Packaging:

The GeForce GTX 560 Ti SC:

Specifications and Features:

• Free version of 3DMark 11 Advanced Edition included. Users can redeem their code when the product is registered on EVGA.com. This is a $19.99 value. For more information, please visit http://www.evga.com/articles/00594/
• Overclocked out of the box – This card offers a performance increase out of the box, attached is the product spec sheet which includes the details on clockspeeds.
• Special Edition EVGA Precision v2.0.1 – We created a special look of our popular software specifically for the EVGA GTX 560 Ti. This utility allows overclocking, monitoring and fanspeed adjustments. It is located on the DVD that ships with your card. For more information, please visit www.evga.com/precision/
• EVGA OC Scanner – EVGA OC Scanner is fully supported by the EVGA GTX 560 Ti. This utility allows you to benchmark, monitor and stress test your EVGA card. For more information, please visit www.evga.com/ocscanner/
• Voltage Tuning with EVGA ELEET– EVGA ELEET allows voltage tuning on a wide range of EVGA cards, including the GTX 560 Ti. For more information, please visit www.evga.com/eleet/
• EVGA Customer Support – EVGA’s acclaimed customer support can be contacted by support ticket, email, and phone. For more information, please visit  http://www.evga.com/support/
• EVGA Community – Active game servers, thriving forums, integrated chat and social networks allow users to ask questions or get help wherever they feel the most comfortable. For more information, please visit www.evga.com/community/
• EVGA Warranty – EVGA offers a variety of warranties to fit your needs, including our exclusive 10 year Extended Warranty. For more information, please visit www.evga.com/community/
• EVGA Advanced RMA Program – EVGA offers this service to help reduce the downtime of a customer’s system by shipping a replacement product first and lets our valued customer deal with the EVGA directly for quick and efficient service. For more information, please visit www.evga.com/ear/
• EVGA Step-Up Program – EVGA protects your investment for 90 days following your purchase. Qualifying products can be upgraded, one time only, to a select greater performing product. For more information, please visit www.evga.com/stepup/

Next: Test System & Overclocking

Packaging:

The Zotac 560 Ti:

Specifications & Features:

Next: Test System & Overclocking

The video card:

Specifications and features:

Next: Test System & Overclocking

Packaging:

Specifications and Features:

The EVGA GTX 570 SC:

Next: Test System and Overclocking

Test Setup

Hardware

• Motherboard: Gigabyte GA-MA790FXT-UD5P
• CPU: AMD Phenom II X4 965 BE (4,2GHz CORE/2,9GHz IMC)
• RAM: 2x 2GB OCZ Platinum (1550MHz 6-6-6)
• Monitor: LG W2600HP-BF 1920×1200 pixels
• GPU: Two ATI HD 5770 1Gb (Crossfire) & one ATI HD 5870 1Gb

Software

• MS Windows 7 Ultimate 32bit
• Catalyst 10.3 February preview (driver package 8.71.3-100205a-095620E)
• Fraps
• Battlefield Bad Company 2
• Aliens vs. Predator
• Borderlands
• Half Life 2
• Call of Duty Modern Warfare 2

The FPS where recorded over a 20 second period using Fraps. Multiple runs where completed for each game, and the average values where then calculated for each game and GPU setup. Vertical Sync where disabled for all tests. And the highest possible quality settings where used, unless else is noted.

Battlefield – Bad Company 2
1920×1200, 8x MSAA, 16x AF, HBAO, DX11

(x-axis = seconds, y-axis = fps)

It’s pretty close, but the 5870 managed to push out a few more FPS. Very much playable on both setups. But guess a few more FPS could always be desired for online play.

Aliens vs. Predator
1920×1200, 16x AF, DX11

(x-axis = seconds, y-axis = fps)

At a first glimpse this looks like extremely poor, or no Crossfire support at all. However, I made sure that both 5770 GPU’s where active during testing. And after some additional research, this does indeed seem to be a reasonable result in AvP on two crossfired 5770s…  The 5870 is really flexing its muscles in this one!

Borderlands
1920×1200, 16x AF

(x-axis = seconds, y-axis = fps)

I where not able to bypass a FPS limit of 62 in this game. But decided to include the chart anyway, as both card setups stayed under that limit most of the time.

Half Life 2
1920×1200, 4x MSAA, 16x AF

(x-axis = seconds, y-axis = fps)

Both setups managed to push out a healthy amount of pixels. And still, quite impressive graphics for such a old title!

Call of Duty – Modern Warfare 2
1920×1200, 4x AA, “Extra”

(x-axis = seconds, y-axis = fps)

Towards the end of the run when things heat up a little, the crossfired pair of 5770s closed the gap to their bigger brother. But the 5870 is still a head nonetheless.

Quick Final Thoughts
Well there isn’t much to say really. But if you have plans to grab one of these cards, then go for the 5870 if you can. But a pair of 5770s is still to be counted on, and it has proven to be a very potent setup. Nonetheless fun for the tinkerer! But I bet two 5870s is even more fun…

Be sure to check out TestFreaks for more expert reviews on the Radeon HD5770 & HD5870.

(Image source)

A pixel must be only one colour (or no colour at all). To determine the colour of a pixel, a sample is taken from the centre of the pixel, and whatever the colour is at that point, the whole pixel becomes. Full screen anti-aliasing (hereafter called AA) is an evolving method of dealing with this problem. One of the earliest methods for performing AA was supersampling. This is a brute force method for applying AA to the entire screen.

In simple terms, supersampling takes all screen space coordinates (which are generated for a specific screen resolution) and up-samples them, usually by at least a factor of 2. This means that for a frame displayed at 1680×1050 and upsampled by a factor of 2, the up-sampled version would be rendered at 3360×2100 in the off-screen buffer, and for every pixel of the original frame we now have four; double in each direction.

Now that there’s a high-resolution frame to work with, we need to down-sample it again to fit the screen. The colours of each of the four pixels are averaged (where the sample is taken in each pixel is determined by on of several methods, which are beyond the scope of this post), and this average colour is used to fill a single pixel in the down-sampled frame, which is now the same size as the original; equal to the screen resolution. Going back to the image above, rather than being either black or white, a pixel could be a shade of grey, which “smooths” the edge.

Unfortunately, it’s not a perfect world. The biggest disadvantage with supersampling is that it is a memory hog and it is computationally intensive, as you will soon see. But enough with the theory for now, lets see what supersampling can do! First, it is important to know the configuration of the test system, so here it is:

• Intel Core i7 920 @ stock (but undervolted!)
• NVIDIA GTX 275 @ stock
• MSI X58 Pro-E
• 3×2GB G.Skill DDR3-1333
• ForceWare drivers v191.03
• OCZ Vertex 30GB (thanks EnJoY! )
• Western Digital Caviar Black 640GB
• Half-Life 2: Lost Coast (for the test application)

No AA

Well that’s just plain ugly. Just look at the fence. Now we will apply supersampling to see what effect it has on the image. Take note of the framerate with each successive increase in AA by looking at the information in the bottom right (ignore everything but the framerate).

1×2 SSAA

Here we have 1×2 Supersampling AA (SSAA). It is not much use as it only affects horizontal edges, not vertical. Since the fence is mostly vertical edges there is not a great deal of difference. The tall rock in the background is also unaffected by this type of SSAA. Also note that the open gate looks almost identical to the previous image. Let’s move on to 2×1.

2×1 SSAA

Now that’s alot better. The fence is much improved, and so are other vertical edges. The vertical bars in open gate are significantly more defined than the previous two images. Unfortunately, this method does not affect horizontal edges, so the shipwreck on the left looks worse than the previous image, and only slightly better than with no AA at all. Both 1×2 and 2×1 SSAA have a negligible effect on framerate as both are over 120FPS still.

2×2 SSAA

Now we’re getting the best of both worlds. Both horizontal and vertical edges are being smoothed (compare the tall rock with the previous two images, and notice that all of it is now smoothed instead of only the top or sides), and the shipwreck looks far better than before. We’re starting to see the impact SSAA can have on framerate now, as we’re down to 88FPS. While that’s still high, it’s a 40% drop over the framerate with no AA.

3×3 SSAA

At 3×3 the image is most improved in the areas of the fence at a sharp angle to the camera, as well as the open gate. Whether this is worth the additional 50% performance drop over 2×2, and 70% over no AA, is up to you. Now on to the holy grail of anti-aliasing.

4×4 SSAA

While having the best image quality, the differences are getting very difficult to point out now. The fence is certainly improved, but little else has changed compared to 3×3. The most notable exception is the framerate, which is in unplayable territory. Of course, since the resolution of the frame in the off-screen buffer is now 16x higher than the screen resolution, that is to be expected. Remember that this game is based on the Source engine, so it is not that taxing on the system. On newer games, the framerate would likely drop to unplayable with the lower SSAA settings even. But it would look nice

That brings me to the end of this post on Supersampling Anti-aliasing. Check back soon for the next article on Multisampling.

Late June/Early July rumors started spreading throughout the various hardware forums that AMD has three new chips coming to the market; Hemlock, Cypress and Juniper, part of AMD’s “Evergreen” family.
Now that samples are hitting the reviewers and some even making it for sale, I will try my best to explain why I do not think its great enough to counter Nvidia’s GT300.

When AMD released their RV770 and farther down the line the RV790 it took the scene by storm, who could have thought a graphics chip could show so much efficiency for the price, targeted to compete
with the GTX260 Core 216 and the GTX275, the 4870 and 4890’s went beyond the call of duty, coming close to the Performance of the bad boy GTX280/285 in some titles, which probably caused a panic within
the Nvidia organization, causing Nvidia to quickly drop prices on their more expensive cards to try to compete with the juggernaut AMD designed.

Now to the hard stuff to write about. It is known in its infancy the HD58xx series doesn’t have full driver optimizations to show what it’s truly capable but I think it will not be as fast as GT300.
I’ve been spending the past few weeks looking over every HD5870 review I could find and the results are very great. Staying close or beating the GTX295 in some titles, and HD5870 crossfire
being the fastest multi-card configuration currently available by beating everything. But I do not think a 100% increase in performance over the older 4870/4890’s is good enough to compete with
the next single card offering Nvidia will show, the GT300 could be weeks, even weeks away from a GT300 release and that could allow them the time to take grasp and embarrass AMD’s offerings.

If GT300 is indeed grossly faster then the HD5870 in single card vs. single card, could AMD bring back the ideology they had with the RV770?

Only time will tell if AMD can hold this performance crown or have to recede to the drawing boards and make a chip to best GT300, thank you for taking the time to read this little rant.

Anyhow, back on topic – It arrived Thursday, in a shiny plain box – After taking it home, I finally got around to installing it over the weekend.

Overall, it would be a nice increase in performance over my 4850, and well – it’d be something to review!

In with the new! There is a Asus Xonar D2x, a P45-DS4P and a Intel E8400 powering the rest.  Plenty of ventilation, too.  As you can see

*Anakin skywalker voice* IT’S WORKING! (And yes, it’s 4.0ghz with 8GB of RAM. After much fiddling..)

So, after much fiddling it’s all good. I have it running!

Starting with the installation – fairly simple. Uninstall old card drivers, put new card in, reinstall drivers. No issues found.

As for performance, here’s a few numbers so far. (I was too lazy to take screenshots of some). Overall, it’s very solid. Much better than my 4850 – all games were played at maximum detail, anti-aliasing and AF enabled in-game. CCC settings were set to default, with advanced on Catalyst AI. From what I’ve noticed, AA is abit sharper – weather this is me moving up from 8.9 drivers, or the card itself is yet to be found, but overall I feel as if IQ has improved somewhat.

For the reviews below, the resolution used was 1920×1200, with all in-game settings set to the highest values possible. After all, what’s the point of doing any lower when you buy a faster card?

I also deliberately was too lazy and left some background programs running – Foobar2000, Firefox with 20 or so tabs, Steam (dur), and the usual array of background apps most run – MSN, so on.
Most reviewers run with none at all – leaving some open does put some extra weight on the CPU, however, and in reality quite a few people do leave crap open all the time – so why not.

I’d consider this overview more of a “What to expect on your PC.” (Even if the overclock is abit much – It’s easily achievable.)

Anyhow, on to the card itself!

The temperatures of the card were quite impressive – peaking at around 75 degrees, idling at 45. The idle temperature is thanks to ATi’s PowerPlay technology – down-clocking the card when it’s not under stress.
These were temps with the fan forced at 28% fan speed through Catalyst – so overall, if you’re a quiet freak, the card isn’t too bad at all. It’s fairly inaudible at 2 meters, and a slight hum when it’s near you.
Bear in mind current ambient temp is around 15 degrees.

The games tested aren’t too many – but these are the main ones I have installed at the moment, and the ones that I tested. FRAPS + in game FPS meters were used.

• Racedriver GRID – 86fps avg, 120max, 62min
• Team Fortress 2 – 97fps avg, 190max, 75min
• Call of Duty 5 – 83avg, 103max, 69min
• 3Dmark 2006 – 15910

As for image quality, it’s good. The key thing I notice at the resolution I play at over the 4850 is the lack of choppiness. The 4850 would stutter every here and again. The 4890? Pumps out pretty damn solidly – Overall minimum framerates are much higher for any game I’ve played so far.

In individual games, I did notice the smoke in screenshots of Racedriver GRID to be much smoother and “realistic”. Once again, drivers or otherwise, it’s an improvement. (You’ll be on 9.3 or 9.4 drivers with a 4890 anyway, so stress not!)
TF2 performance was much smoother overall – the minimum framerate was much higher than my 4850’s 35. (Attributed to the high resolution and AA/AF settings applied.)
3Dmark – eh, it’s all the same, boring synthetic benchmark.

Anyhow, here’s some screenies of TF2 and CoD. Not too sure where my GRID ones went. O_o

TF2 Screenshot 1
TF2 Screenshot 2
Call of duty Shot 1
Call of Duty Shot 2

Overall, I would definitely recommend this purchase to anyone in need of an upgrade from say, a 4670 or 8600GT/8800GT . It’s solid, it runs well. It’s still using beta drivers, which means there’s still performance to be had, and for around 400AUD, getting one in crossfire later on is definitely a performance option for the high-end market.

In retro, for your average joe – stick to the 4870. $100AUD cheaper. I would really only recommend this to those needing a good future-proof upgrade, or cutting edge performance. If you have a 4870, GTX260, or similar – don’t bother. I’d stick with current-gen cards if I were you, seeing as the next series for ATi is only just around the corner. If you really want a boost, grab another card and SLi/Crossfire!

Anyhow, hope you enjoyed the read.
Lardman out.