Rage3D BFG GeForce 7800 GT OC SLi Review Rage3D BFG GeForce 7800 GT OC SLi Review
BFG Tech was one those partner than had their factory overclocked 7800 GT OC ready and on shelves on the day of the launch. The 7800 GT OC, as the OC tag implies, has its core set to 425MHz, 25MHz over stock, and the memory set to 1,050MHz, 25MHz over stock. The board in every other respect is identical to the stock reference design.
In this review I'll be taking a look at two of these cards, retail packaged, sent to me by BFG Tech. Having two of them will obviously let me check out SLi performance, which will naturally be the focus of this review, but I will also run in single card mode and compare directly to a single reference 7800 GT to give readers a reference point for the performance of the 7800 GT OC.
However, where the GTX has 24 fragment pips and 8 vertex units, the GT has 20 pipes and 7 vertex units. The physical card is also shorter than the GTX, by about an inch, and has a different layout and also uses a different cooler.
Although the GT has reduced pipelines and one less vertex shader unit when compared to the GTX, it still retains all the features of the GTX. These include full Shader Model 3.0 support, support for NVIDIA's UltraShadow II, PureVideo, and IntelliSample 4.0 technologies among others and, of course, supports NVIDIA's SLi.
I'm a big fan of charts and tables and not so much with the words, so here's breakdown of the 7800 GT OC specs with comparisons to the other cards used in this review:
| BFG GeForce 7800 GT OC | GeForce 7800 GT | GeForce 7800 GTX | Radeon X1800 XT | Radeon X850XT-PE | |
| Core | G70 | G70 | G70 | R520 | R480 |
| Silicon Process | 110nm | 110nm | 110nm | 90nm low-k | 130nm low-k |
| Transistor Count (millions) |
302 | 302 | 302 | 321 | 160 |
| Core Speed MHz | 425 | 400 | 430 | 630 | 540 |
| Memory Speed MHz | 525 GDDR3 (1.05GHz) |
500 GDDR3 (1.0GHz) |
600 GDDR3 (1.2GHz) |
750 GDDR3 (1.5GHz) |
590 GDDR3 (1.18GHz) |
| Bus Standard | PEG x16 | PEG x16 | PEG x16 | PEG x16 | PEG x16 |
| Bus Width | 256bit | 256bit | 256bit | 256bit | 256bit |
| Pipeline Configuration Textures/Pixels/Z Samples (Per Clock) |
20/16/32 | 20/16/32 | 24/16/32 | 16/16/16 | 16/16/16 |
| Vertex Units | 7 | 7 | 8 | 8 | 6 |
| Peak Memory Bandwidth (GB/s) |
33.6 | 32.0 | 38.4 | 48.0 | 37.8 |
| Pixel Fillrate (million pixels/sec) |
6,800 | 6,400 | 6,880 | 10,080 | 8,640 |
| Texel Fillrate (million texels/sec) |
8,500 | 8,000 | 10,320 | 10,080 | 8,640 |
| API Compliancy | DX 9.0c Shader Model 3.0 |
DX 9.0c Shader Model 3.0 |
DX 9.0c Shader Model 3.0 |
DX 9.0c Shader Model 3.0 |
DX 9.0b Shader Model 2.0 |
| MSRP ($US) | $449 | $449 | $599 | $549 | $449 |
Resolutions
We chose screen resolutions based on the most commonly used displays today. 1280x1024 is the native resolution of most 17" and 19" 4:3 LCD panels and often used on lower-end CRTs (and is comparable to 1280x960, which is the proper 4:3 resolution). 1600x1200 is used on many mid-range to high-end CRTs and also on some LCD panels and is generally considered "high-res" by most gamers. 1920x1200 is the native resolution on many widescreen LCD panels such as Dell's popular 2405FSW LCD panel, and 2048x1536 is the highest resolution the test display can achieve and beyond the abilities of most displays out there. It's useful for giving these cards a workout where CPU performance would have little impact.
Image Quality Settings
For the settings we have tested with no Anti-aliasing or Anisotropic Filtering, with AA and AF separately, and again with both AA and AF enabled at the same time. This way we can see how much of impact these technologies have on 3D performance individually and when used together.
Test System Specs
Benchmark Software
The Windows XP desktop was set to 1280x960 with a 32bit color depth and 85Hz refresh rate for all tests. Refresh rate locks for 3D graphics modes, as supported by both NVIDIA and ATI graphics control panels, was not enabled. V-Sync was forced off via the graphics card control panel as well. All other graphics card control panel settings were left to their default settings unless otherwise noted.
Anti-Aliasing and Anisotropy were applied in the game engine where the options existed. For games that did not support those options natively, the graphics card control panel was used.
Custom batch files were used when possible for automated benchmarking (the details of the commands used are outlined for each test). When manual benchmarking was necessary Fraps version 2.6.4 was used.
Benchmarking was done with Windows set to the "Adjust for best performance" profile, and all unnecessary Windows services and hardware devices were disabled. The latest drivers for each necessary hardware component were installed prior to testing and kept consistent throughout.
The Onboard sound was disabled for all tests unless otherwise noted.
To setup the test machine I installed Windows XP, patched and tweaked it, and installed all the required games, apps, utilities, and hardware drivers needed for the testing procedure except for the graphics drivers. Using Norton Ghost, I then cloned the drive onto a second identical hard-drive. After that I installed the ATI drivers on one hard-drive and the NVIDIA drivers on the other. Testing the videocards was then a simple matter of swapping videocards and hard-drives when required.
For this review I also overclocked my aging A64 3500+ to 2.4GHz from 2.2GHz to try and reduce the CPU bottleneck on these graphics cards.
Ok I lied, I will do some IQ comparisons here, but not the normal style. Instead I will look at NVIDIA's SLI AA as a sort of addendum to the results in my previous reviews. First, a quick once over of SLI AA for the uninitiated.
SLI-AA, as the name implies, is a special AA mode that is unique to NVIDIA's SLi technology. In reality SLI-AA is another SLI rendering mode, and sits in the same options panel (once enabled via a Coolbits) as the other two SLi modes; SFR and AFR. SFR (Split Frame Rendering) splits the frame horizontally and has each card render one half. AFR (Alternate Frame Rendering) on the other hand has each card render alternating frames, one cad takes even numbered frames and the other takes odd numbered frames. Where both SFR and AFR modes increase performance by dividing the rendering workload, SLI-AA increases image quality by dividing the anti-aliasing work load.
Truthfully, NVIDIA's nemesis ATI originally came up with this concept when they announced CrossFire, their answer to NVIDIA's SLI mutli-GPU technology, back in May. NVIDIA however managed to beat them to the punch and include it with a simple driver update well ahead of ATI having CrossFire hardware available.
With NVIDIA's SLI-AA there are actually two modes to choose from, 8x SLI and 16x SLI. Getting them to work requires some simple registry modification (simple depending on your level of experience I suppose. The steps are outlined here). I'll take the 8x SLI mode as an example and attempt to explain how it works.
When using 8x SLI each card renders the same scene with 4x AA, but with slightly offset (jittered) sample patterns so that, when the scenes are combined, the result is in effect 8x AA. However, instead of 4 AA samples and 1 texture sample, the combined result is 8 AA samples and 2 texture samples. In essence, a sort of super-sampled mixed AA mode very similar to NVIDIA's 8xS mode. Here's a nice chart which should help:
With the knowledge of how the 8x SLI mode works we can make a pretty good guess at how the 16x SLI mode works as well. The difference between the two modes is that instead of using 4x AA on both cards as is the case with 8x SLI, 16x SLI uses 8xS on both cards resulting in 16 AA samples and 4 textures samples being used to render the final image, basically giving us true full scene super-sampling.
With that little SLI-AA primer out of the way, let's take a look at image quality and the level of performance you can expect.
First we'll look at image quality. I loaded up a Half-Life 2 saved game position which presents a really good scene for AA comparisons. There are a lot of angles and geometry here, along with a lot of partially transparent polygons where technologies such as NVIDIA's various mixed modes and Transparency AA really shine. Click the image below to launch a javascript application which will let you easily compare AA image quality.
I'll outline my thoughts on the image quality aspect at the end of this page, but first let's take a look at performance. I tested both SLI 8x and SLI 16x on Half-Life 2 and Doom 3 over my standard timedemo tests using all the resolutions we normally run. Here are the results:
SLI AA Summary
After checking out SLI AA for a bit I'm honestly not sure what the point of it is supposed to be. I assumed that after reading the documents the whole point was to improve image quality, but as you can see from the IQ tool above it doesn't do a good a job as you would initially expect. SLI 8x doesn't seem to be anywhere near as good at cleaning up the image as NVIDIA's 8xS mode, and SLI 16x looks to only be about on par with 8xS (though maybe a little better in some cases). However as you can see the performance impact of SLI AA is quite high and certainly wouldn't be useable in most newer games.
One area I think where SLI AA would be of use, and likely remain very useable, is with any of the older games that don't demand as much from the graphics system (or maybe even for the newer slower paced games where high-framerate isn't an absolute necessity). In that case, where a modern CPU would be the limiting factor and not graphics power, SLI AA has a place. For the newer twitch games though you would need a very, very high-spec machine to make use of it.
Ultimately I think SLI AA was more a response to a similar technology that ATI announced with their CrossFire multi-GPU technology than anything else, but I'm not going to take points off for adding a feature, no matter the reason, and even if it's not entirely practical in most situations it's there and it works. NVIDIA just needs to figure out a way to make it easier to access.
Along with SLI AA I also wanted to check out some of the other improvements that NVIDIA has included in the release 80 drivers.
Previously when using SLI you had to use identical cards from the same manufacturer. While I don't think this restriction would have been a problem for anyone planning on SLI at the outset of an upgrade, it would probably have caused problems for someone looking at upgrading to SLI in stages. For example, if you were to buy a single SLI capable card with the intention of upgrading to a full SLI system somewhere down the road, maybe a few months or a year or more later, looking for another identical card could have been a problem if the manufacturer stopped making them or changed them during that time (even different BIOS versions used on otherwise identical cards could break the old SLI).
Thankfully with the release 80 drivers NVIDIA has fixed things up so that you can now mix and match vendor cards at will, making upgrading to SLI much easier.
To test the claim I simply installed the BFG 7800 GT OC and the reference 7800 GT in my system and connected the SLI link. Upon boot the system recognized it as SLI capable and popped up one of those "click here to enable SLI" popups. Once enable, SLI worked as well as it did with both identical BFG cards installed. Not only does mixed vendor work, but from what I can tell the reference 7800 GT is actually automatically overclocking itself to match the BFG 7800 GT OC clock rates. You still can't mix a GT with a GTX (I tried), and apparently if the clock rates between both cards differ by enough then mixing won't work, but mixed-vendors goes a long way toward needed improved flexibility.
Other improvements to SLI include the ability to enable and disable it dynamically without needing to reboot, TV/HDTV support, and further additions to list of games supported in the profiles.
As far as SLI as a technology goes, the biggest issue with it for many people was that you had to use identical cards. This issue no longer exists thankfully, and the update seems to work very well (I haven't run into any problems in this regard, but NVIDIA does mention that sometimes it will not work and a reboot might be required). It's still not as flexible as ATI's CrossFire (where you can mix and match vendors without big restrictions on the GPU), but it's a lot better than it was before.
Another issue that some people have taken with SLI was the need for game profiles in the drivers. The problem is that any game without a profile in the drivers won't work with the performance enhancing modes of SLI (SLI AA should always work). This could certainly a big deal if you play a game that NVIDIA hasn't included a profile for (you can add game profiles yourself, but it requires registry editing and some knowledge of SLI. Certainly not a straight forward task for most people). The good news is that NVIDIA has been working with developers and has helped them get support for SLI right in the game, negating the need for NVIDIA to use game profiles in their drivers. F.E.A.R. and Call of Duty 2 are two such games that include SLI support out of the box with undoubtedly more to follow.
Even ATI saw the success that NVIDIA was having and changed their attitude toward multi-GPU graphics pretty quickly (but not quickly enough some would say). Over the last year and a half NVIDIA has taken SLI from debatable impracticality to unquestionably practical and highly desirable.
Next we'll take a look at general SLI performance in a bunch of different games and compare it to a host of other graphics cards over a bunch of different resolutions and settings. Read on...
F.E.A.R.
The "Computer" option was set to Maximum and the "Graphics Card" option was set to High. Anti-aliasing and anisotropic filtering settings were changed in the Advanced video card options panel. Soft Shadows was disabled for all tests.
Quake 4
I used a custom batch file which automatically runs each demo, resolution, and setting. The commandline used in the batch file is detailed below.
The Quake 4 results are below. Click the text links at the top of the chart to change settings.
Battlefield 2
The results are below. Click the text links at the top of the chart to change settings.
Splinter Cell Chaos Theory
In the second chart set I enabled Shader Model 2.0 for the X850XTPE, which is a new feature of the version 1.04 patch, and Shader Model 3.0 for the X1800 XT, 7800 GTX and 7800 GT.
The third and final chart shows the performance of the SM3.0 capable cards with all the extra SM3.0 features enabled in the game.
Click the text links at the top of each chart to change settings.
Half-Life 2
A batch file was used to automate testing; the command line is below for reference. This batch file was used for each card that was tested. The settings surrounded by < > change for each pass:
The Half-Life 2 results are below. Click the text links at the top of the chart to change settings.
Doom 3
Another batch file was used to automate Doom 3 testing as well. The command line is below for reference. This batch file was used for each card that was tested. The settings surrounded by < > change for each pass:
The Doom3 results are below. Click the text links at the top of the chart to change settings.
Chronicles of Riddick: Escape from Butchers Bay
The Chronicles of Riddick results are below. Click the text links at the top of the chart to change settings.
Far Cry
Again a custom batch file was used to benchmark Far Cry. I used 3 demos included with the newer patches to test performance (from the Cooler, Training, and Volcano levels). Because Far Cry benchmark frame rates can vary between each subsequent pass (sometimes fairly significantly), I ran each of the 3 demos 3 times, then averaged the 9 results to get the final score for the detail level.
The command line for the batch file I used to automate Far Cry benchmarking is below. This batch file was used for each card that was tested. The settings surrounded by < & > change for each pass:
The Far Cry results are below. Click the text links at the top of the chart to change settings.
Pacific Fighters
I benchmarked Pacific Fighters by loading the included "N1K1 vs BeauFighter.ntrk" track and logged frame rates using Fraps from the beginning of the track for 90 seconds. All the in game details were set to their maximum levels, including "Landscape Detail" which was set to "Perfect", enabling Pixel Shaded water. Video was set to the "NVIDIA GeForce 6800/6600/FX/4/3" profile for the 7800 GT and 7800 GTX, and set to the "ATI Radeon X800/9800/9700/9600/9500" profile for the X850XT-PE. Anti-aliasing and Anisotropy were set via the graphics card control panel.
The Pacific Fighters results are below. Click the text links at the top of the chart to change settings.
3DMark03
I benchmarked 3DMark03 using the default settings. I set it up to loop each test three times, it then automatically averaged the results and gave a final score. The results for each test are below.
| 3DMark03 Game Test 1 - Wings of Fury |
||||||||||||||
|
||||||||||||||
3DMark05
Like 3DMark03, I set 3DMark05 up to loop three times to get the final score. I tested 3DMark05 using the default detected settings.
| 3DMark05 Game Test 1 - Return to Proxycon |
||||||||||||||
|
||||||||||||||
Overclocking the card was pretty easy using Coolbits, NVIDIA's built-in overclocking tool (once enabled via the registry). I let it automatically detect the best overclock, and then manually adjusted it as high as possible. The end result was 490MHz core and 1.20GHz memory, which is 60MHz over the default core speed and 95MHz over the default memory speed, a very good overclock indeed. This pushed the 7800 GT OC past the default 7800 GTX core and on par with it's memory speed (430c/1.2m) and, as you can see from the performance chart below, the boost in clock speeds was enough to overcome the missing quad and push the 7800 GT OC past the 7800 GTX in F.E.A.R. performance.
|
[ No AA / No AF ]
[ No AA / 16x AF ]
[ 4x AA / No AF ]
[ 4x AA / 16x AF ]
|
Other than the increased clockspeeds there really isn't anything else noteworthy regarding the BFG 7800 GT OC. It's reference through and through in fact.
The other impression I got, at least when I popped open the box, was that there are no games included in the retail bundle. The good news though is that NVIDIA and Activision recently struck a deal to give exclusive rights to NVIDIA partners, like BFG, which allows them to bundle Call of Duty 2 with their retail cards. A brand new kick ass game like Call of Duty 2 certainly jacks up the added value. There was no actual game or a voucher with these cards, but shipping retail versions should include Call of Duty 2 in some form. I would be willing to bet that if you recently bough a 7800 GT from BFG without CoD2 bundled then BFG would be happy to send along a free copy as well.
Fancy coolers with colored PCBs are all good things to have but I'm not going to slight BFG for keeping costs down. After all, when it comes right down to it the real value of a card comes from performance and image quality. Those two things (particularly the first) are all everyone really cares about, and with the 7800 GT OCs performance increase over the stock 7800 GT there's no reason to settle for something slower.
Speed boosted from the factory with a lifetime warranty and excellent support options, the BFG 7800 GT OC should be on your short list when shopping for a kick ass graphics card this holiday season.
BFG GeForce 7800 GT OC SLI
Plus
Minus
