The biggest mistake ATI made last year was under-estimating the AGP market and focusing too much on PCI-Express platforms. They pretty much admitted that themselves. They had high-end X800 AGP cards out there, but they were all but impossible to find anywhere. Without any available competition for the high-end AGP platform people moved en masse to nVidia's camp and bought up the 6800GT and, to a lesser extent, the 6800 Ultra at a remarkable rate.

With ATI loyalists screaming for AGP boards, ATI finally made a couple of moves to increase their AGP product range. The key move was Rialto, ATI's PEG-to-AGP bridge chip. Rialto is a small chip that sits on the back of ATI's graphics cards and converts the signals coming from native PCI-Express graphics cores into signals that the AGP bus can understand. Basically it allows ATI to design a native PCI-Express chip and put it on an AGP graphics card.

What many people were waiting to hear was the announcement of a new high-end AGP card. Not long after Rialto was revealed, ATI broke the news that they would be making available an X850XT for the AGP bus, but surprised a lot of folks by saying that it wouldn't be using their brand new Rialto bridge chip. Instead, the X850XT would be designed using a new native AGP graphics core based on the PCI-Express R480 core. The result is the R481 core which in all aspects, other than the bus interconnect, is identical to the R480. So the masses rejoiced. ATI has finally started providing high-end AGP parts.

HIS, one of ATI's original launch partners and one of only a few companies to offer a full range of ATI graphics cards, was one of the first partners to announce X850XT AGP cards after ATI unveiled them to the public. They had not just one, but several different models to choose from targeted at a couple of different segments. On the enthusiast level are the two cards that I'll be looking at in this review; the X850XT IceQII Platinum Edition and the X850XT IceQII Turbo. Let's get to it.

Bundles

Both the X850XT-PE and X850XT come with respectable bundles. The only difference between the bundles are the X850XT comes with a CD that contains HIS' iTurbo utility and a DVD containing bonus software and games. The X850XT-PE doesn't need the iTurbo software so that's not included, but for some reason it also doesn't include the bonus DVD. It does however include a PowerDVD 6 CD. I've listed the contents of both bundles below:

X850XT IceQII Platinum Edition	X850XT IceQII Turbo Bundle
HIS X850XT IceQII Platinum Edition Bundle Flatout game CDs PowerDVD 6 CD HIS "Power Up, Gamers" drivers and software CD Video-In/Video-Out cables Component-out cable S-Video cable Composite cable DVI to VGA adapter Molex power cable Installation Manual	HIS X800XL IceQII Turbo Bundle Flatout game CDs iTurbo overclocking software CD Bonus software DVD HIS "Power Up, Gamers" drivers and software CD Video-In/Video-Out cables Component-out cable S-Video cable Composite cable DVI to VGA adapter Molex power cable Installation Manual

The R481 core is built using about 160 million transistors on TSMC's 130nm low-k silicon process. It has 16 rendering pipeline with 6 vertex shader units and a 256bit memory bus. It supports all the same features as the other R4x0 cards including full DX9.0b support, OpenGL 2.0, and ATI's own 3Dc. Right now the R481 core can only be found on the two AGP versions of the X850XT graphics card I'm looking at in this review, the X850XT Platinum Edition and the X850XT.

Both the X850XT Platinum and the X850XT are physically identical (save for a sticker on the back telling you which is which). The only differences between the two are the core and memory speeds. The X850XT-PE has a 540MHz core clock and 590MHz memory (1,180MHz effective) while the X850XT has a 520MHz core clock and 540MHz memory (1,080MHz effective). Both cards use the same Samsung GDDR3 memory modules rated for 1.6ns (625MHz).

The chart below contains the full specs of both cards along with the specs of the other cards I'll be using in this review.

Specifications
	HIS X850XT IceQII Platinum	HIS X850XT IceQII Turbo	Radeon X800XT Platinum	Radeon X800XT	Radeon X800 Pro
Core	R481	R481	R420	R420	R420
Silicon Process	130nm low-k	130nm low-k	130nm low-k	130nm low-k	130nm low-k
Transistor Count (millions)	160	160	160	160	160
Core Speed MHz	540	520 (540 Turbo)	520	500	450
Memory Speed MHz	590 GDDR3	540 GDDR3 (590 Turbo)	560 GDDR3	500 GDDR3	475 GDDR3
Bus Standard	AGP 8x	AGP 8x	AGP 8x	AGP 8x	AGP 8x
Bus Width	256bit	256bit	256bit	256bit	256bit
Pixel Pipelines	16	16	16	16	12
Peak Memory Bandwidth (GB/s)	37.76	34.56 (37.76 Turbo)	35.84	32.00	30.40
Pixel Fillrate (million pixels/sec)	8,640	8,320 (8,640 Turbo)	8,320	8,000	5,400
Texel Fillrate (million texels/sec)	8,640	8,320 (8,640 Turbo)	8,320	8,000	5,400
MSRP ($US)	$549	$449	$499	-	$399

Both of the X850XTs support the following features:

SMARTSHADER HD

• Support for Microsoft® DirectX® 9.0 programmable vertex and pixel shaders in hardware
• DirectX 9.0 Vertex Shaders
  • Vertex programs up to 65,280 instructions with flow control
  • Single cycle trigonometric operations (SIN & COS)
• Direct X 9.0 Extended Pixel Shaders
  • Up to 1,536 instructions and 16 textures per rendering pass
  • 2nd generation F-buffer technology accelerates multi-pass pixel shader programs with unlimited instructions
  • 32 temporary and constant registers
  • Facing register for two-sided lighting
  • 128-bit, 64-bit & 32-bit per pixel floating point color formats
  • Multiple Render Target (MRT) support
• Complete feature set also supported in OpenGL® via extensions

SMOOTHVISION HD

• 2x/4x/6x Anti-Aliasing modes
  • Sparse multi-sample algorithm with gamma correction, programmable sample patterns, and centroid sampling
  • Lossless Color Compression (up to 6:1)at all resolutions, including widescreen HDTV resolutions
  • Temporal Anti-Aliasing
• 2x/4x/8x/16x Anisotropic Filtering modes
  • Up to 128-tap texture filtering
  • Adaptive algorithm with bilinear (performance) and trilinear (quality) options

3Dc

• High quality 4:1 Normal Map Compression
• Works with any two-channel data format

HYPER Z HD

• 3-level Hierarchical Z-Buffer with Early Z Test
• Lossless Z-Buffer Compression (up to 48:1)
• Fast Z-Buffer Clear
• Z Cache optimized for real-time shadow rendering
• Optimized for performance at high display resolutions, including widescreen HDTV resolutions

VIDEOSHADER HD

• Seamless integration of pixel shaders with video in real time
• FULLSTREAM™ video de-blocking technology for Real, DivX, and WMV9 formats
• VIDEOSOAP™ noise removal filtering for captured video
• MPEG1/2/4 decode and encode acceleration
  • DXVA Support
  • Hardware Motion Compensation, iDCT, DCT and color space conversion
• All-format DTV/HDTV decoding
• YPrPb component output for direct drive of HDTV displays
• Adaptive Per-Pixel De-Interlacing and Frame Rate Conversion (temporal filtering)

Display Features

• Dual integrated display controllers
• Dual integrated 10 bit per channel 400 MHz DACs Integrated
• 165 MHz TMDS transmitter (DVI 1.0 / HDMI compliant and HDCP ready)
• Integrated TV Output support up to 1024x768 resolution

Both cards use the same IceQII cooler, which in reality is the same cooler designed by Arctic Cooling and sold in the retail channels as the ATI Silencer 4. It's a double-height cooler, meaning that to install the card you need to have the slot adjacent to the AGP slot free. The cooler is quite a bit larger than many other coolers, including the double-height cooler ATI uses on their own X850XT cards. There is always a chance that, depending on your motherboard, the cooler could get in the way of some components. On my ABIT AV8-3rd Eye motherboard for example, the cooler blocks one of my motherboards USB headers. I can still use it, but it's a very tight fit and I have to remove the card to get at the header.

Basically, the fan draws air from inside your case and forces it through a large plastic housing. Inside the housing is a large heatsink with metal cooling fins attached to a copper base. The base sits directly on the graphics core to draw heat away from it. The air being forces through the housing then picks up the heat from the heatsink and gets directed out the back of your case. Arctic Cooling has been using this basic design now for quite some time on several different VGA coolers and they've all proven to be very effective at cooling.

If you look at the fan spinning you might think that it's spinning backward because the blades aren't angled towards the fans rotation, but away from it. Instead of "digging" into the airflow, the fan instead gently pushes it away through the plastic housing. This is the key to the coolers other notable attribute; silence.

HIS has also slightly modified the cooler to have it match the "Ice" theme by changing the color of the fan and making the whole unit UV reactive. Under a UV light, the cooler glows a bright blue.

Along with the default clock speeds I also decided to test the X850XT IceQII Turbo card in "Turbo" mode, which sets the it to 540c/590m. Even though these clock speeds aren't enabled by default, I believe that the manufacturer overclocked settings are the ones that should be used when comparing these cards. The following is my standard test setup breakdown.

Anti-Aliasing and Anisotropy were applied in the game where the options existed. For games that did not support those options natively, the graphics card control panel was used. V-Sync was also forced off for each test via the graphics card control panel. Otherwise all other graphics card control panel settings were left to their default values.

Batch files were used when possible for automated benchmarking, the details of the commands used are outlined for each test. Manual benchmarking was done using FRAPS 2.5.5.

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 expect for Richard Burns Rally which will not load without sound hardware installed and operational.

Test System Specs

• CPU: AMD Athlon XP 3500+, 2.2GHz (Newcastle core)
• Motherboard: ABIT AV8-3rd Eye, VIA K8T800 Pro
• Videocards: (Check the Specifications chart for more details on these cards).
• HIS X850XT IceQII Platinum Edition
• HIS X850XT IceQII Turbo
• Radeon X800XT Platinum Edition
• Radeon X800XT
• Radeon X800 Pro
• Memory: 1GB (2x512MB) OCZ DDR PC-4200 Platinum Edition @ 400MHz/2-2-2-5 1T
• HDD: Seagate Barracuda V SATA, 7200rpm
• Sound: Onboard, disabled for most tests
• PSU: Enermax Noisetaker EG495P-PE 485W
• Case: CoolerMaster Stacker
• OS: Windows XP Pro SP2

Testing was done on a fully assembled PC.

Benchmark Software

• Splinter Cell: Chaos Theory, v1.02 (DirectX)
• Half-Life 2, Source Engine 7 (DirectX)
• Far Cry, v1.31 (DirectX)
• Doom3, v1.1 (OpenGL)
• Chronicles of Riddick: Escape from Butcher Bay (OpenGL)
• Richard Burns Rally, v1.02 (DirectX)
• Pacific Fighters, v3.04 (OpenGL)
• 3DMark03, build 3.6.0 (DirectX)
• 3DMark05, build 1.2.0 (DirectX)

Splinter Cell Chaos Theory

I benchmarked Splinter Cell Chaos Theory using the benchmarking batch file that comes with the game. Anti-aliasing, anisotropy, and resolution were set within the game. All other settings remained the same.

The location of the batch file is <game dir>\System\Timedemo.bat. Each result is appended to the <game dir>\System\TimeDemoResults.xls file.

The Splinter Cell Chaos theory results are below. Click the text links at the top of the chart to change settings or, if you wish, click the "Show All" link which will show all the charts on the page at once.

Half-Life 2

Testing of Half-Life 2 was done using AnandTech's five recently released Source Engine 7 timedemos (which are available on this page). I ran each timedemo for each resolution and AA/AF setting used in the charts below, and then averaged the results to get the final score. Anti-aliasing and Anisotropy were set on the command line.

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 or, if you wish, click the "Show All" link which will show all the charts on the page at once.

[ No AA / No AF ] [ 2x AA / 4x AF ] [ 4x AA / 8x AF ] [ 6x AA / 16x AF ] [ Show All ]

Doom 3

Doom3 was tested using the built in timedemo, demo1.demo. I benchmarked the default "High Quality" mode, which sets Anisotropy to 8x, and then combinations of Anti-Aliasing and Anisotropy over the resolutions shown in the chart below. Anti-aliasing and Anisotropy were set on the command line.

Another batch file was used to automate Doom3 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 or, if you wish, click the "Show All" link which will show all the charts on the page at once.

[ No AA / No AF ] [ 2x AA / 4x AF ] [ 4x AA / 8x AF ] [ 6x AA / 16x AF ] [ Show All ]

Chronicles of Riddick

Chronicles of Riddick was benchmarked using the built-in timedemo feature. No batch file was used (mainly because I didn't have time to make one). The timedemo used is a recording of the intro sequence. The Shader Model 2.0 path was used for all cards, including the 6600GT, which takes a huge performance hit using the Shader Model 3.0 path. Anti-aliasing and Anisotropy were set in the graphics card control panel.

The Chronicles of Riddick results are below. Click the text links at the top of the chart to change settings or, if you wish, click the "Show All" link which will show all the charts on the page at once.

[ No AA / No AF ] [ 2x AA / 4x AF ] [ 4x AA / 8x AF ] [ 6x AA / 16x AF ] [ Show All ]

Far Cry

Far Cry benchmarking was a little different. In the interest of time (Far Cry takes quite awhile to benchmark) I only tested anti-aliasing on the High setting and did not test anisotropy performance at all. I did however test over three different maps using custom timedemos. I averaged the results from each of those timedemo runs to get the final score. Anti-aliasing was set in the "Configure Far Cry" app.

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 or, if you wish, click the "Show All" link which will show all the charts on the page at once.

[ No AA ] [ High AA ] [ Show All ]

Richard Burn’s Rally

I used Fraps to benchmark Richard Burn's Rally using two custom demos recorded on the Great Britain Chirdonhead II stage with rain. I started Fraps framerate logging as soon as the marshal launched the car and had it automatically stop logging after 90 seconds. I logged each pass 3 times then averaged the results to get the final score for that setting and resolution. The default replay camera was used and sound was enabled. Anti-aliasing and Anisotropy were set via the graphics card control panel.

The Richard Burn's Rally results are below. Click the text links at the top of the chart to change settings or, if you wish, click the "Show All" link which will show all the charts on the page at once..

[ No AA / No AF ] [ 2x AA / 4x AF ] [ 4x AA / 8x AF ] [ 6x AA / 16x AF ] [ Show All ]

Pacific Fighters

I benchmarked Pacific Fighters by loading the included "N1K1 vs BeauFighter.ntrk" track and logged framerates 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 "Maximum Settings" for all graphics cards and sound was disabled in the Pacific Fighter Setup app. 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 or, if you wish, click the "Show All" link which will show all the charts on the page at once..

[ No AA / No AF ] [ 2x AA / 4x AF ] [ 4x AA / 8x AF ] [ 6x AA / 16x AF ] [ Show All ]

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.

Game Test 1 - Wings of Fury Game Test 2 - Battle of Proxycon Game Test 3 - Troll's Lair Game Test 4 - Mother Nature Vertex Shader Pixel Shader 2.0 Overall Score

3DMark05

Like 3DMark03, I set 3DMark05 up to loop three times to get the final score. I tested 3DMark05 using the default detected settings.

Game Test 1 - Return to Proxycon Game Test 2 - Firefly Forest Game Test 3 - Canyon Flight Pixel Shader Vertex Shader - Simple Vertex Shader - Complex Overall Score

To test overclocking potential I used the ATI Radeon tweaking app known as ATITool. It has an automatic overclocking feature which I used to find a base overclock, then I used the manual overclocking features from there to find the absolute maximum.

I managed to get the X850XT from a stock setting of 520c/540m to 550c/595m. That's quite a bit, but when you consider that it's only 10MHz and 5MHz over the iTurbo speed, you can see that HIS is pushing the card pretty hard to get Platinum Edition speeds out of the card.

The X850XT Platinum was a bit more impressive. I managed to get it from the stock speed of 540c/590m to 575c/615m speed. That's 35MHz over the stock core speed and 25MHz over the stock memory speed, not insignificant.

Keep in mind that overclocking success varies from board to board, so there's a good chance that these results might not be inline with what you see.

Below you'll find the performance results for the overclocked settings.

Half-Life 2 Overclocked Performance No Anti-Aliasing / No Anisotropy

1024x768  113.2 113.5 113.2 113.5  1280x960  111.8 111.9 111.5 111.5  1600x1200  106.0 104.6 105.0 103.4   0 65 130 Average Framerate  HIS X850XT-PE @ 575c/615m  HIS X850XT-PE  HIS X850XT @ 550c/595m  HIS X850XT

Half-Life 2 Overclocked Performance 4x Anti-Aliasing / 8x Anisotropy

1024x768  111.8 111.9 111.7 111.7  1280x960  108.0 107.1 107.1 105.7  1600x1200  92.3 89.6 91.1 86.8   0 65 130 Average Framerate  HIS X850XT-PE @ 575c/615m  HIS X850XT-PE  HIS X850XT @ 550c/595m  HIS X850XT

Doom3 Overclocked Performance No Anti-Aliasing / No Anisotropy

1024x768  105.9 104.3 105.0 102.4  1280x960  87.7 84.5 85.5 81.0  1600x1200  67.5 64.0 65.1 60.6   0 60 120 Average Framerate  HIS X850XT-PE @ 575c/615m  HIS X850XT-PE  HIS X850XT @ 550c/595m  HIS X850XT

Doom3 Overclocked Performance 4x Anti-Aliasing / 8x Anisotropy

1024x768  79.6 76.3 77.4 72.5  1280x960  57.8 54.8 55.8 51.6  1600x1200  42.0 39.7 40.4 37.4   0 60 120 Average Framerate  HIS X850XT-PE @ 575c/615m  HIS X850XT-PE  HIS X850XT @ 550c/595m  HIS X850XT

HIS' X850XT Platinum and X850XT are physically identical. In fact, without the little sticker on the back of the card proclaiming which is which, the only way to tell them apart would be to install them and check out their clock speeds.

Thanks to iTurbo, the X850XT even turns into a X850XT Platinum with the click of a button. The only tangible difference between the two cards then is essentially overclocking ability. In my tests, the X850XT Platinum can reach a little higher than the X850XT for both core and memory speeds.

If you're not interested into overclocking then you should stick with the X850XT and use iTurbo which doesn't void your warranty and is easy to install. There's no real reason to shell out the extra $100 for the Platinum version. In fact, the X850XT Platinum Edition has it's memory clocked at precisly 584MHz, 6MHz slower than the X850XT's 590MHz memory clock when iTurbo is enabled. Neither card is cheap by any means, but for $100 less the X850XT is clearly the better buy.

Overall, both of the cards offer excellent performance that can't be beat… at least not on the AGP bus. In the end it may be the AGP bus that is the sticking point with these cards. PCI-Express is here to stay and with all the big motherboard chipset manufacturers such as Intel, ATI, nVidia, and VIA all busy producing PCI-Express based technologies, I would personally find it extremely difficult to spend this kind of money on an AGP based card. The relatively large investment will have to be shelved when the time comes to upgrade to a PCI-Express platform.

---

---

HIS X850XT IceQII Platinum Edition AGP Overall Score 4 out of 5 Plus Extremely Fast Great cooling performance with hardly any noise Nice bundle VIVO Blue UV glow looks cool Minus Very, very expensive

HIS X850XT IceQII Turbo AGP Rage3D Editor's Choice Award Overall Score 5 out of 5 Plus Very Fast iTurbo turns it into a Platinum Edition for $100 less $50 cheaper than the X800XT Platinum Great cooling performance with hardly any noise Nice bundle Blue UV glow looks cool Minus Slightly slower than the X800XT Platinum