EU Microsoft Project Olympus Dual Epyc 7003 Boards - 87,12€ each
- Thread starter Prof Theorie
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
under Linux : as root: flashrom -r bios-bak.rom
exact syntax can vary (depend on flash chip) please more advanced user help here (fast backup in case no spare bios chip laying around ).
hope this works for AMD, but at Intel system you can not get full flashable dump with this.
use hardware USB programmer to be safe.
will do tomorrow a dump with xprog and upload it here.
Decided not to wait....
Fire the one (the right one looking from the back)
mega.nz
Fire the one (the right one looking from the back)
3.64 MB file on MEGA
don't waste your time to get the orginal chip w25q256FV (EOL)
the replacement is w25q256JV
the replacement is w25q256JV
Decided not to wait....
Fire the one (the right one looking from the back)
3.64 MB file on MEGA
Code:
BIOS Date: 04/23/2020 10:19:02 Ver: T6U_3A25
Olympus Epyc W25Q256FV(OTP)@SOIC16 (right).BIN (1/1)
╔════════════════════════════════════════════════════════════════╗
║ AMD ║
╟───┬──────────┬──────────┬────────────┬───────┬──────────┬──────╢
║ # │ CPUID │ Revision │ Date │ Size │ Offset │ Last ║
╟───┼──────────┼──────────┼────────────┼───────┼──────────┼──────╢
║ 1 │ 00800F12 │ 08001230 │ 2018-08-04 │ 0xC80 │ 0x26FE00 │ No ║
╟───┼──────────┼──────────┼────────────┼───────┼──────────┼──────╢
║ 2 │ 00800F11 │ 08001129 │ 2017-07-14 │ 0xC80 │ 0x270B00 │ No ║
╚═══╧══════════╧══════════╧════════════╧═══════╧══════════╧══════╝
BIOS Date: 05/15/2018 16:09:29 Ver: T6U_3A15
Olympus Epyc W25Q256FV(OTP)@SOIC16 (left).BIN (1/1)
╔════════════════════════════════════════════════════════════════╗
║ AMD ║
╟───┬──────────┬──────────┬────────────┬───────┬──────────┬──────╢
║ # │ CPUID │ Revision │ Date │ Size │ Offset │ Last ║
╟───┼──────────┼──────────┼────────────┼───────┼──────────┼──────╢
║ 1 │ 00800F12 │ 08001227 │ 2018-02-09 │ 0xC80 │ 0x26FE00 │ No ║
╟───┼──────────┼──────────┼────────────┼───────┼──────────┼──────╢
║ 2 │ 00800F11 │ 08001129 │ 2017-07-14 │ 0xC80 │ 0x270B00 │ No ║
╚═══╧══════════╧══════════╧════════════╧═══════╧══════════╧══════╝
Last edited:
better is.
i guess they replaced the the BIOS with the initial Naples only before decommissioning...
You mean only naples is supported? I'm too tired to dig into it.
Will need more dumps... maybe other buyer's/user's boards come with milan version.
Last edited:
Got the server with case. Tried to connect power to the power supply for testing by connecting the individual pins using connector like this (not these exact ones need to use connectors which are flat enough that the can fit next to each other) "https://www.reichelt.de/de/de/shop/...MI28vWxN_ekAMVj7CDBx27fT1vEAQYAiABEgKf6_D_BwE" The connectors need to be compressed a little bit using thongs to fit the quite thin Pins (the are designed for 0.8mm thickness pins, but psu pins are more like 0.3-0.4mm).
I connected all three phases (only feed A not backup so 3* 2 Pins) of the power supply to a single phase plus the GND. However I can not get 12V output on the powersupply. There is something happening (Relais clicking) after plugging and unplugging and it is drawing some power and warming up slightly (ca. 45 Watts) but there is no 12V on the output. Anyone tried something similar or got the PSUs to power on?
I connected all three phases (only feed A not backup so 3* 2 Pins) of the power supply to a single phase plus the GND. However I can not get 12V output on the powersupply. There is something happening (Relais clicking) after plugging and unplugging and it is drawing some power and warming up slightly (ca. 45 Watts) but there is no 12V on the output. Anyone tried something similar or got the PSUs to power on?
Last edited:
My guess -
You might be on the right track, there's just more to do.
The 45w is probably ~30w for battery charging + other "always on" components.
You'll get more data after several hours of charging I think.
I think we'll have to figure out the data pins on the PSU connector as well - I suspect the node needs to be told to turn on as it would have been told to turn off before being decomissioned (otherwise, how would the batteries arrive with a good charge?)
Yeah, there is a BLADE_ENABLE Pin on the data lines. First image is PSU sheet, second is from the Rack Manager. Maybe the analog ground pin is the reference point? It being called analog trips me off though.
Attachments
Havok - I was actually just working on a follow-up 
I based my research on the PMDU spec, I see your work is based on the rack manager which is awesome since it all seems to match as far as I can tell...
See pages 29-30 in the link above for more info, but I'll try to summarize it quickly (there's actually two variants because feed A is the primary feed for one half of the rack and feed B for the other).
You could also check out the rack manager block diagram in page 15.
I am pretty sure there is the PSU <-> PMDU connector is not as interesting as we thought, but we should get the PSU to turn on.
Other than power of course (GND + 2x 3-phase feeds), the data pins do:
So, to get 12v out of the PSU, I believe we'd need to:
Pull pin A1 (LR_SELECT) - HIGH, I believe 3.3v, could be 5v though
Pull pins A3 (analog ground), C2 (blade enable), D1 (PSKILL) - LOW (GND)
You could also pull C1 (Blade throttle) high, but I believe there's an existing pull-up in place
Please double check my work and stay safe if anyone tries this. Remember that power electronics can be very very dangerous!
(I edited since I had a bad guess regarding LR_SELECT earlier, sorry!)
I based my research on the PMDU spec, I see your work is based on the rack manager which is awesome since it all seems to match as far as I can tell...
See pages 29-30 in the link above for more info, but I'll try to summarize it quickly (there's actually two variants because feed A is the primary feed for one half of the rack and feed B for the other).
You could also check out the rack manager block diagram in page 15.
I am pretty sure there is the PSU <-> PMDU connector is not as interesting as we thought, but we should get the PSU to turn on.
Other than power of course (GND + 2x 3-phase feeds), the data pins do:
- 6 pins dedicate for blade/node (slot) ID (PMDU provides this signal)
- Blade/node presence (node indicates it is in the slot, active low)
- Blade/node enable (PMDU tells node it is allowed to power-on, but possibly different from BMC level power control, active low)
- Blade/node throttle (PMDU provided signal, I believe they support setting a power throttle per node, to fit more nodes within a limited power budget, active low)
- PSKILL signal (short pin, I believe meant to make sure nothing bad happens during insertion/removal, active high)
- Analog ground - no idea
- LR_SELECT - the rack manager spec accounts for left and right entities within the same blade. I believe this is not important for this board.
So, to get 12v out of the PSU, I believe we'd need to:
Pull pins A3 (analog ground), C2 (blade enable), D1 (PSKILL) - LOW (GND)
You could also pull C1 (Blade throttle) high, but I believe there's an existing pull-up in place
Please double check my work and stay safe if anyone tries this. Remember that power electronics can be very very dangerous!
(I edited since I had a bad guess regarding LR_SELECT earlier, sorry!)
i simply grep'd strings from the binaries, and could only find a reference to Naples
>> strings olympus_bios_right.bin | grep Naples
NaplesPI-SP3 1.0.0.B
thats an old AGESA. grepping Rome or Milan brings up nothing.
let's hope someone else has a newer bios. i'm gonna ask a colleague of mine if i can borrow his eeprom flasher.
out of interest, what tool did you use to get those tables?
btw, there are three revisions / dates on the tables. Are those three different AGESA versions or how should I read those things?
tbh, we also should be able to ask pio? They tested the boards and advertised them as 7003 Milan boards.
Maybe they got a different bios at hands?
Last edited:
revision is microcode.
MC Extractor
one is backup BIOS different version. these are
they also state VGA onboard which is wrong.
even if Milan work, only PCIe 3.0