Review Index:
Feedback

ATI Mobility X800 and X300 GPUs

Author: Ryan Shrout
Subject: Mobile
Manufacturer: ATI
Tagged:

New Technologies in ATI's Mobility Line-up

Before we get into the exact specifications of the new products being announced today, there are a couple of features that ATI developed for their Mobility line that are of particular interest.


PowerPlay 5.0 and Dynamic Lane Count Switching


The first is ATI's PowerPlay 5.0 technology, which is a power saving technology to reduce the amount of power that the GPU uses during use and thus extends battery life — the key goal of nearly all notebook development.  This new revision of PowerPlay keeps the Power-on-Demand, clock gating, and different power modes, but adds something that only PCI Express can offer to graphics.


ATI developed DLCS, or Dynamic Lane Count Switching, that does exactly what it sounds like it does.  Taking advantage of the flexibility of the PCI Express interface and specifications, ATI can save power on the laptop by reducing the number of PCI Express lanes and are powered on at any given time.  With DLCS, the lane width can switch from the full x16 all the way down to x1.  The lane count changes take place during the ATI PowerPlay transitions just like the rest of the features ATI has.  Though our testing hasn't confirmed anything yet, ATI is claiming that DLCS could save as much as 30% of the power the GPU uses during what they call 'typical usage scenarios.'  This probably entails mostly office applications and a bit of 3D or movie playback.



Powerplay 5.0 Overview


This technology makes perfect sense and was easily predictable after all the talk we have heard that the current high end graphics cards don't come close to saturating the bandwidth of a x16 PCIe connection.  Coupling that with the fact that the PCIe bus provides the power to the graphics card, and that the power is scalable with the width of the total PCIe connection, lowering a graphics card to a x1 connection should provide adequate video performance for a lot of users while sucking less of the valuable power in your battery.  Once I can get my hands on a system to test this out, I'll be looking at how much of a real world difference it makes in various scenarios.


HyperMemory


HyperMemory is another feature that ATI has developed, but this one is in use on both their mobile GPUs as well as their new core logic chipsets, such as the Radeon XPRESS 200 chipset we previewed earlier this month. 


In short, by looking at the image above, HyperMemory Technology does what computers with integrated graphics have been doing for some: allow the GPU to use the system's main memory as a graphics buffer.  However, in this case, ATI has made changes that create a dedicated memory channel between the GPU, the chipset and the memory that facilitates faster transfers of the data between memory and the graphics chip, which has been a major problem in on-board graphics technologies in the past.  This connection is possible now that PCI Express bandwidth is being brought to the mobile market. 


 


HyperMemory Overview


Memory on high end discrete video cards is reaching over 1.2 GHz in speed with a dedicated connection between the GPU and the memory, while on-board system memory was limited by the 400 MHz speed (of DDR) and by the latency involved in going through the main core logic system.  HyperMemory Technology attempts to improve this situation and provide the ability to both decreases costs (by not requiring external memory) and providing options for the vendors to choose between on-board memory and system memory for their mobile graphics solution.

No comments posted yet.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.