Palit NVIDIA GTX460 in Surround: 1GB vs. 2GB

Declaration of Interest: I have purchased these cards myself from a reputable UK computer e-tailer; none of this hardware or software has been donated or loaned for the purposes of this review. There are therefore no conflicts of interest which could arise which would bias this review in any manner.

Author: Paradigm Shifter (WSGF Admin & WSGF Member #5 with the top post count (11.6k+))

Introduction

Right, let’s cut to the chase.

Surround resolution testing.

NVIDIA Surround is the NVIDIA equivalent to ATIs Eyefinity technology, which in turn was their version of Matroxes TripleHead2Go unit. With the release of the GF104, affordable, capable NVIDIA Surround is within the grasp of those who, for whatever reason, do not wish to purchase used GTX200 series hardware or a GTX470/480.

When I originally bought two GTX460 1GB cards to run Surround, I was amazed at the performance available in them… but quickly ran into a brick wall when some settings were turned up. Running games in Surround became an exercise in determining what settings I could or couldn’t enable on a “per game” basis and still maintain a playable framerate. It was no easy matter – many times the games played well until one setting was pushed a tiny bit too far, at which point the game would nosedive from a comfortable 30+fps or 40+fps to low teens or in some cases even single digit framerates.

Now, it can be argued that the GTX460, as NVIDIA’s “mainstream” card, simply is not capable of running the vast Surround resolutions at ultra-high details and maintain a decent framerate. My early observations, aided by MSI Afterburner and my G15 LCD, allowed me to create a hypothesis which I will attempt to prove (or disprove) in this review. It was quite simple; whenever Afterburner reported that VRAM usage was close to 100% of the cards VRAM, framerates nosedived – I believed that it was at this point that the system was swapping to system RAM and the latencies and bandwidth decrease resulted in huge framerate losses.

Therefore, let’s propose the hypothesis:

At Surround resolutions, games are heavily VRAM limited.