|
|
 |
|
|
|
 It’s finally with a 2 year delay that Centaur Technology, the CPU subsidiary of VIA, is planning on launching a new processor, the Isaiah.
Just like its predecessors, it is first and foremost a low power CPU. Energy consumption was announced as being identical to that of the C7, with which it is also pin to pin compatible (or at least it seems with the 21x21 mm NanoBGA version). Like the latter, cryptographic acceleration functions, called PadLock, are incorporated into the silicon. In terms of frequencies, it should first attain 2 GHz which the C7 already does.
So what’s new? The Isaiah can still function on the FSB800 but certain models will now support the FSB1333. Engraved in 65nm, it is equipped with the transistors necessary to extend its x86 instructions set to 64bits. There is also virtualization technology. It keeps its two L1 caches of 64 KB and L2 changes to 1 MB.
As for the FPU, the press release explains that Isaiah integrates the world’s fastest floating point x86 calculation unit capable of executing four floating point additions and four multiplies per clock cycle and with a new algorithm that minimizes waiting time. Some will say that a high performance FPU has been somewhat of a myth for VIA/Cyrix/Centaur...
This involves the support of new SSE instructions which is surprising because the C7 already supported SSE3 and SSE4 was only introduced at Intel with the Penryn. VIA documentation remains vague for the moment on this point.
Expected out in this semester, it is described by its creator as having 2 to 4 times the performances of the C7. Now we will just have to see if this is really enough to run the most demanding applications including high definition video and 3D gamesas is promised.
 |
| |
|
AMD took advantage of the launch of its new graphic cards to unveil Hybrid Graphics, a response to Nvidia’s Hybrid SLI or rather its PowerBoost component as this AMD technology only involves the increase of performances by enabling the chipset’s integrated core to work in combination with the graphic card.
The manufacturer already has a technology destined to reduce power consumption by deactivating the graphic card and using the integrated core when the load is minimal, although this is reserved to laptops. AMD told us that the same capability would not be available for PC desktops in the immediate future. Given that it is too much of a ‘’fix it’’ type solution at the current time, the manufacturer prefers not to spend its resources in its validation and practical implementation.
The AMD demonstration platform. You may have noticed that contrary to Nvidia’s solution, it’s the graphic card’s connection that is used.Hybrid Graphics relies on use of an RS780 based platform, the future chipset with a DirectX 10 integrated core. It will be limited to combination with a Radeon HD 2400 or Radeon HD 3400. This is only logical as with a more powerful card the gains would be nil or performances would actually be reduced. Graphic cards are in fact limited to twice the performance of the integrated core in AFR mode (each GPU calculates every other image), which is the most common. If the graphic card has more than twice the power of the integrated graphic core (not uncommon) performances drop. This is also true for Nvidia’s PowerBoost. Thus, these are solutions reserved for the entry level and we believe of lesser interest. Even if all performance gains are welcome on the entry level, spending a few extra Euros on a higher level graphic card seems like a better idea than opting for a multi-GPU system in this segment. |
| |
|
|
Copyright © 1997- Hardware.fr SARL. All rights reserved.
Read our privacy guidelines.
|
|
