With supplies on the X48 still being quite low, we have a new chipset that managed to take the spotlight from GPUs in the recent Computex: Intel's new mainstream chipset, the P45. This is the first upgrade for this segment since the P35, more than a year ago; the X38 and X48 were aimed at the enthusiast, and actually replaced the 975X. So, what you have here, is the third chipset for the Core 2 (the P965 was the first one) that you should really care about.

The Eaglelake platform raises FSB speed (always comparing to the P35) from 333 to 400 MHz (1600 MHz). Easy overclocks of 50% are guaranteed for older dual core CPUs (quads with a little bit of luck) and 20% for the 8000 and 9000 series. Of course, you can always expect to get even better overclocks with adequate cooling. Moreover, the P45 supports CrossFireX up to two PCIe x8 slots (PCI Express 2.0).

To get a better idea of what it is capable of, let's say that in the article linked below, with DDR3 memory, the Asus board reaches 500 MHz while running a Kentsfield, not bad for a "mainstream" motherboard.  When it comes to performance, the ASUS P5Q3 looks to be on par with the X38 and X48, which are supposed to be better. For energy consumption measures of the ASUS Energy Processing Unit and the 65 nm process used to fab the P45, we'll have to wait for another review.

Link: Hardwarezone.

After a long hiatus, it looked like abit was regaining some of it's strength a couple of years ago, but the lack of such successful products as the ones they pulled out at the end of the 90's (along with it's inability to compete in the budget sector) may translate into the end of their business.

According to some, and regardless of abit's denial, if sales of their P45 chipset based motherboards aren't up to meet their goals, those will be the last ones under the abit brand. And what's the goal, you ask? Oh, about six million boards, not too many if we consider that last year they didn't reach half that figure. No, wait, that's what they thought they would be selling before 2008 started, but they're actually selling less this year. No wonder someone at Universal Scientific Industrial (the owners), has gotten a little bit impatient and asked for immediate results.

Let's see who's so bold as to order their products now. You might not be getting any kind of support afterwards.

Link: DigiTimes.

The 780a chipset is here, and alongside it arrives a couple of new technologies which have already been anticipated: Hybrid Power (the ability to work with just the integrated graphics) and GeForce Boost (the ability to join forces with the discrete graphics card you have plugged in).

As it was expected, the AMD 780G is clearly superior when it comes to IGP rendering, so let's focus on the scenario that GeForce Boost brings about. Recall that, because of efficiency concerns, this new technology only works with the GeForce 8400 GS or the GeForce 8500 GT.

Comparing the mGPU + 8400 GS duo (basically an 8400 SLI configuration) with the AMD chipset (+ Radeon 450), nVidia loses again. At least for the moment and regarding the five games selected for the review. With a GeForce 8500 GT, results are so bad that we better leave them for the moment until drivers are a little more polished.

Link: AnandTech.

There aren't going to be a lot of chipsets with a memory controller down the road, and yet Intel has decided to make the X48 as good as an external memory controller can be. Not only is this chipset selected for higher working frequencies, but also to reach them at lower latencies. What some motherboard makers are calling "Performance Level", is nothing more (or less) than the ability to modify the memory controller's Read Delay (tRD) values, something you can do directly and transparently with the ASUS Rampage Formula.

Below, you have the link for a long and illustrative article about this important parameter. It's too dense to summarize in a few words, but if you doubt about the influence of this BIOS setting on your CPU's performance, let's say that lowering it you may get bandwidth from 10,301 to 10,923 MB/s, keeping a QX6950 at 4 GHz (450 x 9) and reducing tRD from 7 to 5. The whole spectrum goes from 14 to 5, so a 400 MHz FSB also enjoys some spectacular gains.

The role of the tRD was first noticed with the P965 and it's "FSB straps", here you see it in all it's glory, promising that all these improvements in the memory subsystem will translate into a real world applications performance increase.

Link: AnandTech.

Since it came out, just in time for the holiday season, not a lot has been said about the performance of the nForce 780i SLI. You wonder why? Let's take a look at this comparison between an XFX incarnation of nVidia's latest and greatest with the Inno3D nForce 680i SLI, to see if something can be said at all. An Abit IX38 Quad GT is also tossed into the mix, just for fun.

So much for the CPU of choice, a Core 2 Duo E6750, not THE Q6600, but it'll have to do. The graphics card is no other than the GeForce 8800 GT (two, if we talk about SLI), so there's nothing to complain about in this department. The OS chosen starts to be ubiquitous (because of DX10): Windows Vista Home Premium (32 bit version, to make it not too complicated).

2D applications, single graphics card, are more or less favourable to the 780i. More if we talk about WinRAR 3.71, less if you look at results from the other two programs. When it comes to games, we could also consider it a draw. So, if you own an nForce 680i SLI, you aren't missing much.

Developed more than a year ago, the version 2.0 of PCI Express still raises some doubts among users. Apart from a bandwidth increase (the old maximum was doubled), what else can you expect from the new bus? Well the frequency stays at 100 MHz but what has been increased is the internal multiplier.

Both ATI/AMD and NVIDIA launched their first generation of PCI Express 2.0 compliant graphics cards several months ago: Radeon HD 3000 series and G92 (65nm). From that moment several complains were heard about compatibility with "old" motherboards. This looks like the manufacturers fault, because both a PCI Express 1.x device installed in a PCI Express 2.0 slot, and a PCI-E 2.0 device installed in a PCI-E 1.x slot, should work (at 1.x speeds).

The aforementioned doubling of the bandwidth will allow us to enjoy as much as there was available in those hard to see, because of their length, PCI-E x32 slots. And we will have it in the so much more convinient PCI-E x16 form factor.

Link: AnandTech.

Yes, it looks like the brains at ASRock have decided to slowly change the philosophy of the company. They will continue to target mainly the entry level market, while increasing their presence in the upper sector until the sales from high-end and performance motherboards reaches 30% of their sales.

Another consequence of this shift of general strategy will be a "slight" price rise for all their products. The intention is to stop being considered, above all, a cheap manufacturer. On the contrary, they want to sell a image of good price/quality ratio.

The starting point will be the 4Core1600P35-WiFi+ which, as you might have guessed, is based on Intel's P35 chipset, but will reach 1600 MHz FSB, nevertheless. To that end, and let's hope something more (with the aid of a specific overclocking program), the board gets solid capacitors.

Link: DigiTimes.

DigiTimes sources at motherboard makers seem to agree on one thing, among the chipsets that will go head to head in the first quarter of this year (AMD's RS780 and Nvidia's MCP 78), it is AMD that holds between a 15 and a 20% performance advantage in 3DMark 05 and 06 tests.

Nvidia has an ace up it's sleeve, the MCP 78U, which should tilt the performance balance in their favor, but due to the higher core frequencies of the graphics engine, the cooling measures that would be needed are just too expensive.

To make things worse, it's possible that the Nvidia product gets delayed until February, while AMD/ATI did their homework, allowing motherboard manufacturers to keep the January 23 date.

Link: DigiTimes.

It looks like, for the time being, AMD is in a good position to continue ATI's tradition when it comes to chipsets with a frugal energy consumption. When the nForce 4 was the best choice, the Xpress 3200 came about, bringing, not only 40 PCI Express lanes, but even lower power requirements than nVidia's solution.

Link: The Inquirer.

The MSI P6NGM is based on nVidia's flagship GeForce 7150/nForce 630i chipset. Being a mainstream motherboard, they mostly followed the reference desing, including all the features by the chipset and even FireWire support through a JMicron JMB381 2-port FireWire-400 controller.

"The board will support the full range of Intel LGA775 processors with FSB up to 1333MHz and comes with two single-channel DIMM slots for up to 4GB DDR2-800 memory [Single-Channel only] NVIDIA's concerns on this aspect were to bring about a more affordable and efficient mainstream mGPU [rather than a top performing one]."

Other features inlcude Realtek ALC888 HD Audio CODEC, Realtek RTL8211B Gigabit PHY, a native HDMI port (HDCP compliant), VGA output, four SATA ports and one IDE. The internal graphics core runs at 630MHz with a maximum 256MB framebuffer support. Expansion slots include two PCI, one PCIe x16 and one PCIe x1.

Before the days of nVidia and ATI, the chipset market had a front row seat for the asian manufacturers. AMD wasn't interested in that business at all and Intel prices left a good profit margin for those wanting to sell cheap. Right now, most people don't bother about a very small price difference and choose products that are out of reach for VIA or SiS to develop. Nevertheless:

"VIA's September revenues reached NT$1.18 billion (US$361.98 million), a drop of 7.6% on month and 28.9% on year. [...] On the other hand, SiS's September revenues rose 41.8% on month and 52.5$ on year to NT$893 million."

What do they sell? Ultra mobile PCs in the case of VIA and any chipset they can to whoever might be interested in saving a penny. And there are lots of them, judging from those figures.

Link: DigiTimes.