Oh look, another EVGA board...

[Blauhung]

go ask around at intel
they stumbled with skulltrail and fell on their face, this time they could finally get it right, but they didnt touch it...

francois?

btw, i think jacob meant that you can use two entirely different cpus as in one with a higher multi than the other. OR two identical ones and run them at different multis. should work... might cause problems with cache and qpi syncing... but i doubt that.

i really hope evga can get i7 chips working on it... that would be so kick4ss

From talking with some people internally, They really did the best they could with skulltrail and honestly it was the fastest thing you could get for a desktop that would serve this community. But the business group that does board design is always under lots of top down pressure to make money, and the Skulltrail project turned out to be a loss. For the time they spent working on it, not enough people bought one to make it worth it.

You could say that the platform just wasn't capable of being something that you would end up turning a profit on, and I venture a guess that EVGA might end up cutting really close to making money on this board as well, but overall it is all about publicity and if the board puts out the best possible #'s for a extended period of time then its done its job. I think as a company, EVGA stands to profit far more from having good publicity surrounding this kind of project as a much higher % of their customers will hear about their success here. For Intel and specificly the group dealing with mobo design and marketing. High end sales is a much smaller % of the boards we put into customers hands and it just ends up not looking like a success to the people watching the bottom line.

It's unfortunate that this is how things work as I'm sure we could have put out something wonderful as well.


Oh and to answer the question...

Do not count on running desktop chips on this
-1st off, you wouldn't be able to run 2 of them because the Xenon chips are linked directly to eachother through their dual QPI's so that you can do a NUMA (Non Uniform Memory Access) type memory architecture. Basically this allows CPU1 to ask CPU2 for something stored in RAM in the event that its not in the RAM directly accessed by CPU1's controller. Without this link, i7's would be at a performance disadvantage as they would have to rely entirely on inteligently placing only stuff that the proper CPU would use in the correct stack of memory otherwise it strikes out and has to run to disk again.
-2nd, running 1 i7 chip in a board like this would be kinda silly

[saaya]

From talking with some people internally, They really did the best they could with skulltrail and honestly it was the fastest thing you could get for a desktop that would serve this community. But the business group that does board design is always under lots of top down pressure to make money, and the Skulltrail project turned out to be a loss. For the time they spent working on it, not enough people bought one to make it worth it.

You could say that the platform just wasn't capable of being something that you would end up turning a profit on, and I venture a guess that EVGA might end up cutting really close to making money on this board as well, but overall it is all about publicity and if the board puts out the best possible #'s for a extended period of time then its done its job. I think as a company, EVGA stands to profit far more from having good publicity surrounding this kind of project as a much higher % of their customers will hear about their success here. For Intel and specificly the group dealing with mobo design and marketing. High end sales is a much smaller % of the boards we put into customers hands and it just ends up not looking like a success to the people watching the bottom line.

It's unfortunate that this is how things work as I'm sure we could have put out something wonderful as well.


Oh and to answer the question...

Do not count on running desktop chips on this
-1st off, you wouldn't be able to run 2 of them because the Xenon chips are linked directly to eachother through their dual QPI's so that you can do a NUMA (Non Uniform Memory Access) type memory architecture. Basically this allows CPU1 to ask CPU2 for something stored in RAM in the event that its not in the RAM directly accessed by CPU1's controller. Without this link, i7's would be at a performance disadvantage as they would have to rely entirely on inteligently placing only stuff that the proper CPU would use in the correct stack of memory otherwise it strikes out and has to run to disk again.
-2nd, running 1 i7 chip in a board like this would be kinda silly

thx for the headsup
yeah i expected this... but tbh im surprised intel planned to make money out of skulltrail1... and the funny part is that doing skulltrail2 would have taken a lot less resources than doing skulltrail1, and it would have performed MUCH better...

its a shame, i heard some top intel managers whine about how they arent as popular and successful with their branding as apple, but they arent willing to invest ANY money whatsoever into doing some propper pr and doing something that helps them to be seen as a company pushing technology to its limits...

the enthusiast group inside intel is tiny and ignored most of the time...

oh and about two i7 chips working on this board... so is a direct qpi link between the cpus mandatory? did intel somehow lock DP operation in the MRC code if there is no direct link between the cpus?
MCM works great on Core2, and thats using a cr4ppy old FSB, and a single FSB for both cpus at that! a dual i7 with one qpi link to the IOH without a direct link between each other would definately work, and it would most likely offer 90% if not more of the perf of a setup with a direct qpi link between the two cpus...

[STEvil]

thx for the headsup
yeah i expected this... but tbh im surprised intel planned to make money out of skulltrail1... and the funny part is that doing skulltrail2 would have taken a lot less resources than doing skulltrail1, and it would have performed MUCH better...

its a shame, i heard some top intel managers whine about how they arent as popular and successful with their branding as apple, but they arent willing to invest ANY money whatsoever into doing some propper pr and doing something that helps them to be seen as a company pushing technology to its limits...

the enthusiast group inside intel is tiny and ignored most of the time...

oh and about two i7 chips working on this board... so is a direct qpi link between the cpus mandatory? did intel somehow lock DP operation in the MRC code if there is no direct link between the cpus?
MCM works great on Core2, and thats using a cr4ppy old FSB, and a single FSB for both cpus at that! a dual i7 with one qpi link to the IOH without a direct link between each other would definately work, and it would most likely offer 90% if not more of the perf of a setup with a direct qpi link between the two cpus...

No, it doesnt work that great really. Works better if you disable the snoop filter 90% of the time. Also the FSB is what killed skull trail. If it could have made 430-440mhz reliably it would have been good but as it is motherboards from Supermicro did better. Even my 5000x system could pull 429mhz if I dry iced the chipset, and that was with only 667mhz ram (crap clockers..).