backtop


Print 45 comment(s) - last by TheMailMan78.. on Apr 18 at 12:07 PM

Intel prepares to take on AMD's Quad FX

Intel hinted at the possibilities of a high-end gaming system based around its Xeon workstation processors earlier this year at CES 2007. The Intel V8 system featured two Xeon Clovertown processors clocked at 2.4 GHz and a single GeForce 8800 GTX graphics card. Although it was a proof of concept to counter AMD’s Quad FX enthusiast platform at the time, Intel has announced the V8 concept will become a reality at its Spring Intel Developer Forum in Beijing, China.

Intel has named its new enthusiast platform Skulltrail. Intel has remained very tight-lipped about its Skulltrail platform except a few basic features. Skulltrail will arrive later this year and accommodate two Core micro architecturequad-core processors. It is unknown if Intel will develop a new platform to accommodate dual LGA775 socket processors or rework its LGA771 socket Xeon workstation platform to accommodate unbuffered DDR2 memory. Intel’s Xeon workstation platform only accepts FB-DIMM memory.

On the graphics processing side of things, Skulltrail will sport four PCIe slots for graphics expansion. Intel doesn’t divulge details on which multi-GPU graphics technology Skulltrail takes advantage of, however, Intel’s current 975X Express supports AMD’s CrossFire technology.

Intel has not released any details on the chipset powering Skulltrail yet. Nevertheless, Intel expects to unveil its next-generation single processor enthusiast chipset, X38 Express, later this year.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

RE: 2 Chips, 4 Dies, 8 Cores?
By KristopherKubicki (blog) on 4/17/2007 10:06:21 AM , Rating: 4
Why?

http://www.dailytech.com/article.aspx?newsid=6928

There's no performance difference and it's cheaper for multi-chip packages.


RE: 2 Chips, 4 Dies, 8 Cores?
By Alexvrb on 4/17/2007 7:52:19 PM , Rating: 2
Although I would tend to agree with that sentiment in general, I did not see any conclusions about performance differences in that link. It said it was more efficient from a development standpoint, as you don't have to redesign the whole chip, just the die in question. This does allow for cheaper, faster development of new chips, as some dies may be upgraded and others completely unchanged.

However, with regards to performance, that very much depends on the specific implementation of the multi-die chip (or multi-chip module). With multiple dies, your inter-die communication has to be well thought out. Some chips may do it just as well as a single-die chip, and there would be equal performance. However, if the two dies had to communicate over the FSB, this could provide a performance disadvantage. In the future I'm pretty sure all MCM packages will have a better approach and there will not be a difference in performance. But that doesn't mean there is never a difference, because it really depends how they do things.


"A lot of people pay zero for the cellphone ... That's what it's worth." -- Apple Chief Operating Officer Timothy Cook

Related Articles













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki