Google translation of 3DCenter article [HERE] . The latest rumors about NVIDIA’s GF110 chip mentions now 512 shader units and 128 texture units on a 512-bit DDR memory interface. This is somewhat surprising, as this against the GF100-chip (512 shader units, 64 TMUs and 384-bit DDR memory interface)  gain in computing power wouldn’t be very high, but rather a significant gain in TextureFillRate (+100%) and memory bandwidth (33%) . It also implies for  shader clusters: The GF100 this chip has 32 shader units and 4 TMUs in a shader cluster, the smaller Fermi chips GF104, GF106, and GF108, however, have 48 shader units and 4 TMUs in a shader cluster. In a GF110-chip would include a shader cluster with 32 shader units and 8 TMUs – alternatively 16 shader units and 4 TMUs .

Possible new high-end nVidia Chips
GF100 GF100b new stepping of the chips with lower power consumption, thus the possibility of a graphics card with the full 512 shader units and possibly a little more time to little or no power loss supplement
Potential compared with GeForce GTX 480: +5% to +10%

GF110 with 512SE new chip with 512 shader units and 128 TMUs to a 512-bit DDR memory interface, without the GF100 ballast with further redesign of the shader cluster, the area with 430-480mm ² smaller than GF100, probably even higher taktbar as this with the same or smaller power loss
Potential compared with GeForce GTX 480: +20% to +30%

GF110 with 576SE new chip with 576 shader units and 96 TMUs to a 384-bit DDR memory interface, without the GF100 ballast for GF104-model, the area with 430-480mm ² smaller than GF100, probably even higher taktbar than this with the same or smaller power loss
Potential compared with GeForce GTX 480: +20% to +30%

GF110 with 768SE new chip with 768 shader units and 128 TMUs to a 512-bit DDR memory interface, without the GF100 ballast for GF104-model, the area with 550-650mm ² larger than GF100, resulting in higher power dissipation, the target market of graphics chips would be rather then the dual-chip solutions from AMD
Potential compared with GeForce GTX 480: about +50%

The real question is, of course, will like to get such a chip from nVidia more performance if the processing power at all or only slightly more likely (than a higher clock) is increasing – in fact this would be illogical. However, the source of new information on the GF110-chip can be trusted – it is the same which had been previously noted by us called the specifications of the AMD chips Barts and the Cayman, which is now apparently out to be correct. That the targeted power increase is only about a higher Texture Fill Rate and more memory bandwidth is also not entirely inconceivable, if one views the previous chip GT200 GeForce GTX looks 200 Series: There were 240 shader units, 80 texture units and a 512 bit DDR memory interface – if you double that of the chip hardware units since it is in fact the GF110-chip to the above data.

Maybe nVidia designs generally favor a higher Texture Fill Rate and more memory bandwidth, several previous nVidia chips interpret any case out. The changes in the shader cluster with the smaller Fermi chips GF104, GF106, and GF108 was already in the direction to give these chips per Texture Fill Rate computing power for the journey. Also plays high-end projects from nVidia Anyway, the point of the chip area and thus the overall manufacturability of a role is fairly certainly that the GF110 not GF100 ballast for professional applications with the bear, and thus very good saves chip area compared to this – all too much longer for new hardware units fit but as yet not into it so as not to land quickly on the size of a GF100 chips (or more). Maybe it was more economical nVidia ago by the chip area to build in more texture units rather than more shader units.

Also this solution is therefore better performance and even easier to implement than a previously discussed GF110 option with 576 shader units and 96 TMUs to a 384-bit DDR memory interface. In both cases, the performance forecast is +20% to +30% compared to the GeForce GTX 480, which should probably also be sufficient to come to the Cayman chip from AMD still can barely stand up to. His tape-out of the GF110 chip to the rest have long been behind and unconfirmed, but not implausible information to the chip and sales cycle rates are even finalized already, which would allow for a launch later this year. However, could the GF110 nVidia chip also deliberately delayed so that only AMD has to reveal to the Cayman chip and you can then immediately counter correspondingly better.

Shortcuts: How the PC Games Hardware reports should come along just the overclockers CPUs of the Sandy Bridge generation with smaller features : For the processors with activated multiplier both missing VT-d and TXT, both are support technologies for hardware virtualization. However, it remains the basic technology for hardware virtualization in the form of VT-x received, so that it seems as though Intel withheld only professional virtualization solutions, the K-models, not the hardware support of virtualization for domestic use but . prevent That being said, Intel has in the past third quarter despite a prior sales warning again posted record sales and profits: According to Heise Online the first time 11 billion U.S. dollars quarterly sales and first made nearly three billion U.S. dollars quarterly profit – this is far from any crisis. Remains to be seen whether these figures were actually reached on the basis of a booming market – or perhaps only on the basis of a higher market share from Intel to the detriment of a declining market share from AMD.

TechPowerUP:

NVIDIA to Counter Radeon HD 6970 ”Cayman” with GeForce GTX 580
AMD is undertaking its product development cycle at a breakneck pace, NVIDIA trailed it in the DirectX 11 and performance leadership race by months. This November, AMD will release the “Cayman” GPU, its newest high end GPU, the expectations are that it will outperform the NVIDIA GF100, that is a serious cause for concern, for the green team. It’s back to its old tactics of talking about GPUs that haven’t even taken shape, to try and water down AMD’s launch. Enter, the GF110, NVIDIA’s new high-end GPU under design, on which is based the GeForce GTX 580.

The new GPU is speculated to have 512 CUDA cores, 128 TMUs, and a 512-bit wide GDDR5 memory interface holding 2 GB of memory, with a TDP of close to that of the GeForce GTX 480. In the immediate future, there are prospects of a more realistic-sounding GF100b, which is basically GF100 with all its 512 CUDA cores enabled, while retaining its 384-bit GDDR5 memory interface, 64 TMUs, and slightly higher TDP than that of the GTX 480.

• READ MORE (Source): NVIDIA to Counter Radeon HD 6970 ”Cayman” with GeForce GTX 580 | techPowerUp

• READ MORE (Source): News des 13. Oktober 2010 | 3DCenter.org

• READ MORE (Source): NVIDIA GeForce GTX 580 with 512 Cores in December 2010? – 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming – Geeks3D.com