On 2017-04-26 at 15:25, Rene Shuster wrote:
Maybe even FSF-endorsed. Great news! Spread the news.
The H8SCM is a nice find, but unfortunately probably not eligible for FSF certification.
The board has an old AGP-based Matrox G200eW GPU. In addition to having a non-free VBIOS [1], the GPU has a "WARP Engine" – a RISC coprocessor core with non-free microcode (loaded by Linux [2]) to do triangle setup operations [3].
So, a couple of blobs that would be rather non-trivial to reverse engineer and replace. The board could surely run headless without them, but the FSF most likely wouldn't certify it since users could be tempted to load the blobs to get video output and graphics acceleration.
[1]: https://www.coreboot.org/Board:supermicro/h8scm#Extract_VGA_BIOS [2]: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/driv... [3]: https://en.wikipedia.org/wiki/Matrox_G200
Is the Asus KCMA-D8 socket C32 motherboard eligible for FSF certification?
On 04/27/2017 07:30 AM, Patrick 'P. J.' McDermott wrote:
On 2017-04-26 at 15:25, Rene Shuster wrote:
Maybe even FSF-endorsed. Great news! Spread the news.
The H8SCM is a nice find, but unfortunately probably not eligible for FSF certification.
The board has an old AGP-based Matrox G200eW GPU. In addition to having a non-free VBIOS [1], the GPU has a "WARP Engine" – a RISC coprocessor core with non-free microcode (loaded by Linux [2]) to do triangle setup operations [3].
So, a couple of blobs that would be rather non-trivial to reverse engineer and replace. The board could surely run headless without them, but the FSF most likely wouldn't certify it since users could be tempted to load the blobs to get video output and graphics acceleration.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 04/28/2017 03:15 AM, PeerCorps Trust Fund wrote:
Is the Asus KCMA-D8 socket C32 motherboard eligible for FSF certification?
As far as I am aware, yes.
- -- Timothy Pearson Raptor Engineering +1 (415) 727-8645 (direct line) +1 (512) 690-0200 (switchboard) https://www.raptorengineering.com
Hey OP so you know I wouldn't hold your breath on this, getting even serial output on this board is being difficult I assume due to the fact that the actual serial port connects to the WPCM-450-M BMC chip.
If you want this for some project or whatever I would pay for a professional to do it (such as tim)
Hi!
Thanks for the feedback on this. It is a decent motherboard, but it wasn't for anything critical. It just seemed like a nice low barrier (cost wise) entry to a desktop. It might be nice however if some of the other Supermicro Opteron boards could be made coreboot capable. Supermicro seems to have a lot of boards that remain in production for quite some time after other vendors have EOL'd theirs.
On 05/02/2017 11:47 AM, Taiidan@gmx.com wrote:
Hey OP so you know I wouldn't hold your breath on this, getting even serial output on this board is being difficult I assume due to the fact that the actual serial port connects to the WPCM-450-M BMC chip.
If you want this for some project or whatever I would pay for a professional to do it (such as tim)
On 05/02/2017 05:05 AM, PeerCorps Trust Fund wrote:
Hi!
Thanks for the feedback on this. It is a decent motherboard, but it wasn't for anything critical. It just seemed like a nice low barrier (cost wise) entry to a desktop. It might be nice however if some of the other Supermicro Opteron boards could be made coreboot capable. Supermicro seems to have a lot of boards that remain in production for quite some time after other vendors have EOL'd theirs.
They have some great boards yeah, their latest were 2014 vs asus's 2010 models and some of them have better onboard nics, great onboard SAS and a dual northbridge setup so more pci-e lanes.
If you have the cash they can be ported, the only issue is the crappy onboard graphics that are text mode only without blobs but they're so terrible you wouldn't be using it for X anyways - they only support 1280x1024 and it is really laggy.
The other issue is that the ASUS boards will be getting OpenBMC in a few months, so there is no point really in buying anything else again unless you have the money for another port.
On 04/27/2017 12:30 AM, Patrick 'P. J.' McDermott wrote:
On 2017-04-26 at 15:25, Rene Shuster wrote:
Maybe even FSF-endorsed. Great news! Spread the news.
The H8SCM is a nice find, but unfortunately probably not eligible for FSF certification.
The board has an old AGP-based Matrox G200eW GPU. In addition to having a non-free VBIOS [1], the GPU has a "WARP Engine" – a RISC coprocessor core with non-free microcode (loaded by Linux [2]) to do triangle setup operations [3].
So, a couple of blobs that would be rather non-trivial to reverse engineer and replace. The board could surely run headless without them, but the FSF most likely wouldn't certify it since users could be tempted to load the blobs to get video output and graphics acceleration.
Yeah. AGP? I had thought it was PCI behind a PCI-e>PCI bridge
What could be nice is replacing the crappy matrox chip with another better one that is compatible, does anyone know if that is possible? is it a standard bga or what not?
The only issues with the board I have seen is the lack of PCI-e lanes on the SR5650, but for $30 you really can't go wrong. Configs: x16 and x4, x16 and x8 physical. x8, x8 and x4 - x16, x8 and x8 physical.
When I get the cash I will get one of the latest and greatest 2014 G34 boards from supermicro and see about porting that as they have a dual northbridge layout there is plenty of PCI-e lanes, for instance the H8QG7-iLN4F is the best G34 board.