Wednesday, May 26, 2010

IT News HeadLines (Elite Bastards) 26/05/2010


Elite Bastards
Elite Bastards review: ASUS M4A89GTD PRO/USB3 motherboard

AMD's recent Phenom II X6 launch has arguably put the company back in the game in the eyes of some enthusiasts, but what motherboard should you pair a new AMD CPU with? We take a look at ASUS' M4A89GTD PRO/USB3 which makes use of AMD's newest 890GX chipset (complete with integrated graphics) and SB850 south bridge, topped off with CPU core unlocking functionality and USB 3.0 support. That makes for lots of tasty buzzwords, but how is the actual motherboard itself?

Despite the large leap in nomenclature, the basic configuration of the 890GX mirrors that of its direct predecessor the 785G exactly - Indeed, it's actually the same 55 nanometre, 200 million transistor chip, albeit speed binned to ensure that any 890GX parts are the cream of their respective crop. These means that what we have here is an AM3 socket supporting part which can handle up to DDR3-1333MHz officially, with sixteen PCI Express 2.0 lanes reserved for graphics (with support for two electrical 8x slots for a CrossFire configuration if needed) and six PCI Express 2.0 lanes for other devices.

The reason for this speed binning we just mentioned is all about those integrated graphics, allowing the 890GX to sport a GPU labelled the Radeon HD 4290 by AMD - This GPU supports DirectX 10.1 with a core clock speed of 700MHz (compared to 500MHz for the 785G's GPU), while holding the same complement of forty Stream Processors as its predecessor, complete with the same UVD 2.0 video decoding engine with support for dual-stream decoding and a full complement of video outputs (minus audio bitstreaming sadly, although audio pass-through for HDMI remains intact). Although the Radeon HD 4290 shares system memory for its frame buffer, this part also features 128MB of DDR-1333 for its Sideport cache, designed to improve performance by offering the GPU at least some dedicated memory to work with.

ASUS M4A89GTD PRO/USB3 motherboard review

View or post comments.


Read More ...

NVIDIA's VP for CUDA and PhysX moves to AMD?

If true, this is quite the coup for AMD...

Word has reached KitGuru that Manju Hegde, nVidia’s VP for CUDA and PhysX, is moving to AMD. What can we infer from the situation, when Vidia’s own VPs seem to believe that Fusion is the Future?

The former professor of electrical engineering for Washington and Louisiana State Universities, founded Ageia in 2002 and has as responsible as anyone for putting physics and GPGPU centre stage in the evolving graphics market.

As nVidia confirms, Hegde has been on a mission to improve game play and Jen Hsun Huang was so impressed with the physics guru’s talents that, after buying Ageia out, Hegde was given a critical role in leading the development and deployment not only of PhysX, but also CUDA.

KitGuru has the full story.

View or post comments.


Read More ...

NVIDIA GeForce GTX 465 preview

The fact that a GeForce GTX 465 SKU is on its way is hardly one of NVIDIA's better kept secrets, but how will it perform? This preview using a pre-release sample of the board should give us some clues.

The name is rather odd, by the way. It would be logical for the new card to get a model number of 460 but Nvidia uses the digit 5, probably because the name “GeForce GTX 460” is reserved for a card with a simplified and even cheaper version of the GF100 core. Technically, the GeForce GTX 465 is the result of deactivation of one more processor cluster. Its GPU has only three such clusters left. The number of stream multiprocessors is reduced to 11, the number of ALUs, to 352. This is yet another odd number because the GeForce GTX 460 had been supposed to have 384 CUDA cores on board. The GF100 is already inferior to its predecessor in terms of texture-mapping units, and there are only 44 of them left in the GeForce GTX 465 – a ridiculously small number even compared with the GeForce GTX 275. This dramatic reduction in GPU resources will surely affect the performance of the new card in 3D applications. It might be justified if it helped improve the frequency potential. However, the GPU clock rates are the same as with the GeForce GTX 470, namely 607/1215 MHz. The RBE subsystem is all right: it is equivalent to that of the G200b-based cards and represented by 32 raster operators. The memory bus is only 256 bits wide, which is not good at all, even though the GeForce GTX 465 uses GDDR5. The GeForce GTX 275 had a peak memory bandwidth of 127 GBps with GDDR3, but the new card is slower at 102.7 GBps. Thus, we cannot expect the GeForce GTX 465 to perform fast at high resolutions and the reason is its low memory frequency of only 802 (3208) MHz. That’s not serious for modern GDDR5 memory chips.

X-Bit Labs have the full preview.

View or post comments.


Read More ...

Futuremark announces 3DMark 11

Although DirectX 11 has been with us for a little while now along with compatible hardware, and we've even seen a handful of DirectX 11 game titles emerge, one thing has been missing from our experience of this latest iteration of Microsoft's API thus far - A new version of 3DMark. We've known that it's been coming for a little while now, but today Futuremark have officially announced 3DMark 11, together with some screenshots, information and a High Definition video trailer.

You can find all of these details on our 3DMark 11 announcement page, but if you're too lazy to click, check out that video below.

View or post comments.


Read More ...

No comments: