Thinkpads Forum

Ok, we're hearing about these x64 8GB RAM upgrades for the T61-series. And now for something completely different.

Yes, there are yet other ways of going bananas with RAM in a notebook -- but they don't seem to be working out well for me.

I'm just going to give you this lil' anecdote before I RMA my RAM, in case anyone has a clue how to fix the problem here.

I picked up some of the Kingston Hyper-X 800MHz CAS4-4-4-12 SODIMM memory to see what would happen on my 667MHz T61p.

Here's the press release:
http://www.kingston.com/asia/press/2008/product/09a.asp

Kingston's press release said that this memory had two BIOS pre-sets, 800 4-4-4-12 and 667 3-4-4-10. In theory, this should work.

Now, let me just mention to you that there's misinfo out there about this chipset. Someone at Kingston screwed up and released one version of their catalog that suggests that KHX6400S2ULK2/2G is actually 4GB of RAM. It's not. It's 2GB. You'll see a lot of vendors out there listing this as 4GB, and my vendor had no clue either. It appears that Kingston is just skrewing the vendors for their own mistake when the first customer comes back irate. (Seriously! Shame on them.)

In my case, that may have temporarily been the silver lining in this whole experiment. As you'll probably know, the chipset in the T61 series hardware drops back from 667 to 533 when it has over 3GB installed. So, in effect, a 4GB set of this RAM might go all the way to CAS 2. Gnarly? . . . Dooooood . . . . .

That's what I was itching to check out, anyway. When I found out that I had only received 2GB, I thought I could try it out a little more safely. I mean, would the fixed settings of a laptop BIOS actually handle CAS 2? Likely not.

So, then I was dealing with 667 CAS 3 -- and Kingston promotes that the RAM has a BIOS preset of 3-4-4-10 when downclocked to 667.

Guess what?

It didn't work.

At all.

It took me until 2 am in the morning just to figure out that when I was getting 2048MB and not 4096MB, BIOS wasn't giving me bad readings. Kingston had just majorly hosed me. But the more pressing problem was that the 3-4-4-10 setting wasn't cutting it. I'd get that dreaded 0xc0000221 error code when Windows started to boot. I fiddled with the RAM's seating, and nothing worked. If I stuck in a stock stick of 5-5-5-18 alongside a 3-4-4-10 stick, it would boot. Not under any other conditions. The BIOS is 2.19 7LETB9WW. Pretty sure that's newest, or next-to-newest.

Before I RMA, anyone have any suggestions on what to do with the RAM to get it working?

Furthermore, how do you get satisfaction in your complaint to Kingston? (And keep them from screwing the vendors.)

Finally, if you're interested, there was an excellent survey done on Madshrimps.be about DDR2 performance differences. Just interpolating from 800 CAS 4 to 667 CAS 3 and from 667 CAS 4 to 533 CAS 3 (a rough proxy for what it would have been like going from 800 CAS 4 to 533 CAS 2), it looks from their testing as though there would only have been a 3-7% speed performance gain over CAS5-5-5-18 667 if I'd gotten 4GB of the 800 CAS 4 RAM. Here's the report. Good stuff!

http://www.madshrimps.be/?action=getart ... rticID=472

Welcome in the Hyper-X world

OK, I know I'm using Da Forumz to compensate for my inate idiocy, but ...

Did you mean I'm blazing some new trails of stupidity, or that Hyper-X will be kicking some butt when I get this sorted?

I'm just going to repeat the crucial point (pardon the pun):

This RAM has a 667Mhz BIOS preset of 3-4-4-10, so it ought to work, dingit!

Well, while my ThinkPad is one generation older (same hardware as T60P), just like yours it supports up to 667 MHz RAM. Within these 667 MHz I don´t believe there is really much that can be done on laptops even by tweaking latencies. Having Hyper-X installed in my ThinkPad, I don´t see even the smallest performance gain, so basically I think Hyper-X is just waste of money unless one has a desktop machine or a gaming laptop (as in both cases the hardware is quite different from the one in regular laptops and hence specs are different), but then again one would buy another type of Hyper-X So, to sum up, if I was you, I would return it and ask for refund (if still possible).

Cheers

Marin

A side note: Mine is 4-4-4-10, 667 MHz, so I don´t really believe a 3-4-4-10 would make much of a difference even if you ever manage to bring it to that.

Marin,

Yeah, that's kind of what I expected. It would still be nice to get this RAM to work, but you're right, this 3-4-5 stuff is still a gimmick. I'm wishing for 3-7% gain, but really it's probably somewhere between 'meh' and infinitessimal.

Interesting Hardware Bistro piece I remember reading that substantiates your point on exactly the RAM kit that you purchased. The difference on HyperX is close to absolute zero, even on the synthetic benchmarks. (I love how the reviewer is saying, yes, he spots the difference, and he'd consider buying this RAM. Paid review, no doubt.)

http://hardwarebistro.com/index.php?opt ... CL4-Review

Incidentally, I just ran Memtest, and it really didn't like it. The first stick crashed the program on Test 3 -- black screen, restart. The second stick got an interrupt by Test 2.

It said it was running at 3-4-4-10.

Another Memtest update. Running one HyperX stick with a stock RAM stick alongside resulted in ZERO errors.

This is something else -- an SPD or BIOS problem, it seems.

Hi

Plese do me a favour and make a SPD dump of that HyperX RAM

Just download spdtool from here:

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

In SPDTool click on "file" and there on "read" choose the stick and after that "save"

please upload that output file somewere so i can take a look into it and compare it to a pair of 2gb hyperx that i have running here.

BIG THX

Prema

Hello Prema,

Perfect timing! Pardon the pun.



I'm working on this over at Notebook Review as well.
http://forum.notebookreview.com/showthr ... ?p=4300458

The SPD dumps are posted there. I've got two memory slots, but THREE(!) modules to read. I presumed at first that one of them was the Robson Turbo Memory, but I don't know. Maybe DDR2 has two modules per stick? (Then why only 3 and not 4?) If the first interpretation is the right one, then, as I say there, the HyperX is probably Module 5's SPD, because it doesn't say either Thinkpad or 6459CTO (the T61p model number) on the hex table.

A few good people helping there, but it would be great if you checked in there, as you have some HyperX as well.

If that's not going to work for you, PM me and I'll give you my email address.

Big Mike on that forum has done this before, but he can't make head nor tails of this data. He thinks SPDTool is reading from the wrong address. If you know what to look for and can sort this out, please be our honored guest! We're kind of stuck. I can't even find a CAS Latency recorded in the RAM!

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.

maybe try to do the dump from a 32bit xp

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.