DDR3 DDR4 RAM price trends thread.
- Thread starter BLinux
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
I paid $250 ea for 32gb RDIMMs (not LR) during the spike about couple of years ago and it was a decent price. Going rate for 16gb at that time was $125 and 32 was over $300.
lel, forget about finding reasonably priced DDR4 32GB UDIMMs.
Reminds me of the shipping dock losing over $15k in 64GB dimms. It was sitting under a notebook of procedures for checking in POs.
I picked up 8x16gb (sk hynix HMA42GR7AFR4N - UH) used for USD$280 locally (AU), one of the dimms was dead, seller gave discount and ended up with 7 for $220. Cheapest i can find that particular part to have full set is USD$100 for the single 16gb stick. Cant quite bring myself to spend it though atm.
Last edited:
Be careful when you buying 1st generation 32 GB DDR4 Samsung M393A4K40BB1-CRCxxx from spring 2016! 80% of RAMs can pass Memtest Pro, however, it generates 1-bit corrected error. This testing one by one is killing me.
Result summary
Result summary
| Test Start Time | 2020-08-27 02:04:56 |
| Elapsed Time | 23:31:09 |
| Memory Range Tested | 0x0 - 2080000000 (133120MB) |
| CPU Selection Mode | Parallel (All CPUs) |
| ECC Polling | Enabled |
| # Tests Passed | 56/56 (100%) |
| ECC Correctable Errors | 630 |
| ECC Uncorrectable Errors | 0 |
| Test | # Tests Passed | Errors |
| Test 0 [Address test, walking ones, 1 CPU] | 4/4 (100%) | 0 |
| Test 1 [Address test, own address, 1 CPU] | 4/4 (100%) | 0 |
| Test 2 [Address test, own address] | 4/4 (100%) | 0 |
| Test 3 [Moving inversions, ones & zeroes] | 4/4 (100%) | 0 |
| Test 4 [Moving inversions, 8-bit pattern] | 4/4 (100%) | 0 |
| Test 5 [Moving inversions, random pattern] | 4/4 (100%) | 0 |
| Test 6 [Block move, 64-byte blocks] | 4/4 (100%) | 0 |
| Test 7 [Moving inversions, 32-bit pattern] | 4/4 (100%) | 0 |
| Test 8 [Random number sequence] | 4/4 (100%) | 0 |
| Test 9 [Modulo 20, ones & zeros] | 4/4 (100%) | 0 |
| Test 10 [Bit fade test, 2 patterns, 1 CPU] | 4/4 (100%) | 0 |
| Test 11 [Random number sequence, 64-bit] | 4/4 (100%) | 0 |
| Test 12 [Random number sequence, 128-bit] | 4/4 (100%) | 0 |
| Test 13 [Hammer test] | 4/4 (100%) | 0 |
| Last 10 Errors |
| [ECC Error] Test: 13, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 13, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 13, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 13, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 13, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 13, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 13, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 12, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 12, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
| [ECC Error] Test: 10, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: Yes, Syndrome: N/A, Channel/Slot: 0/0 |
Could this be a system compatibility thing? I've had motherboards in the past that would do this with some memory modules (ECC corrected errors), while the same modules worked fine in other boards.
I don't think so. I have modules with the same part number from another source produced in late 2017 and it work smoothly.
Faulty modules came from local datacenter, working modules from this Greek guy. I have Asus Z10PA-D8 and lot of Haswell/Broadwell Xeons.
faulty PN:
M393A4K40CB1-CRC4Q, dates: 1551, 1607, 1616, 1617; IDT buffer (Lenovo P/N: 00NV205 OPT: 46W0833 FRU: 46W0835)
working PN:
M393A4K40CB1-CRC4Q, all have manufacture date 1709, Montage buffer (HP: HPE PN: 809083-091, replace PN: 819412-001)
Faulty modules came from local datacenter, working modules from this Greek guy. I have Asus Z10PA-D8 and lot of Haswell/Broadwell Xeons.
faulty PN:
M393A4K40CB1-CRC4Q, dates: 1551, 1607, 1616, 1617; IDT buffer (Lenovo P/N: 00NV205 OPT: 46W0833 FRU: 46W0835)
working PN:
M393A4K40CB1-CRC4Q, all have manufacture date 1709, Montage buffer (HP: HPE PN: 809083-091, replace PN: 819412-001)
Well, different registers/buffers on these, that's exactly the kind of problem I'm talking about and I've had in the past. Samsung modules (that was DDR3, though) with identical P/N worked fine only if they had Inphi registers. Same modules with IDT registers were all showing ECC corrected errors. That was on Gigabyte GA-7PESH2 boards. But I had issues like this with at least a couple boards in the past, afaik with an Asus board as well (Z9 generation).
One of the reasons I stay away from Asus server boards is because they're quite picky when it comes to RAM. Not saying it's 100% your case, but might be worth checking those modules in another brand's system, like Supermicro or something. There's a chance they're not faulty after all.
One of the reasons I stay away from Asus server boards is because they're quite picky when it comes to RAM. Not saying it's 100% your case, but might be worth checking those modules in another brand's system, like Supermicro or something. There's a chance they're not faulty after all.
Remember if you have a Skylake Xeon or newer system you can repair the faulty memory modules. Was built in to many systems. Post Package Repair. If you have a system running you can test it with soft PPR and then change to hard PPR when you are confident its fixed. Then export the memory to an older system.
It's not just Asus boards, had the exact same problem with Gigabyte dual 2011 boards as well, and somewhat similar issues with other brands in the past. Afaik memory compatibility depends on the way the BIOS is done. There are things in there that the board manufacturer has to program: sub-timings, for example - even the most overclocker-friendly desktop boards don't always allow changing every single sub-timing there is, and set some automatically; server boards generally don't allow playing with timings at all. So some set those to safer values, and some might mess it up a little bit so not every module out there will be stable or even boot. There are probably other things in there that are different between brands/boards. Some brands code their BIOSes better than the others, and this results in better memory compatibility. I think I've had some sort of memory compatibility issues with intel s2600ip as well, those board just didn't like Micron 16gb modules with IDT registers that I had at the time, wouldn't boot with them at all (or rather they would boot, but the memory slots had red/orange LEDs next to them, and those modules weren't recognized). The same modules worked fine in supermicro x9drd boards. There could also be electrical differences, different layouts of dimm slots on the board could result in some modules not working properly, I guess.
I don't know, I thought memory compatibility was a known thing for decades, there are lists of modules that are guaranteed to work posted for pretty much every motherboard out there, on the official board maker websites. They wouldn't be posting those if memory compatibility was not a thing. While most "not on the list" modules still work fine, it's not always the case. And I've had the exact same problem as JanCerny before, with ECC corrected errors in memtest (all produced 0 "regular" errors, only those 1-bit ecc corrected). Thought the modules were faulty at first, but then they passed all the tests in other brand's boards.
Dammit I want to see these prices again! Crazy that over a year later prices are nearly double for the same thing.
