Lian Li with a PCIe Sata II adapter model IB-01

The card seems new. At least its new to me. If it would work in unRaid it gives us 5 sata and 1 esata ports in a 1x pcie slot.

http://www.lian-li.com/v2/en/product/product06.php?pr_index=345&cl_index=2&sc_index=10&ss_index=117

I contacted their sale department and they pointed my to the JMicron site.

http://www.jmicron.com/Index.html

I could not relay the function to a particular chip so I've asked for confirmation from Lian Li. I suspect they are using one of JMicron's Port Multiplier chips to acheive that port density.

A quick google found others reporting success with JMicron in general but those seems to be linked to on board host controllers. JMicron does claim host controller support in the 2.16.18.2 kernel.

Anyone here have success or failures with JMicron?

The JMB363 chipset works on unRAID.  It is on the ABIT AB9 PRO.  PMP support worked when I tested it also.

According to the silk screen it is a jm362 and jm393, so as you suspected a two port sata II PCI-e chip with a port multiplier on the internal port to give five internal sata ports.

Did Lian Li give any idea on pricing?

boy, u got great eyes..

I'm waiting on Lian Li to let me know where I can buy. I'll post up whatever they provide.

Thanks,

db

According to the silk screen it is a jm362 and jm393, so as you suspected a two port sata II PCI-e chip with a port multiplier on the internal port to give five internal sata ports.

 

Did Lian Li give any idea on pricing?

 

How slick. Good eyes too!!!.

boy, u got great eyes..

 

I'm waiting on Lian Li to let me know where I can buy. I'll post up whatever they provide.

 

Thanks,

db

This has merit worth testing... do let us know!

FWIW, two of the Addonics port multiplers claim to use the JMB393 Chipset.

http://www.addonics.com/products/host_controller/tutorial_pm.asp

boy, u got great eyes..

Not too bad yourself, you found a new take on a sata controller.

http://www.addonics.com/products/host_controller/tutorial_pm.asp

Never knew you could daisy chain PMs, cheers Weebo. Now if only the Chinese would start knocking these PMs out for $20 a pop like they do 2 port sata cards.

It took a few emails but here are the responses from the Lian Li Sales Specialist answering my emails. I was not able to a get a direct answer on cost but I'm still trying.

Kaygee nailed the chips!

"The series number of the JMicro chipset is JMB393." and "It’s JMB 362."

"IB-01 is a latest product for last month and just delivered this month.

All our distributors will not receive the item until the mid of November.

I think you may need to wait for about half month owing to the delivery time.

You may also call our distributor for the exact availability date."

They did provide a list of outlets but its the usual suspects. I felt weird posing up the list. I didn't want the thread to look like a sales promotion. Finding the card once availble won't be tough. Lian Li is well promoted.

So now its a waiting game.

cheers db

This looks interesting.  When this card comes out and I need to expand my array past the built-in ports on my Abit Ab9 Pro I will probably end up grabbing one of these to put in my PCI-e x1 slot.  It is one of the higher density SATA cards I seen that fit into the x1 slots.

Thanks for doing all the detective work!!

Keep in mind, we don't know how well this PM will be supported in linux.

We know the silicon image PM's work ok.

http://www.jmicron.com/JMB362.html

http://www.jmicron.com/Product_JMB393.htm

Keep in mind, we don't know how well this PM will be supported in linux.

I'll ask how well its supported in Linux and if anything is needed since their site doesn't specify. It won't guarantee compatibility but it will give us a starting point. 

Found this too

RobJ mentions the non SIL PM Boards being command based thus reducing efficiency.

http://lime-technology.com/forum/index.php?topic=1475.15

From my tests, and even documentation  on the Silicon Image website, the throughput per disk is around 60MB/s.

I've yet to test this with a pair of 1.5TB 7200RPM 32MB cache seagates, but with other drives this is what my tests showed.

I saw this average with both command and FIS based PM's. The Silicon Image was slightly faster, but not by much with my hardware.

Hopefully this new Lian Li card will be very cost effective.

I may have to hold off on building my first box until we know more - busy month anyway.  With my old asus p5b-e I could get 21 drive support with just the mb and 3 of these 1x boards (assuming I don't need a 1x for a new nic.)  Even 16 drives easily would be a nice base to build on.

JMicron claim their PM is using FIS switching so should be comparable with SIL PM. Hardware PM solution so drivers should not be an issue (famous last words...).

An interesting read.

http://en.wikipedia.org/wiki/Port_multiplier

http://www.serialata.org/technology/port_multipliers.asp

Makes sense, I used to get around 65MB/s on my PMs under windows. 85MB/s per port without a PM connected.

I'm not sure I agree about the cost effective part. In the UK a 5 device PM is about the same cost as a 4 port PCI-E card like a 1430SA. I guess they are cost effective compared to high port density RAID cards but they have super fast io processors caching mechanism and onboard memory. Hardly apples to apples. 

Where they are usefull is adding ports to existing systems that dont have more available bus space. 

Lian Li was having a tough time with my questions so I went to the JMicro

"Thanks for your mail.  We are testing on the Slackware 12.1.  Let you know what we see here ASAP.  We have JMB36x + JMB393 Linux Driver base on the 2.6.30.  The Slackware uses the 2.6.24.5 kernel.  We will update information to you later."

If I recall correctly the beta7 is using 2.6.30.x . Is this base driver something we would need to add ourselves?

db

If I recall correctly the beta7 is using 2.6.30.x . Is this base driver something we would need to add ourselves?

Correct. 

If I recall correctly the beta7 is using 2.6.30.x . Is this base driver something we would need to add ourselves?

 

 

Correct. 

I thought jmicron was already supported, so there would not be a need to add anything?

If I recall correctly the beta7 is using 2.6.30.x . Is this base driver something we would need to add ourselves?

 

 

Correct. 

 

I thought jmicron was already supported, so there would not be a need to add anything?

I was expecting the host controller to be the go between the PM and the OS as well. Once the host was supported which it seems like it might be I would think nothing would need to be added. Based on JMicron repsonse it seems that might not be the case.

I will continue to post their communications. Hopefully I can get my hands on a IB-01

db

Indeed PM should be h/w independant, assuming the port is pm capable.

I guess the question is really whether firstly does current unraid driver support jm362(it should support jm36x). Next question, is it pm capable which it should be since 2.6.24 with sil pms (ty39 Tejun Heo san). Last question would be, does jm pm require driver updates (suspect this answer is 2.6.30 from jm response).

Looks rather promising...

Indeed PM should be h/w independant, assuming the port is pm capable.

It's not totally, Please do not assume so. It is maturing, but PMP support still has issues on some Marvell chipsets.

Earlier kernel versions with unRAID had some issues on my JMB chipset on the ABIT AB9 PRO.

I remember someone here bought the addonics PMP's. They seemed to have the JMB chipset. He said they were slow.

I'm sure it is improved now. But test the water before diving in.

 

I remember someone here bought the addonics PMP's. They seemed to have the JMB chipset. He said they were slow.

 

I'm sure it is improved now. But test the water before diving in.

I will ask JMicron to confirm speed as well.

Thank you all for the contributions.

(off topic alert - move on)

At first I read Kaygee's great eyes so I though checking it myself.

http://lian-li.com/v2/tw/product/upload/image/ib-01/ib-01-01.jpg

...at first my brain told me to check the ICs themselves (although Kaygee very clearly says "on the silk screen")... impossible. At this point a "wow" was starting to grow (in fact doubt too as it is really impossible to read the chip text on this resolution photo).

Then I realized that the IC numbers were clearly mentioned just over the PCIe port. Clearly as in "size of this forum text" clear.

Great that he noticed the text yes, but "great eyes"?

Latest email from JM

"Yes, our JMB36x + JMB393 (as model IB-01) are work with Slackware 12.1.

Following are our test platform environment.

M/B: GA-EP43-DS3R

CPU: Intel Core 2 Duo 2.8GHz

Chipset: ICH10R"

cheers, db