XGI/Club3d Volari Duo V8 256mb


Review: Stuart "Veridian3" Davidson
Editor/Design: Allan "Zardon" Campbell

 


Update- Reactor 1.02.05 (non WHQL)

On 11/2 XGI released Reactor version 1.02.05 on their site. Considering the concerns we experienced with Reactor 1.0 and 1.01.51 we thought it may be interesting to see if they changed the Volari experience.

As a reminder here is the test setup from the original review, which was also used for this update:

REVIEW SETUP
Hardware:
Club3d XGI Volari V8 Duo 256mb
Intel Pentium 4 2.4c @ 3.0ghz (250fsb)
Abit AI7
2x256mb DDR3200 CL 2.5 (twinmos)
IBM Deskstar 120gxp 40GB 7200rpm Hard Drive.
Sony CRX300E DVD/CDRW
Sony Floppy Drive
Antec True Power 550W
AOC 19” 9GLR CRT

Software:
Windows XP, SP1a
Direct X 9.0b
Reactor 1.01.51
Reactor 1.02.05
Intel Chipset Drivers

------------------------


Aquamark:
Reactor 1.01.51 – Triscore= 42236
Reactor 1.02.05 – Triscore= 42562

A small increase in this new driver. No noticeable IQ changes.

 

Unreal Tournament 2003:

Reactor 1.01.51
Reactor 1.02.05

The benchmark results for Unreal Tournament show an increase of 2fps in botmatch. There was also a noticeable increase in IQ for the benchmark. The interference we noted in the original review is now gone and the benchmark completed a full run with no sections skipped. Unfortunately we were still able to see missing textures in actual gameplay with 1.02.05 (exact same places as the 1.01.51) and on some textures there were noticeable rolling lines. We also took the time to test some actual gameplay in UT2003.

Reactor 1.02.05 – 1024x768 0xAA 0xAF
Single Player Level 1 - Avg: 35.775 - Min: 22 - Max: 62
CTF Chrome - Avg: 63.427 - Min: 44 - Max: 102
CTF Magma - Avg: 38.718 - Min: 27 - Max: 63

As you can see, scores in actual gameplay are considerably different to the flyby and botmatch results. Whilst not unplayable there were some dips below and very close to the acceptable 25fps level.

Max Payne 2
Reactor 1.01.51 – Would not run.
Reactor 1.02.05 – Would not run.

As with the older drivers attempting to run Max Payne 2 with 1.02.05 resulted in a crash on the Duo v8. We have spoken to Remedy and they have confirmed that this crash does not occur on the v8 single chip cards however they don’t have a Duo to test on. XGI had informed us that this bug was fixed in a newer driver when we originally reviewed the card. This appears not to have been the case.

Need For Speed: Underground 1024x768 0xAA 0xAF
Reactor 1.01.51 – Avg: 10.3 - Min: 7 – Max: 15
Reactor 1.02.05 - Avg: 6.8 - Min: 1 - Max: 16

Disapointingly in NFSU there was a drop in performance making the game even more unplayable.

Prince Of Persia: The Sands Of Time 1024x768 0xAA 0xAF
Reactor 1.01.51 – Avg: 32.95fps - Min: 11fps - Max: 41fps.
Reactor 1.02.05 – Avg: 22.6fps - Min: 19fps - Max: 26fps

Prince of Persia’s Color Diffusion bug from Reactor 1.01.51 has now been fixed. This makes the game playable without ini editing which is a bonus. The game is not really playable at 1024x768, 800x600 is more realistic. On the IQ front the game is about 30%-40% too bright by default on the Volari and requires some tinkering with the display settings to make it acceptable.

Call Of Duty 1024x768 0xAA 0xAF
Reactor 1.01.51 – Would not run.
Reactor 1.02.05 - Avg: 57.493 - Min: 33 - Max: 65

The bug which caused COD to crash on the Volari during the training mission has been resolved in the new driver and the game is now playable. Reasonable scores for the card however the IQ is disappointing. The intro glitches are still evident and game appears to default to 16bit for many textures.

Farcry Demo
Reactor 1.01.51 – Would not run.
Reactor 1.02.05 – Would not run.

Farcry continues to crash during Precache when attempting to run Farcry on the Volari V8 Duo.
The following games were not tested in our original review and so feature no 1.01.51 results.

Breed Demo
Reactor 1.02.05 – Would not run

Breed gets as far as selecting single player then during the level load the game crashes to desktop.

Halo v1.4 1024x768 0xAA 0xAF
Reactor 1.02.05 – Avg: 43.631 - Min: 12 - Max: 100

Halo did successfully run however there were major Image Quality concerns. Firstly despite selecting 1024x768 the resolution was clearly not correct and appeared more like 800x600. There were also major texture problems, no shading on water and low quality sky are 2 examples. Despite the reasonable average FPS figure the fps drops and IQ made the game completely unplayable on 1.02.05

X2: The Threat 1024x768 0xAA 0xAF
Reactor 1.02.05 (with shadows) - Avg: 18.489 - Min: 9 - Max: 27
Reactor 1.02.05 (without shadows) - Avg: 26.743 - Min: 18 - Max: 36

X2 was almost playable at 1024x768 with shadows off in the game options. The menu screen objects were jerky however the game was less so. There were also some texture issues on both settings with meteors and ships etc displaying flickering textures. With shadows enabled the game was unplayable. Finally the game suffered from noticeable changes in texture quality. Instead of a gradual increase in detail as you approach objects there were jumps in IQ which only served to make the reduced quality all the more noticeable.

RTCW: Enemy Territory 1024x768 0xAA 0xAF
Reactor 1.02.05 (spectator mode, central Europe map) - Avg: 29.631 - Min: 15 - Max: 91
Reactor 1.02.05 (gameplay, central Europe map) - 22.742 - Min: 10 - Max: 71

Disappointing figures again from the Volari, even in this game which uses a very old engine.Quite unplayable when out in the open.

 

Other Bugs:
In our short time testing 1.02.05 we also noticed two rather annoying bugs, one old and one new.

As with the 1.01.51 driver DVD playback is still overly bright. XGI informed us that this is due to a video setting being applied twice (once by the drivers and once by the playback software, in this case Power DVD 5). Disapointingly despite being aware of this bug weeks ago, and knowing the cause, XGI have yet to fix this.

A new bug in 1.02.05 resulted in a desktop refresh being required after exiting many 3D applications. If you did not refresh the desktop (open a window on fullscreen for example) the desktop backdrop was not displayed and the refreshrate for the game and not the windows setting was used.

 

Other useful performance figures:

MDolenc’s Fillrate Tester -
--------------------------
Display adapter: XGI Volari Family
Driver version: 6.14.10.1020
Display mode: 1024x768 A8R8G8B8 85Hz
Z-Buffer format: D32
--------------------------

FFP - Pure fillrate - 1397.547852M pixels/sec
FFP - Z pixel rate - 1033.910034M pixels/sec
FFP - Single texture - 213.602646M pixels/sec
FFP - Dual texture - 244.201492M pixels/sec
FFP - Triple texture - 231.460052M pixels/sec
FFP - Quad texture - 135.523285M pixels/sec
PS 1.1 - Simple - 669.259460M pixels/sec
PS 1.4 - Simple - 182.736908M pixels/sec
PS 2.0 - Simple - 417.008362M pixels/sec
PS 2.0 PP - Simple - 417.005585M pixels/sec
PS 2.0 - Longer - 242.962524M pixels/sec
PS 2.0 PP - Longer - 242.989548M pixels/sec
PS 2.0 - Longer 4 Registers - 242.972504M pixels/sec
PS 2.0 PP - Longer 4 Registers - 242.963303M pixels/sec
PS 2.0 - Per Pixel Lighting - 19.593424M pixels/sec
PS 2.0 PP - Per Pixel Lighting - 19.593660M pixels/sec

 

Shadermark 2.0

click for full size

 

Painkiller Demo:
1024x768 - all ingame display quality options set to high/on. Trilinear selected.

No AA/AF:
Avg: 17.536 - Min: 5 - Max: 33

Very jerky, almost in slow motion to view.( Unplayable/unenjoyable basically) No major texture bugs seen. Only issue was that when changing the options in the game menu and clicking apply the card crashed and the system needed restarted. When restarted the options had been applied.

UT2004 Demo:
UT2004, 1024x768 no aa/af, game settings maxed.

UT2004 Deathmatch
Avg: 50.913 - Min: 35 - Max: 97

UT2004 Assault
Avg: 33.622 - Min: 19 - Max: 51

UT2004 CTF
Avg: 57.536 - Min: 35 - Max: 111

Issues: Basically fine in DM and CTF, assault was a bit ropey FPS wise. CTF had major texture issues on shadows... it made them grey...the screenshot bug (*cheat) is in this too. So the IQ shots don’t represent the real IQ that you get in this game.

 

Conclusion
Again, lets look at the positives first. Call Of Duty is playable, as is Prince of Persia (at 800x600 with some tweaking) – two big wins for the XGI driver team. There were however some worrying drops in performance with the new drivers, Need For Speed is the best example. 3 of the major issues from 1.01.51 are also still evident in 1.02.05. Max Payne 2 and Farcry still refuse to run and DVD playack is still unwatchable with default settings. In the added tests the signs are still not promising, older engines (RTCW:ET) and new engines (Halo/X2) are still underperforming and it could be argued in some cases, bug ridden. And others (Breed) wont even run. The screencapture bugs are also still in these drivers. This means that it is almost impossible to capture the IQ issues as the Volari produces distorted captures or automatically adjusts the IQ to a higher level when it detects a screenshot in progress. Having spoken to XGI after the original review I felt quite upbeat about future drivers and the possibility of fixes however none of the fixes mentioned have made it to the released driver. Its clear that there is just too much to be done in order to save this product whether it be at mainstream or enthusiast levels.

 

 

 

Click here to go to application and install page Click here to go to pcmark2004 page Click here to go to the results page Click here to go to the conclusion page

 
Click here to go to application and install page Click here to go to pcmark2004 page Click here to go to the results page Click here to go to the conclusion page