HD3870, HD3850 review - Is RV670 the new mid-class DX 10?

Written by Димитър Динчев a.k.a. Veseliq

Monday, 04 February 2008

  Page 5 - Problems met, test chosen

Due to the variety of hardware we used in our test it was totally predictable we did ran in bunch of problems. On the last page you saw the initial list of video cards that we had to compare through the tests. But the MSI 8600GTS couldn't fit in the MSI P35 NEO 2 - FR motherboard (the big passive radiator of the card, standing on top of it and connected to the core with few heatpipes, was overshadowing 3 of the RAM slots). The reason that happened was becouse the memory had a rather high radiators, which is a shame since that card is the target HD3850 is aimed at and tis review is supposed to reveal how is the new ATI holding out against it. Yet testing the card with a single Ram module would make the tests completely inadequate, so that was the reason that got the 2600XT out of the review, having nothing to compare it to.

On the other hand the 2900GT (unfortunately our only AMD/ATI representative of the Radeon 2900 family) had a horribly strange problem, that I found no information about. The card worked with Catalyst 8.1 at 511MHz Core / 1022MHz Memory frequencies. The Control Panel integrated overclock tool of the driver gave us PASSED when testing if the card could run the on the default frequencies, but when pushing the button to set these values (650MHz core / 1600MHz memory) nothing happened. I try Catalyst 7.12, but he whined that is unable to work with the current adapter (Again - odd), however after a few hours of efforts i just put the card back in the box, reducing the final list to:

Sapphire HD3870 512MB GDDR4
Sapphire HD3850 512MB GDDR3
Sapphire HD3850 256MB GDDR3
Albatron 8800GT
MSI NX8800GT - T2D512E - OC Edition
MSI NX8800GTS - T2D512E - OC Edition
XFX 8800Ultra
XFX 8800GTS 640MB Alpha Dog Edition (slightly overclocked version of 8800GTS)

For the more specific problems during the tests i decided to leave a filed with notes, in which i described what problem was created by which card. Also i wrote down the best combination of settings on which the game is played flawlessly. For the performance tests we choose 4 games, supporting DirectX 10, manifested by use of different effects that intensely load the cards squeezing everything out of them. The list could be much longer, but that would be hardly beneficial. The idea was to use engines that use HDR (High Dynamic Range) effects, Shader Model 4.0 effects and so on.

All games were tested at maximum details on resolution of 1280x1024 with 4x Anti Aliasing, 1680x1050 with 4x Anti Aliasing, and with AA turned off.