Unfortunate. Guess gotta use just 1 8259CL until I can find a pair of socket covers, ahem 4108s.
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
some systems just need standby power(e.g. intel S2600xxx) those who need to be powered need a CPU installed in order to get the VRM controller powered and alive.
if your system does not shut down i a few seconds, you can program with 2x 8259CL,
the VRM controller need powered only.
That's what I ended up doing, it did work but it felt hella sketchy.
Thanks for the great tutorial and software!
I may have spoken too soon, when it gets to Proxmox it takes a long time to boot (after POST) and it gives some errors that boil down to udev timeout and the networking doesn't work, anyone else ran into this with 8259CLs?
is there a solution? and does it indicate a issue with the server or the CPUs?
Does that mean: It already support 255 ?
But when I put 8259cl on the board, I620-g30 can't load into BIOS, and the BMC show the following :
i don't know the board, not the VRM controller (they are all different)
sorry i can' help, use MCP2221a and my app.
so you can give yourself the answer.
I've tried doing this too, but there must be something wrong.
I'm not sure whether the bus where the VRM is located is the same one you're accessing with i2ctool.
You might be accessing a different device instead.
I've successfully applied the mod to a X11DPI-NT v2.00 and can run 2x 8259CL. Phenomenal work @RolloZ170 !
I want to buy and mod more dual socket, EATX boards, but am afraid of limitations that preclude using high power CPUs, especially like
Does anyone know if there are any revisions to avoid for these dual socket boards?
X11DPH
X11DAI
EP2C621D16-4LP
EP2C622D16-2T
I noticed the 2 Supermicro ones don't have mounting holes for VRM heatsink, while Asrock does. But that's not a big issue, especially in servers where air flow is stronger at the front. According to RolloZ170, the bug is random short circuits that causes Intel Agent's power stage (on the CPU ?
) to receive the full 12V.
@jbhoodatt do you know what version of X11DPH you modded? It should be reported by dmidecode.
@RZSN @xiaguanbo what version of X11DAI you used?
I want to buy and mod more dual socket, EATX boards, but am afraid of limitations that preclude using high power CPUs, especially like
https://forums.servethehome.com/index.php?threads/xeon-8259cl.33334/post-342553
Does anyone know if there are any revisions to avoid for these dual socket boards?
X11DPH
X11DAI
EP2C621D16-4LP
EP2C622D16-2T
I noticed the 2 Supermicro ones don't have mounting holes for VRM heatsink, while Asrock does. But that's not a big issue, especially in servers where air flow is stronger at the front. According to RolloZ170, the bug is random short circuits that causes Intel Agent's power stage (on the CPU ?
) to receive the full 12V.@jbhoodatt do you know what version of X11DPH you modded? It should be reported by dmidecode.
@RZSN @xiaguanbo what version of X11DAI you used?
Last edited:
all X11DPH have 6+1 power stages for first CPU, you are fine with it.
if the X11DAi does NOT have no VRM heatsink holes, should not happen because 'A' = WS class motherboard, don't buy old rev.
only X11DPi up to vev. 1.21 is affected by that VRM bug. if you run the baord in a server chassis you are fine with cooling
(most want to run them in PC/WS case without continous VRM cooling)
Whoops, I was looking at a blurry ebay stock photo that didn't have VRM holes.
I've taken a sample of board revisions I see from pictures. It could be biased due to many ebay listings using generic photos.
I don't see any major version change so that seems to rule out a grievous mistake like sending 12V to the CPU. Thanks for confirming they're OK @RolloZ170
x11dai
1.02 USED - Supermicro X11DAi‑N LGA3647 DDR4 E‑ATX Motherboard w/ IO Shield | eBay
1.01 ONE New X11DAI-N Via DHL or FedEX | eBay
1.02 1PCS used X11DAi-N Via DHL or Fedex | eBay
x11dph:
1.1 Supermicro X11DPH-i LGA3647 C621 Server +2x Intel Xeon Gold 6138 1.8Ghz 20c/40t | eBay
1.1 DEFECTIVE Supermicro X11DPHT LGA 3647 EATX Server Motherboard 2x 4114 10 G3D | eBay
1.1 Supermicro X11DPH-i LGA3647 C621 Server ＋Intel Xeon Gold 6138 1.8Ghz 20c/40t | eBay
I found the I2C VR header on I620-G30 too. There are 3 pins, so how to connect them?
And the voltages from right to left and middle to left are both around 3.5V.
Last edited:
1st, find which pin is ground with a multimeter - the clock and data lines float high. Then just use trial and error (try both ways) for the clock and data line. There's no risk of a short circuit because the maximum current an I2C master or device can sink is limited by the resistors.
But you need to check if the idle voltage on the clock and data lines is between the allowable ranges of the I2C master you use (e.g. 3 to 5.5V for MCP2221A). If it's less, then it won't work. If it's more, then the I2C master will be destroyed.
But you need to check if the idle voltage on the clock and data lines is between the allowable ranges of the I2C master you use (e.g. 3 to 5.5V for MCP2221A). If it's less, then it won't work. If it's more, then the I2C master will be destroyed.
Last edited:
OK, then that means the left pin is ground. On my X11DAI, SDA & SCL are 3.3V and so is SCL & SDA on my MCP2221A. Hmm, that's unexpected. I was expecting only 1 of them to be pulled up. Then I guess it's only safe if the idle voltage are within ~0.5V of each other. The I2C voltage must not exceed the supply voltage of any device on the I2C bus, or you could get either temporary or permanent latch up - a short circuit that makes the device very hot.
Last edited:
SCL and SDA are unused pulled up to 3.3V suppy (if the supply is 3.5V you see that)
just try a scan: 50/50 chance.