PC100 that is actually PC133

Message

iann070 · #1 Post by **iann070** » Wed Apr 18, 2007 4:51 am

I have a T21 with 384MB of PC100 RAM (128+256MB).
I just bought a stick of 256MB RAM so I can upgrade to 512MB.
The eBay seller advertised the stick of RAM as:
"256MB IBM ThinkPad PC100 Laptop SDRAM SODIMM FRU33L3070"
They listed the T21 as compatible, so I bought it.

Well, it arrived today and on one side of the stick of RAM there is indeed an IBM sticker with the following:

256MB PC100 CL3 3.3V S N (32Mx64) OPT:33L3069 FRU:33L3070

However, on the otherside there is a Micron sticker with the following:

MT16LSDF3264HG-133E4 PC133S-333-542-Z
256MB, SYNCH, 133MHz, CL3

So I assume that it was made by Micron as a PC133 stick and IBM have deemed it compatible as a PC100 and stuck their sticker on it, or the seller is trying to sell my a PC133 stick as a PC100 stick.

Anyway, I installed it and the BIOS did a memory check and passed all 512MB of RAM, but when it tried to boot from the hard drive, I got a BSOD and a message telling me a problem occured with Windows, it then checked the file system and reported it found no problems, then tried to boot again.
After another RAM count passed OK, I got an "Invalid System Disk" error.
After several reboots I took the "faulty" RAM out and booted from an XP Pro install disc and ran the Recovery Console, then ran FIXBOOT to fix the boot sector.
On the next boot, everything was OK and I have booted several times into Windows with the old RAM sticks.
But when I install the new stick I get the BSOD again, so I can only assume that it is because it is a PC133 stick, and although it passes the memory test in BIOS, it falls over when it loads Windows.

So what do you guys think?
Have I been sold a stick of PC133 disguised as PC100?
Would IBM rebadge PC133 as PC100, or is the stick simply faulty.

I'll wait a day or two and see what you guys think then see if I can send it back to the seller for an exchange or refund.

Thanks in advance.

#2 Post by **RealBlackStuff** » Wed Apr 18, 2007 6:21 am

You can use PC133 RAM from Crucial in a TP 20-21-22 (see also the 'Sticky' at the top of this forum).
But the crucial (no pun intended) point is the CAS Latency or CL.
To work in your TP, it must be CL2 and your purchase is CL3. Won't work unfortunately.
So send it back for an exchange or a refund.

DooMinick · #3 Post by **DooMinick** » Wed Apr 18, 2007 6:47 am

Don't have any experience with RAM in thinkpads, but in my desktops any PC133 CL3 RAM (even no-name) could be run with CL2 at 100MHz. Even now I'm running PC100 CL2 RAM in a 133MHz system (PIII 800EB, i440BX) with CL3 and passing few hours of memtest86 without any error, but it's Memory WorX.
In conclusion, I think it could be faulty RAM or SODIMM socket more likely than too high CL. Try exchanging your old module with new in the same socket, it worked in my 600E (second socket needs re-soldering).

Purcy · #4 Post by **Purcy** » Wed Apr 18, 2007 7:08 am

I believe that is what I am running in my T22, as I purchased the same FRU#33L3070 and it had an IBM sitcker on one side and a Micron sticker on the other side. Here is what it says on the Lenovo site for T20, T21 & T22 systems:

ThinkPad 256MB SDRAM SO DIMM
Note: Not supported on ThinkPad A21e (2628, 2655) and A22e (Intel Celeron) systems. Optional part#33L3069 FRU part#33L3070 3.3 Volt, 144-pin, non-parity, 64-bit, 100 MHz SDRAM memory

Hopefully Ray Kawakami will post on this, he is the memory expert.

DooMinick · #5 Post by **DooMinick** » Wed Apr 18, 2007 7:51 am

The best you could do for now is making a memtest86 liveCD (or bootable floppy, as you wish) and check your new module alone in the socket used by your old module. That test eliminates the bad socket.

iann070 · #6 Post by **iann070** » Wed Apr 18, 2007 7:54 am

Purcy wrote:I believe that is what I am running in my T22, as I purchased the same FRU#33L3070 and it had an IBM sitcker on one side and a Micron sticker on the other side.

Yes, but on mine the IBM sticker says PC100 and the Micron sticker on the other side says PC133, so I'm guessing someone is trying to pass off a stick of PC133 as PC100, although I would have thought they would have taken the Micron sticker off first.

iann070 · #7 Post by **iann070** » Wed Apr 18, 2007 8:02 am

DooMinick wrote:The best you could do for now is making a memtest86 liveCD (or bootable floppy, as you wish) and check your new module alone in the socket used by your old module. That test eliminates the bad socket.

I don't think it's the socket as I already use both sockets, (128MB in one socket, 256MB in the other) and everything is fine.
I simply want to swap out the 128MB stick with the new 256MB stick to give me the maximum 512MB.
I have tried my existing 256MB stick and the new 256MB stick in the two sockets, the old stick works in both socket while the new stick causes problems in both sockets, so it can't be the sockets.

#8 Post by **RealBlackStuff** » Wed Apr 18, 2007 8:25 am

DooMinick,
it's useless to compare desktop memory with laptop memory.
Older desktops 'eat' any memory more or less that you throw at them.
IBM Thinkpads however, especially the older ones, are very finicky about CL speed.
The T20/T21/T22 MUST take CL2, either PC100 or the newer Crucial PC133, otherwise no go!
The T23s take PC133 CL3, so iann070's memory would fit a T23 perfectly.

iann070 · #9 Post by **iann070** » Wed Apr 18, 2007 8:43 am

RealBlackStuff wrote:The T20/T21/T22 MUST take CL2, either PC100 or the newer Crucial PC133, otherwise no go!
The T23s take PC133 CL3, so iann070's memory would fit a T23 perfectly.

If the IBM sticker on my RAM states: "PC100. CL3. FRU:33L3070", and the T21 takes the same (FRU:33L3070), then surely the T21 can take CL3, unless IBM use CL2 and CL3 both with the same FRU number.

DooMinick · #10 Post by **DooMinick** » Wed Apr 18, 2007 8:49 am

RealBlackStuff wrote:DooMinick,
it's useless to compare desktop memory with laptop memory.
Older desktops 'eat' any memory more or less that you throw at them.

I know it's quite different, but some basic rules are the same.

The T20/T21/T22 MUST take CL2, either PC100 or the newer Crucial PC133, otherwise no go!

Does it check SPD for CL at the highest supported speed?

#11 Post by **rkawakami** » Wed Apr 18, 2007 10:58 am

Purcy wrote:Hopefully Ray Kawakami will post on this, he is the memory expert.

I know enough to be dangerous

.

I tracked down a very similarly marked Micron module:

MT16LSDF3264HG-133E4 PC133S-333-542 (Note: no "Z")
256MB, SYNCH, 133Mhz, CL3

and on the other side:

IBM P/N Z H/C S/N: 38L2983
256MB PC100 CL3 3.3V S N (32Mx64)
OPT: 33L3069 FRU: 33L3070

I had a problem recently with this module in a T21. Actually I was in the process of selling it to another member here when it failed memtest86+ testing. The strange thing was that when I received it I plugged it in a 600E (a 66Mhz system) and it worked fine. At any rate, it appears to be an intermittent issue as it seems to be working at the moment in the same T21 this morning. So far it has passed after 40 minutes of testing. It's possible that there's a temperature related problem (fails when hot).

At any rate, I am under the impression that the T21 requires the use of CL3 enabled modules. I'll try to verify this today by taking the module to work and re-programming the SPD settings and see what works (or doesn't). What it sounds like to me is that there may be some bad memory locations since the system goes through POST. You should also make sure that the new module is the only one installed in the system. The best thing to try is to boot a copy of memtest86+ or PC Doctor. If the system POSTs, you should be able to load either diagnostic program and run some tests.

memtest86+
.ISO file of PC Doctor for T2x

If the system doesn't POST with just the new module installed and you are still within the time frame for a refund, then you should contact the seller and get your money back or a working replacement.

iann070 · #12 Post by **iann070** » Thu Apr 19, 2007 3:15 am

Thanks rkawakami, my new stick of RAM also seems to have an intermittent problem like your (almost identical) stick of RAM.

Here are the results of memtest86+

With the 2 original sticks of RAM (128+256MB) installed, I ran memtest86+ for nearly an hour with no problems, then I did the following:

Slot 1 = New Micron 256MB PC133
Slot 2 = Empty
POST counts RAM OK
RAM failed memtest86+ after only 2 seconds

Slot 1 = Empty
Slot 2 = New Micron 256MB PC133
POST counts RAM OK
RAM failed memtest86+ after 30 minutes

So it seemed to fail straight away in slot 1 and fail after 30 minutes in slot 2, could it be a slot thing?
But then the new RAM started to fail in either slot after only 2 seconds, even after leaving the T21 turned off for several minutes between each test.

BTW, why would IBM stick a PC100 sticker on Micron RAM that is PC133?
Does that mean if you ordered FRU:33L3070 from IBM you could get either PC100 or PC133.

#13 Post by **RealBlackStuff** » Thu Apr 19, 2007 3:23 am

No offense meant, but you are wasting time!
Get that stick returned for exchange or refund before the warranty/exchange period expires!

iann070 · #14 Post by **iann070** » Thu Apr 19, 2007 3:37 am

RealBlackStuff wrote:No offense meant, but you are wasting time!
Get that stick returned for exchange or refund before the warranty/exchange period expires!

No offense taken, I only got the stick yesterday and didn't want to jump the gun by returning it straight away, after all, the problem could be with my T21.

But yes, you are right, and I have already emailed the seller about an exchange, (the seller seems to deal in RAM so they should be able to exchange it for proper PC100.

#15 Post by **rkawakami** » Thu Apr 19, 2007 4:15 am

Computer POST checks (especially if the "quick boot" option is used) do not throughly test the memory like a dedicated diagnostic program (memtest86+ or PC Doctor). A difference in the amount of time to failure can be attributed to a number of reasons (temperature, test pattern being used and yes, possibly even a slot issue). That it *does* fail in either slot tells me that there's something wrong with it.

I can't speak for IBM on why they would put a 33L3070 FRU on both a PC100 and PC133 module. However, from my point of view, if a system uses a 100Mhz memory bus, then modules tested and guaranteed to the 133Mhz spec should automatically run without any problems in the 100Mhz system; module "density" being the same (a topic which has been explained many times before; ask if you need clarification) and CAS latency (CL) being the same. Obviously the 100Mhz parts were available first. IBM designed in those parts for their systems. A few years later, the PC133 spec (and parts) came along. Those faster parts will still work if they are "slowed down" a little bit when installed in a PC100 system. IBM does/did many "qualification" exercises when specifying what memory modules can be used in any particular system. A module maker will submit samples to IBM and extensive testing is done to verify that they work as desired. Only then will they say that module "X" is qualified for use in the system. As the FRU had already been assigned to modules that would work in the PC100 systems, I suppose they decided to keep using it even for the PC133 modules.

To further explain the CL spec without getting too much into the nitty-gritty: CAS latency is the number of clock cycles that must occur after the read command is given, until the data is ready to be read from the memory. It is a fixed number that is built into the memory controller on the motherboard. My understanding is that the IBM PC100 systems use/require CL3. I was not able to perform the CL experiments on my Micron module and T21 system last night due to a family emergency. I will try to get to them again before the end of the week. It is something that I have been wanting to verify for awhile. Modules that are marked CL2 should automatically work in systems which use CL3. The reason being is that the module was tested at the CL2 condition and it passed (or so we hope!), therefore if given the additional time is takes for another clock cycle to occur, the data would still be ready. The reverse is not true. If the module is marked with the CL3 designation, there's no guarantee that it will operate correctly at the CL2 setting. The system would in effect be asking for the data before the module could deliver it. I had explained this before in much detail in another thread but I don't have time to search for it at the moment. I'll try to update this post in the morning (well, later this morning).

iann070 · #16 Post by **iann070** » Thu Apr 19, 2007 9:18 pm

Ray
Please don't go to all this trouble for me, you have been more than helpful already, thanks.
We can safely say that my RAM is faulty by running memtest86+, or is incompatible with my T21, although it may work perfectly in another T series laptop.

Speaking of memtest86+, I ran this program on my Pentium 3 desktop with 512MB RAM and it found errors, I let it run for about 30 minutes and I think there were about 242 errors before I rebooted.
The thing is, my desktop is a very reliable PC, it never crashes or freezes and I do push it to it's limits when I'm authoring DVD's from my video camera, so I was quite surprised to see the RAM failing the memtest.

david_sa · #17 Post by **david_sa** » Thu Apr 19, 2007 9:50 pm

Seems to be a defective module or laptop issue

I have the exact ibm-micron pc100-pc133 module, it works perfectly on T21, A20p, 600x... even combined with some 256mb pc100 modules: kingston ktm-tp390x, pny, viking ... and with a Crucial 256mb pc133 CL2 low density module.

I combined CL2 and CL3 modules, and worked perfectly (ibm/micron 256mb CL3 + crucial 256mb pc133 CL2).

Try a bios upgrade, it could enable some pc133 compatibility. My T21's bios version: 1.16 (kzet34ww)

pianowizard · #18 Post by **pianowizard** » Thu Apr 19, 2007 9:59 pm

Correct me if I got this wrong, but I think some (or all?) CL3 PC133 sticks become CL2 when they are slowed down to PC100 in a 100MHz motherboard. If this is true, then the OP's CL3 PC133 stick is supposed to work. It really is just defective.

#19 Post by **rkawakami** » Thu Apr 19, 2007 10:17 pm

iann070 wrote:Ray
Please don't go to all this trouble for me, you have been more than helpful already, thanks.

It's really no trouble. I've been curious about how the system uses the memory modules. It's a perfect excuse for using the tester at work and playing around with the SPD bits to see what happens. I've already re-programmed a couple of modules that had bad information programmed into it.

As far as your desktop system is concerned, I'm surprised that if memtest86+ has detected some errors you haven't seen any evidence of it when the OS is running applications.

I was not able to get to it today (playing travel agent at home for the wife) but I hope to do some poking around tomorrow.

@pianowizard: The CAS latency (CL) of the module is a timing specification which indicates how many system clock cycles need to occur before you have valid data. A PC133 CL3 module can in fact be slower than a PC100 CL2 module. Going from CL3 to CL2 (at a given PC speed) actually means that the memory needs to be faster (3 clock cycles gives more time for the memory to be "ready" than 2 clock cycles).

edit: I found the post I made several months ago about CL:

http://forum.thinkpads.com/viewtopic.ph ... 151#232151

Purcy · #20 Post by **Purcy** » Thu Apr 19, 2007 10:27 pm

Well I haven't opened up my slots yet to check but I really think I have the same sticks as the OP. I found this site that sells IBM memory and here that FRU# 33L3070 is shown as 100/133Mhz:

pianowizard · #21 Post by **pianowizard** » Thu Apr 19, 2007 10:35 pm

rkawakami wrote:@pianowizard: The CAS latency (CL) of the module is a timing specification which indicates how many system clock cycles need to occur before you have valid data.

Ray, don't worry, I understood that very well. In fact, I explained that to someone on this board several months ago, trying to convince him that the absolute latencies for CL5 PC2-5300 and CL4 PC2-4200 are identical. Unfortunately, that guy had zero background in math and couldn't understand my calculations.

What I was asking in my last post was, whether a CL3 PC133 module becomes able to send data in 2 clock cycles instead of 3 when it downgrades to 100MHz. Another similar situation is, when we put a CL5 PC2-5300 into a PC2-4200 machine, it slows down to 533MHz and the CAS latency becomes 4 cycles.

BTW, I've also seen RAM sticks with both PC100 and PC133 stickers.

#22 Post by **rkawakami** » Thu Apr 19, 2007 10:55 pm

PC133 CL3 = 22.5ns access time (7.5ns x 3)
PC100 CL2 = 20.0ns access time (10ns x 2)

So, a PC133 CL3 has a chance of not working properly at PC100 CL2 timings. Oh, and the module must have the CL2 bit set in the SPD in order for it to be recognized as working at that condition (if the motherboard checks).

#23 Post by **RealBlackStuff** » Fri Apr 20, 2007 2:59 am

To add to the confusion: Crucial sell 256MB modules for a T21 with these specs (CT484656):
SDRAM, PC133 • CL=2 • Non-parity • 133MHz • 3.3V • 32Meg x 64
Kingmax advertises their T21-modules as PC100/CL2. These can come in 2 flavours: 16-chip (16x8) TinyBGA and 8-chip (32x8) regular.

Which brings me to a question to the OP: how many chips has your module?
It might be of the High-density variety vs. the (required) Low-density!

iann070 · #24 Post by **iann070** » Fri Apr 20, 2007 7:01 am

RealBlackStuff wrote:Which brings me to a question to the OP: how many chips has your module?
It might be of the High-density variety vs. the (required) Low-density!

It's double sided with 8 chips per side, and as it has an IBM sticker on it with their part number FRU:33L3070 printed on it, then surely it must be low density.

ludox · #25 Post by **ludox** » Thu Apr 26, 2007 5:07 am

david_sa wrote: Try a bios upgrade, it could enable some pc133 compatibility. My T21's bios version: 1.16 (kzet34ww)

That seem an interesting response.

I just bought a second (or more) hand T21 and want up my 128Mo to at least 384Mo and it appears that I must be carreful for choice of module.

iann070 could say if his bios is the last one ?

Another question : did the 512Mo limit real ? Or 512Mo modules could be recognized ?

Purcy · #26 Post by **Purcy** » Thu Apr 26, 2007 7:04 am

ludox wrote:
david_sa wrote:

Another question : did the 512Mo limit real ? Or 512Mo modules could be recognized ?

Hello Ludox, welcome to the community. If what you are asking is if there is a 512MB limit for the T21, then yes, that is correct. 512 is the highest size of RAM you can use in that model.

#27 Post by **RealBlackStuff** » Thu Apr 26, 2007 11:35 am

Or to be exact: 2 x 256MB modules.

iann070 · #28 Post by **iann070** » Fri Apr 27, 2007 3:00 am

ludox wrote:
david_sa wrote: Try a bios upgrade, it could enable some pc133 compatibility. My T21's bios version: 1.16 (kzet34ww)
That seem an interesting response.

I just bought a second (or more) hand T21 and want up my 128Mo to at least 384Mo and it appears that I must be carreful for choice of module.

iann070 could say if his bios is the last one ?

Another question : did the 512Mo limit real ? Or 512Mo modules could be recognized ?

Yes, BIOS is the latest. Windows System Info says
BIOS Version/Date: IBM KZET34WW (1.16 ), 28/04/2004

I already sent the RAM back for an exchange so will see what I get back.

ludox · #29 Post by **ludox** » Fri Apr 27, 2007 3:40 am

Ok,

So I search also a memory module.

This that you try had 16 chips or 8 chips ?
Double sided ?

Edit : answer is up. Sorry.

PC100 that is actually PC133

PC100 that is actually PC133

Who is online