elmor

Same process we use will work on any vendor

Don't think it will work on Skylake unfortunately, the issue is XP has no way of seeing USB devices even during install. The workaround we use loads the XP iso to RAM and adds a driver during F6 phase to emulate a CD drive.

What is it that you can not?

Hey, check the OC guide here http://forum.hwbot.org/showthread.php?t=142991 You'll also need "Tools and files for W7/XP USB installation".

Great job Ian, 85k CPU score on full run!

Links should work better now These bioses are "one to rule them all", meaning they will work with all OS including XP.

Try this: https://www.dropbox.com/s/jdmf5jri6nmegnk/eVc%20v1b5.exe?dl=0 Hmm, OK. Don't have a card so can't test it myself. Can you try a register scan? Device address 0x06 Start 0x01 End 0xFF Don't check SMBus

Still such a fun benchmark Old rules don't apply, need to relearn for Skylake.

Madness

xxbassplayerxx: Wouldn't work for the EVGA card, let me get you an updated version. Smoke: How about monitoring, any read out?

So the device scan is from the HOF power card? Do you get any error?

Hey, try this: https://www.dropbox.com/s/x1pagutk5h2s7ju/eVc%20v1b5.zip?dl=0 Pick "IR3595 test". Depending on the VID setting voltage might be a bit lower. Check which readout is correct, the 6.25mV or the 5mV version.

Cheers Trouff, updated the link

Hey Smoke, I'm on it just need a bit of time to get data on how this device works. Don't have any unit with it to try right now.

Cheers guys, still a lot to be had out of this system but 4-ways a **** as usual GPU slims from Roman, hit him up for details

Interesting Nice work Ronaldo

Sorry about no reply Masterchief, completely missed that comment. I'll look into VT1556. xsuperbxg: There's experimental support for up6208 which should be similar, I'll look into that as well. Should be connected properly since you get a readout. (Busy with Computex right now, will have more time next week)

Oh, and:

eVc actually has support for ePower 2.0

AOD414: http://www.aosmd.com/pdfs/datasheet/AOD414.pdf AOD412: http://www.ic72.com/pdf_file/a/71730.pdf

Part of it is that you added some sensing modifications we suggested to improve droop conditions, could be part of the reason why this happens. Those mods 1.8V set on the power card would give us 1.76V loaded on the GPU side (just below 2G core, almost 500A average current).

Are you sure Petri? Sampsa's back in action competing for the Finnish crown http://hwbot.org/submission/2775019_sampsa_superpi___1m_pentium_2_celeron_300a_slot_1_2min_43sec_916ms

Massive score Vince

That's correct. You need to supply +12V and +5V from the PSU in some way, easiest to use a standard Molex or floppy connector (in the picture you supplied).

Looking forward to see the results! All is still experimental so you're bound to find some bugs and instruction typos, that's what you guys are here to help with. The remote sense inputs are for displaying a more accurate output voltage on the power card. By default the output voltage is measured at the designated contact points at the front of the card, but because of the high current flowing through the wires to the GPU the actual input voltage will be difference because of droop during load. The most accurate reading will be from the decoupling capacitors (MLCC) at the back of the GPU, so just solder a wire from the positive remote sense input to the positive (GPU Voltage) side of the capacitor and the negative input to the negative side (Ground). If there's a voltage on this input it will automatically be used for reporting rather than the on-board sense. The Hotwire connectors (six in total, three at the top and three at the bottom) are supposed to help you adjust the voltage of auxillary VRMs (think on-board memory and PLL). Hotwire channels 1 & 4 come with increased ranges and each step causes a bigger change, typically suited for a GPU VRM or similar. In the end which channel will fit best depends on the specific VRM design and the resistor values used in its feedback loop. One step could be just a few millivolts on one design and several hundreds on another. I'd say you should try with channel 2,3,5 or 6 first and if the change is too small switch to channel 1 or 4. I'll update the instructions with more clear information.