Information Technology and Software Development

We still need AMD, don’t you know?

Thu, 17 Apr 2008 07:24:40 +0000

Column: The voice of IT experience

Despite the fact that when it posts its results tomorrow, those results will not make silicon investors shriek with glee, no one in the industry – not even Intel – wants AMD to fail.

This is all a little bit like a game of European chess. The Red King is obviously Hector Ruiz, while the White King is, without a shadow of a doubt, Paul Otellini.

Identifying the other 30 pieces on the board with real human beings is harder. Who, for example is the Red Queen, and who the White Queen?

The bishops are obviously the PR spinners to the left and to the right of the monarchs – ready, once the pawns make an opening, to go diagonally wherever there is a diagonal place to go.

The knights – or horses, as my dad used to call them – are ready to pursue the strange gambit of going two up and one across. These are probably the field area engineers.

The rooks, also called the castles by uneducated oiks like me, normally wait for the end game before they move vertically and horizontally on the pixellated board.

Meanwhile the infantry – the poor bloody foot soldiers – are forced to advance and can never retreat unless they cross the whole board to transform into any piece they want to be, except pawns, of course.

The long game can get remarkably tedious. Geniuses in this frankly Asian board game can apparently tell what’s exactly going to happen if there are only a handful of pieces left on the board. Sooner or later a king will be toppled. And if only the kings are left, then it’s an endless stalemate.

Sometimes it’s good to know when to give up, to surrender, knowing that you can live to fight another day on the chequered panel.

When there are only two competitors on the board, at least spectators can throw lots and gamble on the outcome, rather like those who invest on the short and long game on Wall Street. If there’s only one team, then the game is frankly not worth betting on or playing.

After tiring your brain out by thinking about the long and short of it all, you might fancy playing Fox and Hounds, where one piece is the cunning Reynard pitted against four baying hounds.

Sorry for this long excursion into board games. But if there’s only one player left on the board, and it’s Chipzilla, we may as well all go off and play quoits, just like Alice in Wonderland, and using pink flamingoes as rackets.

We’re all praying for the White Rabbit to make an appearance, glance at his pocket watch, and panic because the spring has bust and the clock is overclocking, with one day spinning by in the space of one real minute.

Maybe Microsoft will help bale out AMD. It’s certainly not in Microsoft’s interest for its arch enemy Intel to win, win and win again.

Source: http://www.fudzilla.com/index.php?option=com_content&task=view&id=6871&Itemid=1

RV770 is already in production

admin — Wed, 09 Apr 2008 07:58:30 +0000

Test one

We learned that RV770 GPU already went to production and there is a big chance to see R700 generation much earlier than anyone has expected. TSMC is already doing a pilot production and the chips are, as we reported earlier, developed in 55nm machitecture.

The way it looks now, there is a strong possibility that R700 should show its face at Computex (June 3) while the launch itself might be shortly before or after Computex. We still don’t have the final details.

RV770 is the based on R700 and R700 consists of two RV770 chips. From what we know, RV770 should be at least fifty percent faster than the RV670.

If all goes well there is a big chance that R700 will launch in Q2 but rather later than earlier. We are still in the dark about the real details of D10U 30 as this GPU(s) has a similar schedule and is ought to launch in Q2.

Source: http://www.fudzilla.com/index.php?option=com_content&task=view&id=6721&Itemid=1

The fastest card in the world is 3870 X2

admin — Mon, 07 Apr 2008 11:42:51 +0000

Says Virtualmark

We were surprised to learn that Futuremarks new Virtualmark benchmark makes you believe that a heavily overclocked Radeon 3870 X2 is the fastest graphics card around.

Virtual mark actually compares the scores of your machine to similar systems and usually gets you some approximate score. At least its fast but I wouldn’t count on much accuracy.

The machine with Core 2 Extreme QX9650 quad core at 5820MHz with Radeon 3870 X2 probably in Crossfire X scores 32513 Virtual marks. This looks to be the fastest card around and we suspect that we are talking about two dual chip Radeon 3870 X2 and that both CPU and GPUs are heavily overclocked.

The common reason tells us that Geforce 9800 GX2 in quad SLI should be the fastest system around, but not according to Virtualmark. I guess that ATI simply clocks and scores better.

Source: http://www.fudzilla.com/index.php?option=com_content&task=view&id=6684&Itemid=1

Intel FUD’s Phenom Tri cores

admin — Mon, 07 Apr 2008 11:32:08 +0000

Getting scared aren’t we?

It looks like OEMs are responding to the Tri core better than Intel expected. That can explain why is Intel going after AMD’s Tri core processors claiming its dual cores are actually better. Intel understands that OEMs will send a message that three cores are better than two, and this is the how majority of the market will accept it.

Intel has sent a funny little document where it claims that its Pentium E series of Dual cores performs and scores better than Tri core. To spice it up, Intel did mention that Tri core is not design to be tri core, but it’s rather a CPU with one core defective or disabled.

We did test one of these first OEM cores and it scores almost as fast as a Quad core. You can check that review here. Once AMD cranks up the speeds of its tri cores, we believe Intel might find itself in trouble.

Furthermore, Intel tells that in System 2007 its E2220 scores a point better than a Phenom 8600 2.3GHz Tri core. To sweeten it up Intel “Ranges are performance estimates” for Phenom 8600 and 8400, which means that Intel actually didn’t have these CPUs in hand but it still decided to send this rather questionable presentation and base its scores on estimates.

Here is one of the images.

Source: http://www.fudzilla.com/index.php?option=com_content&task=view&id=6683&Itemid=1

RV670 rev 12 are getting close

admin — Fri, 28 Mar 2008 13:37:17 +0000

Bit better overclock, but better yields

We learned that RV670 revision 12 chips will start showing up in late April or early May. The new chips will be easier to produce and you can expect a bit higher overclocking scores, but don’t expect miracles.

This is a step that will affect the yields and price of the chip itself and this might be a long awaited solution to get enough of RV670 based products to the market. This was one of the biggest problems for ATI since it launched, as its products are affordable and perform great for the price.

RV670 rev 12 chip is the last step before AMD moves to a RV770 series a brand new 55nm marchitecture.
Source:

Radeon HD 3850X2 coming, Radeon HD 38×0 prices drops

admin — Tue, 25 Mar 2008 13:18:28 +0000

Radeon HD 3870X2 uses two RV670 chips on a single PCB. Both Radeon HD 3870 and Radeon HD 3850 are based on the very same chip, the only difference is the frequency. Think about for it a short while and you realize that launching Radeon HD 3850X2 would be quite a simple task. Simply downclock Radeon HD 3870X2, which is exactly what AMD has done and will present the new card around Computex in June.

By then, CrossFireX performance should be up, and this would be an easy way for AMD to empty out stock of any residual RV670 chips as the R700 series should be around the corner. That and the new, low prices.

At the same time, we’re still waiting for confirmation on the new and faster dual-GPU cards which is suppose to sport the new A12 revision. Maybe Radeon HD 3890X2 coming too?
Source: http://www.nordichardware.com/news,7497.html

Exclusive: ASUS EAH3850 Trinity - three GPUs with one card

admin — Tue, 25 Mar 2008 13:14:52 +0000

We knew that CrossFireX would bring new multi-GPU setups, but we did not expect anyone to actually use more than two cores with a single card. ASUS proved us wrong, as EAH3850 Trinity is a tri-core card. It has three RV670 cores cooled by a heatpipe construction and a water block, all in a single card. The card is still just a concept and it hasn’t been decided when or if it will be released, however we do know that it works. When benchmarks with the card will surface remains to be seen, but keep an eye open the upcoming weeks.

Below are some pictures of the card, which makes at least our mouths water
As you can see, ASUS has chosen to use a kind of modular design where the two cores are located on the back and one on the front, each on their own daughter card. The cores are directly cooled by heatsinks and heatpipes that transfers the heat away from the cores to the heatsink located in the rear of the card, which in turn is cooled by a water block.

Four DVI ports which can be used with up to four monitors, when CrossFire is deactivated.
Along with the card comes a big water cooling unit which is fitted into two empty 5.25″ slot
Source: http://www.nordichardware.com/news,7543.html