|
|
 |
|
|
|
 According to information gathered by the The Inquirer, TSMC has had to delay the introduction of 40nm. In March, the company indicated that it hoped to produce the first wafers before the end of June. It is now a question of February/March 2009 if everything goes well, which represents an non-negligible delay.
Thus, Altera, which is generally one of the first to benefit from the Taiwanese manufacturer’s production lines, will not offer 40nm chips this year. According to our colleagues, other large companies are concerned including Nvidia. However, given the fact that they are only starting to use 55nm with the G92b (the GTX 200 is still produced in 65nm), this actually shouldn’t be too detrimental. Unsurprisingly, ATI will also be affected and consequently we shouldn’t expect to see any 40nm GPUs from them before the first quarter of next year. This is nothing too dramatic in the long run because 55nm will still enable offering interesting GPUs for some time and the Intel Larrabee will only arrive much later. |
| |
|
 Pat Gelsinger, the first to have carried the title “Chief Technology Officer” at Intel and the current Vice-President and General Manager of the Digital Enterprise Group made a series of interesting statements yesterday in San Francisco in a preview of celebrations that will take place next month for an event marking the 40 years of Integrated Electronics Corporation, aka Intel.
According to him, starting in roughly 2012 the number of companies that have their own "Fabs" will begin to significantly drop off with the transition to 450mm wafers. This figure will eventually be less than 10 and he estimates a capital of more than 1 billion dollars to be able to be part of this very select group. You may recall the creator of the x86 has planned on partnering up with Samsung and TSMC for this next generation of wafers.
Concerning the size of circuits, he thinks someday it should be possible to go under the 10nm bar even if it is not yet known how. However, there is no hurry as his firm will first offer 32nm, 22nm and then 16nm if everything goes well. As a corollary, this means that Moore’s law, which at any rate applies in a very "flexible" way to current needs, should still remain relevant for some time. |
| |
|
 While the Core 2 Extreme QX6800 and QX6850 will soon be retired, Intel should launch a "standard" Q6850 on July 20th.
Set at 3 GHz with 8 MB of L2 cache and an FSB1333, on the technical level it will essentially set itself apart from its bigger 65nm Extreme siblings by its blocked multiplier. Its price, still unknown, should logically reflect its statue of being an end of series CPU which enables selling off some of what comes off of 65nm production lines. Moreover, it should be far from the $999 usually asked for in the high end. We will just have to see if its TDP will be maintained at 130W or lowered and if Intel will take advantage of its introduction to reduce the prices of certain other existing models. |
| |
|
 Finally! After a few very long months, a new "big" GPU has arrived. Nvidia’s GT200 pushes GeForce 8 architecture further and promises to replace the GeForce 9800 GX2 by a card equipped with a single GPU.
> Product review: The Nvidia GeForce GTX 280 & 260 |
| |
|
|
Copyright © 1997- Hardware.fr SARL. All rights reserved.
Read our privacy guidelines.
|
|
