RAM or SPAM? What seems to be the problem here?

Hello, I'm not an idiot, but I play one on a PC.

I just tried out Kingston's HyperX 800MHz CAS 4-4-4-12 RAM on a laptop with a T7500 Core2Duo, which is 2.2GHz, and has this weird bus rate that I can never figure out -- some say 667, other say 800.

It's a ThinkPad T61p. The reason for checking it out is that Kingston claims to have an SPD preset for this memory of CAS 3-4-4-10 at 667MHz. (This is the 2GB set, so unlike the 4GB set, it won't start to drop all the way to 533MHz CAS 2, which is probably a good thing.)

(For those of you who are trying to make sense of this gibberish, join the club.)

Anyway, good to go, I thought.

Didn't work.

At all.



The OS would start the boot, and immediately grind to a halt, with the 0xc0000221 error code. Tried reseating the RAM and all that, and it just didn't want to giddyup.

As soon as I tried it with one 5-5-5-18 1GB RAM stick alongside the HyperX stick, it booted.

I have to say that I cannot begin to make sense of base clocks and FSBs and voltages and all that. I own a laptop, dad gummit! So, anyone care to take a stab at walking me through what might be going on here? And just be ready -- I might actually ask a question if the explanation isn't clear enough yet.

Thanks for the help!

 

Have you run a memtest on that ram?
http://www.memtest.org/

That would be my first course of action, that memory should have no problems running at 3-4-4-10 at 667mhz. Also, what voltage does that memory draw? Should be 1.8V, if higher, that could be the problem.

PS - The bus of the T7500 is 800MHz, just to clarify that for you.

 

Thanks for the Help, Hep!

I just ran across a comment that says the T61p "doesn't support 800 MHz." No idea what that means.

The voltage is 1.8V, so we should be OK there.

I'll run it now. I've never had to -- always expected to see one bad RAM stick, not two, so it's not my first guess, but you might be right.

Look for a reply in a while.

 

Might not be your problem but it could be that the SPD tables in the ram aren't very detailed, or your notebook doesn't read all the entries, and it tries to pick a speed and timings set that the ram isn't capable of supporting, when you put the slower ram in that has a good SPD that it likes it will use that data to pick its timings (it will always go to the slowest stick installed for the settings). You can check the SPD settings with CPUZ if you boot it like you did with one of each, there's a tab where you can browse the installed memory sticks and read their SPD tables. If the table is no good there are ways to change the SPD on the ram, but it's somewhat risky.

 

Big Mike, thanks too. I'll check out CPU-Z next, but if you don't mind my asking, you're going to have to clarify a bit what I'd be changing on the SPD tables -- I know it's risky, but just so I understand what's up here, right?

Incidentally, CPU-Z will give you separate readings for each stick, right? Otherwise, as I said, I can only boot with 1 stick in, complemented by the old RAM. If it doesn't tell you the readings separately, I'm either not going to be able to boot, or I'm going to get some dirty readings.

* * *

Well, just ran MemTest, and it really didn't like these sticks!

The first stick just flat-out crashed the whole system on Test #3. Screen went black, and so on. The second stick stopped on Test #2

I'm not sure how to interpret the interrupt report for the second stick, cause it looks like a bunch of random crap, but here it is:

Incidentally, is there a print function?

Type: Divide
PC: 000111e2
CS: 00000010
Eflag: 00210046
Code: 00000000
eax:00000064
ebx: 0001db40
ecx: 00000000
edx: 00000000
edi: 0001ee80
esi: 00000064
ebp: 00000064
esp: 0001eb74

It's got the stack to the right of it. Too much to type, but it's just a bunch of random address crap.

Along the bottom, it's another random set of crap:
CS: EIP: f7 7f 30 89 c6 83 ee 0c 6a 00

I'll run CPU-Z, but the basic data from the Memtest screen, by the way, says:
Intel Core 2 2.394 MHz
Chipset: Intel GM965/GL960
FSB: 200 MHz DDR2
RAM: 332 MHz (DDR 665) / CAS: 3-4-4-10 / Single Channel

Can you guys explain the CPU number here, by the way? I don't see how you get a clean multiplier up to 2.2 GHz with that figure.

 

Okay, you're right that it's an infinitely small chance that both sticks are bad, so I don't think that's the issue. Sometimes when memtest acts real dodgy like this, downloading an older version fixes the problem. It's weird, I keep around two Memtest CDs because some computers just won't work with one but are fine with the other.

Truthfully, I do not know what that means beyond the fact that it is failing tests. Never cared to learn more, as a bad stick of RAM is a bad stick of RAM, no fixing one issue.

You get 2.2GHz by multiplying 200MHz by the max multiplier of the T7500 (11.0) to get 2200MHz, or 2.2GHz. The FSB is quad pumped on these chips though, to give you an actual throughput of 800MHz. Kind of like how your memory is really running at 333MHz, but it's DDR2 (Double Data Rate, Revision 2) so it's actually double pumping your memory. I don't understand where the 2.394MHz is coming from there - surely a misread or mistype, or both. (T7700 would be 2.4GHz, though CPU-Z used to read my T7500 as a T7700)

-When you boot with one of the old sticks, and one new, you have a stable system - correct?
-If yes to above, does the system pass a memtest with an old stick and a new stick? If it fails, make note of the ranges that fail (in megabytes). You can then calculate which stick had the failure on it.
-What you heard about the T61P not supporting PC2-6400 is correct. However, it should not be a problem, as it will simply downclock the memory to PC2-5300 which the system is designed to take.

 

Hep, this is great. You're being incredibly patient to answer these questions. Of course, it's not like you pick this up in your formal education, so how would you know this stuff it to begin with except by asking? So I dunno ... maybe it's just getting harder to sift out clear answers on the Google, as the internets-tubes spawns more and more Ted-Stevens-like mangled explanations.

http://www.youtube.com/watch?v=f99PcP0aFNE

One question, as I head off to test this stuff again (on your suggestion) with an older Memtest, and a mix of old and new RAM sticks ... You say,

Not sure I remember seeing anything about ranges per se. Maybe I'll notice the second time through, but if not, can you explain what you mean?

* * *

Big Mike,

I looked in GPU-Z, and it looks like, at least for the stock RAM that works, the SPD tab is all greyed out. Am I likely to get more detailed info from the other sticks? What am I missing?

Thanks both of you.

Oh yeah, and Hep, I have no idea how that decimal slipped into the CPU figure. Still, I'm surprised to see the T7500 running at 2.4 GHz rather than the stated 2.2 GHz. Not that it matters. Everyone else with an Intel is probably unofficially overclocked by the same amount. (What's Intel doing, anyway? Subsidizing the competition?)

 

Haha, Ted Stevens is such a fool. Thankfully most people no longer take him seriously after all the negative media attention he has gotten.

Memtest86 will have a top section with all the info on the test, and on the bottom, it will have all failures. The deliminator between the two will be a report of how many passes you've had as well as failed ranges (IIRC, I do this like 10 times a day it should be a better picture in my head). Either way, on the far left you're going to have it mentioning where the failure was in the format ###M. If you put in 2x128MB of memory, test, and it fails on 165MB, then you know that the second DIMM is bad. Just an example, this saves me time when working on many computers and I can begin a test and walk away. Of course, I always test with the supposedly failed stick removed to confirm. I digress.

 

The part where I had to rewind just to make sure he actually said what I heard him say is when he blurts out: "My staff sent me an internet this morning." I'll cut him some polyester slacks for being an octagenerian or whatever, but he was seriously, seriously bumbling through that speech. Is it that one where someone interrupts him and says, "Oh, I'm sorry Mr. Chairman, I thought you were done."



Alright, well this is getting interestinger and interestinger.

Right off the bat, there was the first bright spot of this entire misadventure: Even with a stick of 5-5-5-15 in alongside the faster stuff, Memtest recorded it running at 4-5-5-15.

Not toooo shabbay.

The (verbose) driver loading screen at boot went by noticeably faster, by the way.

I backed up and did version 2.01 of Memtest instead of 2.10. Individually, both sticks crashed the program, but they ... umm ... took a little longer, I guess. (Yippie skippy.)

Now, to get back to your suggestion Hep -- and then I have to get some shut eye -- I had time to run at least one of the Kingston sticks with a stick of stock RAM alongside, and, as you might have wondered by now, there were ZERO errors putting the RAM through its paces. Good enough for me for right now. Sounds like this is tripping up over something else. I can always come back and run Memtest again through a couple of passes in a row if nothing else is explaining it.

Big Mike -- you still out there? Just to keep it rolling on the SPD front, Mike might be on the right track with the SPD flash, or the way the BIOS is reading it, since the RAM seems to be checking out OK by Memtest. But again, CPU-Z is telling me nothing. I checked the SPD panel and it's all greyed out. (Is this telling me that there's a pay version?) You guys ever look at SPD-Tool? Seems like the info should be visible there, but this is the first time I'm wading into SPD territory, so if you haven't taken a look at its features, it would be a real help if you could take a peek and tell me what readings to look for. Here's the latest version (I think):
http://www.techpowerup.com/spdtool/

Hasta mañana, düdz!

 

I agree, the problem is likely being caused by incorrect timings. I do not know the specifics on how do to this on your laptop, but I will follow up on the thread and provide help where I can.
The only place I've ever seen SPD settings be changed reliably is the BIOS.

 

I was kind of out of it last night so i didn't want to give you a bad answer. So you're saying that CPU-Z won't show you anything in the SPD tab when you get it to boot with a stock stick and one of the kingstons, it fails memtest with the kingston ram in, but it runs ok when you have one of the original 1gb sticks in with one of the kingstons? Do I have that all correct? I have a program specifically written for reading and writing ram SPD information, but I'm having a hard time finding it, I'll search around this afternoon and see if I can locate it.

When I spoke about the SPD the problem I've run across in a few PCs is that generally the ram stick has 2-3 timing tables, say 200mhz ddr, 266mhz ddr and 333 mhz ddr for a generic PC5300 stick. At each memory speed there is a table of cas/ras/etc values telling the computer what timings to use. If for some reason your ram doesn't have a particular timing at the exact speed its running the ram, or doesn't properly read the timings for that speed, the BIOS will sometimes just guess based on the timings in the stick. If the low timing has pretty agressive settings sometimes it will interpolate too low of a timing set for the inbetween speed and cause issues. This isn't as prevalent an issue anymore and most equipment tends to use looser timings to ensure compatibility, but it sounds like it could be your issue if you're getting errors with brand new memory that is rated for a faster speed than the computer will allow. With that program I found you can back up all the SPD tables and write a new one with just a single entry of your target speed (pc5300 3-4-4-10 I think is what you said your kingstons are rated for at 667, so you'd just write that as the sole SPD forcing the computer to hopefully use it, or write several entries all with the same timings, you just have to make sure to change the settings back if you ever use it in another computer)

 

Thanks for replying, Big Mike. And thanks too, Hep. Sounds like we're all on the same page. Still, I feel like I'm about to read that legal clause that MS posts everytime they mention their Regedit tool.

Do you guys feel confident that we're on the right track, or can you think of a hardware forum where the Grey Wizards hang out? Maybe I should consult with the oracles first before continuing the conversation here. I think I'm in good hands. But that doesn't mean that this is something you run into every day. If this is something you guys don't see too often, then I'd guess you guys also know where you'd go to get your own answers.

Couple of things.

First of all, before we go the SPD route, help me understand this: In terms of the electronics, in what sense are timings like 3-4-4-10 "aggressive"? How does it throw things off? And what exactly is it with timings of this kind that a laptop might not handle but that a desktop would take in stride?

I read once that voltage comes into play here, but I don't have a clue how.

Second, when it was crashing Memtest, Memtest was reporting that it was running at 3-4-4-10. Are you saying, then, that what's not working here is that the SPD table needs to supply info in addition to the 3-4-4-10 that it doesn't have? Or are you saying that the 3-4-4-10 needs adjustment?

Big Mike, I mentioned a tool called SPD-Tool above that I think does pretty much exactly what you're referring to. Any chance that's the one, or is there a professional tool?

Thanks!

 

If your laptop is set to be running the memory at a lower latency than the memory is designed for, it's going to cause instability. It's on the same lines of overclocking, you're forcing the memory to do more work than it's capable of doing. It tries to keep up, but eventually crashes.

 

Basically the numbers 3-4-4-10 are the number of clock cycles that the computer has to wait before performing specific tasks in memory. If the wait time is too short the memory can't do the next thing its asked to do and you get an error. One thing that wasn't clear, when memtest threw all the errors was that with just the kingston or was that still with the old memory and one stick of kingston? Also do you have the part # of the kingston ram? That will help. I think I actually used Thaiphoon, same idea, different writer. Giving us a print screen or the like of what either one of those programs pulls off the sticks as the existing SPD would also help.

 

Santa Rosa based laptops while having 800mhz FSB cannot run 800mhz memory they automatically downclock to 667mhz.

 

Nobody is trying to run ram at 800mhz, his ram won't run at 667 which it supposedly has an SPD setting for and should run at fine.

 

With more time on NBR you will realize the sad fact that half the people providing "help" have no idea what they are talking about and/or have not even read the original question.

 

Memtest said 3-4-4-10 when it was running by itself -- that's when it was crashing all the time, and 4-5-5-15 with the stock RAM alongside. So, given that the stock RAM is timed at 5-5-5-15, that means Memtest was giving us an actual measure of the RAM performance, rather than just telling us that in theory this is what the memory is set to run at, correct?

See, that's what I'm not getting. The RAM seems to be running at its intended preset, but still crashing.

Looks like Thaiphoon is $25 and hasn't been updated since Vista came out.
http://www.softforall.com/Utilities/System/Thaiphoon_Burner09140600.htm

Well, using SPD-Tool, I took three readings -- not two! There are 3 memory units installed, according to the app, and it's hard to tell one from another. What I can tell you is that the one that says "Thinkpad" and the one that says "6459CTO" are both the original modules. I'm assuming that one is the RAM, and the other stock piece is going to be Intel Robson NAND ("Turbo Memory").

I think that means that the HyperX is the "module 5" file, but you guys can probably tell better than I can.

Honestly, none of the three even show a CAS Latency -- just have no idea what I'm looking at. There is a long discussion thread for SPD-Tool at the Tech PowerUp forum where this app is on the server. Maybe that is where I'm going to have to spend an hour or two.

Let me know what you think.

The files are attached. Just grab a copy of SPD Tool and you should be able to open them.

 

Thaiphoon is free for 15 days
http://cbid.at.tut.by/html/download.html
Only thing that sucks is my virus scanner hates the file, I've used plenty of programs from these guys before, its detected by a generic trojan pattern, I think it doesn't like it because it uses a backdoor to get into the memory to read the SPD and what not. These files you gave me from SPDtool look like they're read from the wrong address, and SPD tool doesn't work on my P35 chipset pc at all, thaiphoon works great, think maybe its newer, you might want to give it a shot and get a dump from it. Says it works under vista32 also. It honestly sounds like 3-4-4-10 or one of the subtimings that the kingston sets at 667 is just too fast a timing for the ram, ie the ram doesnt run correctly at its rated speed at 667?

 

why don't you just change the ram since it's giving you so many problems ?

I'm running 4-4-4-12 and have no problems.

the part where the CPU did read at 2.4GHz is the intels IDA - it runs one core only at higher speed (12x multiplier), rather than 2 cores at 2.2GHz

 

Miro,

Thanks for explaining that about the CPU -- I'd never heard that. I'll check into it more later.

About using different RAM, yes, I agree, and I'm probably going to RMA these ones. But I want to know what's going on with them while I have them.

Big Mike,

I tried Thaiphoon and Thaiphoon Viewer, and neither one would run. (Couldn't install system driver error.)

What kind of a name is Showshock Softnology, anyway? All I know is, I'm not about to run his "Central Brain Identifier."

Before I give up, can you see a way to get SPDTool to read from the right addresses?

 

They're Russian what can I say, lol. I'll see what I can come up with on SPDtool or if there's something else out there. You might want to call Kingston though it definately seems like there's a problem with the ram running correctly in your laptop.

 

For getting Taiphoon to run in vistaX64 you could right click the taiphoon.exe and run it in xp compatibility mode or maybe try to install a 32bit xp


Download the latest spdtool (0.63) from here and re-do the dumb

http://forums.techpowerup.com/showthread.php?t=20349

See if you can pull the turbo memory from the rig before you dump again.

Don´t know if there is a dos spd dumping tool...
One more thing that you can try is to make a new user account on windows, so it is not loading much while booting then try to boot with that one memory and make the dump.

 

Hello Prema,

Thanks for joining us here.

Big Mike, Hep and Miro have been really helpful with the walkthrough as well, as you can see by reading the thread.

By way of introduction, Prema caught a related thread about this over at the ThinkPads forum and has some of these HyperX SO-DIMMs himself. He was kind enough to offer to help by comparing the SPD dumps -- if we can get it to work.

* * *

In this instance, creating a different user account isn't going to be of any use because the RAM doesn't remain stable even long enough to get the driver loading finished at boot ... or even started.

I re-ran SPDTool, and I'm afraid that I'm getting exactly the same readings for all three modules. (Again, I'm convinced that the HyperX is Module 5 of the three, if you haven't checked it out yet.) Seems like totally incomplete information.

Sorry! I know there are two forums where SPDTool is discussed -- one's Tech PowerUp, and the other is Xtreme Systems. I read through the entire TPU thread last night, and there was no info of any use, except a tangential question about whether this program actually works with SO-DIMMs, though it was definitely written around the DDR2 specification. I'll check out the Xtreme Systems forum next. Seems like the creator, Wizzard, hasn't been heard from in a while.

Prema, what did you get when you ran SPDTool? Could you upload it here so I can take a look? I'm hesitant to just flash my RAM with your SPD table, since it depends to begin with on whether SPDTool is working correctly anyway.

Are your HyperX sticks, by the way, the 800MHz, and are they 2x1GB or 2x2GB?
(I wasn't sure from your Thinkpads post.)

As to Thaiphoon in compatibility mode, it didn't work. I got the same system driver error message. Won't load. Can you get it to load on your machine, out of curiosity?

* * *

I'm assuming that even if you can stick around to answer those questions, Prema, we're almost at an impasse with the tools at our disposal.

I'll try at Xtreme Systems and at Tech PowerUp, but are there any other forums you know of?

 

I don't think you'll be able to get thaiphoon to run easily under x64 as it probably has an unsigned driver... Unfortunately most of these tools only work well under Windows XP and some work ok under Vista x32

 

At least as of 2006, someone on the TPU thread for SPDTool complained that it seemed not to work for 1GB SODIMMs, and at Xtreme Systems, back in 2006, someone posed the question (but got no answer) as to whether it was even able to read past the first of the DDR2 modules. I can't tell at this stage -- would have to finish reading the two threads over again, and then contact the author.

As for Thaiphoon ...

Well, it might be worth installing XP on a different partition and adding that to my boot entries, but the trouble I've heard that comes with adding XP to boot entries .... Any recollection of the precautions needed there? It's like, install XP first, then Vista, right?

But you're right, Big Mike. It's the unsigned driver ban. Installing XP is overkill. I'm going to check the intertubes for a workaround. Surely someone's thumbed their nose at this. I've got a second OS partition that I need to reinstall over anyway, so I'll find a workaround and test it there. I'll get back in a little bit with an option or two.

 

Well if you install XP AFTER vista it overwrites the boot sector with an XP boot, you can just insert a vista dvd and do the "repair my computer" thing and repair the boot sector with it pretty easily. You could also consider BartPE or one of the other "XP from CD/flash drive" installs of XP to run it from. Or if you have an x32 vista I think they're less stringent about the driver issues.

 

Ok, there are a bunch of workarounds for unsigned drivers in x64.

The first is F8 at boot up. (I trust that still works. Have never tried.)

The second is to write a test certificate for any particular unsigned driver you need on a permanent basis. There's a great AnandTech thread about this.
http://forums.anandtech.com/message...ORDFRM=&STARTPAGE=2&FTVAR_FORUMVIEWTMP=Linear

I don't know if this is a different option or not, but you can disable driver signature enforcement. If that's permanent, it's a really, really bad idea.
http://pc-hardware-networking.blogspot.com/2008/08/windows-server-2008-working-with_13.html

The last is to find a torrent copy of Linchpin Labs' Atsiv, which had a signed driver that disables unsigned driver disabling, if you can follow that. (All I can say is, someone must have had a lot of big brothers and sisters as a kid .... " I know you are, but what am I?")

Just to add to the tit-for-tat, MS eventually revoked their signature, and they've gone under. Maybe that means it won't work anymore. (Man, the useless information you run into.)

ComputerWorld and The Register did some good stories on Atsiv, if you're curious. The problem is that it works like a rootkit, and its code, I'd imagine, could probably even be reverse engineered by rootkit coders.
http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9028440
http://www.theregister.co.uk/2007/07/30/vista_driver_tool/

So, it looks like the F8 method is the way to go, duh, unless it doesn't actually let Thaiphoon install a driver in the first place.

Off to check it out....

 

CRAP!



Didn't work.



F8 works, but Thaiphoon still gives the same farkin' error.



GRRRRR.



Do you think Thaiphoon is flexible enough to run off of BartPE? Have you ever tried it?



Sayyyy .... You know what? If we're going this route, there's GOT to be a Linux LiveCD that reads SPD Tables, don't you think? Let me check on that.

 

I was going to mention that but being a Unix n00b I didn't have the first clue where to look for a LiveCD compatible SPD reader/writer. And I would think Thaiphoon would run ok off a PE disk just because it installs the driver on the spot when it runs from what I can tell, so you don't need to be able to reboot post driver install i dont think.

 

I'm still looking for a LiveCD.

Just discovered BartPE Builder will not work with x64 systems.

 

Yes it will, just not 64 bit OS's...
I doubt you're running a true 64 bit chip like an itanium, you're probably running a core 2 duo or turion 64. Therefore, you're backwards compatible to 32 bit.
Anyway. Just build a PE based on XP 32 and you'll have no issue.

 

My understanding of it is that the Builder program itself doesn't run under Vista x64, even if you have the XP installer disc. At least that was what the BartPE Builder site says. Is there a different way of running the Builder? Can it run off of a USB stick? Doubtful, but I'll have to check, I guess.

Well, I have been through several more Windows-based SPD readers, and none of them are giving me more detailed results about what's on the RAM. The only thing I'm able to get is a report off the trial version of Everest Ultimate that gives the timings from the Northbridge chip:

North Bridge: Intel Crestline-PM PM965 ]

North Bridge Properties:
North Bridge Intel Crestline-PM PM965
Intel Platform Santa Rosa
Supported FSB Speeds FSB533, FSB667, FSB800
Supported Memory Types DDR2-533 SDRAM, DDR2-667 SDRAM
Maximum Memory Amount 4 GB
Revision 0C
Package Type 1299 Pin FC-BGA
Process Technology 90 nm
Core Voltage 1.05 V
In-Order Queue Depth 12

Memory Controller:
Type Dual Channel (128-bit)
Active Mode Dual Channel (128-bit)

Memory Timings:
CAS Latency (CL) 4T
RAS To CAS Delay (tRCD) 5T
RAS Precharge (tRP) 5T
RAS Active Time (tRAS) 15T
Row Refresh Cycle Time (tRFC) 43T
RAS To RAS Delay (tRRD) 3T
Read To Precharge Delay (tRTP) 5T
Four Activate Window Delay (tFAW) 17T
Write CAS Latency (tWCL) 3T
Refresh Period (tREF) 7.8 us


There is one line in here that's interesting, though:

Supported FSB Speeds FSB533, FSB667, FSB800
Supported Memory Types DDR2-533 SDRAM, DDR2-667 SDRAM

Note that it's saying "Supported Memory Types" is a separate consideration from "Supported FSB Speeds."

Anyone know the reason the SPD Table is inaccessible? Is it
1) The mixed RAM?
2) The fact that it's 800MHz RAM and the motherboard doesn't like it?
3) A limitation in the BIOS? (This is the ThinkPad T61p, latest BIOS.)

I'm going to put my two slow sticks in (each a different brand) and that'll settle whether it's the mixed RAM.

 

I would suspect your issue with most programs is a lack of support for the SMBus your computer uses. This is basically a dedicated bus that gives access to all the information such as speeds, timings, temperatures etc about your computer a way to be addressed by software. I would suspect that everest is correct and your timings are 4-5-5-15, unfortunately you can't get the timings from the other ram aside from what memtest said which was what 3-4-4-10? If that's the case I'd lean towards both of those being correct figures and the ram for whatever reason just isn't running right at DDR667 at 3-4-4-10 timings, its possible there's a bad sub timing but theres no real way to assess that without a much more advanced low level program for reading further SPD data, which I'm not sure there is one readily available.

 

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

perhaps I've spoken too soon, there are some linux apps listed in there that might work with a livecd.

 

The strange thing is that Everest is a commercial program that probably supports as many different systems as possible. Just looked up the driver (?) for SMBus (machine.inf? pci.sys? which?), and it's dated 2006. Could it be an out-of-date driver, instead of the programs themselves? 2006 is before the Thinkpad T61p even came out. This is confusing, though, because under Device Manager, I checked online for a driver update, and it said that the latest driver is installed. It also said that there is no driver needed for this SMBus controller.

[ System devices / Intel(R) ICH8 Family SMBus Controller - 283E ]

Device Properties:
Driver Description Intel(R) ICH8 Family SMBus Controller - 283E
Driver Date 6/21/2006
Driver Version 6.0.6001.18000
Driver Provider Microsoft
INF File machine.inf
Hardware ID PCI\VEN_8086&DEV_283E&SUBSYS_20A917AA&REV_03
Location Information @system32\drivers\pci.sys,#65536;PCI bus %1, device %2, function %3;(0,31,3)
PCI Device Intel 82801HBM ICH8M - SMBus Controller

Device Resources:
IRQ 11
Memory FE227400-FE2274FF
Port 1C60-1C7F

I do see, however, that there are some 2008 hotfixes that mention machine.INF as one of their components. Looks as though the most recent version is May or July of this year. Is there a way to add the INF/SYS file without installing the whole hotfix? The reason: I don't know if you remember, but at some point here I mentioned that I'm having Windows Update issues. No hotfix has successfully installed since July, and it's not a malware issue. Even the WU team has given up and transferred me to the Vista tech support team. (Yes, it's time for a clean install soon, but I'm of the "it never hurts to learn from troubleshooting school.)

 

FYI, Big Mike, there are some Linux tools that seem to do this, but I am not sure yet whether I should be using them. All the same, you might be interested in them.

The basic toolchest is i2ctools, and I believe it is now incorporated into the standard Linux kernel, which, if true, means it would be available in all the Live CDs.

Here's the manual:
http://www.lm-sensors.org/wiki/i2cToolsDocumentation

This Live CD, Inquisitor, also looks interesting for hardware diagnostics. It includes i2ctools, but it's got a really low-level, no-buttons GUI, so I think you have to know what you're doing at the command line.
http://www.inquisitor.ru/about/index.html

 

If you posted this I missed it, do you have the part # of that kingston ram? Should be on the label as well as the package. I wanted to look at the detailed specs for them.

 

Well, there are two kits -- a 2x2GB and a 2x1GB. I have the 2x1, and Kingston has taken down the spec sheet for it.

Hard to explain the logic. I think it isn't the RAM per se, but that they had a "catalog malfunction" that led people to list this as a 4GB set. They want to keep this a "vendor" problem rather than something Kingston needs to reimburse for. Still having trouble making sense of the disappearance of the spec sheet. But here is the 4GB spec sheet.
http://www.valueram.com/datasheets/KHX6400S2ULK2_4G.pdf

The 2GB version is KHX6400S2ULK2/2G. I'll look around for a copy not on the Kingston site.

FEATURES:
Power supply : Vdd: 1.8V ± 0.1V, Vddq: 1.8V ± 0.1V
Double-data-rate architecture; two data transfers per clock cycle
Bidirectional data strobe(DQS)
Differential clock inputs(CK and CK)
DLL aligns DQ and DQS transition with CK transition
Programmable Read latency 4, 3 (clock)
Burst Length: 4, 8 (Interleave/nibble sequential)
Programmable Burst type (sequential & interleave)
Timing Reference: 4-4-4-12 at +1.8V
Edge aligned data output, center aligned data input
Auto & Self refresh, 7.8us refresh interval (8K/64ms refresh)
Serial presence detect with EEPROM
PCB : Height 1.180” (30.00mm), double sided component

PERFORMANCE:
Clock Cycle Time (tCK) CL=4 2.5ns (min.) / 8ns (max.)
Row Cycle Time (tRC) 57.5ns (min.)
Refresh to Active/Refresh Command Time (tRFC) 127.5ns
Row Active Time (tRAS) 45ns (min.) / 70,000ns (max.)
Single Power Supply of +1.8V (+/- .1V)
Power TBD W (operating per module)
UL Rating 94 V - 0
Operating Temperature 0o C to 55o C
Storage Temperature -55o C to +125o C

 

I think at the end of the day either this ram isn't compatible with your laptop or it's just not capable of the advertised 3-4-4-10 timings at ddr 667... 3-4-4 is REAL FAST at 667, especially for a laptop module. There's hardly any desktop ram that performs better than CAS4 at pc5300 on the rated speed, some good ram can be run at that timing but very few sticks come factory rated that quick. I'd say its time to call kingston for an RMA, and or to see if they know any compatibility issues with that ram and your notebook?

 

SPDtool It is compatible, i have tested 6 different DDR2 So-DIMMs here and it could read all of them.

I have 2 x 2GB of the 667 version of the HyperX.
But we can not alter your sticks until we get a proper original dump of the SPD timings instead of that trash read-out...

My guess here is that SPDtools is incompatible with your board or that 64bit Vista... i think Thaiphoon is the way to go here, but you would need a 32 bit vista or any xp version.

I have a 90MB XP Live, which boots from a usb stick... drop me a PM if you would like to test with that.

 

Yeah, I agree. Matter of fact, I just sent in a tech support request to teh Kingston.

What I don't get is, why would they put the 667 preset in there without doing thorough testing? It's a SO-DIMM; it's made for laptop use where BIOS is inflexible and overclockers take the Darwin Award every year. Someone like me would obviously really love to use this HyperX stuff, and just imagine that they've gained 3-5% performance from the extra latency boost, even if they don't in fact get any real improvement. The Kingston engineers should have some idea of the average constraints on a laptop that takes only up to 667 -- and I can promise you that this T61p is no slouch. I can't fathom how you could assume, under these conditions, that the preset was good to go.

Maybe they'll send me an SPD flashing tool and a timings profile that'll actually work!

Sigh... NOT.

Big Mike, thanks for hangin' in there through this. I've gotten a huge edumacation!

Just watch: Someone will check in with this thread two months down the road and say, "ROFL! All you gotta do is start with the battery and the power supply plugged in, and it runs smoove."

.... Truth be known, I never tried that, since I had to swap out teh HyperX so many times to begin with. But don't hold your breath.

Happy holidays to you, and keep your alerts active for this thread for when I hear back from the manufacturer.

 

Wow. Just caught Prema's post. Ok, PM coming your way, friend. Thanks!

 

Just got the information that Thaiphoon is NOT running on 64bit Vista at all.

Tried SPDtool and Thaiphoon on a virtual xp and both don´t work.

On a BartPe XP Boot cd only SPDtoll will run (and give proper dumps) and Thaiphoon refuses to run at all.

 

Just got off the phone with Kingston.

They confirmed that this SPD Table mystery is an SMBus controller problem, courtesy Lenovo. Lenovo's BIOS isn't reporting the SPD Table to the OS properly.

He said there is a 667 setting for this RAM, but it's apparently too aggressive. I should have asked him -- why would you sell these sticks to work on a 667 machine if they don't?

I can feel free to change the SPD settings, but it'll void the warranty.

 

Can you feel free to return your not working ram for a replacement? lol It's not the machines fault that their ram doesn't work at the SPD timings.

 

Yeah that has nothing to do with your laptop. It's just sh*tty memory.

 

We agree on the verdict on this so-called performance RAM then?

Well, Prema was good enough to let me borrow his BartPE ISO and I re-ran SPDTool from the USB that BartPE can read. (Incidentally, BartPE can't read SATA / PATA controllers.) That was shortly after the Kingston tech called.

As Gomer Pyle used to say, "Sur-PRISE sur-PRISE sur-PRISE!" The readings on all three(?) RAM modules didn't change an iota. No good data on the Kingston stick. That means, I guess, that the BIOS is bad on the T61p. Even BartPE needs the T61p BIOS. Three days to figure out there's nothing more you can figure out.

But, basically, since MemTest showed the memory running at 3-4-4-10, it's following the preset, just unforgivingly so.

Clearly not RAM but SPAM.

Thanks for stickin' with me through this, y'all. I didn't learn a single thing about adjusting timings, but I got a great education about everything else. +1 all the way around.

Variations on a theme. Enjoy!

http://www.youtube.com/watch?v=xAno18rbkoc
http://www.youtube.com/watch?v=tFovNSSjg7k
http://www.youtube.com/watch?v=YwQAggKGd_0
http://www.youtube.com/watch?v=IxOtSWYPJqQ
http://www.youtube.com/watch?v=ODshB09FQ8w
http://www.youtube.com/watch?v=Ix081prSiNc
http://www.youtube.com/watch?v=zjqZ0aIAgFM

Hope to see you around soon.

Happy Holidays!

a.k.a.

 

Ya'll come back now ya hear!