Sapphire HD3870 with 7900GTX cooler

Keywords: ATI Radeon HD3870 better cooler
The new Radeon HD 3800 series supports the same mounting holes as the old 7900GTX so, what better way than to cool it, than recycle the old 7900GTX cooler? This cooler is known for being quite silent, while performing above average, fruit of it's 80mm fan. This move keeps R&D costs down and consumers happy.
It's a noticeable upgrade from what is the stock HD3870 card, which is not the most efficient of designs.
For someone who doesn't care about the dual-slot design, this is very good choice, and while some may argue that better coolers are available, this one is a proven, cheap and silent design.
The clocks are the standard ones, as are the 512MBs of RAM provided.

Expect to pay around 200eur for this new offer.

Ars reviews the Macbook Air, now with the SSD

After yesterday's review of the Macbook Air, Ars Technica has took the top model for a spin. The one with the 1.8GHz Core 2 Duo and solid state drive, that costs a whopping $1300 more.
Not really an improvement worth the extra fee, not battery or performance-wise.
"Mainstream" SSDs still have a long way to go if they want to be worth the higher cost, which will, hopefully, also go down at the same time.
Keywords: Macbook Air SSD review performance

Ars reviews the Macbook Air

Ars Technica has an in depth review of the new laptop from Apple, one definitely worth a look.

They state various problems, from the very slow 1.8" hard drive to the erroneous claims about battery capacity.
While Apple claims that the laptop can work for four and a half hours on a full charge, the reviewer only was able to squeeze two and a half out of it. This is unforgivable in a laptop of this price and type. Apple should think of addressing the problem soon since you can't carry a spare battery with you either.

The charging time is of five hours, according to the reviewer. That's probably due to the small wattage of the AC adapter. Nothing that can be easily be fixed by buying a better one, since the MagSafe connector placement is a bit non-standard.

The RAM is soldered onboard, but that is a compromise that had to be made for it to have this kind of thickness.
There's definitely some room, even need, for improvement.
Keywords: Macbook Air review ram battery

AMD's 780G chipset is able to overclock past 1GHz

Keywords: AMD 780g overclock 3d mark 2006 rs780
Gamer on a budget, be delighted. The 780G from AMD will be able to provide integrated graphics with decent performance, at least to some extent.
2000 3DMarks in the 2006 version will be attainable with some effort.

J&W is a Chinese motherboard manufacturer that has been able to tweak their AMD 780G motherboard to withstand an amazing, aircooled, core speed in excess of 1GHz while overclocking. That is double of the stock clock, 500MHz, and is certainly due to the recent process technologies that AMD has been employing in their chipsets.

Intel is leading in the CPU world, but when it comes to IGP technology, AMD has been first for some time and plans to further dilate that advantage. Not just in raw processing speed but power consumption figures have also been encouraging.
Nvidia is being very slow to answer and some reports claim that the MCP78 is having some problems at it's highest clock speed.

These news bring very good expectations for AMD's upcoming Fusion processor.

Nvidia to acquire AGEIA

Nvidia has made official that it plans to acquire AGEIA Technologies, provider of the PhysX middleware and hardware accelerators with the same name, also known as PPUs.

Intel had already acquired Havok a short time ago, in a move to broaden it's assets in the field of physics processing for games.
Nvidia has now made a similar move, hoping to provide decent physics acceleration in their line of GPUs. That is planned to happen, in mass, by the time that DX11 sees the light of day. Microsoft has already announced that they will provide APIs for physics acceleration in that release, although no timeframe is yet certain for the release of the new version.

The press report is available at Nvidia's site.

Vista SP1 and Windows Server 2008 ready

News.com reports that Microsoft has already finished work on both pieces of software and they will soon be ready for deployment.

Microsoft will be delivering Vista SP1 to customers in mid-March, while 2008 Server will be officially released in the 27th of February.

The original report is available here.
Keywords: Windows Vista SP1 release date

Intel Silverthorne supports HyperThreading, more Tukwila details

Yesterday, Intel unveiled more details about it's new Silverthorne processor, for ultra mobile applications, together with some new information about it's upcoming Tukwila processor for servers, part of the Itanium family.

The more recent details about the Silverthorne processor indicate that it features HyperThreading support and also hardware virtualization technology. Intel also claims that the CPU's instruction set architecture is fully compatible with the Core 2 and that it has a power consumption from 0.5W to 2W at a top clock speed of 2GHz. Final clockspeeds are still subject to official confirmation.

The return of HyperThreading - planned also for a big comeback with Nehalem - is a surprise, but in today's multithreaded landscape it will be very useful. The current scenario is far better than the one that the Pentium 4 3.06 faced when it came to life, while the performance increase we could expect from using HT in this core might even surpass the optimistic 30% from that time.
The reason for that is that when the processor stalls, or is unable to have it's two execution units in utilization, it may pull another thread for execution. In this case however, being an in-order core, and if proper compiler optimization was not well executed, the CPU will stall much more frequently than a typical out-of-order CPU. That brings a bigger performance increase to the table then what was possible with the P4.

Tukwila is also an interesting chip, and a big one, fruit of it's two billion transistors.
It features the typical Itanium ISA, IA64, but brings some improvements which will also come with Nehalem, like dual integrated memory controllers, QuickPath interconnect - similar to AMD's HyperTransport - and a whooping 30MB of L2 cache. Hyperthreading is also back and, being a quad-core, allows for running 8 threads at the same time, in hardware.
The expected TDP for Tukwila is 130W, because the processor is still manufactured at the old 65nm process. Reasons for that were outlined by Intel's CTO, Justin Rattner:

Rattner said that the decision to stick with 65nm was "probably" due to reliability concerns. "90 percent of the latches on Tukwila are SER-hardened," he said, referring to error-resistant logic circuits designed to shrug off the effects of cosmic rays. As components in processors get smaller, they are more prone to electronic errors caused by radiation, he said. Rattner added that it was possible that the next generation of Itanium would skip 45nm and go straight to 32nm.

Tukwila will come towards the end of 2008, in a LGA 1366 derived socket format, and will feature socket and chipset compatibility with Nehalem based Xeon CPUs, due to the use of the new QuickPath interconnect.

The original news post is available at ZDNet.
Keywords: Intel Silverthorne HT hypertreading tukwila

CrossFire X benchmarks - RV670 times four

What kind of speed would a Core 2 QX9770 @ 4.2GHz and two HD3870X2 cards provide? One worth as gaming machine, and then some.

The quad CrossFire driver peeped out of it's hole and my colleagues at FPSLabs grabbed it for a quick test, to check if it's worth spending 800eur in graphics subsystem.

The driver is working well in FEAR and a bit worse in Call of Juarez, providing a 77% and 50% increase over one card, respectively. 3DMark2006 is worse at 15%, but that was already a little CPU bound with two RV670's, let alone four.
More results would be welcomed but that will have to wait. For now AMD/ATI looks poised to regain the performance crown while Nvidia doesn't strike back with the 9800GX2.
March will be quite the month.

XFX 8800GS XXX overclocking impressions

As I mentioned a few days ago, the XXX cards from XFX are usually very overclockable samples, but now I had the opportunity to confirm that claim.
The above shot was taken running the card with the stock aircooler, while running Linux with the latest Nvidia drivers and Coobits for Linux.
This result of 773MHz for the core and 998MHz, or 1996MHz, for the RAM was taken by using the automatic overclocking function, so improvement is likely.
This is a very good result, especially for the RAM which gets to almost 2GHz from the stock 1.6GHz. The increase in core speed is also quite impressive, considering it was already up to 680MHz in the stock overclock.

The card definitely features a better PCB than the ones used for most 8800GTs on the market, while the stock cooler has definitely been stripped down to a more simple design.


More coming soon.

Nvidia's GT200 based card to consume 250W

Nvidia's next generation card, based on the chip codenamed GT200, will consume in excess of 200W, 250 to be precise.
Various sources have stated that the chip will be manufactured at 65nm, the same as current G9x chips. Nvidia usually doesn't risk to bring high-end chips with new a new process, preferring to deliver them manufactured in the process with which they have already some experience, common practice since the GeForce FX5800 fiasco.

The wattage claim seems valid since current G92 chips already feature 754 million transistors and consume around 100W. The GT200 card is rumored to surpass the one billion mark while aiming to bring more performance than the upcoming 9800GX2.
Considering that the 9800GX2 is expected to surpass 200W of power consumption, while delivering double the performance of an 8800GTS with 512MB, they look like reliable numbers. Nvidia's partner, TSMC, isn't pulling any 45nm high-K, low leakage, process out of it's "hat" in the next few months and this will be an high-end - to the extreme - graphics card.
The power consumption of the high-speed RAM set to be used will also play a role in these numbers.

A release date for the GT200 is rumored a few months after the 9800GX2, around June-July, possibly earlier.

More details should come around CeBIT, in early March.
Keywords: nvidia GT200 power consumption

Seagate 7200.11 500GB review

Keywords: Seagate 500gb 7200.11 review raptor
Next Level Hardware has reviewed the new Barracuda 7200.11, in the 500GB flavor.
Featuring capacities of up to 1TB and 32MB of cache, the new HDD is one of the top performing models currently available that has a 7200RPM spindle speed.

The new model manages to challenge Western Digital's 150GB Raptor - a 10k RPM drive - while it also manages to overtake it in non random access bound usage.
Raid 0 benchmarks are also available, where two drives simply destroy any hopes of the single Raptor. One of the only only exceptions is the time it took to boot windows, due to the higher Random Access Time(RAT) of each single drive. The RAID 0 RAT results are very good but probably not accurate or windows booting should have also been faster that the Raptor.
The other, not so good results from a RAID 0 of 7200.11 drives, place it in the same level as a single raptor, making them a good option for someone looking for both harddisk space and speed.

The new Seagate drive isn't all alone though, the 7K1000 from Hitachi also is able to challenge the Raptor, in a single drive environment. It has been available for almost a year and also features 32MB of cache.

Both the 7K1000 and the new 500GB 7200.11 are very capable drives, providing a lower cost per GB and a good performance alternative, leaving the Raptors as the choice for the hardcore enthusiast looking for only the best, without looking at cost.

Intel readying Core 2 Duo E8600

As AMD gets ready to launch triple core Phenoms, Intel prepares to strike back at any potential threat. The current rumor is that a new E8600 Core 2 Duo model is coming, it's based on the "Wolfdale" core and runs at 3.33GHz with an 1333MHz FSB.

While still a rumor, 3.33GHz are more than easily achievable with the new cores, especially in dual cores, so take this as a serious possibility.

This is not something really needed right now, probably not even before the B3 Phenom X3 makes a debut, but Intel will have it ready nonetheless, so reviewers can have something palpable to compare the X3 against.
By doing this, Intel is also able to keep the quad-cores priced higher and have the dual-cores, which are cheaper to manufacture, fighting with AMD's Phenom X3. The performance should still weigh to the blue side due to the low clocks the Phenom X3 will have.

No word yet on pricing.
Keywords: Intel core 2 duo e8600 3.33GHz 1333

Asus releases list of boards compatible with 1600MHz FSB CPUs

Asus has revealed a list of boards that support the upcoming 1600MHz FSB processors, the first of which will be the QX9770, a 3.2GHz 45nm "Yorkfield".
No big surprises since they are all based on the latest chipsets from Intel. It was known that most of them could work with the new FSB, without official support by Intel. Kudos to Asus for providing it, when they weren't obligated to.
Here's hoping for others to follow.

Support for other boards is not out of the question, since P965 chipsets are capable of surpassing this kind of FSB without much trouble. The P5B Deluxe, for example, is starting to become a legend among Intel 775 users - the new support of 45nm CPUs has given a new life to the board.
At the same time, 1600MHz FSB for P965 boards, and the like, while possible for some, shouldn't come around due to the huge amount of boards that Asus would have to certify. The 1333MHz FSB support for that generation of boards was already a gift.

List of boards supported and needed BIOS version:


The original news post is available at TechConnect.

Ubuntu flavored XPS M1330 reaches the US - Priced just right

Keywords: Ubuntu dell price xps m1330
After a rather strange pricing strategy by Dell in Europe, the United States have now also received Ubuntu as an option for the XPS M1330.
The offer is limited to a single color, Tuxedo Black, which hopefully won't be a problem when the offer matures.

The price is just right this time around - it's priced $45 cheaper than the Vista option.
While some people may argue that the price could be brought further down, the fact is that this is a good discount for people who would probably never use Windows anyway. Also, now you have the option of buying the more "sensible" retail version, which won't get legally stuck with the same computer if you decide to upgrade later.

Meanwhile, the situation in Europe has indeed evolved and the price difference is now of £50, considerably bigger than in the US. Take a look at the Ubuntu and Windows customization pages.
The only visible difference, for now, is that the Vista version packs the fingerprint reader for free while it is not available as an option with Ubuntu. It is due probably to the lack of drivers for Ubuntu and is something that most people won't miss either.

The new pricing allows for both customer savings and probably also some small gains for Dell or, at least, to help pay for R&D.
Kudos for Dell Europe for bringing the price down, to the deserved spot.