Monthly drivers report and popular 3D
accelerators comparison

Monthy 3Digest is the project of iXBT.com and Digit-Life.com dedicated to informing you about the performance of a large number of graphics cards under Windows XP in the past month.

This issue is dedicated to drivers released from the 15th of February to the 15th of March 2007. Traditionally our 3Digest informs you about the best price/performance ratios available in the market. Another very important part of any issue is cards' 3D graphics quality shown on the example of a wide selection of games.

It's our 82th issue of 3Digest.

This digest is dedicated to CPU dependency of graphics cards. This month we added AA+AF benchmarks.

Attention! This digest contains benchmarking results for PCI-E cards only! The final AGP benchmarking was conducted in March 2006!

CONTENTS

1. Testbed configurations and test software
2. Tested videocards list
3. What's new in 3Digest
4. Test summary diagrams and digest bottom line
5. 3D accelerator rating calculations for the end of the month
6. Screenshot gallery
7. Game Quality Section
8. Game Quality Control Section

1. Testbed configuration

• Intel Core2 Duo (775 Socket) based machine
  • Intel Core2 Duo Extreme X6800 (2930 MHz) (L2=4096K) CPU;
  • Intel Core2 Duo Extreme X6800 CPU (overlocked to 3400 MHz by changing the multiplier to 13), 1066 MHz FSB and memory;
  • Intel Core2 Duo E6400 (2133 MHz) CPU;
  • EVGA nForce 680i SLI motherboard on NVIDIA nForce 680i;
  • 2 GB DDR2 SDRAM Corsair 1066MHz (CAS (tCL)=5; RAS to CAS delay (tRCD)=5; Row Precharge (tRP)=5; tRAS=15);
  • WD Caviar SE WD1600JD 160GB SATA HDD.

• AMD Athlon64 X2 4000+ (AM2 Socket) based machine
  • AMD Athlon64 X2 4000+ (2000 MHz) (L2=2048K) CPU;
  • Foxconn motherboard on NVIDIA nForce 590 SLI;
  • 2 GB DDR2 SDRAM OCZ 800MHz (CAS (tCL)=5; RAS to CAS delay (tRCD)=5; Row Precharge (tRP)=5; tRAS=15);
  • WD Caviar SE WD1600JD 160GB SATA HDD.

The testbed has Windows XP Professional Service Pack 2, DirectX 9.0c installed. Dell 3007WFP (30") monitor was used.

The list of benchmarks is here

2. Graphics card list

Below is a list of graphics cards that we tested linked to the respective 3Digest feature description pages.

You can see the quality of 3D produced by tested cards in our monthly updated Screenshot gallery.

Card list:

1. PCI-E-ATI RADEON X1300 256MB DDR 2.8ns (450/550 MHz)
2. PCI-E-ATI RADEON X1300 512MB DDR2 3.0ns (450/500 MHz)
3. PCI-E-ATI RADEON X1300 PRO 256MB DDR 2.5ns (600/800 MHz)
4. PCI-E-ATI RADEON X1300 XT 256MB DDR2 2.0ns (500/900 MHz)
5. PCI-E-ATI RADEON X1650 PRO 256MB DDR 1.4ns (590/1380 MHz)
6. PCI-E-ATI RADEON X1650 XT 256MB DDR 1.4ns (600/1400 MHz)
7. PCI-E-ATI RADEON X1650 XT 256MB DDR 1.2ns (625/1530 MHz)
8. PCI-E-ATI RADEON X1800 GTO 256MB GDDR3 2.0ns (500/1000 MHz)
9. PCI-E-ATI RADEON X1950 PRO 256MB GDDR3 1.4ns (580/1400 MHz)
10. PCI-E-ATI RADEON X1950 PRO 256MB GDDR3 1.2ns (635/1480 MHz)
11. PCI-E-ATI RADEON X1950 PRO 512MB GDDR3 1.4ns (573/1360 MHz)
12. PCI-E-ATI RADEON X1900 XT 256MB GDDR3 1.2ns (625/1450 MHz)
13. PCI-E-ATI RADEON X1900 XT 512MB GDDR3 1.2ns (625/1450 MHz)
14. PCI-E-ATI RADEON X1900 XTX 512MB GDDR3 1.1ns (650/1550 MHz)
15. PCI-E-ATI RADEON X1950 XTX 512MB GDDR4 0.9ns (650/2000 MHz)
16. PCI-E-ATI RADEON X1950 XT 512MB GDDR4 0.9ns (625/1800 MHz)
17. PCI-E-NVIDIA GeForce 7300GT 256MB GDDR2 2.8ns (350/666 MHz)
18. PCI-E-NVIDIA GeForce 7300GT 256MB GDDR3 1.4ns (500/1400 MHz)
19. PCI-E-NVIDIA GeForce 7600GT 256MB GDDR3 1.4ns (560/1400 MHz)
20. PCI-E-NVIDIA GeForce 7600GS 256MB GDDR2 2.5ns (400/800 MHz)
21. PCI-E-NVIDIA GeForce 7600GS 512MB GDDR2 2.0ns (550/950 MHz)
22. PCI-E-NVIDIA GeForce 7600GS 256MB GDDR3 1.2ns (450/1100 MHz)
23. PCI-E-NVIDIA GeForce 7800 GTX 256MB GDDR3 1.6ns (430/1200 MHz)
24. PCI-E-NVIDIA GeForce 7900 GT 256MB GDDR3 1.4ns (450/1320 MHz)
25. PCI-E-NVIDIA GeForce 7900 GS 256MB GDDR3 1.4ns (450/1320 MHz)
26. PCI-E-NVIDIA GeForce 7900 GS 256MB GDDR3 1.4ns (480/1400 MHz)
27. PCI-E-NVIDIA GeForce 7900 GS 256MB GDDR3 1.4ns (540/1500 MHz)
28. PCI-E-NVIDIA GeForce 7950 GT 512MB GDDR3 1.4ns (550/1400 MHz)
29. PCI-E-NVIDIA GeForce 7950 GT 256MB GDDR3 1.4ns (600/1430 MHz)
30. PCI-E-NVIDIA GeForce 7900 GTX 512MB GDDR3 1.1ns (650/1600 MHz)
31. PCI-E-NVIDIA GeForce 7950 GX2 2x512MB GDDR3 1.4ns (500/1200 MHz)
32. PCI-E-NVIDIA GeForce 7950 GX2 SLI 2x2x512MB GDDR3 1.4ns (4x500/1200 MHz)
33. PCI-E-NVIDIA GeForce 8800 GTS 640MB GDDR3 1.2ns (512/1600 MHz)
34. PCI-E-NVIDIA GeForce 8800 GTX 768MB GDDR3 1.1ns (575/1800 MHz)
35. PCI-E-NVIDIA GeForce 8800 GTX SLI 2x768MB GDDR3 1.1ns (2x575/1800 MHz)
36. PCI-E-NVIDIA GeForce 8800 GTS SLI 2x640MB GDDR3 1.2ns (2x512/1600 MHz)

Archive (graphics cards, information on which is no longer updated)

3. March 2007 news and current matters

Added a number of new graphics cards (marked "NEW").

• 3.1. The following drivers were released from February 15, 2007 to March 15, 2007:
  
  
  • NVIDIA ForceWare for all NVIDIA-based cards:
    
    
    • Windows XP: 97.92 for GF8800 series;
    • Windows XP: 93.81 for GF6xxx/7xxx series;
    • Windows Vista 32bit: 101.41;
    • Windows Vista 64bit: 101.41;

    
    

  • ATI Catalyst for all ATI-based cards:
    
    
    • Windows XP: CATALYST 7.3
    • Windows Vista 32bit: CATALYST 7.3
    • Windows Vista 64bit: CATALYST 7.3
      

  
  
  

• 3.2. This month we used 97.92 drivers for NVIDIA-based card summaries; 7.1 drivers for ATI-based card summaries.
  
  

4. Summary graphics card performance charts for the latest drivers available at the moment of testing in March 2007

All willing to see the complete results in Excel 2003 format can download this RAR 3.0 or this ZIP archive.

Attention! Due to technical issues not all benchmark results are published. However, provided are results for new games, which give good understanding of CPU dependency.

Also, due to technical issues the incorrect "No AA, No AF" lines were added to the charts. The correct information is provided in chart headlines: AA 4x + AF 16x.

• 4.1. Splinter Cell Chaos Theory (No HDR)
  • All resolutions on one page, AA 4x + AF 16x (High Quality for ATI)
    • 1024x768
    • 1280x1024
    • 1600x1200
    • 2560x1600

  
  
  

• 4.2. Splinter Cell Chaos Theory (HDR)
  • All resolutions on one page, AA 4x + AF 16x (High Quality for ATI)
    • 1024x768
    • 1280x1024
    • 1600x1200
    • 2560x1600

  
  
  

• 4.3. Call Of Juarez
  • All resolutions on one page, AA 4x + AF 16x (High Quality for ATI)
    • 1024x768
    • 1280x1024
    • 1600x1200
    • 2560x1600

  
  
  

• 4.4. Serious Sam II (No HDR)
  • All resolutions on one page, AA 4x + AF 16x (High Quality for ATI)
    • 1024x768
    • 1280x1024
    • 1600x1200
    • 2560x1600

  
  
  

• 4.5. Serious Sam II (HDR)
  • All resolutions on one page, AA 4x + AF 16x (High Quality for ATI)
    • 1024x768
    • 1280x1024
    • 1600x1200
    • 2560x1600

  
  
  

• 4.6. Archive (information from the previous 3Digests absent in this issue)
  
  
• 4.7. Notes on graphics cards performance with the latest drivers:
  
  
  • If a particular card is not in the summary, it means the given benchmark didn't work or worked incorrectly with it.
  • We should underline that, like before, WinXP drivers for NVIDIA 8800 series introduce a number of serious artefacts in some games, not to mention the inoperative SLI (e.g., Call Of Juarez.) Therefore you can see that performance of these SLI configurations is low (even lower than that of single cards.) Though the new beta drivers, which are promised to fix these problems, have already leaked to the Web, we didn't benchmark them for our comparison with the previous issue of 3Digest (tests without AA+AF) to be correct.
  • Speaking of our impressions from testing. Unlike the similar examination conducted nearly 2 years ago, here we can clearly see that modern games have a very heavy impact on graphics cards, so CPU influence is somehow lower. Take a look at, say, Call Of Juarez, which places all the load on a graphics card. Our testing also visually separated cards to strongly CPU dependent, moderately CPU dependent, and lightly CPU dependent. Of course, you should consider that benchmarks were run at maximum quality settings possible ingame, which increased the load on graphics cards. In general, we can state that these AA+AF benchmarking results are rather unique, because nearly all cards (except for the 8800 series) demonstrated almost the same performance on different platforms and with different CPU clock rates.
  • Also consider that we have already had the test results for the base Core2 Duo 2.13 GHz processor. And its performance was 15% higher than that of the A64 X2 5000+. Therefore, all other Athlon64 X2 processors (except for the 5200+) should get within the range between the A64 X2 4000+ and the Core2 Duo E6400. And you can estimate the difference percentage yourself.

5. 3D accelerator ratings calculations for the end of the month

Attention! Since we have only examined CPU dependency of graphics cards and haven't conducted tests without AA+AF, we haven't made any ratings! Therefore we can only pe-publish those from December 3Digest.

Calculations were based on gaming benchmark results as well as the results from 3DMark06 new tests (ATI CATALYST 6.12, NVIDIA ForceWare 96.97).

All willing to conduct their own rating calculations with own values are invited to download this Excel XP table in RAR 3.0 or ZIP format.

The method of rating calculation is described here.

Calculations were conducted considering the following assumptions:

1. Percentage of performance and quality demand ratio:
   • performance priority - 60%
   • quality priority - 40%

1. Only the HIGHEST resolution available in games was used to calculate prospect ratings.
2. The usability rating indicates performance and features of a card (numerator) related to its price (denominator).

According to your wishes, we separated ratings into two groups: the first gathers all benchmarks, and the second contains only the gaming tests without m3DMark05/06. Disbelieving the synthetic tests, many think the second group is of higher importance.

Core2 Duo X6800 platform:

"Usability" Rating:

GAMING ONLY usability rating

GENERAL usability rating

"Prospect" Rating:

GAMING ONLY prospect rating

GENERAL prospect rating

Ingame quality

6. Games screenshot gallery

All test screenshots are compiled into a gallery where you can qualify the work of a chosen videocard. All shots were taken at the same 1024x768 resolution at the maximum quality. Screenshot gallery

7. Game Artefacts Gallery

Game Quality Section

8. Game Quality Control Section

Game Quality Control Section

The PSU for the testbed was kindly provided by HIPER

The Dell 3007WFP monitor was kindly provided by NVIDIA

Vitaly Milov (vmilov@ixbt.com)	Danil Gridasov (degust@ixbt.com)

Andrey Vorobiev (anvakams@ixbt.com)