• Please review our updated Terms and Rules here

Machinist X99 Motherboards?

It's nice having a PS/2 port over USB because PS/2 has a dedicated interrupt so the CPU can't ignore it, vs USB polling. And you're not having to waste a USB port for HID devices.
Many years ago, I worked on a team supporting about 6000 windows servers at a customer site. They had them set so that you could generate a BSOD and memory dump by pressing and holding right ctrl, and then pressing scroll lock twice. The servers with PS/2 keyboards were way more likely to respond to that sequence than the servers with USB keyboards, especially when they were half locked up already anyway. And if there wasn’t already a USB kvm dongle attached, the USB only server wouldn’t respond at all.
 
From what I understand, these Machinist boards use chips from servers sent to China for recycling. Which explains the server RAM in them. Actually, I applaud the re-use of old server hardware.
Are you saying that the Machinist boards uses recycled chips? I am not against recycled hardware, but why not just buy used computers if we end up using used hardware anyway?
 
Are you saying that the Machinist boards uses recycled chips? I am not against recycled hardware, but why not just buy used computers if we end up using used hardware anyway?
The Machinist boards use recycled consumer desktop chipsets with modified BIOS to support the LGA-2011/V2/V3/V4 CPUs. Workstation/Server boards with the actual X79/X99 or server chipsets are much more expensive used.
You do lose some features vs the real workstation / server boards, but it is much less expensive and you get a newly manufactured motherboard.
 
Not all Machinist boards use recycled chipsets. My Machinist LGA 1155 board uses and Intel B75 express chip which allows for M.2 SATA but only PCIE 2 x8 on the PCIE16 slot. They are meant for applications other than gaming (industrial use).

There are niche makers who make motherboards with recycled parts because they are easy to find and the company is too small to get chips directly from Intel anyway. Those boards use stolen BIOS from other boards.
 
Bought the aforementioned motherboard (14 core) with 32GB memory, 600W PSU, cheap video card and cooler shipped for about $189. (Honey had some coupons that knocked $15 off the total). If it all turns to slag, I won't cry--it's still a business expense.
 
It is a bit strange seeing PS/2 mouse and keyboard sockets on a motherboard this late.
The Asus Crosshair VI has them.
Bought the aforementioned motherboard (14 core) with 32GB memory, 600W PSU, cheap video card and cooler shipped for about $189. (Honey had some coupons that knocked $15 off the total). If it all turns to slag, I won't cry--it's still a business expense.
What are your plans for the new board?
 
They aren't usually high end chipsets, but you can still get the latest platforms with onboard serial and parallel ports if you look hard enough. Native PCI is long gone though. Everything is done via PCIe-to-PCI bridge chips that usually work.
 
What are your plans for the new board?
Mostly tinkering and web browsing--just to satisfy my curiosity. The shame is that it's currently destined for a NIB tower case with 3 5.25" drive bays and a 3.5" floppy bay. Maybe I can find a flimsy new case for cheap with none of the above to stick it in.
 
There are PCIE x4 SCSI cards out there with support for at least Windows 10 x64 if you want to put some drives in the bays (not easy to find nor cheap, look for ATTO UL5D as an example).
 
I do prefer the non-IME (for Intel) and non-PSP (for AMD) chips for my personal systems, though I do have a spyware-enabled laptop that I use mostly for YouTube.

The AM2 and some AM3 socket AMD chips were the last to not have the PSP functionality. Intel chips had the IME much earlier, I think. The A8-6600K in my desktop is fine except that YouTube does bog down the system a bit (about 40% CPU utilization).

Then I run Linux or FreeBSD on them, which is about as secure as you can get these days for anything connected to the Internet (behind a firewall of course).

The PSP: https://en.wikipedia.org/wiki/AMD_Platform_Security_Processor , Intel Management Engine: https://en.wikipedia.org/wiki/Intel_Management_Engine
 
I do prefer the non-IME (for Intel) and non-PSP (for AMD) chips for my personal systems, though I do have a spyware-enabled laptop that I use mostly for YouTube.

The AM2 and some AM3 socket AMD chips were the last to not have the PSP functionality. Intel chips had the IME much earlier, I think. The A8-6600K in my desktop is fine except that YouTube does bog down the system a bit (about 40% CPU utilization).

Then I run Linux or FreeBSD on them, which is about as secure as you can get these days for anything connected to the Internet (behind a firewall of course).

The PSP: https://en.wikipedia.org/wiki/AMD_Platform_Security_Processor , Intel Management Engine: https://en.wikipedia.org/wiki/Intel_Management_Engine

Is there any proof whatsover that these are not just BMC/IPMI on a chip for desktop?
Why aren't server companies accused of putting backdoors in their systems? Out of band management is a heavily used feature in server space for decades now. Low end servers come with shared NIC for mgmt and Ethernet. I don't see any problem here until someone traces rogue packets with Wireshark in the Ethernet wire.

As far as I've read ME cannot interfere/coexist/tap into network drivers and TCP stacks in the OS or any data for that matter. There is nothing pointing to that kind of rogue capability and there is no indication such function ever occurred in runtime. It would be a heavily complicated scheme where ME would need to have a lot of data on exact implementation of OS you're using and would have to coexist with its model of locking and paging in a such seamless manner that no performance hit can be experienced. Dare I say this is impossible.

And if they did it, you take your FreeBSD or Linux kernel and reshuffle a minimal part of code to make the hyper-intelligent-rogue-ME completely blind. If it starts with sniffing out a structured piece of kernel memory, and knows how to proceed in sniffing from there, maybe all it would take is simply reorder struct members and recompile.
 
Back
Top