ATI Radeon VE 32MB

by Matthew Witheiler on February 19, 2001 3:09 AM EST

The Chip

Unlike previous Radeon incarnations, the Radeon VE core only shares some features that other Radeon cores have. Let's see what ATI has chosen to drop in order to bring dual display capability to a Radeon powered card.

The core remains based on a 0.18 micron manufacturing process, meaning that ATI is able to sell the card running at a 183MHz core like other retail Radeon products. The core also continues to include ATI's memory bandwidth saving HyperZ technology. We already discussed HyperZ in depth, but in summary the technology essentially decreases the amount of data that must travel over the already crowded memory bus. HyperZ proved to give a big push in the case of the Radeon SDR, as the memory pipeline was half the size of the already memory bandwidth limited Radeon DDR. The technology will play even a larger role in the case of the Radeon VE, but more on that in a minute.

The chip continues to include support for three bump mapping techniques (emboss, dot product 3, and environment mapped bump mapping) as well as support for inverse discrete cosine transform, or iDCT for short. iDCT has been a specialty of ATI for quite some time now, and has gained them quite a following in the home video crowd. As we showed in our DVD Roundup, iDCT is able to significantly reduce the amount of work that a CPU has to do when performing DVD playback.

Aside from the aforementioned features, the Radeon VE shares little in common with the higher performing Radeon chips, securing its spot as a true budget card. First, and perhaps most importantly, is the deletion of one of the Radeon's two rendering pipelines. You may recall that the original Radeon chips possess two rendering pipelines capable of processing 3 texels per clock for a total of 6 texels per cycle. The Radeon VE core has a decreased total output of 3 texels per clock due to its single rendering pipeline. In actuality, however, most games will only be able to take advantage of 2 of the Radeon VE's 3 rendering pipelines. This is due to the fact that the vast majority of games out there use only dual textures, leaving the 3rd texel slot unused. Upon the release of the Radeon DDR, we mentioned that this feature would be nice in the long run as games switch to using three textures per pixel, however this transition seems to be taking longer than anticipated. In contrast, the GeForce2 MX can render 4 texels per clock when in a dual texture game and 3 texels per clock when in a trillinear textured game.

The second thing that the Radeon VE looses when compared to the rest of the Radeon family the 128-bit wide memory bus. In the Radeon VE, the memory bus width is shortened to only 64-bit wide and is used with DDR memory. This gives the Radeon VE effectively the same memory bandwidth as the 128-bit wide SDR memory Radeon SDR card. This, however, is half the memory bandwidth offered on the Radeon DDR card, with its 128-bit wide DDR memory bus.

Thirdly, the RAMDAC on the Radeon VE is actually crippled when compared to the other Radeon cards. Many praise the Radeon DDR, SDR, and All-in-Wonder cards for their superior 2D image quality. This is most likely due to the fact that these cards use ATI's updated 360MHz RAMDAC to power 2D image output. The Radeon VE, on the other hand, uses a downgraded 300MHz RAMDAC. In our subjective tests, the Radeon VE possessed about the same image sharpness as the GeForce2 series cards, which many say is less crisp than the 2D output of the aforementioned Radeon cards.

Finally, the Radeon VE's core lacks a technology that ATI make quite a scene about upon the release of the Radeon DDR: the Charisma engine. The Charisma engine was ATI's T&L engine that allowed transforming, lighting, and clipping calculations to be performed on chip, reducing CPU utilization. It is a shame that the budget oriented Radeon VE does not possess these features, as this card aimed at lower speed CPUs that will take any break they can get. Then again, in actual game play the single 3 texel pipeline will hold the Radeon VE back long before T&L could begin to make a difference.

The most obvious reason for cutting the Charisma engine out of the chip is the same reason that one of the texture pipelines was cut from the chip: to reduce die size. How much space was saved by cutting these components out is unknown, as ATI would not give us a number. It can be estimated, however, that some significant space was saved, as there was no reason to do so otherwise.

By looking at the Radeon VE's features, you may notice that it shares many things in common with another recently announced ATI product: the Radeon Mobility. Although we were just able to comment on some what the Radeon Mobility is about in our preview last week, the Radeon VE's specs look almost identical to those of the Radeon Mobility, sans the power management technology. Although no notebooks take advantage of the Radeon Mobility yet, the Radeon VE should give us somewhat of a vague glimpse of how the Radeon VE should perform in some instances.

Index The Card
Comments Locked


View All Comments

Log in

Don't have an account? Sign up now