Logo   Banner   TopRight
TopUnder
Transparent
Lucid Hydra 200 Multi-GPU Performance Revealed
Transparent
Date: Nov 11, 2009
Section:Graphics/Sound
Author: Mathew Miranda
Transparent
Introduction


About a year ago, Lucid started making waves in the graphics space with claims of being able to revolutionize multi-GPU computing. It promised consumers the ability to add any graphics card, unrestricted by model or vendor, to an existing set up and achieve highly efficient load balancing with near linear performance increases. This option presumably provides consumers the flexibility to buy an ATI graphics card, install it next to an NVIDIA model on the same motherboard, and see a boost in graphics rendering performance close to the sum of both individual components.

Today we have news of this technology finally coming to life. The Hydra 200 processor is currently in mass production and will soon be featured on retail motherboards, starting with MSI's Big Bang Fuzion P55 motherboard. While this processor will initially be implemented on high end boards, it is actually better suited for the much larger mainstream market segment. Instead of replacing old hardware, upgraders and system builders alike can presumably optimize their existing graphics performance by adding a new affordable videocard.

Recently, we sat down with Lucid and had the chance to get some hands-on time with their product to see how well it performed. Accordingly, we installed multiple graphics card configurations on Lucid's test system and ran several gaming benchmarks to find the truth about Hydra's capabilities and finally lay some of the rumors to rest. So what's the verdict? Read on as HotHardware gives you an exclusive look at one of the PC industry's most intriguing technologies.   


MSI's Big Bang Fuzion Motherboard featuring Hydra 200 Processor

Lucid Hydra Implementation
ASIC Hardware



"The implementation of the Hydra engine is done in ASIC hardware with the support of a software driver. The ASIC is located between the Northbridge and the GPUs, as shown in the figure. The Hydra engine scales performance of multi-GPU configurations from any GPU vender and will scale relative to each GPU's individual performance. In other words, the GPUs do not need to be identical. The Hydra ASIC  handles all connectivity between the CPU, the GPU, and between the GPUs through a full-duplex wired speed implementation. As such, the solution is connector free and does not require any GPU to GPU connector." - Lucid



Hydra Engine Graphics User Interface

Hydra's control panel is simple to use and provides users the ability to enable or disable installed graphics cards within the operating system. With Hydra enabled, the GUI can be minimized to the system tray while graphics performance is automatically optimized.

Transparent
Test Setup and 3DMark06

HOW LUCID CONFIGURED THE TEST SYSTEM: We tested the graphics cards in this article using a unique setup provided by Lucid. The main components consist of a Gigabyte GA-EX58-UD3R motherboard powered by Intel's Core i7 920 quad-core processor and 2GB of OCZ DDR3 RAM. Of course, this particular motherboard does not feature Lucid technology so as a result, the graphics cards were installed on a special evaluation board featuring the Hydra 200 chip. The evaluation board was connected via PCIe card installed on an x16 slot on the GA-EX58-UD3R motherboard. We were told that this test setup simulates the performance of the Hydra 200 when integrated on a mainboard.



 

Hydra 200 Test System
Core i7 Powered

Hardware Used:
Intel Core i7 920 (2.66GHz)

Gigabyte EX58-UD3R
(X58 Express)

Radeon HD 4890
Radeon HD 4890
GeForce GTX 260
GeForce GTX 260

OCZ 2GB DDR3-1333

Relevant Software:
Windows 7 32-bit

Benchmarks Used:

3DMark06
Operation Flashpoint: Dragon Rising
Call of Juarez Benchmark
Stormrise

Lucid had several graphics cards and video games available for testing. Since time was a limiting factor, we could not test every videocard and game at our disposal but came up with five configurations and chose a handful of benchmarks in order to provide comparable results.

To find our baseline scores, we ran the benchmarks using an HD 4890 and GTX 260 individually. Next, we added identical graphics cards to the setup to determine scaling performance. In other words, we tested a pair of 4890's and two 260's together. The final combination consisted of a GTX 260 and HD 4890 simultaneously.

Some may wonder why we failed to use MSI's Big Bang Fuzion in our testing. For our meeting, Lucid actually had a separate test bed ready that featuried MSI's motherboard, an HD 4890, and a GTX 260. Initially, we sat down for a few minutes and played Operation Flashpoint: Dragon Rising on the system and acknowledge that it played very smoothly. Unfortunately, we must wait to see the performance numbers from MSI's Big Bang platform as they won't be released until the product has launched. 

Futuremark 3DMark06
Synthetic DirectX Gaming


At first, we wanted to run both 3DMark Vantage and 3DMark06 to produce additional performance scores that would support our gaming tests. Unfortunately, we experienced frequent crashing as driver issues prevented us from completing Vantage runs, so we stuck with just the 3DMark06 numbers as it proved to be a lot more stable than Vantage. At any rate, the results coincide with our expectations as the dual HD 4890 configuration provided the highest 3D Marks, followed by the mixed ATI / NVIDIA combination, and then the dual GTX 260's. 
Transparent
Operation Flashpoint, Call of Juarez, and Storm Rise

Operation Flashpoint: Dragon Rising
Real World Gaming Performance

 

We activated FRAPS, set the resolution to 1920x1200, and cranked up the image quality settings in the recently released Operation Flashpoint: Dragon Rising. Here, impressive scaling took place as we saw an 89% performance increase when going from a single HD 4890 to dual card configuration. The transition from one to two GTX 260's in tandem also provided an effective 74% increase in performance.
 

Call of Juarez
DirectX 10 Benchmark

 

The Call of Juarez benchmark was one of the first to show off DX10 features while offering precise repeatability. Once again, we see exceptional scaling when going from single to dual card configurations. Two GTX 260's demonstrated an 85% performance increase over a single card, while two HD 4890's produced an 83% gain above an individual card's score. 
 

Stormrise
DX10 Gaming Performance

 

Stormrise verified our prior GTX 260 performance results with 85% scaling once more. However, two HD 4890's in tandem fell short of previous outcomes with only a 44% performance gain, much lower than the percentage we saw in the other games. Its worth noting that the ATI / NVIDIA combination consistently provided the second best scores throughout testing while generally meeting our expectations of this configuration.

Transparent
What's Next?

Performance Summary: After running Lucid's test bed through an assortment of tests in several graphics configurations, we have a better idea on how well the Hydra 200 performs. Using the components available to us, we saw impressive scaling in most instances, which peaked at 89% in dual-GPU mode. The mixed ATI / NVIDIA combo ran pretty well and fell within the expected range of performance throughout testing. 

 

We think it's important to keep in mind that this was just a preview of Hydra's potential. Testing was performed using an evaluation board that obviously is not a retail product available to consumers. Rather, the setup allowed us to see the capabilities of Lucid's new technology before any Hydra equipped motherboards actually hit the streets.

 

Although there were driver issues that prevented us from completing some benchmarks, we've been informed that the problems will be resolved over time as Lucid's driver development team works to squash bugs with a wide range of hardware and software. With that, we are definitely looking forward to future Hydra-enabled motherboards and the potential new levels of performance that can be reached by using them.   




Content Property of HotHardware.com