Well finally got around to creating that VM for the install, so got it booting and running now.
But for some reason i can not the display driver/Opencl install properly on boot up.

OpenCL driver doesn't throw an error but the AMD driver refuses to load if you start the .msi(says it needs catalyst install manager).

Launch a miner app(Phoenix 1.48 in my case) and it stalls for a while before throwing an error about the OpenCL wrapper not being found.

Can i integrate motherboard drivers into the bootup x86.wim file? Or would it mess up cross-platform compatibility?

Bit lost how to solve this, since no drivers can be installed after the fact, since they're lost at reboot  

Edit: Dpinst reports error on AMD driver install when run manually.

X86/i686 support would be nice at some point, can't boot this on my VIA C7 board

I also  second all the suggestions to disable screensaver.


Oh and thanks for all your hard work and dedication, much appreciated mate 

My inner geek feels so fulfilled now 

Interesting, nice to see how it's supposed to work, though you may want to re-edit it a bit to speed things up a little

I had the TFTPd server and PXE booting working, couldn't get the Make_PE3.cmd/exe working on XP x86.
The script/installer makes a lot of hardcoded references to local win7 system files, which causes it to fail when unpacking install.wim from the Disc/ISO source. Tested with two different versions of Win7 SP0.
Initial error was Windows media player(wmplayer.exe) was missing "First find Win7 files".
For now i suggest adding a check if server/install OS is NOT Win7 X86 and throw an error.

I'll retry when i find a spare Win7 x86 machine   

Looks nice, I'll try it out later on one of my older test rigs, having  PXE booting Windows based miners might be fun 

Very nice job!
Finally got to break the 400Mhash/s barrier on my HD5850, an increase of about 8-10Mhash/s over POCLBM kernel. 

Indeed the EC2 compute boxes aren't that powerful for Bitcoins.
But the figures quoted are a little lowballed, i got 70-80Mhashes/s on each Tesla GPU and about 1.6Mhashes/s per 'CPU' core ~170-190Mhashes/s total.

Though it was still a fun idea you had there Kefir 

I used Amazons 'spot pricing' to bring the cost down to around $0.70-0.80/hour, though that still doesn't seem to offset the cost enough, even when i had the miners autostart with the instance to run automatically during the off-peak hours 

Esp. when my HD5850 easily does >370Mhashes/sec at a cost of about $0.10/hour.