Thinkpads Forum

Ok, machine is a 2645-45u, running latest BIOS.

A couple weeks back, I took the system to a computer show to see about upping the ram from 288MB to 544MB (switching the 128s to 256s).

The memory dealer had two low-density, PC133 SODIMMs, but when I installed them, they didn't seem to work exactly right. The BIOS indicated that there was 544MB installed, but only 420MB was usable (going from memory here, so numbers are approximate). Switching the modules around didn't help. Needless to say, I didn't buy them.

Unless I'm missing something, if the BIOS sees all the memory, shouldn't I be able to access it all if everything is working right?

These were Kingston units, and I don't seem to recall them having much of a reputation hereabouts. My 128s are not Kingston, and they work flawlessly.

I will try to do a search later on this, but in the mean time I'll leave this question open for whomever wishes to comment.

Rob

They were probably high density, which has 4 chips on each side. You need low density, 8 chips on each side.

No, these were low-density - I could see eight chips per side.

Rob

It's the fact that they are PC133 memory that's the root of the problem Rob.

While there's some PC133 memory that will work in the 600X series, e.g. the latest PC133 lo-density memory from Crucial, most of it won't.

Regards,

James

We appear to have a touch of confusion. The 2645-45U is NOT a 600E or 600X, it's a straight 600.

That aside, you seem to be in agreement that not all RAM is created equal.

Rob

Rob Mayercik wrote:That aside, you seem to be in agreement that not all RAM is created equal.

If possible, check to see what the CAS latency (CL) is on the module. My guess is that you put in CL3 modules while the 600 is most likely looking for CL2. PC133 CL2 modules do exist but it can be hard/more expensive to find.

It is a WELL KNOWN FACT that some old TP600 can detect only half of the memory installed into the second DIMM slot. Your system seems to fall into this case.

In this case, you have 32(On board)+256(1st slot)+128(2nd slot)=416MB, which is just about the amount you saw.

This has nothing to do the memory config such as CL2/3 or PC100/133. This is an architecture limitation of TP600 systemboard so there is nothing you can do about it.

shined wrote:It is a WELL KNOWN FACT that some old TP600 can detect only half of the memory installed into the second DIMM slot.

Hmm.. 2645-45U has i440BX chipset (according to twbook). I was not aware of any problems associated with the 600 in memory expansion (only have 600E and Xs).

shined wrote:This is an architecture limitation of TP600 systemboard so there is nothing you can do about it.

Do you know exactly what the problem/limitation is? If the on-board memory is disabled, can both memory slots be populated with 256MB modules and will the full 512MB be usable?

shined wrote:It is a WELL KNOWN FACT that some old TP600 can detect only half of the memory installed into the second DIMM slot. Your system seems to fall into this case.

In this case, you have 32(On board)+256(1st slot)+128(2nd slot)=416MB, which is just about the amount you saw.

This has nothing to do the memory config such as CL2/3 or PC100/133. This is an architecture limitation of TP600 systemboard so there is nothing you can do about it.

Just a second there, Shined - I agree that some 600s fall into this category, but mine is not one of them.

I am currently running 32 (on board) + 128 (1st slot) + 128 (2nd slot) and getting all 288MB both recognized and available. I should think that if the 128MB module is being completely recognized, the 256MB module should be as well. I think in my case the CL2 vs. CL3 is the more likely scenario, or the Kingston-branded modules just aren't "happy" in my particular machine.


Rkawakami, I don't have the modules currently - since the computer didn't completely recognize them, I didn't buy them. For my future reference, is there something on the module (or a sticker on the module) that I can use to determine this information?

Rob

Rob Mayercik wrote:We appear to have a touch of confusion. The 2645-45U is NOT a 600E or 600X, it's a straight 600.

Sorry Rob, for some reason I thought we were discussing a 600X.

You need PC66 RAM for that machine to happily max out at 544MB.

While the PC133 RAM you tried plays better with the 32MB of 66MHz RAM thats on the MB than PC100 would, there's a glitch in the 440BX chipset that prevents the machine from using all of the installed PC133 memory, even though all of it is seen on BOOT.

I remember going through a series of tests years ago on my old 51U with differing speed SODIMMs and the only way to get it to run happily was with PC66 memory IIRC. I reported the results in the old Forum as well as repeating them here. Try searching for 2645-51U.

Regards,

James

Rob Mayercik wrote:For my future reference, is there something on the module (or a sticker on the module) that I can use to determine this information?

Most modules will have a "CLn" designation on the label so it's very easy to tell the CAS latency. Others may have a "-2nn" or "-3nn" suffix (for CL2 or CL3 respectively) after the speed rating (PC66, PC100, PC133).

Also for some clarification on "low density vs high density" memory modules: 256MB memory modules may contain 8 (high density) or 16 (low density) individual memory ICs. This really means that the design technology and architecture used between them is totally different (32Mbit vs. 16Mbit). The higher density memory generaly requires more refresh cycles than the lower density memory in order to operate properly. If the memory controller inside the computer assumes that only a certain type of refresh rate is always going to be used and if it cannot adjust itself or the memory modules to use a different rate, then problems can arise. Mostly, the high density modules will fail POST. Sometimes they can pass POST and seem to work fine until a random bit failure hits and the system crashes. And still other high density modules may work without any problems at all, even though they are technically being operated way out of spec. I do not know if the 600/E/X series systems are able to self-adjust the memory timings for either the CAS latency or the refresh rate. That is a function of how the Intel 440BX chipset is designed and/or implemented. Perhaps somebody with better knowledge about the inner workings of the system can provide that answer.

Apologies for the bump on this older thread, but having just read another RAM thread with lots of info, I came up with a question that more appropriately belonged here rather then there.

After reading that other thread (had to do with PC133 and 600X machines), I did a quick google search for PC133 256MB CL2 SODIMMS, and got several hits.

Earlier discussion in this thread suggests that CL2 is a major factor in getting 256MB memory modules working properly in a 600-series machine (note: NOT 600E or 600X). Density is also listed as a factor.

James: you said previously that only PC66 256MB modules would work "happily" on a 600. Do you happen to remember whether you ever had occasion/opportunity to try PC133 CL2 modules?

Of course, I seem to be turning up several hits on 256MB PC66 modules too, but I'm still interested in the above for information's sake.

Rob