VOGONS


First post, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t

I've been sorting through my huge pile of SIMMs and DIMMs. Many of the sticks aren't labeled with their speed or capacity, so I'm attempting to identify them. The SIMMs are generally easy to figure out... its the 168pin DIMMs that are quite baffling.

I'm finding a surprisingly wide range of speed ratings on the SDRAM chips, and some brands (Samsung) don't even seem to have the ns speed rating on their DRAM chips, at least not in the way others do.

On the sticks marked PC-100 I'm finding markings like these printed near the end of the long part numbers: 8, 8A, 8PC, and on a Samsung "128MB 100Mhz CL3" labeled stick the chips say "022 K4S280832A-TC1L" ... then on a Hyundai PC-100 labeled stick "TC-20S" and on a generic looking "ATL PC100" stick and an "IBM FRU" PC-100 stick there are NEC chips that say "-A10-9JF" toward the end of their part numbers.

I have a ton of un-labeled very low capacity (16MB to 32MB) sticks that have chips with markings like: 10, 10S, 10K, TC-10, T10, G12 (Samsung) and a Samsung stick that has a label that says "32MB SDRAM 9ns" with chips that say "-G10" and don't have a 9 anywhere on them. 😕 10ns should equal 100Mhz... but why do the PC-100 labeled sticks appear to have 8ns chips? The ones marked with "12" may even be 83Mhz rated? Then I find actual 83Mhz rated SDRAM and the chips say -10 on them. 🤣

Marked PC-133 sticks generally say: 7, 75, 7G, 7.5, 7H, and some Hyundai "PC-133 marked" sticks who's chips have no recognizable speed markings, just "HY57V64820HG 005ITA T-H" , and a Kingston ValuRAM PC-133 256MB stick with Hynix chips that say "244A A HY57V56820BT-H". 7.5ns should be 133mhz, so I guess these aren't too mysterious. Are the 7Mhz ones more likely to provide overclocking headroom?

I also have one PNY stick with Toshiba chips that have markings that look more like those of EDO chips "tc51v17805BNTS-60". The 16 chips also only have 28 pins despite being flat like the SDRAM chips on other DIMMs. What kind of board would use 60ns EDO RAM on a 168pin DIMM?

Is there any rhyme or reason to this? I'd love to be able to look at a DRAM stick and identify the capacity and speed capabilities. I've mostly gotten to that point with 30pin and 72pin SIMMs and DIPs, but DIMMs are just all over the place.

Any insight on this situation would be much appreciated. I genuinely want to know how this works.

Also, I was reading recently about "SPD", and I was surprised to see that DIMMs have had this kind of information stored in those tiny little single chips all along. Is there a reliable DOS program that will always 100% read the information contained within these chips? Something like this would certainly help identify preferred RAM settings.

Now for some blitting from the back buffer.

Reply 1 of 5, by shamino

User metadata
Rank l33t
Rank
l33t

Keep in mind that the timings of SDRAM are complex. It's not a fixed cycle time for every mode of operation.
Try looking up some datasheets for the RAM ICs. That will clear up the meaning of the speed markings.

I've only seen -10 rated SDRAM chips on PC83 modules. I don't remember if they can also handle PC100 CL3 - but I don't think anybody ever marketed PC100 modules with that latency.
-8 chips are good for PC100 CL2, but not PC133.
-75 chips are good for PC133 CL3 or PC100 CL2.
-7 chips are good for PC133 CL2. These are hard to find.

Sometimes modules will have faster chips than what the module is advertised for. For example, later PC100 modules typically have -75 chips on them, which means they are perfectly fine at PC133 CL3.

Reply 2 of 5, by lazibayer

User metadata
Rank Oldbie
Rank
Oldbie

You can google datasheets for the chips on the RAM sticks and they will tell you the speed of the chips. You can read or even reprogram the SPDs with SPDtool under windows. Some 440FX boards, some Macs, and some VIA socket 7 chipsets take 168 pin EDO DRAMs.

Reply 3 of 5, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

These are hard to find.

For example, most late Micron memory is PC133 CL2 capable, even if markings say otherwise. Also I have two M.tec sticks with 6ns memory which are capable to run at 150mhz CL2.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 4 of 5, by shamino

User metadata
Rank l33t
Rank
l33t
The Serpent Rider wrote:

These are hard to find.

For example, most late Micron memory is PC133 CL2 capable, even if markings say otherwise. Also I have two M.tec sticks with 6ns memory which are capable to run at 150mhz CL2.

Almost every Micron IC I've seen on PC100 and PC133 modules (other than early PC100 ones) have been marked with the -75 rating. Occasionally I ran into the sexy -7E chips in unexpected places, even on "PC100 only" modules, but they have been a rare find (for me at least). I used to have some 440BX compatible 16Mx8 256MB modules with -7E chips on them, but I apparently sold them off. I really regret that, because they're ideal for making an optimized BX133 system.

I don't think I've tried overclocking the -75 rated chips. In the past I wasn't willing to overclock RAM ICs, but I guess I should try that if the occasion arises. If they have a high success rate then maybe I'm in okay shape - I have several modules with -75 rated chips from Micron, Elpida, and Infineon. I'm guessing Elpida is the weakest of those.
I think in order for me to trust them at a 133CL2 overclock, I'd have to see them pass memtest at 138-140MHz or so. Otherwise I'd be worried about them being on the edge of instability.

Reply 5 of 5, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t

Thanks for the info guys! I looked up lots of datasheets but I found very very few for SDRAM chips. For whatever reason, its much easier to find datasheets for EDO and FP RAM chips.

I'll have to sort through my chips a bit more thoroughly at some point. I actually have several -7 rated sticks, so its nice to hear they are somewhat more rare. It's amazing how hard it is to find PC-150 and other overclocking-focused SDRAM. I remember people using them back in the pre-DDR days for the highest of the high end SDR systems, but the market was so small, its no surprise I've never actually seen a PC-150 stick in person.

Now for some blitting from the back buffer.