Intel Skulltrail Preview @ [H]

Sunin said:
Nehelam(sp) or whatever you call it processor is going to lessen the need to code software, games, etc for multi cores. The quad core natively splits the processing of a single thread among multiple cores. @ 2.6ghz it pulled an 8sec SPI.
Click to expand...

Proven a hoax. We'll have to wait a little longer for that kind of performance. ;) Anyway, saying hyper-threading will help there is laughable, as Super PI is the quintessential single-threaded performance benchmark, and no slight-of-hand will remove the need to finish the previous step before going onto the next one with that kind of mathematical calculation.

Sunin said:
Still very interesting technology. Hmm imagine an 8 core chip based on Nehelam? /drool
Click to expand...

You don't have to imagine, it's already on the roadmap. 8 cores, 16 threads. It won't be in the first wave of Nehalem late this year or first quarter next year (unless AMD pulls a rabbit out of its hat), but will probably be there in the second wave. I wasn't too impressed with last time Intel did hyper-threading, so let's hope they've improved it since then, but 8-core without hyper-threading sounds droolworthy enough to me!
 
farfromhome said:
You don't have to imagine, it's already on the roadmap. 8 cores, 16 threads. It won't be in the first wave of Nehalem late this year or first quarter next year (unless AMD pulls a rabbit out of its hat), but will probably be there in the second wave. I wasn't too impressed with last time Intel did hyper-threading, so let's hope they've improved it since then, but 8-core without hyper-threading sounds droolworthy enough to me!
Click to expand...

Well Hyperthreading worked very well in the Pentium 4 and Pentium D processors. Systems using them always felt faster and more responsive than their AMD counterparts. Without it the Pentium 4 often performed very badly. So I think Hyperthreading was well executed before with Netburst.
 
Man, just when I was starting to plan on building a new Graphics Workstation this summer. Guess I should wait until 2009... aaugh...
 
does this mean we'll see Intel / nVidia SLi boards come out for us mere mortals?
 
So cooling these monsters off will not be an easy task and so this platform won’t be an ideal candidate for silent computing
Click to expand...

omg u sed silent computing :p
 
Yeah but if i had millions of dollars and i wanted to use this to make my dream machine... id have to use a usb tv card instead of a pcie one.... cas after i fill up all the pcie 16 slots up with OHNOES!!

SLOT1 - 9800gx2

SLOT2 - 9800gx2

SLOT3 - Areca raid card

SLOT4 - one of this http://www.fusionio.com/

there'd be no room to plug in nothing...


anyone care to donate hardware for this build would be appreciated ... :D

oh I plan to use it for you know emails, messenger maybe a little youtube..
 
That video owned.
Dan_D, my question is when Francois Piednoel said FB-DIMM on octo-core is more important than DDR2/DDR3 what did you think? He said they're better because none of the 8 cores are "starved"
 
McCartney said:
That video owned.
Dan_D, my question is when Francois Piednoel said FB-DIMM on octo-core is more important than DDR2/DDR3 what did you think? He said they're better because none of the 8 cores are "starved"
Click to expand...

Which is true as long as your running quad channel. The problem with the press kits is, they only sent out 2 FB-Dimms, way to go Intel...
 
So you're saying that 4 sticks indeed increases performance to the exponential factor that this system is supposed to perform at? That's what I thought.

With FB Dimms being cheap (you can get 4 2 gig sticks at 800 mhz for roughly 400 bucks) that isn't an excuse by intel *rolleyes*
 
McCartney said:
That video owned.
Dan_D, my question is when Francois Piednoel said FB-DIMM on octo-core is more important than DDR2/DDR3 what did you think? He said they're better because none of the 8 cores are "starved"
Click to expand...

I can agree with that statement to a point. Yes FB-DIMMs give you more bandwidth and in theory greater stability. I won't deny the bennefits of FB-DIMMs. More bandwidth is really only needed in server and workstation applications. Which in truth will be the applications that need the extra bandwidth. Those are also the applications that will use the additional cores. As far as gaming and overclocking goes, I think the use of FB-DIMMs hurts the platform. It also makes the platform less attractive from a cost perspective. Again this is a nice super high end platform that can do it all, but it doesn't match other offerings in gaming performance, but it is reasonably close. In the testing I did you'd be hard pressed to notice the difference between the Skulltrail and a comparable SLI machine in actual gameplay testing. In any case you'd really only buy Skulltrail if you can afford it and you need it for workstation or server type applications and you want the box to double as a gaming rig. So in a way FB-DIMMs might be the way to go, but I'm not entirely convinced of that. There is some performance give and take going with them over conventional models. I suspect they went with FB-DIMMs so they didn't have to make a new chipset for the D5400XS motherboard. It allowed them to use the 5400 chipset instead as is with no modification.

ND40oz said:
Which is true as long as your running quad channel. The problem with the press kits is, they only sent out 2 FB-Dimms, way to go Intel...
Click to expand...

Yes that was dissappointing. There were a couple of reviews out there done with quad channel but unfortunately all the articles I saw used quad channel modules at 667MHz. So even with the quad channel results out there, we haven't seen what the board can really do in regard to memory performance.
 
You must log in or register to reply here.
Back
Top