Home  |  News  |  Reviews  | About Search :  HardWare.fr 

MiscellaneousStorageGraphics CardsMotherboardsProcessors
Advertise on BeHardware.com
Review index:
Report on the Fall 2007 IDF
by Damien Triolet
Published on October 11, 2007

We now know the latest details concerning the release of the Core 2 Penryn on the 12th of November, first starting with the Core 2 Extreme QX9650. This quad-core processor will replace the current Core 2 Extreme QX6850.

A quad-core Penryn without the heatspreader and a Penryn wafer.

You may recall, Penryn processors represent a small evolution, in other words a « Tick » versus the « Tock » which is the introduction of an entirely new architecture. The Penryn thus brings in the 45 nm production process signifying a big evolution because the transistor itself has undergone major modifications. This is something that hasn’t happened for a long time. We won’t go into the details again which were already covered here.

Slight improvements compared to the current Core 2 have already been described in March. Simply remember the two main points: its L2 cache goes from 4 to 6 MB and SSE4 makes an appearance.

The Core 2 QX9650 will be set at the same frequencies as the current QX6850, including its FSB. The gain in performances will then come from small architectural modifications and should vary between 5 and 15% according to the numbers given by Intel. This is nice but not earth shattering, and actually we are awaiting a Penryn set at 3.33 GHz and not 3 GHz like the current high end. No one doubts that the reason for this non-increase in frequency is due to AMD’s lack of competitiveness in the high end. In addition, the Phenom isn’t expected to really compete with the Core 2 in this domain, mainly due to its lower frequencies. So what would be the purpose of launching a faster processor in November?
Surprised by AMD’s announcement of the 4x4, Intel’s reaction was via the V8 platform whose main purpose was to show that they too could enter into the popular battle for who can be the biggest. While this platform is unequaled in terms of processor power, it suffers from a number of disadvantages, because it’s entirely based on server components ; server motherboard, Xeon and FB-DIMM memory. In short, it’s "general public" but doesn’t really fit into this category.

If having a Xeon instead of the Core 2 didn’t weigh on the conscience of power users, the price of the platform as well as the impossibility to do the least bit of overclocking or tweaking rapidly scared away the few people who might have opted for such a system.

With Skulltrail, Intel claims to have taken into account all of the criticism towards the first V8 version and will now offer the desired platform. So are we done with the Xeon? The FB-DIMM which has high heat production, a lower performance and is more expensive? No. Intel did hear the criticism but not for this. Actually, it appears they continue to create here but without really coming up with a product that is well adapted for this very high end platform. What does change is the motherboard. Now, even if it’s based on a server chipset, it will be completely oriented to power users.

Overclocking will be up to expectations and moreover, Intel has a SkullTrail platform with Penryns pushed to 4 GHz (not bad !). The only thing left is the price, which could scare some away and the fact that this extreme platform will especially be devoted to very high end machines of large PC manufacturers.

The V8 SkullTrail pushed to 4 GHz with an FSB 1600.

One particularly interesting item is that the SkullTrail motherboard that is produced by Intel will be CrossFire and SLI compatible with 2, 3 or 4 graphic cards! This is a first and the interest of this platform grows significantly thanks to this detail, although this only aims for a certain niche of well-off gamers. To offer SLI support, Intel had to place two MCP nForces on its motherboard, which only serve in relaying the graphic ports. In reality, Nvidia requires the presence of nForce chipsets solely to justify the marketing which says that SLI only functions with Nvidia chipsets. Of course from a technical point of view, this is false. It would have been better to see Intel and Nvidia agree on a license rather than seeing a PCB uselessly polluted by these nForce chipsets.

The SkullTrail motherboard. Note the two nForce chipsets next to the PCI Express ports to justify SLI support on a non-Nvidia chipset.

Whatever the case, this platform should be the most flexible on the market in terms of graphics. And moreover, it seems that Intel is saying to Nvidia ; "It appears you are capable of producing Quad-SLI… Well, we are waiting!".

<< Previous page
Introduction, process

Page index
1 | 2 | 3 | 4 | 5 | 6 | 7
Next page >>
The Nehalem  

Copyright © 1997- Hardware.fr SARL. All rights reserved.
Read our privacy guidelines.