Workings of chip 200 of Hydra LucidLogix

[Sanju C]

Let me tell you that I am trying to learn the motherboard on chip level. So gathering much information as possible about each and every components present on the motherboard, where graphic card are also important. And let me tell you frankly that only on this site, I am getting detailed information about an asked queries. So I am posting here again. Now I want to know about the workings of chip 200 of Hydra LucidLogix. Before posting here, I have searched on web, but as usual I didn't find any described information. So lastly thought to post over here. I am just knowing that the chip LucidLogix is primarily a PCI Express controller that includes a RISC processor capable, with the help of a driver installed on the machine to perform load balancing graph, that is to say, divide graphics tasks required to create an image between several cards. But how does the chip 200 of Hydra LucidLogix works exactly? Curious to know about it and hoping that like always you people will help me.

[Saaarc]

If the motherboard from MSI offers some unique features, it was already sold on the market under the name Big Bang Trinergy. What makes the uniqueness of the Fuzion version is of course its use of the chip LucidLogix Hydra 200. What this chip first? This is a PCI Express controller with 48 rows. A link connects the chip to the x16 processor, while the remaining 32 lines are shared, in the case of the Big Bang Fuzion, three PCI Express 16x. Depending on how the slots are filled, they will work at 16x or 16x + 16x + 8x + 8x. So far, nothing very revolutionary indeed the Big Bang Trinergy, it uses a PCI Express nForce 200 for controlling its three PCI Express. The Hydra chip incorporates, however, a small RISC processor clocked at 300 MHz, whose role is to perform a "load balancing" chart.

[Rail racer]

The concept of load balancing is widely used in computing, particularly in the area of the network. For those who have never heard of, you have already used it without knowing it in the downloads section of lucidlogix. To offer fast downloads, they have multiple servers in parallel, all synchronized with each other. A system constantly checks the use of each server and when you want to download a file, you are automatically redirected to one of the least busy servers. Lucid chip attempts to apply a similar concept to graphics resources. You know, as well as ATI NVIDIA provide their own implementations of multi-GPU technology. They have in common to have fairly strict rules of operation: they are activated only if it has identical cards. NVIDIA for example, impossible to mix a GeForce GTX 470 with a GTX 480, while it will not mix a Radeon HD 5770 and AMD HD 5830. If these limitations exist, it is primarily because the render method of choice for SLI and Crossfire technologies is called Alternate Frame Rendering (AFR).

[Pollock]

As its name suggests, the idea is to calculate the images alternately to the first card, then the second. The division of tasks is extremely simple, graphic card, calculating an image on both. And that's where we understand why we find this restriction maps identical: if there was a faster card than the other, an image of two would be calculated more quickly. Yes, because we have a visual impression of jerks, which is more pronounced when the performance gap between the maps increases. If history SLI and Crossfire enabled other rendering modes (Split Frame Rendering, or Super Tiling ATI), they have gradually disappeared. Less effective even at that time (only the pixel shader was actually separate the geometry to be calculated entirely on two cards), they are now limited to the support of the oldest games as incompatible with some techniques used to achieve modern effects such as reflections or HDR (Render To Texture) which then require large transfers of data between two cards, limiting any benefit of these rendering modes.

[RogerFielden]

To allow different cards to work, the method "AFR" favored by the SLI and Crossfire is not applicable, then. And methods as the SFR (cut the image into two pieces, by adjusting the amount given to each card depending on the complexity of calculation) or SuperTiling (identical to the SFR, except that the screen is cut checkerboard for a fairer distribution: Avoiding a card calculates a sky blue and the other to have all the work of vegetation) are not resistant to modern rendering methods. Result is a new way that tries Lucid down a little lower in the abstraction: DirectX calls directly analyze and distribute the "objects" dynamically on maps. Clearly, instead of cutting the screen into two (two blocks, or a checkerboard), it cuts the image based on its content: trees and sky on a map, grass and the character on the another, etc. ... Hope these examples were not much sensible, but still you can understand the logic.

[StephanieRice]

Result is a new way that tries Lucid down a little lower in the abstraction: DirectX calls directly analyze and distribute the "objects" dynamically on maps.

To achieve this magic, it must first come to interfere before the drivers from ATI and NVIDIA. That is what Lucid with an additional driver. Specifically, we first install the drivers for each of its cards and finally you simply install the driver Lucid over the whole. To get to intelligently cut, Lucid has developed a system supposed to determine on the fly automatically cutting load. The company also constantly repeated that its system is universal and works with all DirectX applications. Here is the format used by Lucid to explain how it works :

The bulk of the system is the software part by the pilot. What the scheme does not quite correctly, that's the end of the process. Items calculated on other cards are returned to the main board on which you have connected your screen, the image is reconstructed and displayed.

[Quattro]

Pilot Lucid proposed to interface a small control panel that sets out some options, primarily the ability to enable and disable the technology. In case it disables only the primary display adapter (which is connected on your screen) is actually used. This allows to circumvent some limitations of the system. The other option is the control panel to display a Hydra logo on the top left of the screen when the technology is enabled. LucidLogix emphasizes the universal aspect of its solution, however, found inside the control panel management profiles. If an application is in the list, technology Lucid activated if it is not, it is disabled. The concept seems a bit odd, especially since there is no special setting on which we can influence. The pilot, for some of these profiles contains heuristics pre-calculated to aid in decomposition, but it is unclear which ones.