Part of the talk detailed Intel's next generation 32nm process which is currently working on SRAM.
At the Intel chipset and CPU meeting today bit-tech
had a chance to sit down with some lovely people at Intel and get answers to our niggling questions about the new 4-series of chipsets, formerly known as, Eagle Lake.
First off, let's talk about process technology because it is something that Intel is pushing hard. Everywhere you go is 45nm this that and everything which isn’t bad, but it has been typically limited to CPUs.
That's not the only process that Intel is pushing though, because the entire 4-series range is now built on 65nm technology, as opposed to 90nm with the 3-series. So it turns out that X48 will
be different to X38 after all...
In effect, this means that while Intel has delayed its X48 for many months it’s had time to not only re-spin it but tweak it quite considerably.
We were told that it’s had some quality time with engineers and not only have the internal timings have been tightened but it’s been developed and virtually assured to run at the faster front side bus than even a QX9770 – yes, that means the chipsets are virtually guaranteed to offer a considerable overclock, obviously providing everything else is up to take the strain. This should put considerable weight on Nvidia’s upcoming 790i SLI DDR3-based chipset, although that still has SLI and 3-way SLI in its favour of course.
X48 also has XMP support, an alternative to Nvidia’s EPP2 – both are DDR3 standards that should actually provide more compatibility and performance from these profiles, but in actual fact it just divides the market with what should be wholly interchangeable modules. Obviously ANY DDR3 can run in either system, but on our end – having to get XMP and
EPP2 performance DDR3 memory to test these boards is a sheer pain in the backside, and it'll also be an annoyance for anyone that switches from one platform to the other.
On the software front, Intel is not being left (too far) behind – while AMD continues to make strides with its OverDrive software and Nvidia has not only its Tweaking Utility but also its ESA monitoring now, Intel’s Extreme Tweaking Utility will work with all 4-series discrete chipsets (X48, P45 and P43) to allow in-Windows overclocking and stability testing.
Intel told us that the software was designed to allow even the greenest user to have a go, without having to venture into the world of the BIOS. It’s not meant to be a replacement of BIOS tweaking for hardcore users and Intel knows that there are plenty of Taiwanese companies that work extremely hard to maximise their own BIOS potential.
However, the company also said it offered the software code for these companies to tweak and work with to suit the options they wanted to include – this could make it potentially far more open and varied than AMD’s or Nvidia’s software. From the demonstration it looked extremely well constructed and intuitive, but still provided quite a few options to play around with.
Intel’s G45 is also an evolutionary step for the company – supporting the full range of HD formats for acceleration as well as “DirectX 10”, the company acknowledged that its gaming performance should certainly improve with its new drivers launched when the chipset arrives, but the G45 was also a stamp for the new Vista Premium certification. Basically, by June 1st only graphics chipsets that are DirectX 10 compliant can obtain Vista Premium certification.
A few Intel executives basically told us that if people want to game properly
, they should buy discrete graphics – their chipsets are designed for general productivity and now movie/video watching too, even up to HD resolution. However, the biggest change should be the price – Intel is set to match the competition from the AMD 780G and Nvidia GeForce 8200 IGPs. Considering the AMD 780G is very inexpensive at best, this could mean a very interesting and positive market for the consumer with plenty of competition.
Intel is also pushing the mini-ITX platform for its G45 and G43 very strongly – whether it’s through its partners or with its own boards. In the UK and Europe expect this to mean partners, such is the way of the channel. Intel itself is developing four mini-ITX boards – two of these will be G45 and both will include HDMI and HDCP, the other two will also be embedded solutions with a CPU soldered to the board – this is more for OEMs.
Q45 mini-ITX boards for business with vPro will also be available for small form factor, low cost, low power solutions. Intel explained where before developing for mini-ITX cost companies significant amounts of money, as Intel is working with its partners to push the format hard that $100-$200 price premium should drop to just $20 over any other motherboard.
Finally, we found out the P45 and G45 are exactly the same chip – just with certain parts of one turned off and other parts turned on instead. We were assured the P45 was still optimised as a discrete part, where it’s possible to open up avenues of data travel that the G45 prohibits, but the actual manufacturing of both is just a single die.
While we may think of this was wasteful at a die level – it’s also why the 65nm P45 part looks as big as the current 90nm P35, at least it allows Intel to shift the production of either SKU, without having to start a new wafer from scratch.
The 65nm process should make the chipsets more overclockable, need less voltage and more energy efficient – we’ll be sure to test all these claims when they are released around May.
Are you looking forward to G45, P45 or X48? Let us know your thoughts in the forums