Groups | Search | Server Info | Login | Register
Groups > alt.comp.hardware.pc-homebuilt > #43823
| Date | 2025-02-01 07:31 -0500 |
|---|---|
| From | bad sector <forgetski@_INVALID.net> |
| Subject | Re: The build is ON |
| Newsgroups | alt.comp.hardware.pc-homebuilt |
| References | (4 earlier) <1mcqpj1nb8d16322tabso5m9ciivs6ldfb@4ax.com> <3cqcnZlT2-wFoQD6nZ2dnZfqn_SdnZ2d@giganews.com> <vnjvom$3piao$1@dont-email.me> <IJ-cnb0qLO5oPgD6nZ2dnZfqn_GdnZ2d@giganews.com> <vnkt5v$22cn$1@dont-email.me> |
| Message-ID | <XRGdnfjS362MiQP6nZ2dnZfqn_WdnZ2d@giganews.com> (permalink) |
On 2/1/25 05:31, Paul wrote: > On Fri, 1/31/2025 11:32 PM, bad sector wrote: >> On 1/31/25 21:09, Paul wrote: >>> On Fri, 1/31/2025 4:11 PM, bad sector wrote: >>>> I'm also posting about this on the asus rog forum as towbar. >>>> >>>> Yes, it's been about 12-15 years. I had to bury my younger son in the middle of this one, the same one with whom we did two identical builds that last time. Being more than a little scatter-brained I need a Department-of-Corrections addendum here: >>>> >>>> The power comes ON but the fans stop, i.e. a shutdown when the yellow Q-LED light illuminates. That's the second one from inboard and the manual just says "DRAM". It took me MONTHS to get an OK from Asus about my ddr5 ECC cards being approved for this board, another month to actually get them. >>>> >>>> https://www.amazon.ca/dp/B0BH95T7K7?ref=ppx_yo2ov_dt_b_fed_asin_title >>>> >>>> I've reseated them several times, also tried just either one in slot-1 alone. They're otherwise in slots 2 & 4. >>> >>> ( ASUS X870E ProArt Creator WiFi ) >>> >>> ******* >>> >>> The picture seems wrong to me. >>> >>> https://www.amazon.ca/dp/B0BH95T7K7?ref=ppx_yo2ov_dt_b_fed_asin_title >> >> Amazon vendors are almost as bad a alibaba ones for putting up wrong pictures. The description in the ad seems to be what I ordered >> >> Kingston Server Premier 32GB 4800MT/s DDR5 ECC CL40 DIMM 2Rx8 Hynix M Server Memory - KSM48E40BD8KM-32HM >> >> but what they delivered is KSM48E40BD8KI-32HA >> >> Return initiated >> >> What they sent me is shown by Kingston on that x870e ProArt _page_ under the Server-Premier tab as being _therefore_ compatible and >> >> KSM48E40BD8KI-32HA (what they sent me) >> DDR5 4800MT/s ECC Unbuffered DIMM CL40 2RX8 1.1V 288-pin 16Gbit Hynix A >> >> >> >> >> The weird thing is that the KSM48E40BD8KM-32HM that I "ordered" is NOT on that page at all. So did I order a fake but got delivered two lesser but RealKingston cards? >> >> And if these are OK for my board/CPU combo, the what's puttint that light on? >> >> >>> ******* >>> >>> https://www.kingston.com/en/memory/search/model/110021/asus-proart-x870e-creator-wifi >>> >>> "This system supports ECC or non-ECC memory depending on which processor is installed. >>> >>> Mixing ECC and non-ECC modules is not recommended. Please refer to manual or >>> CPU specifications for additional information." >> >> >> The one they sent me is also shown as CPU/Chipset compatible >> >> *** AMD Ryzen 9 7000/8000/9000 Series AMD X870E *** >> >> my CPU is an AMD ryzen 9 9900X >> >> >>> ******* >>> >>> The Kingston part, >>> >>> https://www.kingston.com/en/memory/search/discontinuedmodels?partid=KSM48E40BD8KM-32HM >>> >>> Part Number: KSM48E40BD8KM-32HM Replaced by KSM48E40BD8KI-32HA <=== not important at the moment >>> >>> https://www.kingston.com/datasheets/KSM48E40BD8KM-32HM.pdf >>> >>> The replacement: >>> >>> https://www.kingston.com/datasheets/KSM48E40BD8KI-32HA.pdf >>> >>> The difference between the two is A-die versus M-die of Hynix. >>> >>> I got a picture of a 48GB stick. I was mainly interested in whether the >>> chip markings are all the same (4x8+1x4 as a part of a 72-bit scheme >>> versus 5x8 used on 80-bit setups). >>> >>> KSM56E46BD8KM-48HM >>> >>> https://www.ebay.com/itm/204907615991 >>> >>> https://i.ebayimg.com/images/g/8doAAOSwxLVmvqWS/s-l1600.png >> >> Here's the one that they sent me: >> >> https://trixtar.org/temp-pub-buffer/badram.png >> >> NB. I don't fucking believe it! A do-not-remove sticker >> right ON the chips that need cooling. Is there a special >> oriental version of the Darwin award? >> >> >> >> >> >>> ******* >>> >>> Summary: >>> >>> 1) Check CPU spec is for the ECC ones. >> >> as far as thge Kingston page and what I was delivered yes >> >> >>> 2) Check the Asus Support page for the tested memory table. >>> See if the tested memory table includes mention of the CPU model used to test. >>> If CPU model is never mentioned, then their ECC results (that is, if any are offered), >>> are not valid without the CPU info as some sort of "hint" the combo worked. >> >> On the Asus Ryzen 9 page @ >> https://www.asus.com/motherboards-components/motherboards/proart/proart-x870e-creator-wifi/helpdesk_qvl_memory?model2Name=ProArt-X870E-CREATOR-WIFI >> >> There are only 4 Kingstons shown; they all begin with KF and they're all 2x16gb >> >> So what gives with the conflict with the also Asus mobo page above? >> >> >> >>> 3) When going out on a limb like this, *always* buy a single stick >>> of the smallest capacity regular UDIMM (in this case NOT AN ECC), >>> as test materials for your motherboard verification. This gives you >>> materials for checking the end two slots (one channel at a time) for >>> successful channel or slot operation. >> >> I've never seen such shit in my life! You have to write a freakin' thesis to find a compatible ram card? >> >> >>> Since you got a DRAM code, via the handy state LEDs, even without >>> a SPKR connected, we know it's a-beeping right now, and it's >>> not happy. This means either the BIOS code is too old for the CPU >>> (check Support page for the CPUSupport document and verify versus >>> whatever visible info is offered on the mobo). Or, there is some >>> error in the knowledge about what an AMD CPU supports for ECC sticks. >>> >>> The amd.com site, you may need to do a site search in Google: >>> >>> site:amd.com xxxx CPU >>> >>> and try to get to their spec page. At one time, you could trace down >>> all sorts of useful stuff there, like a PDF for a bulldozer, but >>> the info today is a lot harder to find. >>> >>> Paul >> >> It doesn't look good. I'll send these ones TOO back for now. >> The horror show has been going on since August. > > The RAM you ordered is discontinued, and the RAM that shipped > is the replacement. Hynix apparently switching from "M" die to "A" die, > and the BOM reflects the change on the new product. The "dimensions" > logically, the timings, none of that is changed. > > [Picture] > > https://i.postimg.cc/c43nnCyd/badram-comparison.jpg > > I got pictures of the newer product from here. Both their picture > and your picture, the RAM looks like UTT, when you would expect the > Hynix logo to be on it, if it was tested by Hynix. At this speed, > probably 100% of the chips can make JEDEC speed, and it doesn't > even need a speed test as such. But it should still receive > a stuck-at test (the individual chips may have received that > already as well). In any case, this is the module that ships, > when you order using the older part number. > > https://geizhals.de/kingston-server-premier-dimm-32gb-ksm48e40bd8ki-32ha-a3110692.html > > ******* > > ASUS X870E ProArt Creator WiFi > > AMD ryzen 9 9900X > > > https://www.amd.com/en/products/processors/desktops/ryzen/9000-series/amd-ryzen-9-9900x.html > > (connectivity) > > 2x2R DDR5-5600 Two, dual-sided modules and you are running DDR5-4800. Good so far. > > Using your memory page from Asus, selecting ECC (UDIMM) gets these. > Changing the Hynix die, would not affect the result. > > https://www.asus.com/motherboards-components/motherboards/proart/proart-x870e-creator-wifi/helpdesk_qvl_memory?model2Name=ProArt-X870E-CREATOR-WIFI > > Kingston KSM48E40BD8KM-32HM 32GB N/A 4800 4800 DS SK Hynix 40-39-39-77-116 1.1 1,2 > > ******* > > There is a small table on this page, showing some ECC results. > Some boards having multi-bit ECC, and the worst board, failing to POST. > > They tested with MTC10C1084S1EC56BD1R > > https://www.techspot.com/review/2900-amd-x870-motherboards/ > > This is a hell of a way to sort it all out. > > https://forum.level1techs.com/t/x870e-ecc-support/217870/11 > > "Also for DDR5, there are 2 versions of ECC, 72bit and 80bit. > "Only for buffered DIMMs though (RDIMMs). ECC UDIMMs are all 72-bit." > > "I’m pretty sure if they ran anything, it wasn’t more than > > wmic memphysical get memoryerrorcorrection > > in a windows terminal. Hardly evidence as we’ve seen > Windows reporting ECC and returning 6 when it isn’t working. > Proper test with memtest86 and edac-util/mcelog would solve that." > > "sudo lshw -class memory > > configuration: errordetection=multi-bit-ecc " > > There is no mention of ECC in the BIOS release notes. > > https://www.asus.com/motherboards-components/motherboards/proart/proart-x870e-creator-wifi/helpdesk_bios?model2Name=ProArt-X870E-CREATOR-WIFI > > Version 0225 2024/09/23 \ > Version 0409 2024/09/23 \ > Version 0505 2024/09/27 \ > Version 0605 2024/10/28 \ > Version 0606 2024/11/05 / > Version 0706 2024/12/05 / > Version 1001 2025/01/13 / > Version 1003 2025/01/23 / > > Summary: I don't see anything particularly wrong, except... it doesn't work. > > If the board wanted to post with ECC disabled, it should be able to do that. > > According to the Level1techs chit-chat, it should come up with the ECC disabled > on the first try. It could be throwing an error, because some part of the > memory isn't working (unlikely). It could be throwing an error, because > the BIOS does not understand ECC (QVL denies this, but we can trust QVL > only as far as "starting" is concerned -- the bench tester may not be > doing a thorough job, they're just as interested in "crashing" as in > anything else). > > OK, I'm sitting at your desk now. This is my strategy... :-) > > 1) Pick up a small DDR5 DIMM, non-ECC, UDIMM, DDR5-5600 (margin testing). > A single 8GB would be enough for socket testing. > 2) Plug in DIMM, see if it comes up. > 3) Flash up BIOS if it works or if it does not work. > Depends on BIOS flasher type as to how rash and careless > I would want to be at this point. My ten year old board has a > flasher that does not use CPU/RAM. The AMD boards here, use CPU/RAM and > thus, it is best if the RAM is "good" during the flash up operation... > 4) Now that we have tested as far as we can with the latest BIOS, > power off, pull test DIMM, install two jolly ECC DIMMs and retest. > 5) After ECC DIMMS come up, enter BIOS and "Enable" ECC. Reboot. > Issue the sudo lshw command as described in Level1techs. > > Paul Thanks, I'll get on it Monday but out here in the sticks with only 2 or 3 shops (all with low volumes) it doesn't look good so it's likely to be another delay :-(
Back to alt.comp.hardware.pc-homebuilt | Previous | Next — Previous in thread | Next in thread | Find similar
The build is ON bad sector <forgetski@_INVALID.net> - 2025-01-30 11:36 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-01-30 12:32 -0500
Re: The build is ON Paul <nospam@needed.invalid> - 2025-01-30 14:49 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-01-30 17:21 -0500
Re: The build is ON John B Smith <crasso@nycap.rr.com> - 2025-01-31 15:41 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-01-31 16:11 -0500
Re: The build is ON Paul <nospam@needed.invalid> - 2025-01-31 21:09 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-01-31 23:32 -0500
Re: The build is ON Paul <nospam@needed.invalid> - 2025-02-01 05:31 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-02-01 07:31 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-02-01 08:39 -0500
Re: The build is ON Paul <nospam@needed.invalid> - 2025-02-01 12:52 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-02-05 10:18 -0500
Re: The build is ON Paul <nospam@needed.invalid> - 2025-02-05 12:47 -0500
Re: The build is ON bad sector <forgetski@_INVALID.net> - 2025-02-06 08:49 -0500
Re: The build ..chatGPT scrapped the board and/or the CPU Paul <nospam@needed.invalid> - 2025-02-06 16:01 -0500
Re: The build ..chatGPT scrapped the board and/or the CPU bad sector <forgetski@_INVALID.net> - 2025-02-06 20:20 -0500
Re: The build ..chatGPT scrapped the board and/or the CPU Paul <nospam@needed.invalid> - 2025-02-07 00:02 -0500
Re: The build ..x870e ProArt-Creator in Limbo bad sector <forgetski@_INVALID.net> - 2025-02-28 09:19 -0500
Re: The build ..x870e ProArt-Creator in Limbo Paul <nospam@needed.invalid> - 2025-02-28 10:45 -0500
Re: The build ..x870e ProArt-Creator, Scene-1, take N bad sector <forgetski@_INVALID.net> - 2025-03-04 07:41 -0500
Re: The build ..x870e ProArt-Creator, Scene-1, take N Paul <nospam@needed.invalid> - 2025-03-04 13:37 -0500
Re: The build ..x870e ProArt-Creator, Scene-1, take N bad sector <forgetski@_INVALID.net> - 2025-03-21 12:49 -0400
Re: The build ..x870e ProArt-Creator, Scene-1, take N Paul <nospam@needed.invalid> - 2025-03-21 13:52 -0400
Re: The build ..x870e ProArt-Creator, Scene-1, take N bad sector <forgetski@_INVALID.net> - 2025-03-21 14:39 -0400
Re: The build ..x870e ProArt-Creator, Scene-1, take N bad sector <forgetski@_INVALID.net> - 2025-03-21 16:17 -0400
Re: The build ..x870e ProArt-Creator, Scene-1, take N Paul <nospam@needed.invalid> - 2025-03-21 23:18 -0400
Re: The build ..x870e ProArt-Creator, Scene-1, take N bad sector <forgetski@_INVALID.net> - 2025-03-22 07:52 -0400
Re: The build ..x870e ProArt-Creator, Scene-1, take N Paul <nospam@needed.invalid> - 2025-03-22 13:23 -0400
Re: The build ..x870e ProArt-Creator, Scene-1, take N John B. Smith <crasso@verizon.net> - 2025-03-23 09:09 -0400
The never-ending Asus FUBAR ___Re: The build ..x870e-Creator bad sector <forgetski@_INVALID.net> - 2025-05-05 20:25 -0400
Re: The never-ending Asus FUBAR ___Re: The build ..x870e-Creator Paul <nospam@needed.invalid> - 2025-05-06 03:37 -0400
Re: The never-ending Asus FUBAR ___Re: The build ..x870e-Creator bad sector <forgetski@_INVALID.net> - 2025-05-06 18:13 -0400
Re: The never-ending Asus FUBAR ___Re: The build ..x870e-Creator Paul <nospam@needed.invalid> - 2025-05-06 19:17 -0400
Re: The never-ending Asus FUBAR ___Re: The build ..x870e-Creator bad sector <forgetski@_INVALID.net> - 2025-05-10 09:46 -0400
Re: The never-ending Asus FUBAR ___Re: The build ..x870e-Creator Paul <nospam@needed.invalid> - 2025-05-10 13:58 -0400
csiph-web