to fix prices is illegal, and there were court cases and convictions involved...
0 0 [Posted by: taltamir  | Date: 08/18/10 01:47:56 AM]
Reply

5. 
Who in God's name needs this? There's no practical difference between 1 GHz and 2 GHz DDR3 memory. Who cares about benchmarks?

All we got from DDR3 were high prices. There was a point where I paid areound 40$ for a 2x2 GiB DDR2-800 CL5 kit.
And now I'd have to pay 110$ for a standard 2x2 GiB DDR3-1333 CL9 kit, and it's not at all faster.

Sure, graphics cards need high speed memory, but they have different standards. GDDR3 and GDDR4 were based on DDR2, and GDDR5 is based on DDR3, higly modified though.

Well, maybe in 5 years DDR4 will actually make a difference, but I really doubt it. And one module per channel? Unless they start making 4 GiB or 8 GiB modules this will really suck.
Maybe in 5 years we'll have quad channel memory controlers in lowend CPUs. We'll see, but I'm definitely not looking forward to that.
0 0 [Posted by: Harry Lloyd  | Date: 08/18/10 03:51:25 AM]
Reply
- collapse thread

 
You mean 1Ghz DDR2 vs 2Ghz DDR3... )
0 0 [Posted by: TAViX  | Date: 08/19/10 02:39:10 AM]
Reply

6. 
why not 5GHz and make the current data bus width 4 fold wider at 5GHz?? and re engineer the whole PC architecture and ATX standard and just call it Generation 2, literally tripling quadrupling speeds
0 0 [Posted by: mike1101  | Date: 09/02/10 06:55:59 AM]
Reply

[1-6]

Add your Comment




Related news

Latest News

Tuesday, July 15, 2014

6:11 am | Apple Teams Up with IBM to Make iPhone and iPad Ultimate Tools for Businesses and Enterprises. IBM to Sell Business-Optimized iPhone and iPad Devices

Monday, July 14, 2014

6:01 am | IBM to Invest $3 Billion In Research of Next-Gen Chips, Process Technologies. IBM to Fund Development of 7nm and Below Process Technologies, Help to Create Post-Silicon Future

5:58 am | Intel Postpones Launch of High-End “Broadwell-K” Processors to July – September, 2015. High-End Core i “Broadwell” Processors Scheduled to Arrive in Q3 2015

5:50 am | Intel Delays Introduction of Core M “Broadwell” Processors Further. Low-Power Broadwell Chips Due in Late 2014

Wednesday, July 9, 2014

4:04 pm | Intel Readies New Quark “Dublin Bay” Microprocessors. Intel’s “Dublin Bay” Chips Due in 2015