LRDIMM buffer chipset makers

Who makes LRDIMM buffer chipsets ?

UPDATE: 06/15/2012 – added Inphi comments on competitors
UPDATE: 06/15/2012 – added IDTI comments on Inphi/JEDEC
UPDATE: 06/29/2012 – institutional interest declines
UPDATE: 07/06/2012 – non-viability of LRDIMMs

The same players who make RDIMM buffer chipsets (that are used by memory module makers to make RDIMM memory modules).

LRDIMM buffer chipsets are currently only made by Inphi.

IDTI has prudently (after what you read below) postponed LRDIMMs to end of 2012 (i.e. skipping Romley and targeting the Ivy Bridge series according to their conference call).

Texas Instruments has not been interested in LRDIMMs – possibly related to settlement in Netlist vs. Texas Instruments a couple of years ago.

Inphi, IDTI and TI market share and TI exit from LRDIMM space

Inphi comments at the Stifel Nicolaus conference (Feb 10, 2011):

– Inphi and IDTI split 80% of the buffer chipset market for RDIMMs
– Texas Instruments 10%-15% of the buffer chipset market for RDIMMs

– Texas Instruments not interested in the LRDIMM space

Transcript not the highest quality – but points to the section of the audio discussing these issues:

http://www.veracast.com/stifel/tech2011/main/player.cfm?eventName=2133_inphic
Stifel Nicolaus
Technology, Communications & Internet Conference 2011
Inphi Corporation
2/10/2011; 4:25 PM
Mr. John Edmunds
Chief Financial Officer

DISCLAIMER: please refer to the original conference call or transcript – only use the following as guidance to find the relevant section

at 24 minute mark ..

(comments on the competitive landscape)

in servers .. because of the qualification cycles. . there really are some incumbent competitors .. like IDT and TXN (Texas Instruments) .. and so the 3 of us tend to split the market.

IDT and Inphi would probably share 80% of the market. TXN would be somewhere in 10-15% range.

TI is not developing an LRDIMM to our knowledge and .. uh .. their interest level seems to wax and wane at times.

We go head to head with IDT – we respect them as competitors and we think market is going to want multiple suppliers.

It’s not a market that somebody from outside can come into easily just because of the long qualification cycles and the fact these are getting deployed across a wide range of SKUs (stock keeping units ?) .. the OEMs that (unintelligible) the memory module makers don’t want to qualify multiple suppliers because they have to deploy them across a wide set of SKUs ..

IDTI and TI exit from LRDIMM space for Romley

Rich Kugele of Needham and Company comments on Netlist’s Q3 2011 conference call stated that:

– IDTI and Texas Instruments seem to have exited the LRDIMM space (for Romley)

Transcript not the highest quality – but points to the section of the audio discussing these issues:

http://viavid.net/dce.aspx?sid=00008EC4
Netlist Third Quarter, Nine-Month Results Conference Call
Thursday, November 10, 2011 at 5:00 PM ET

DISCLAIMER: please refer to the original conference call or transcript – only use the following as guidance to find the relevant section

at the 17:35 minute mark:

Rich Kugele at Needham:

Uhm .. you know I just want to talk a little bit about the LRDIMM market as it relates to Romley.

And uh .. or the next-gen from Intel.

Uhm .. can you just talk about how big that market is – I know that in recent months we’ve seen a few competitors actually exit that ..

.. space .. uh .. from TI (Texas Instruments) and IDTI .. uh .. just outright ..

.. difficult time figuring out how many units that market actually is and how competitive the solutions are or aren’t.

Compare with IDTI position last year

Last year IDTI said their LRDIMM product was superior to Inphi’s and they would expand into the LRDIMM space.

IDTI also found Inphi’s claims that they wrote the JEDEC specification for LRDIMMs “a bit amusing”.

We have sampled the world’s highest performance and in fact the world’s only JEDEC compliant LRDIMM and that is sampling today.

In fact, it’s a little bit amusing to me that our primary competitor in .. in LRDIMM .. uh .. has made a lot of noise about the fact that they actually WROTE the LRDIMM spec .. uh .. but IDT is actually the only company today that has a JEDEC compliant LRDIMM product.

So we are very confident that we are going to .. uh .. continue to expand our share in LRDIMM just as we did in DDR3 and DDR2 and AMD before that.

Transcript not the highest quality – but points to the section of the audio discussing these issues:

http://ir.idt.com/eventdetail.cfm?eventid=95812
May 9, 2011 1:30 PM PT
IDT Fourth Quarter and Fiscal Year End 2011 Financial Results

DISCLAIMER: please refer to the original conference call or transcript – only use the following as guidance to find the relevant section

at the 26:40 minute mark ..

Blain Curtis of Barclays Capital:

Hey guys. Thanks for taking my question. If you can just talk about some of the strength (?) being in servers, where you saw last quarter and then for the guidance – are you (unintelligible) with LRDIMM and how are you looking at that rollout going forward ?

Answer:

We are seeing very good strength in servers. Intel reported very good strength in servers, so unit demand is going up and our content is going up as I covered in the prepared remarks.

Right now the action is in DDR3.

LRDIMM is really a 2012 story.

We have sampled the world’s highest performance and in fact the world’s only JEDEC compliant LRDIMM and that is sampling today.

However .. uh .. if I was you I would focus the attention on DDR3 because that is really the .. uh .. the technology that is rolling out today.

I mentioned in the prepared remarks that we’ve got increasing market share in DDR3 .. uh .. a year ago on the same (conference) call I remember we were all sitting here and there were a lot of questions about whether or not IDT was going to lose share in DDR3.

Uh .. that clearly has not been the case – we have increased our market share to well above 50% in DDR3 .. uh .. now there are similar questions about whether or not we are gonna lose share in LRDIMM and it’s kinda like deja vu all over again.

at the 28:10 minute mark ..

In fact, it’s a little bit amusing to me that our primary competitor in .. in LRDIMM .. uh .. has made a lot of noise about the fact that they actually WROTE the LRDIMM spec .. uh .. but IDT is actually the only company today that has a JEDEC compliant LRDIMM product.

So we are very confident that we are going to .. uh .. continue to expand our share in LRDIMM just as we did in DDR3 and DDR2 and AMD before that.

After having developed an LRDIMM product, we now find IDTI is not competing in the LRDIMM space for Romley.

Inphi (IPHI) – legal risks

Inphi has been an aggressive opponent – they had challenged Netlist patents at the USPTO in order to stay the proceedings in Netlist vs. Inphi. However, recently the USPTO confirming ALL claims in the ‘537 and ‘274 Netlist patent reexaminations. This is a very STRONG signal of things to come. Because any claims which survive reexam cannot be challenged ever again in court.

Once Netlist vs. Inphi resumes, it will become very hard for Inphi to have a case. Separately Inphi has dropped it’s retaliatory suit against Netlist unilaterally (I have examined the court docs in Inphi vs. Netlist docs and my feeling is that proceeding would have led to invalidation of two of Inphi patents – for double-patenting – basically same patent were filed using different authors and sent to two different examiners at the USPTO – as a blatant case of patent inflation !).

See the section entitled “Inphi’s “intent to deceive” the USPTO” in the article:

https://ddr3memory.wordpress.com/2012/05/30/legal-issues-with-lrdimms-repeating-metaram-2/
LRDIMMs similarities with MetaRAM
May 30, 2012

Some background on Inphi

If you remember MetaRAM, you will recall MetaRAM had claimed in court docs that they had “destroyed” all infringing product. They conceded IP in Netlist vs. MetaRAM and then MetaRAM promptly went out of business.

Check out the seobythesea.com blog for history on NLST/Inphi/GOOG and MetaRAM:

http://www.seobythesea.com/?p=3097
Google to Upgrade its Memory? Assigned Startup MetaRAM’s Memory Chip Patents
By Bill Slawski, on November 20th, 2009

It is not surprising that Inphi has hired former MetaRAM CEO Fred Weber as “Technical Advisor” – with the result that Inphi is going down the same road.

Inphi IPO a couple of years back was based on their slam-dunk role in creation of LRDIMMs. That was the story painted at that time (Inphi is a component maker and makes other products as well).

Their co-founder CTO Gopal Raghavan and CEO Young Sohn have left and a former partner of Khosla Ventures Ford Tamer is now the CEO of Inphi. Incidentally, Khosla Ventures was also a backer of MetaRAM.

UPDATE: 06/29/2012 – institutional interest declines

With the decline of institutional ownership for Inphi, the possibility emerges (since Inphi was almost completely insider and institutional owned after their IPO) that the company could be taken private (by the VCs who supported Inphi and who also supported MetaRAM earlier in much the same circumstances – i.e. of infringement of Netlist IP – MetaRAM eventually conceded IP to Netlist and went out of business).

https://ddr3memory.wordpress.com/2012/06/29/financial-institutions-retreat-from-romley-lrdimm-story/
Financial institutions retreat from Romley LRDIMM story
June 29, 2012

LRDIMMs – performance clarity in early 2012

Since LRDIMMs were set to arrive with Romley rollout (March 2012), more information started to become available in early 2012 – partly due to the Inphi LRDIMM blog:

http://lrdimmblog.inphi.com/

LRDIMMs not only have issues of legitimacy – to add insult to injury, the LRDIMMs delivered by Inphi have not been able to deliver 3 DPC at 1333MHz on HP servers (perhaps because they use 4Gbit x 2 DDP memory packaging which may interfere with load reduction).

In addition because of the high latency issues with LRDIMMs (because of the asymmetrical lines to the centralized buffer chipset), 16GB LRDIMMs cannot compete with 16GB RDIMMs 2-rank using 4Gbit DRAM die – even at 3 DPC (since both options support 3 DPC at 1066MHz).

This leaves the 16GB “load-reduction” space completely to 16GB HyperCloud – which is now being offered through both IBM and HP – and supports 3 DPC at 1333MHz.

HP/Samsung at the IDF conference on LRDIMMs video (Inphi LRDIMM blog) have said they will not be pushing 16GB LRDIMMs (for this reason).

IDF conference on LRDIMMs video available on main webpage:

http://lrdimmblog.inphi.com/
Webcast of HP, Samsung, ANSYS, Intel and Inphi presentation at IDF 2011 for HPC applications

However they would push the 32GB LRDIMMs because they have some utility vs. 32GB RDIMMs (which will only be available in 4-rank for a few years because 8Gbit DRAM die won’t be available for a while if ever).

This is because 32GB RDIMMs (4-rank) experience greater slowdown (more below).

16GB memory module market

Currently you only need “load-reduction” at 3 DPC with Romley (as 16GB RDIMMs 2-rank are cheap and work well at 1 DPC and 2 DPC).

3 DPC usually used for virtualization/data centers (lots of memory per server for all the VMs you want to run) and does experience slowdown and needs “load reduction” in order to run at normal speed.

For more information about the non-viability of 16GB LRDIMMs:

https://ddr3memory.wordpress.com/2012/06/19/why-are-16gb-lrdimms-non-viable/
Why are 16GB LRDIMMs non-viable ?
June 19, 2012

UPDATE: 07/06/2012: non-viability of LRDIMMs

On the non-viability of LRDIMMs in general:

https://ddr3memory.wordpress.com/2012/07/05/examining-lrdimms/
Examining LRDIMMs
July 5, 2012

32GB memory module market

When 32GB RDIMMs arrive, they will be 4-rank.

32GB RDIMM that are 2-rank will not be producible until 8Gbit DRAM die appear (which can be years or never in the future – the investment required to go to 8Gbit DRAM die are huge and only Samsung supposedly maybe capable of that – from what I gathered from Netlist comments in CCs).

If you use 4-rank, the speed slowdowns are more severe – and start showing up not only at 3 DPC but also at 2 DPC (and possibly 1 DPC).

For illustrative purposes, here are some numbers from IBM for their latest Romley servers (IBM shows 4-rank delivering 1066MH at 1 DPC, and a pitiful 800MHz at 2 DPC):

http://www.redbooks.ibm.com/abstracts/tips0850.html
IBM System x3650 M4
IBM Redbooks Product Guide

Table 5. Maximum memory speeds:

RDIMM – dual-rank (2-rank) – at 1.5V
– 1 DPC at 1333MHz
– 2 DPC at 1333MHz
– 3 DPC at 1066MHz

RDIMM – quad-rank (4-rank) – at 1.5V
– 1 DPC at 1066MHz
– 2 DPC at 800MHz
– 3 DPC not supported (because 4 ranks x 3 DPC = 12 ranks which exceeds the 8 ranks per memory rank limit of current systems)

LRDIMM – at 1.5V
– 1 DPC at 1333MHz
– 2 DPC at 1333MHz
– 3 DPC at 1066MHz

HCDIMM – at 1.5V
– 1 DPC at 1333MHz at 1.5V
– 2 DPC at 1333MHz at 1.5V
– 3 DPC at 1333MHz at 1.5V

For more information about the non-viability of 32GB RDIMMs (4-rank):

https://ddr3memory.wordpress.com/2012/06/20/non-viability-of-32gb-rdimms/
Non-viability of 32GB RDIMMs
June 20, 2012

For more info on the market for load reduction and rank multiplication:

https://ddr3memory.wordpress.com/2012/06/05/market-for-hcdimmhdimmshypercloud/
Market for HCDIMMs HDIMMs
June 5, 2012

For industry estimates of attach rates for load reduction:

https://ddr3memory.wordpress.com/2012/06/06/market-opportunity-for-load-reduction/
Market opportunity for load reduction
June 6, 2012

The need for load reduction technology

So with 32GB RDIMMs (4-rank), it becomes impossible to ignore the speed slowdowns – not just at 3 DPC, but even at 2 DPC, and possibly even 1 DPC.

Which means ANY server requiring even modestly large amounts of memory using 32GB memory modules, will be affected by the high memory loading penalty at 32GB.

Thus the need for “load reduction” technology.

The need for rank multiplication technology

In addition, with 4-rank (since 32GB RDIMMs will only be available in 4-rank as explained above), you cannot populate more than 2 DPC – because of the “8 ranks per memory channel” limit for current systems.

So 32GB RDIMMs will not be usable at 3 DPC.

Thus the need for “rank multiplication” technology.

HyperCloud – latency superiority over LRDIMMs

LRDIMMs have a “5 ns latency penalty” compared to RDIMMs (from Inphi LRDIMM blog).

NLST HyperCloud have similar latency as RDIMMs (a huge advantage) and have a rather significant “4 clock latency improvement” over the LRDIMM (quote from Netlist Craig-Hallum conference – October 6, 2011):

http://www.netlist.com/investors/investors.html
Craig-Hallum 2nd Annual Alpha Select Conference
Thursday, October 6th at 10:40 am ET

http://wsw.com/webcast/ch/nlst/

Question:

at the 23:35 minute mark:

(unintelligible)

Chris Lopes:

Inphi (IPHI). Good question. How is HyperCloud different from what IPHI is offering.

IPHI is a chip company – so they build a register.

The register is then sold to a memory company.

And the memory company builds a sub-system with that.

And that’s the module they are calling an LRDIMM or Load-Reduced DIMM.

The difference is that the chip is one very large chip, whereas we have a distributed buffer architecture, so we have 9 buffers and one register.

Our register fits in the same normal footprint of a standard register, so no architectural changes are needed there.

at the 24:35 minute mark:

And our distributed buffers allow for a 4 clock latency improvement over the LRDIMM.

So the LRDIMM doubles the memory. HyperCloud doubles the memory.

LRDIMM slows down .. the bus. HyperCloud speeds up the bus.

So you get ours plugged in without any special BIOS requirement.

So it plugs into a Westmere, plugs into a Romley, operates just like a register DIMM which is a standard memory interface that everyone of the server OEMs is using.

The LRDIMM requires a special BIOS, special software firmware from the processor company to interface to it.

And it’s slower.

Does that answer your question ?

As a result of this latency issue, 32GB LRDIMMs thus underperform the 32GB HyperCloud (CMTL benchmarks for LRDIMM vs. HyperCloud):

http://www.netlist.com/products/hypercloud/whitepapers/hcdimm_vs_lrdimm_whitepaper_march_2012.pdf
HyperCloud HCDIMM Outperforms LRDIMM in  ‘Big Data’ & ‘Big Memory’ Applications 
Whitepaper 
March 2012  

In addition, 32GB LRDIMMs cannot deliver 3 DPC at 1333MHz (perhaps because they use 4Gbit x 2 dual-die packaging memory (DDP) which may hinder load reduction efforts).

As a results, not only is the HyperCloud memory lower latency (i.e. better), but also delivers 1333MHz at 3 DPC (where LRDIMMs cannot).

Legal risks for LRDIMM

Plus, HyperCloud is not saddled with the risk of recall that LRDIMMs are (legal issues above).

This means these markets will be owned by HyperCloud memory modules:

– 3 DPC market at 16GB
– 3 DPC and 2 DPC market at 32GB

On the risk factors for LRDIMM:

https://ddr3memory.wordpress.com/2012/06/05/lrdimms-future-and-end-user-risk-factors/
LRDIMMs future and end-user risk factors
June 5, 2012

https://ddr3memory.wordpress.com/2012/06/15/why-are-lrdimms-single-sourced-by-inphi/
Why are LRDIMMs single-sourced by Inphi ?
June 15, 2012

On DDR4 borrowing from LRDIMM use of Netlist IP in “load reduction” and “rank multiplication”:

https://ddr3memory.wordpress.com/2012/06/08/ddr4-borrows-from-lrdimm-use-of-load-reduction/
DDR4 borrows from LRDIMM use of load reduction
June 8, 2012

https://ddr3memory.wordpress.com/2012/06/07/jedec-fiddles-with-ddr4-while-lrdimm-burns/
JEDEC fiddles with DDR4 while LRDIMM burns
June 7, 2012

First proprietary memory to gain wide-spread industry adoption

Netlist has said in recent conference call that they are going to be the first “proprietary” (although HyperCloud is plug and play and not proprietary in terms of use but in terms of IP) memory that will be adopted by the industry.

LRDIMMs in contrast require the BIOS to have explicit support for LRDIMMs.

From NLST’s Q1 2012 CC:

http://seekingalpha.com/article/592411-netlist-s-ceo-discusses-q1-2012-results-earnings-call-transcript
Netlist’s CEO Discusses Q1 2012 Results – Earnings Call Transcript
May 15, 2012

With today’s joint announcement with HP, HyperCloud is now available for sale on HP’s top-selling newly-released Gen8 servers. HyperCloud has also been designated as an HP Smart memory. This coupled with availability on IBM System x servers makes HyperCloud the first ever proprietary memory technology to be widely adopted by the mainstream server market. HP and IBM account for a majority share of the server market and they combine for even a larger share of the world’s high-density server memory consumption.

Advertisements

11 Comments

Filed under Uncategorized

11 responses to “LRDIMM buffer chipset makers

  1. Pingback: LRDIMMs future and end-user risk factors | ddr3memory

  2. Pingback: JEDEC fiddles with DDR4 while LRDIMM burns | ddr3memory

  3. Pingback: Patent trolls at the JEDEC gate ? | ddr3memory

  4. Pingback: Why are LRDIMMs single-sourced by Inphi ? | ddr3memory

  5. Pingback: Market opportunity for load reduction | ddr3memory

  6. Pingback: Non-viability of 32GB RDIMMs | ddr3memory

  7. Pingback: Financial institutions retreat from Romley LRDIMM story | ddr3memory

  8. Pingback: Examining Netlist | ddr3memory

  9. Pingback: Examining LRDIMMs | ddr3memory

  10. Pingback: A second-source for HyperCloud ? | ddr3memory

  11. Pingback: Inphi reports Q2 2012 results | ddr3memory

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s