Updating Wiwynn Open compute bios for E5 V2's Edit. Gave up trying. :(

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
=======================================================================
EDIT: Gave up trying to get it to work, even with the help from a couple of bios guru's from several bios modding sites, the best I was able to do is update the V1 microcode to a much newer version.

So I recommend if you want to run V2 Xeons buy a Quanta version.
=======================================================================


The reason I want to do this is some of the V2 Xeons are almost as cheap as the V1's now. I know I could just buy a used Quanta's for a little over $100 off of ebay but since I have three Wiwynn's I would save money if this works.

Plus i'm bored

I am using a cheap CH341A USB programmer and the flashing program on a Windows 7 pc.

First try: The easy way, flash a Quanta open compute BIOS to a spare W25Q128BVFG bios chip.

Result: Did not work, results in a endless reboot.

I was not really expecting it to work but hoped it would since the Quanta and Wiwynn are supposed to have been built to the same open compute standard.

Next is to add the V2 microcode to the latest Wiwynn bios image.

Its been years since I modded a bios so there is some relearning to do.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Little update, went to the Win-Raid forum and read some how-to's .

First test is to update the V1 cpu microcode to a newer version just to see if updating the microcode will work.

Flashing now....

Oh and I do not currently have any V2 cpu's so i'll need to order a couple of cheap E5-2603 V2's to test it out.

EDIT: That did not work, powered up and the fans went full on and no boot.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Looks like my cheap usb programmer crapped out or I messed up both spare chips.

Started having issues in Windows so I went and stuck it in my Ubuntu box since Flashrom in Linux was able to see it but it no longer does.

I tried to flash the latest offical bios back on two chips to see if it was my modded bios causing the earlier issue but it does not detect it any longer.

Code:
sudo flashrom --programmer ch341a_spi -w EP_P09RG.02.bin
flashrom v0.9.9-r1954 on Linux 4.18.0-17-generic (x86_64)
flashrom is free software, get the source code at https://flashrom.org

Calibrating delay loop... OK.
No EEPROM/flash device found.
Note: flashrom can never write if the flash chip isn't found automatically.
So until I order another flash programmer and some extra bios chips tonight i'm stuck.


EDIT: I just pulled a bios chip out of a working open compute node to test to see if its the bios programmer or the chip, its definitely the programmer.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
I just ordered a pack of five new bios chips.

Instead of messing with a cheap universal programmer I found one that is a SOIC16 SOC16 programmer with the included socket but will accept other adapters for other style chips.

It was about $40.00 but it comes with a software package that looks nicer than the ones you have to download of the internet like the cheap programmer.

Hopefully I can flash the bios with less of a hassle.
 

Attachments

Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
I found some new information on the CH341A programmer I have, it seems the one that is black and gold supply's the full 5 volts from the usb strait into the chip to be flashed.

My bios chips are are rated 3.3 volts so that might explain the issues I have been having reading a chip after its been flashed a few times.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
I received the programmer Monday but they shorted me the "Hat" with the socket, so I contacted them and they shipped it out and I received it today.

Really nice quality and the flashing program is MUCH better than the ones for the cheap programmer.

The first thing I did was see it it would read and write to the chip that the first cheap programmer would no longer detect.

It sees it fine and I am able to erase it and flash it.

So I took the latest wiwynn bios that I had downloaded and flashed it, did not work.:(

Fans would go to %100.

So I reflashed the spare bios chip with the copied bios from the good chip just to make sure it works and i'm not doing anything stupid.

And it gives me a good bios as a baseline to mod.

Success !!

The first step is done, flashing the unmodded bios to a chip to make sure it works and it does.:)

Now to get up to speed on modding it buy adding the V2 microcodes and see if it still boot on a V1 cpu since I want to be able to use either version.

Edit: I am taking the time to verify everytime I flash. Also still have not recieved the 5 chips that I ordered. I plan on having one modded and unmodded bios chip for each node incase I need to go back to a factory bios.
 
  • Like
Reactions: JSchuricht

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Seems like the winbond 25Q128BVFG bios chips are actually one Cypress S25FL129P and the other shows up as a MXIC MX25L128 marked as a Winbond chip.o_O

When I plug in an original open compute bios chip the programing program shows it as a winbond but when I plug in the replacements the program shows Cypress or MXIC.

So maybe I have some fake chips. :eek:

On second thought maybe they are real and the cheap programmer damaged them and it confuses the new programmer program.

I have zero issues with the original chips but these replacements I bought months ago seem to have issues either do nothing or the fans go %100o_O


So I'll wait on the 5 I bought that are due in Monday and hopefully they are good to go.
 
Last edited:

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
I robbed a bios chip from an unused node so I can keep at it.

First test: Flashed the latest quanta bios and gave it a go, did not work.

Was not expecting it to but I wanted to be sure because if it would have it would be the easiest way to run V2 cpu's.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Second test:

Add the V2 cpu microcode to the baseline bios .bin file and make sure it boots.

Success !!

Boots with two E5-2620 v1 cpu's.

Why V1 cpu's? well because I want to test the bios to see if it would work without changing anything else so if it did not then it means I am doing something wrong plus I do not have any V2 cpu's handy.

Next is to add a hard drive and install Ubuntu server 18.04 and run some tests and if that works I'll go ahead and order a couple of cheap V2 cpu's.

So far so good. :D
 
  • Like
Reactions: Mehmet

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Just to add, every-time on these open compute servers whenever I have flashed a bios either with a chip programmer or via booting into windows 7 and using a windows program to flash after the reboot it gives an error "Bios checksum error .....rebooting" then it reboots with out error the second time.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83


Just a quick install with Ubuntu 18.04.1 since that was on a handy flash drive. May give 19.04 server a go instead of updating this install.

Output of lscpu:



Seems like i'm half way there.:)
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
I ended up just installing Ubuntu server 19.04 to try it out.

Also I placed a couple of low bids on some low end V2 cpu's on ebay.

Hopefully I will win them at a low price.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Still had to do " echo "options mei-me disable_msi=1" | sudo tee /etc/modprobe.d/mei-me.conf" so it wont hang on reboot but I expect that on every Ubuntu install.

19.04 seems to run ok, ill keep testing with the V1 cpu's until I receive some V2's.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Well I just won on ebay four E5-2620 V2 cpu's for $25.00, thats $6.25 each and about $1 or so cheaper than the cheapest E5-2603 V2 listed as a buy it now, so that covers one server. Supposed to be in next Saturday.
 
  • Like
Reactions: JSchuricht

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Small update, the node with the updated cpu microcode has been running perfectly.

I went ahead and downloaded the latest microcode, was using an older one that came with the program I am using.

I'm updating the cpu microcode and any other code that there is an update and that turned out to be the network chipsets.

Will flash shortly and test.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Well that worked so far.

Changed out the bios chip that was flashed with the updated cpu microcode, well updated up til mid 2018 dated microcode, with the latest cpu microcode from intel plus the latest code for the network chips.

Those were the only available updates but now it has the latest cpu microcode that has all the current fixes.

I still not learned how to change the displayed version or the date so it shows the same version number and date as the last official bios update, not a big issue but it would be nice to have it display a more current date.

My procedure:

I shutdown the server then pull the power plug then unlatch the node then I slide the node out a few inches and remove the video card I have been using, then remove the cr2032 battery then open the bios chip socket then swap with the latest flashed version.

Next I reinsert the battery then install the video card then slide the node in and latch it then power it up.

At first boot it complains about a bios checksum mismatch then it reboots, now it shows the bios splash screen without the error and then I go into the bios and load the "optimum settings" in the "boot" page screen then exit and reset.

I next go back into the bios and set the date and time and go through the various bios pages to make sure everything looks like I want it and also change the boot order to boot from the hard drive first.

Then save and reset and let it boot into the OS.

I am running Ubuntu 19.04 server, and to test it a little without heating up my house I installed "cmatrix" and fired it up for some eye candy while I let the server run.

Only thing left is to receive my E5-2620 V2 cpu's and make sure it will run.

As soon as I announce it does work with V2 xeons and if anyone wants a copy of the bios to test just pm me.

I expect most of these servers are just mostly used in home labs since they are cheap and outdated and most would be happy to have the capability to run V2 cpu's for more cores and less power use.

After I do that i'll still need to learn how to change the date and version number on the bios.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
I received my four E5-2620 V2 cpu's and installed them into one node with the updates bios.

Does not boot.:(

Thats the bios with all the updated microcode, so I am flashing the first bios I made with the older V2 microcode and with out the other updates.
 

Klee

Well-Known Member
Jun 2, 2016
1,289
396
83
Next is to try with only the latest V2 microcode, removing the V1 codes.