M11x clock generator ICS9LPRS387BKLF

FYI, for those interested in these type of things the clock generator in the M11x is a ICS9LPRS387BKLF.

The reason I know this is that when my M11x arrived the right speaker didn't work. After a little surgery the speaker was repaired. While I was in there I took a few pics and found the clock gen. I took a quick swipe with SetFSB, but it doesn't have support for that particular chip. I guess if someone is motivated enough they could give it a try or work with the creator of the software to implement support.

 

That's great Duffy, thank you for going to that trouble. It's unfortunate that SetFSB doesn't support that particular PLL. I also notice after a quick Google that the clock generator is the same in the M11x as the Dell Mini 10v. I know some of the Mini guys like OC'ing their netbooks so I'll have a look through the Dell Mini forums tomorrow. I may find some helpful information there.

 

Unfortunately the creator of SetFSB, Abo, No longer supports the product and will not update it.

+1 for finding it. Any chance you got a pic?

 

I received a response from CPUCool in relation to addding support for the M11x's PLL:

I have e-mailed IDT and hope to hear from them soon. I'll keep the thread updated with my endeavours.

 

RPG, I'd rep you, but I've been to generous over the last 24 hours. How ever, wanted to say thanks for your efforts. Lets keep our fingers crossed.

Only thing I could find online is this PDF
http://www.idt.com/products/getDoc.cfm?docID=18639419

 

No problem DR650SE. Thanks for having a look around, there are a lot of datasheets for different clock generators on the IDT website but not for this one. I just hope they're prepared to give it to me!

 

Yeah I did some digging for info when I first opened my machine last week. Strange how you can't really find a data sheet for the device. That is so weird. Who makes something like that and doesn't publish any data about it?

 

Here you go. I also included a picture of the inside of the right speaker. The silver disk is the permanent magnet which in my case had come loose and was rattling inside the enclosure. I had to crack open the sealed speaker, reglue the magnet, and glue the case shut. Works fine. No telling how much money I saved Dell.

 

Is the SetFSB source code available? I'm sure there's plenty of saavy programmers even here at NBR that might be able to incorporate the proper parameters?

 

I think it is but according to SetFSB's webpage, the code is entirely in Japanese! I also decided to e-mail Abo about it just to see what information he could offer on the subject, and it doesn't look good:

I will wait and see what IDT say and send it on to CPUCool, but in lieu of the above information, it might be all in vain. ​

 

At least if you get the data sheet, it might pit a pin mod option on the table.

 

Can you please select PLL diagnosis in setfsb and then click on the diagnostics tab.

If you upload a screenshot of that, I may be able to help.


By the way, you could overclock in linux if you get nowhere in clockgen.
Just PM nando4.

 

I've never pin-modded a PLL before, so maybe you can try it out first. The modification is irreversible though isn't it? i.e. It will constantly run at the higher bus speed?

 

I don't have a M11x (yet), but I do have access to a Mini 10v with the same PLL. Here is a screenshot:

EDIT: Removed due to duffyanneal's screenshot posted below (directly from M11x).

 

Yea, Pin mod is pretty much irreversible. I suppose if you were determined you could reverse it. It would look similiar to this. I did this up when researching a pin mod for my Inspiron 1720. It involves disabling the TME and connecting it to the VSSA pin. I put that diagram together after looking at the Data Sheet.

In my opinion it's not worth it simply because it voids the warranty.

The reason I wouldn't do it (yet) to my 1720 is the 1720 is still running strong and never gives me problems. (other then shocking me to death continuously due to worn paint) I still use it every day as my main computer.


Edit: Links for PLL pin mod example. Should take it from 200MHz to 266MHz

Picture of PLL (1720), With Diagram

Pin Config for 1720

I hope there is another route for this

 

And you can OC?

 

Thanks for the pic. I'd say you'd need some steady hands and some good eyesight to do it properly! That would be a concern for me also, with such a new machine, a warranty could prove invaluable.

Nope, tried pretty much every PLL on every program I could get my hands on (SetFSB, ClockGen, CPUCool) to no avail.

 

In the M11X you can take it from 1.3GHz to 1.6GHz Through a BIOS option(on/off). Would this indicate the potential for futher OC? Or is it still indicative of a locked TME?

 

Here you go.

 

Thanks Duffy. It's interesting to note the clock speed reported from SetFSB is 1733.5MHz. I presume CPU-Z reports your CPU at 1600MHz?

 

Yep (10 char)

 

If TME is bit 6 of byte 9 then both notebooks have TME enabled and need to pin mod to overclock.

But it's impossible to know for sure without the datasheet, so good luck, I hope somebody can get their hands on it.

 

Any more news on this? If we somehow can get the m11x beyond 1.73 GHz, this machine would probably hold up well against the next generation of games. It's feels a bit frustrating to be stuck at 1.73 GHz. Would a hacked bios allowing more overclock be the only option?

 

Unfortunately, IDT (the manufacturer of the clock generator in the M11x) haven't got back to me. I've just sent another e-mail now but this time directly from my e-mail address as opposed to their online enquiry form. I've marked it as important so hopefully I'll hear from them soon.

If we don't get the datasheet, then unfortunately a hacked BIOS is really the only other option I can think of.

 

Or maybe somebody could take a risk and try to pin mod without the datasheet.

 

hahahahahaha

 

Was wondering about this as well. I have no clue what bios code looks like, but I am curious if the overclock from 1.3 to 1.73 GHz was planned and supported by Alienware all along, or if there is room for more, but they left it out because of instability problems? Would be nice to push this chip towards 2 GHz...

I found this interesting thread:

http://forum.notebookreview.com/showthread.php?t=430230

At least there is some kind of successor being worked on to SetFSB.

 

Someone needs to figure this out.... I want to OC this little thing.

 

Bad news on the datasheet guys:

 

**Angry fist pump** Kyle! Grrr... Kyle?

 

LOL.

Oh come on Kyle...just share!!

 

C'mon Kyle!

width='480' height="385"><param name="movie" value="http://www.youtube.com/v/J1maHfe-Row&hl=en_US&fs=1&"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/J1maHfe-Row&hl=en_US&fs=1&" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width='480' height="385"></embed></object>

 

lol, if only Kyle knew

 

If you look at similar datasheets, you can see that the TME pin is usually the same.

So somebody just has to try it.

I don't have a datasheet for the PLL in my GT627 but I gave it a go. Though my mod failed and the notebook didn't post, so yeah...

At least after I reversed the mod, the notebook worked fine.

Anyway, if your PLL is a 64pin MLF package, TME is probably pin 11.

Just use the datasheet for ICS9LPRS365 as a reference.

EDIT: Sorry, from the pic it looks like the PLL has 72 pins.
Let me try to find a similar PLL.

Hmm, that's probably why my pin mod failed, because I didn't count the number of pins, I assumed 64 but there is 72.

Well at least I'm in the same boat as you guys.

EDIT#2: I can't find any datasheet for a 72-pin PLL, seems like the same problem as they had in this thread:
http://forum.notebookreview.com/showthread.php?p=5988306

I'm going to also send an email. Maybe other people could do the same and we could get have some luck.

EDIT#3: Email sent, if they give me the same email that rpg-XPS got, I will just ask for any 72-pin PLL datasheet.

 

So this clock generator is only for sale to Alienware? I thought I've seen it online for sale, just with no datasheet.

 

Man if only there were a software OC that would work. This CPU is really holding back my bench scores.

 

Someone keep at this, we need some cpu overclocking... We should take up a collection and pay one of those guys, the clockgen author or the guy who made cpufsb..

 

IDT wont reply to me. The guy was out of office.
I just got some auto reply.

I'll try to send another email from another address.

 

Apparently if you email http://www.partnoic.com/icku-M11X.html They give you the Data sheet. Might be the wrong type of Data sheet..Just throwing it out there.

 

I've read tha m11x is a 1.3/200Mhz FSB->1.73/266Mhz FSB overclock via bios. The bios could be doing this a number of ways:

1. altering the BSEL pins of the CPU
2. changing the PLL BSEL pins
3. changing the PLL's software programmable registers

I would anticipate it would be 1 or 2.

In any case, can someone post the setfsb Diagnosis window when running 1.3Ghz and when running 1.73Ghz? Specifically we want to see register 13 and 14 (OC/OD). That would identify if the PLL is being programmed for the higher speed. Also some of the readback configuration registers can confirm if the BSEL pins are being changed.

BSEL would also mean the FSB's correct RAM SPDTable is being used. Something you can confirm by checking Everest RAM timings.

 

Someone do as nando4 says. What would be our options if it were 1) or 2), nando4?

 

heading out the door for work, but if no one's posted the screenshots by the time I get back this afternoon, will get them up for nando... and thanks for offering to help!

 

If you find that setfsb doesn't display the additional byte 13/14 (OC/OD), since you'd be using some other random PLL, Linux users can use i2cdetect/i2dump as shown here to grab PLL registers for the 1.3Ghz and 1.73Ghz states.

 

It looks like it's not displaying the additional byte.

 

Choose a PLL like ics9lprs365 and see if that provides the additional data bytes. Specifically 13 and 14 (OD/OW). In any case, I can see what the bios is doing by referring back to the ics9lprs355 datasheet with likely the same configuration registers.

Byte 0, bit 7/6/5 tells us the FSLC/FSLB/FSLA settings on the PLL.

@1.73Ghz, Byte0=01=0/0/0=266Mhz FSB
@1.3Ghz, Byte0=41=1000001=0/1/0=200Mhz FSB

So what they have is some bios controlled gate connecting to the FSLB pins on the PLL itself. FSLC pin that sets 333Mhz operation is likely NOT gated since the chipset only support 200 and 266Mhz operation so would be tied to GND (0).

BSEL mod to CULV CPU not locking to lowest multiplier

The FSLx pins are tied to the BSEL pins on the CPU. Now what is unique here is that normal non-CULV CPUs usually lock you into the lowest multiplier when applying a BSEL mod. Here it is not. Does this apply to all CULV CPUs? Perhaps Intel never bothered to do this since modders are unlikely to try BSEL mods to BGA soldered CPUs.

Software overclocking instead?

You'd want software overclocking to give you finer 1Mhz overclock granularity. Byte 9=E5, so bit 6=1, which usually means it's TME locked, ie: no software overclocking. To get around that would require (i) disabling TME with a pinmod and (ii) sending the right data to the PLL to overclock it. The latter *used* to be done by setfsb, but the author is no longer developing it. So could consider using the grub2 bootloader overclock (see sig) instead.

To progress down this path would require the PLL's datasheet so you know what pin to pinmod.

One other random attempt. *Could* try the TME workaround applied here when running @1.3Ghz (200Mhz FSB), using the FSB tables here to see if you can alter the FSB. Then would just need to build a table of correct values to higher FSB overclocks.

 

Or try ICS9LPRS113AKLF since that is a 72-pin PLL.

 

OK, cool, here we go.

I tried that also but unfortunately it said "PLL Byte Error".

 

Well I was going to do this as well, but I opened SetFSB and then clicked "Get FSB" just for kicks. It' locks up my M11x to the point that even holding the power button won't work. I have to unscrew the bacl plate and unplugh the battery from the mother board. Did it a couple times . RPG have you had this problem?

 

You can see that byte 0 changed. So the FSLB bit is changed to overclock.

The PLL's software programmable registers are not being used to OC.
Bad news if you ask me.

 

I think the SU7300 could handle 2.13GHz sufficiently cooled (even if it didn't I've used programs before that were able to lower the CPU multiplier, which should work with the SU7300). Who would we look for to reverse engineer the BIOS?

I'll have a look at the grub2 bootloader and see what it can do.

Thanks for all your help.

I haven't actually, but I only clicked "Get FSB" on the clock generators that nando and moral hazard suggested (and PLL diagnosis). Did you choose a different PLL that might have caused it to crash?

I had a feeling it wasn't going to be easy.