Soltek SL-65KV2 is socket 370? Must be something wrong with model number you've given...

Other than that - all memory runs at the same frequency, so what's probably happening is that frequency is reduced when you install 2 sticks for whatever reason. That obviously reduces performance. You can override it by configuring memory settings manually in the BIOS if it allows that. But be careful and do run memtest just to be sure - memory errors are most annoying thing ever.

My mistake, it is actually the Soltek SL-75KAV Socket A motherboard.

Are they actually identical, or do they just look identical?

Ie, if you have only one in at a time, does a program like Sandra or CPU-Z report the same parameters under memory SPD for each one?

I have seen sticks in the past that looked the same, memory chip numbers and all, but then actually had different SPD data programmed. It could be a different batch, it could have been intended for sale at a lower price (for the same product), or it could have not passed to be marked as a 133 (unlikely but still a possibility I guess).

There should be somewhere in bios that has a ram speed setting (like auto - manual), or possibly "set by SPD", try turning this to manual/off and see what that does. If it solves the frequency issue but instability pops up try to set cas to 3 instead of 2, as most sticks are fine at running cas2 at 100mhz but may not be stable that way at 133 - SPD would usually take care of that but obviously not if disabled

https://www.techpowerup.com/forums/threads/sp … orys-spd.20349/
The link to the program itself is dead.
But you can find it on the Internet.
If both strips work at 133 MHz, then reprogram the SPD from the working one to the other.

Locked SPD - only for genuine Samsung and Hynix.

I've realized that the 8,2xx 3dmark 03 score was before I broke an antec power supply yesterday by setting the dip switches to 140 fsb instead of 133 fsb (a tiny overclock, but it was enough to break an old Antec true power 750w PSU with a 24 pin to 20 pin adapter attached. I'm now using a different weaker old ASUS power supply and its scoring 7,900 with either stick and with both sticks in.. both running at 133. I think that perhaps the much weaker 250w PSU is limiting the performance. Thoughts?

Socket A systems famously didn't do any thermal throttling, so power delivery shouldn't affect performance: either it's enough and stable or it's insufficient and unstable

If you were messing around with settings overclocking stuff, I'd far sooner expect that any difference in scores is due to different settings at the different times you benchmarked.

A score of 8212 again, I tried the manual instead of auto DRAM settings in the BIOS and choosing 133 (the only option). I also enabled bank interleave [2]
CPU-Z was already reporting the ram running at 133mhz. Strange indeed but I'm glad it's working properly now.

The system never booted at 140mhz fsb (it broke the PSU on the first attempt immediately after changing the dip switches) , the original 82xx score was at 133fsb/2000mhz just like today.

Elixium wrote on 2025-05-17, 15:12:

A score of 8212 again, I tried the manual instead of auto DRAM settings in the BIOS and choosing 133 (the only option). I also enabled bank interleave [2]
CPU-Z was already reporting the ram running at 133mhz. Strange indeed but I'm glad it's working properly now.

The system never booted at 140mhz fsb (it broke the PSU on the first attempt immediately after changing the dip switches) , the original 82xx score was at 133fsb/2000mhz just like today.

I have had a random (I think an Asus slot1) board on which if you messed with the settings and on the next boot something happened preventing it from posting successfully it would silently set some settings to default. An unsuccessful overclock would do it, but I remember accidentally pulling the plug during a boot and it doing the same thing, almost sounds like something like that may have went on here

As for the power supply failure- I find it really unlikely that being the result of an overclock. Those things are protected and usually withstand even a dead short. It probably just happened to fail during power on...just as they usually do