[an error occurred while processing this directive]
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.
This issue is dedicated to drivers released from the 15th of November 2005 to the 15th of December 2005. 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 67th issue of 3Digest.
December 2005
Graphics architectures: bottom line for the year
Relative calmness
The year 2005 was relatively calm in the architectural field, excluding the obviously late and slightly spoiled release of ATI's new architecture supporting SM3 and corresponding chips. Actually, we must admit that industry somehow reduced its performance acceleration - maybe while expecting Vista OS along with new hardware requirements.
Alternatively, we can say that modern industry unambiguously points to the "from hardware to human" trend. More and more often we observe the development and deployment of not only new processor cores with improved performance and clock rates, but also new user interfaces, design, and other little things that make everything more comfortable. And it's all correct, because it's high time to turn to people that will more likely pay for a new opportunity to control media from their couches than for a new CPU or GPU rating that, due to lagging software industry, has almost no effect on actual sensations of users' daily interaction with PC. According to the conservation law, such focusing on the outside, on the usage models slows down the inside progression - architectures and their features. This is exactly what we are observing now.
On the other hand, you could notice a significant revival on the consoles market. The time has come for console generations to change thus we are being offered numerous announcements, including the most expected solutions: Microsoft Xbox 360 and Sony PlayStation 3. NVIDIA and ATI are again sharing the world, since in Microsoft's product we can see the new architecture from ATI (that differs from what we have on PC), while Sony, not to invent the wheel again, partners with NVIDIA that doesn't risk and adapts the successful G70 architecture for the console needs. No one can say about the lull here, though the most interesting battles - for user wallets - are still ahead.
So, since we have discussed the market trends, let's get back to the object and consequences:
Key year events
- The release of NVIDIA G70. This novelty, despite its totally new name, could be called NV47. At least to show its actual place in the product hierarchy. The redesigned and improved NV45 with more pixel pipelines, PCI-E support, new FP16 frame buffer and texture features, and slightly faster ALUs in pixel pipelines. Is that all? But this turns out to be more than enough, considering the good clock rate potential and refined manufacture.
- The release of ATI CrossFire - the answer to NVIDIA SLI.
- The announcement of Xbox 360 architecture - the Xenos chip - the descendant of R500 project (don't confuse with R520) and the prototype of ATI's Vista offering (that meets WGF 2 requirements).
- The announcement of PlayStation 3 graphics architecture. Being almost identical to G70 from graphics point of view, PS3 also hosts all NVIDIA's developers tools: SDKs, Cg shader compiler and other.
- Finally, the announcement of R520, RV530 and RV515 - ATI's long-anticipated desktop architecture with SM3 support and, as we found out, relatively contradictory capabilities. On the other hand, it offers good branching efficiency and long shaders and WGF 2.0 potential. On the other is strange performance disbalance, especially in the middle end; a noticeable lag from NVIDIA's announcement; and some performance lag as well, including the high-end. 16 pixel pipelines versus 24 of G70 and totally unacceptable clock rates, considering G70's potential.
The successful evolution from NV45 into G70 and PS3
NVIDIA G70 is a really successful product. It's not just NV45 overclocked and having more pixel and vertex pipelines. It's redesigned, improved rasteriser, updated and strengthened geometry engine, including the long-expected alpha textures antialiasing and integrated HDTV interface. Though architectural differences of pixel and vertex processors are not that great, the thought-out ALU optimization in pixel processors, rasteriser, triangle unit, geometry engine post-TCL buffer and texture caches result in a good performance gain at the same clock rate as NV45 have. This is especially noticeable in complex scenes and tasks. Besides, the increased amount of pixel and vertex processors and new clock rates plus good overclocking potential made it easy for NVIDIA to answer to R520 with a new G70-based card with higher clock rates that demonstrated who was the high-end king. It also demonstrated that all efforts to implement SM3 early were not in vain and provided a certain reserve that, by the middle of 2005, turned into a advantage, both performance, and economic. The same solution, without dedicating significant R&D resources, also obtained a deserving place in PlayStation 3. What can we say. Good work, people!
Now they can release pared-down modifications like G71, 73, etc. along with a higher-end model (with 32 pipelines, why not) and steadily develop a WGF 2.0 solution in parallel.
Xenos of Xbox 360 - tomorrow now
Two key points of interest of this accelerator are its memory and
unified shader processors handling both pixel and vertex shaders depending
on load (say hi to WGF 2 and Vista,
former Longhorn). Thus we have two control units processing code
of two shaders, each of which can be either pixel, or vertex. Such
a unit controls an array of 16 shader processors. All in all, there
are three arrays, each managed by either first, or second unit ("shader
interp" on the chart). Therefore, all chip's ALUs are loaded
almost all the time automatically solving the issue of pixel and vertex
load balance inside GPU! Many vertices - processors will handle them
more, many pixels - more pixel work. Also, two control units provide
more effective shader switching (a consuming operation). It's interesting
that while there are 48 processors, there are 64 processed states
(or threads, but still it's simpler to speak of 64 objects, be it
a vertex or a pixel). Such redundancy allows permanent processor load
even in case early shader output and fewer instructions in a branch.
Such a chip has resource utilisation close to perfect even in case
of conditional transition shaders. This is the future of graphics,
the things that we are to see in computers when Vista and first WGF
2 accelerators come. And Xbox already offered it for Christmas. Formally,
Xbox accelerator is not fully compatible with WGF 2, but the differences
are minimal. Besides, WGF 2 specs weren't completed at the moment
this GPU was announced. Anyway, ATI got a perfect field to hone future
technologies for Vista. Moreover, under Microsoft's "maternal
care".
There is integrated (!) memory that is meant for the most critical data - depth buffer and frame buffer color values. 10 megabytes might not be enough for every mode, but spooling to and from local memory in large blocks helps. Anyway, typical scenes are not rendered chaotically, but, more likely, consequently. Besides, interaction is supported by a wide and fast interchip bus (since the accelerator hosts two chips in one package). Such organisation allows pretty effective frame buffer operation and makes antialiasing almost free of resource charge in all modes. It's interesting that the second GPU with 10-megabyte buffer also contains special logic. Actually it hosts writing, blending and depth check units that were relocated from the main chip. This provides very wide path and fast access to this buffer's data, makes depth checks in MSAA modes free of charge even from the angle of chip interaction.
So, Xbox gives a lot hints about future graphics architectures. Also pay attention to integrated (or rather attached) memory and unified shader processors.
R520 - spoiled before spinned
It's not a secret that ATI delayed the release of its new generation. Whether the reason was the new 90nm process and efforts to remake R520 to provide necessary clock rates and yield, or some other considerations like market lull before Vista and WGF 2.0. But a fact is a fact. The NVIDIA's competitor arrived earlier and we had enough time to get to know about its evolutional, not revolutional character in relation to NV4X family. But here the situation differs - by another process technology and architecture.
The architecture turned out to be foremost on the one hand and arguable on the other. It arouses a feeling of some intermediate incompleteness. As if R520 was not a standalone product, but a stop on the way to the bright future of WGF 2.0 and R600 or whatever it's called. As if it was just a half of the platform, the second half being Xenos of Xbox 360. Put together and coupled with obvious, but missing, they would make a complete perfection. But having scattered efforts to various architectures, console and desktop, ATI initially made a seemingly logical step, but the task turned out to be heavier than expected. And the company could not perform it well within given time due to simple "specialist crunch". Again and again global management mistakes chase Canadians like a ghost.
On the one hand, here's the cool scheduler "black box" that efficiently solves fundamental problem of branched shader processing. On the other hand, why shader processors are not unified? This would require almost no significant changes in this architecture. But still, only Xenos of these two has unified shader processors. Or take this floating-format frame buffer that handles blending and even MSAA, but strangely does not filter textures. As a result, everything is FP16, except for the most critical task - texture filtering. Why implement MSAA in this mode then and why not make the concept complete? Or these vertex processors - these lack texture access. What a pity. Still, we must note they recommend to use pixel pipelines with rendering into vertex buffer instead. The problem of absent texture filtering is also solved. And the amount of pixel pipelines is not enough for an assured leadership. The amount of texture modules in RV530 is an obvious bottleneck.
We can continue this list with lesser issues, but the bottom line is the same - strong "intermediancy" odour confirmed by the not very successful benchmark results (at least not successful enough to speak of ATI's leadership). Therefore we are waiting for an optimised and better pipelined (we hope) R580. Waiting for R600 that (we hope again) will not only combine the best of Xenos and R520, but also support them with other parts missing. And for now we have a nice new product perfectly spoiled before it could spin. Its architecture is promising, especially considering the console partnership with Microsoft that sketches Vista's WGF 2.0 basics, but is still flawed. The disadvantages do not allow us to call R5x unsuccessful, but its capabilities are not that revolutionary to call it a leader. Formally, ATI caught up with NVIDIA having founded a promising architectural basis along the way - basis for WGF 2.0 and R600. Will it pay off, if you purchase R520? Or should you wait for R580? This is the most vital question for the ATI camp, and company's homework after this lesson should make this new architecture shine anew. Let's wait and see!
And again, now we can only put R520 and Xenos together and imagine benefits that might be waiting for us in R600 - the architecture designed for WGF 2.0.
Ok, let's here finish the introduction and the summary of the year 2005...
And present our readers with a new version of RightMark3D.
It already features very complex shaders Shaders 3.0 with lots of branches, etc. This is a beta, so please be so kind and email Alexander Medvedev, the project coordinator, if anything.
Now let's get back to the 3Digest.
PLEASE NOTE! For this issue we tested only PCI-E graphics cards! The AGP-based benchmarks are provided in the November 2005 issue!
CONTENTS
-
Testbed configurations and test software
- Tested videocards list
- What's new in 3Digest
- Test summary diagrams and digest bottom line
- 3D accelerator rating calculations for the end of the month
- Screenshot gallery
- Games quality section
- Game Quality Control
1. Testbed configuration
- System on Athlon 64 (939Socket)
- AMD Athlon 64 4000+ (L2=1024K);
- ASUS A8N32 SLI Deluxe on NVIDIA nForce4 SLI X16;
- 2 GB DDR SDRAM 400MHz (CAS (tCL)=2.5; RAS to CAS delay (tRCD)=3; Row Precharge (tRP)=3; tRAS=6);
- WD Caviar SE WD1600JD 160GB SATA HDD.
Testbeds have Windows XP Professional Service Pack 2, DirectX 9.0C installed. IIYAMA Vision Master Pro 514 (22") 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 respective 3Digest feature-description pages.
You can see the quality of 3D graphics produced by tested cards
in our monthly updated Screenshot gallery.
Card list:
- PCI-E-ATI RADEON X300 128MB DDR 5ns (325/400 MHz)
- PCI-E-ATI RADEON X550 256MB DDR 4ns (400/500 MHz)
 PCI-E-ATI
RADEON X550 XT 128MB GDDR3 2.0ns (400/900 MHz)- PCI-E-ATI RADEON X600 PRO 128MB DDR 3.3ns (400/600 MHz)
- PCI-E-ATI
RADEON X1300 256MB DDR 2.8ns (450/550 MHz)
- PCI-E-ATI RADEON X1300 PRO 256MB DDR 2.5ns (600/800 MHz)
- PCI-E-ATI RADEON X700 256MB DDR 2.8ns (400/700 MHz)
- PCI-E-ATI RADEON X700 PRO 256MB GDDR3 2.0ns (425/860 MHz)
- PCI-E-ATI
RADEON X1600 PRO 256MB DDR 2.0ns (500/900 MHz)
- PCI-E-ATI RADEON X1600 XT 256MB DDR 1.26ns (590/1380 MHz)
- PCI-E-ATI RADEON X800 GT 256MB GDDR3 2.0ns (475/990 MHz)
- PCI-E-ATI RADEON X800 128MB DDR 2.8ns (400/700 MHz)
- PCI-E-ATI RADEON X800 GTO 256MB GDDR3 2.0ns (400/990 MHz)
- PCI-E-ATI RADEON X800 XL 256MB GDDR3 2.0ns (400/990 MHz)
- PCI-E-ATI RADEON X800 XL 512MB GDDR3 2.0ns (400/990 MHz)
- PCI-E-ATI RADEON X850 XT 256MB GDDR3 1.6ns (520/1080 MHz)
- PCI-E-ATI RADEON X850 XT PE 256MB GDDR3 1.6ns (540/1180 MHz)
- PCI-E-ATI RADEON X1800 XT 512MB GDDR3 1.26ns (625/1500 MHz)
- PCI-E-ATI
RADEON X1800 XT PE 512MB GDDR3 1.26ns (700/1600 MHz)
- PCI-E-ATI RADEON X1800 XL 256MB GDDR3 1.4ns (500/1000 MHz)
- PCI-E-NVIDIA GeForce 6600LE 128MB DDR 3.6ns (300/550 MHz)
- PCI-E-NVIDIA GeForce 6600 128MB DDR 3.6ns (300/550 MHz)
- PCI-E-NVIDIA GeForce 6600 512MB DDR2 3.8ns (300/550 MHz)
- PCI-E-NVIDIA GeForce 6600 256MB DDR2 2.5ns (350/800 MHz)
- PCI-E-NVIDIA GeForce 6600GT 128MB GDDR3 2.0ns (500/1000 MHz)
- PCI-E-NVIDIA GeForce 6800 256MB DDR 2.8ns (350/600 MHz)
- PCI-E-NVIDIA
GeForce 6800 XT 256MB GDDR3 2.0ns (425/1000 MHz)
- PCI-E-NVIDIA GeForce 6800 GS 256MB GDDR3 2.0ns (425/1000 MHz)
- PCI-E-NVIDIA GeForce 6800 GT 256MB GDDR3 2.0ns (350/1000 MHz)
- PCI-E-NVIDIA GeForce 6800 Ultra 256MB GDDR3 1.6ns (425/1100 MHz)
- PCI-E-NVIDIA GeForce 6800 Ultra SLI 2x256MB GDDR3 1.6ns (2x425/1100 MHz)
- PCI-E-NVIDIA GeForce 6800 GT SLI 2x256MB GDDR3 2.0ns (2x350/1000 MHz)
- PCI-E-NVIDIA GeForce 6800 SLI 2x256MB DDR 2.8ns (2x350/600 MHz)
- PCI-E-NVIDIA GeForce 6600 GT SLI 2x128MB GDDR3 2.0ns (2x500/1000 MHz)
- PCI-E-NVIDIA GeForce 6600 SLI 2x128MB DDR 3.6ns (2x300/550 MHz)
- PCI-E-NVIDIA GeForce 7800 GT 256MB GDDR3 2.0ns (400/1000 MHz)
- PCI-E-NVIDIA GeForce 7800 GT 256MB GDDR3 2.0ns (450/1100 MHz)
- PCI-E-NVIDIA GeForce 7800 GTX 256MB GDDR3 1.6ns (430/1200 MHz)
- PCI-E-NVIDIA GeForce 7800 GTX 256MB GDDR3 1.6ns (459/1300 MHz)
- PCI-E-NVIDIA GeForce 7800 GTX 256MB GDDR3 1.6ns (486/1350 MHz)
- PCI-E-NVIDIA GeForce 7800 GT SLI 2x256MB DDR3 2.0ns (2x400/1000 MHz)
- PCI-E-NVIDIA GeForce 7800 GTX SLI 2x256MB DDR3 1.6ns (2x486/1350 MHz)
- PCI-E-NVIDIA
GeForce 7800 GTX 512MB GDDR3 1.1ns (550/1700 MHz)
- PCI-E-NVIDIA
GeForce 7800 GTX 512MB GDDR3 1.1ns (580/1760 MHz)
- PCI-E-NVIDIA
GeForce 7800 GTX SLI 2x512MB DDR3 1.1ns (2x580/1760 MHz)
Archive
(videocards, information on which is no longer updated)
3. December 2005 news and current business
Added a number of new graphics cards to the list.
We replaced the motherboard in the testbed with the NForce4 SLI X16.
We removed the BF2 benchmark since ingame benchmarking means provide unreal results and we are not going to use FRAPS to maintain test objectivity.
- 3.1. The following drivers were released from November 15, 2005 to December 15, 2005:
- NVIDIA ForceWare for all NVIDIA-based cards:
- Windows XP: 81.98 WHQL, 82.12 beta;
- ATI Catalyst for all ATI-based cards:
- Windows XP: CATALYST 5.12, 5.13
- 3.2. This month we used 81.98 drivers for NVIDIA-based card summaries; 5.13 drivers for ATI-based card summaries.
4. Summary diagrams of graphics cards performance with the latest drivers for December 2005
- 4.1. Far Cry, Research
- 4.2. Splinter Cell Chaos Theory
- 4.3. Half Life2, ixbt01
- 4.4. DOOM III
- 4.5. 3DMark05 MARKS
- 4.6. Chronicles of Riddick, demo 44
- 4.7. F.E.A.R.
- 4.8. Call Of Duty 2 DEMO
- 4.9. Archive
(information from the previous 3Digests absent in this issue)
- 4.10 Notes on videocards' 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 recommend you to visit unofficial product support websites, if you come across any "peculiarities". The summary charts were formed using results obtained on all testbeds. We believe no comments are required and you will understand what is what by yourself. :)
5. 3D accelerator ratings calculations for the end of the month
All willing to conduct his own rating calculations with his 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:
- Percentage of performance and quality demand ratio:
- performance priority - 60%
- quality priority - 40%
- ONLY 1600x1200@32bpp resolution was used for the prospect rating;
- The usability rating indicates performance and features of a card (numerator) related to its price (denominator).
- Athlon 64 4000+ platform:
Usability rating (rating from the previous 3Digest is in parentheses):
Usability rating
| 01. |
RADEON X1600 XT 256MB 590/1380 |
(09) |
| 02. |
RADEON X1600 PRO 256MB 500/800 |
(--) |
| 03. |
GeForce 6600 DDR2 256MB 350/800 PCI-E |
(02) |
| 04. |
RADEON X1300 PRO 256MB, 600/800 |
(19) |
| 05. |
GeForce 6800 GS 256MB 425/1000 PCI-E |
(01) |
| 06. |
GeForce 6800XT 256MB 425/1000 PCI-E |
(--) |
| 07. |
GeForce 6600GT 128MB 500/1000 PCI-E |
(05) |
| 08. |
GeForce 6600 DDR2 512MB 300/550 PCI-E |
(03) |
| 09. |
GeForce 7800GT 256MB, 450/1100 |
(04) |
| 10. |
RADEON X1300 256MB, 450/550 |
(--) |
| 11. |
GeForce 7800GT 256MB, 400/1000 |
(06) |
| 12. |
GeForce 6600 128MB 300/550 PCI-E |
(16) |
| 13. |
GeForce 6800 256MB 350/600 PCI-E |
(12) |
| 14. |
RADEON X1800 XL 256MB 500/1000 |
(14) |
| 15. |
RADEON X1800 XT 512MB 625/1500 |
(07) |
| 16. |
GeForce 7800GTX 256MB, 459/1300 |
(10) |
| 17. |
RADEON X800GTO 256MB 400/990 PCI-E |
(15) |
| 18. |
RADEON X700 PRO 256MB 425/860 PCI-E |
(22) |
| 19. |
GeForce 7800GTX 256MB, 486/1350 |
(11) |
| 20. |
GeForce 7800GTX 256MB, 430/1200 |
(08) |
| 21. |
RADEON X700 256MB 400/700 PCI-E |
(21) |
| 22. |
GeForce 6800 SLI 2x256MB 350/600 |
(25) |
| 23. |
RADEON X800 GT 256MB 475/990 PCI-E |
(17) |
| 24. |
RADEON X850 XT 256MB 520/1080 PCI-E |
(32) |
| 25. |
GeForce 6600GT SLI 2x128MB, 500/1000 |
(20) |
| 26. |
RADEON X800 128MB 400/700 PCI-E |
(18) |
| 27. |
RADEON X1800 XT PE 512MB 700/1600 |
(--) |
| 28. |
GeForce 6800 GT 256MB 350/1000 PCI-E |
(13) |
| 29. |
GeForce 6800 Ultra 256MB 425/1100 PCI-E |
(23) |
| 30. |
GeForce 7800GT SLI 2x256MB, 400/1000 |
(24) |
| 31. |
RADEON X800 XL 256MB 400/990 PCI-E |
(26) |
| 32. |
GeForce 6600LE 128MB 300/550 PCI-E |
(29) |
| 33. |
RADEON X550 256MB, 400/500 |
(35) |
| 34. |
RADEON X550 XT 128MB, 400/900 |
(--) |
| 35. |
GeForce 6600 SLI 2x128MB 300/550 |
(30) |
| 36. |
GeForce 6800 GT SLI 2x256MB 350/1000 |
(28) |
| 37. |
GeForce 7800GTX 512MB, 580/1760 |
(--) |
| 38. |
GeForce 7800GTX 512MB, 550/1700 |
(--) |
| 39. |
GeForce 6800 Ultra SLI 2x256MB 425/1100 |
(31) |
| 40. |
GeForce 7800GTX SLI 2x256MB, 486/1350 |
(27) |
| 41. |
RADEON X850 XT PE 256MB 540/1180 PCI-E |
(33) |
| 42. |
RADEON X800 XL 512MB 400/990 PCI-E |
(34) |
| 43. |
GeForce 7800GTX SLI 2x512MB, 580/1760 |
(--) |
| 44. |
RADEON X600 PRO 128MB, 400/600 |
(36) |
| 45. |
RADEON X300 128MB, 325/400 |
(37) |
Prospect rating (the previous rating is in parenthesis):
| 01. |
GeForce 7800GTX SLI 2x512MB, 580/1760
|
(--) |
| 02. |
GeForce 7800GTX SLI 2x256MB, 486/1350 |
(01) |
| 03. |
GeForce 7800GT SLI 2x256MB, 400/1000 |
(02) |
| 04. |
GeForce 7800GTX 512MB, 580/1760 |
(--) |
| 05. |
GeForce 7800GTX 512MB, 550/1700 |
(--) |
| 06. |
RADEON X1800 XT PE 512MB 700/1600 |
(--) |
| 07. |
RADEON X1800 XT 512MB 625/1500 |
(04) |
| 08. |
GeForce 7800GTX 256MB, 486/1350 |
(03) |
| 09. |
GeForce 6800 Ultra SLI 2x256MB 425/1100 |
(05) |
| 10. |
GeForce 7800GTX 256MB, 459/1300 |
(06) |
| 11. |
GeForce 6800 GT SLI 2x256MB 350/1000 |
(09) |
| 12. |
GeForce 7800GTX 256MB, 430/1200 |
(07) |
| 13. |
GeForce 7800GT 256MB, 450/1100 |
(08) |
| 14. |
GeForce 7800GT 256MB, 400/1000 |
(10) |
| 15. |
RADEON X1800 XL 256MB 500/1000 |
(11) |
| 16. |
GeForce 6800 SLI 2x256MB 350/600 |
(12) |
| 17. |
GeForce 6800 Ultra 256MB 425/1100 PCI-E |
(13) |
| 18. |
GeForce 6800 GS 256MB 425/1000 PCI-E |
(15) |
| 19. |
GeForce 6800 GT 256MB 350/1000 PCI-E |
(14) |
| 20. |
RADEON X850 XT PE 256MB 540/1180 PCI-E |
(16) |
| 21. |
RADEON X850 XT 256MB 520/1080 PCI-E |
(17) |
| 22. |
RADEON X1600 XT 256MB 590/1380 |
(18) |
| 23. |
GeForce 6800XT 256MB 425/1000 PCI-E |
(--) |
| 24. |
GeForce 6800 256MB 350/600 PCI-E |
(22) |
| 25. |
RADEON X800 XL 512MB 400/990 PCI-E |
(20) |
| 26. |
RADEON X800 XL 256MB 400/990 PCI-E |
(21) |
| 27. |
GeForce 6600GT SLI 2x128MB, 500/1000 |
(19) |
| 28. |
RADEON X1600 PRO 256MB 500/800 |
(--) |
| 29. |
RADEON X800GTO 256MB 400/990 PCI-E |
(23) |
| 30. |
RADEON X800 GT 256MB 475/990 PCI-E |
(25) |
| 31. |
GeForce 6600 DDR2 256MB 350/800 PCI-E |
(26) |
| 32. |
GeForce 6600GT 128MB 500/1000 PCI-E |
(24) |
| 33. |
RADEON X1300 PRO 256MB, 600/800 |
(28) |
| 34. |
RADEON X800 128MB 400/700 PCI-E |
(27) |
| 35. |
GeForce 6600 SLI 2x128MB 300/550 |
(29) |
| 36. |
RADEON X700 PRO 256MB 425/860 PCI-E |
(30) |
| 37. |
GeForce 6600 DDR2 512MB 300/550 PCI-E |
(31) |
| 38. |
RADEON X700 256MB 400/700 PCI-E |
(32) |
| 39. |
RADEON X1300 256MB, 450/550 |
(--) |
| 40. |
GeForce 6600 128MB 300/550 PCI-E |
(33) |
| 41. |
GeForce 6600LE 128MB 300/550 PCI-E |
(34) |
| 42. |
RADEON X550 XT 128MB, 400/900 |
(--) |
| 43. |
RADEON X550 256MB, 400/500 |
(36) |
| 44. |
RADEON X600 PRO 128MB, 400/600 |
(35) |
| 45. |
RADEON X300 128MB, 325/400 |
(37) |
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. Games artefacts gallery
Games quality section
8. Game Quality Control!
Game Quality Control
We thank all the companies that provided us with graphics cards and other equipment for our 3Digest, including:
HIS (Peter Yueng personally)
Bench'emAll! and personally Alexander Kondratyuk AKA Render for helping us optimize bechmarking.
|
|
 |
|
|
|
