Introducing AMD’s new HD 6870 and HD 6850 vs. GTX 460
AMD Graphics and Nvidia are locked in a perpetual battle to one up each other in what can only be described as a “graphics war”. Nvidia had issues with introducing their Fermi DX11 architecture and video cards and AMD beat them to the market by over six months with the first DX11 video cards about this time last year. In April of this year, Nvidia launched their GTX 470 and GTX 480 which were criticized for being hot-running, power-hungry and loud although they offered somewhat higher performance than AMD’s HD 5870 and HD 5850. However, a few months later, Nvidia’s midrange GTX 460 turned out to be a very successful reworking of GF100 Fermi into GF104 that scaled well, ran cool, had good thermal characteristics, overclocked well and no doubt ate into AMD’s DX11 90% marketshare.
To combat GTX 460, AMD is releasing HD 6000 series with HD 6870 and HD 6850 being debuted this week. This is not AMD’s high end which is to come later on this year, but rather their upper-midrange which is renamed from HD 58×0 series and it is designed to take on and surpass Nvidia’s most successful Fermi card, their “sniper”, the GTX 460 in all of its variants. We also say goodbye to the ATi brand name and logo; the new cards are branded AMD Radeon.
ABT was represented by this editor at AMD’s Press Day at the famous LA Exchange, just a week ago, Thursday and saw AMD’s vision unfold further for us. Since we are going to focus on the HD 6870 and the HD 6850’s performance in 22 games, we will only give you the barest outline of their 5 hour presentation. We do see that the reason that they chose downtown Los Angeles is symbolic of their increasing commitment to the movie industry and they have partnered up with several Hollywood movie studios to increase productivity by using AMD hardware and know how. They also used the presentation to introduce their support for 3D in PC gaming and 3D for video playback.
AMD’s Press Event was called “Believe Your Eyes” and they laid out their vision for the world’s press. AMD feels that Fusion is uniquely suited to conquer the world and they stress the “firsts” they have accomplished, including being first to bring DX11 GPUs to market very quickly and successfully. They are quite proud of their marketshare and do not intend to allow Nvidia to make inroads – especially with the GTX 460. AMD points out the advantages of their Eyefinity which now allows more displays to be driven off of a single card rather than Nvidia’s competing 2D Surround solutions which require two similar video cards running in SLI to power it.
A rose by any other name …
Today, AMD Graphics is proud to introduce an improved version of their successful HD 58×0 series with similar performance, but now as the new HD 68×0 series and on the same 40 nm process as the 5000 series. The advantage is higher performance at lower prices. AMD appears intent to go after Nvidia’s GTX 460 by offering more performance at a similar price. This is a strategy that has worked for AMD in the past and they call it “aiming for the sweet spot”. The HD 68×0 series will also support a more flexible form of Eyefinity and it is also going to support 3D PC gaming and 3D video playback. What may be confusing to many is that HD 6870 and HD 6850 are slower than HD 5870 and HD 5850 respectively and yet are replacing them. AMD’s goal is to give more gamers the ability to experience HD 58×0-type performance in a less expensive, smaller and even less power-hungry video card.
The codename for AMD’s new HD 58×0 cards is “Barts”. AMD has not forgotten the high end gamer as they will be releasing “Cayman” – HD 6900 series – which will have a further set of improvements and refinements over the Barts HD 68×0 video cards that we shall be evaluating in this review. And HD 5700 series will remain unchanged for now. In many ways, a picture is worth 1,000 words and here is AMD’s continuation of their video card strategy:
We see changes only to the upper midrange and at the high end. AMD is continuing the HD 5700 series for now unchanged as they evidently feel unchallenged by Nvidia’s new offering, the GTS 450 which we reviewed here against HD 5750. We see the HD 5800 series diverge into 3 streams – the “Antilles”, an “X2” video card at the highest end as a successor to the current dual-GPU HD 5970; the “Cayman”, which will use AMD’s fastest single GPU to succeed and surpass the Cypress HD 5870 and we would expect that it would take on Nvidia’s GTX 480 video card directly.
Here is the latest round of our testing HD 5870 versus GTX 480 in our reviews here (overclocked versus super-overclocked). You can follow the links backward through several other of our GTX 480 vs. HD 5870 testing, back to our first testing of these cards in April. We can see that the GTX 480 is consistently faster than even the overclocked HD 5870 PCS+ and there is some need for AMD to address this with a faster and also more efficient GPU. There will also be a Cayman video card to likely beat Nvidia’s second fastest video card, the GTX 470.
With this launch, AMD is directly targeting Nvidia’s GTX 460. The HD 6870 (SEP $239.00) will take on GTX 460-1GB and the HD 6850 (SEP $179.00) will take on GTX 460-768MB video cards. In a preemptive move, Nvidia lowered the prices of their GTX 460-1GB to $199 and their GTX 460-768MB to $169. Even the GTX 470 was reduced to $259 so we can see that AMD’s launch has got a strong reaction from Nvidia.
What’s new in HD 68×0?
Since “seeing is believing” is AMD’s theme for this launch and it is all about the 3 “eyes”, we shall briefly cover them:
Under Eyedefinition, we see a further subdivision with more efficient tesselation; there is mention of a tweaked engine, offering up to 2x the tessellation performance of the HD 58×0 GPUs – an area where AMD was perceived weak in comparison to Nvidia’s Fermi GPUs in heavily tessellated benchmarks and games. We also see mention of enhanced architecture for efficiently using GPU compute and for improvement and performance in games. We also note improvements in Anisotropic Filtering (AF) and a new Anti-Aliasing mode – morphological AA.
The HD 5870 is pictured (below) in the middle of the two HD 68×0 cards. There are 2x DVI ports and one HDMI port plus two mini-DisplayPorts which are DP version 1.2. This is important because of the new Eyefinity features that now allow for daisy chaining of displays and for using a new hub, much like using a USB hub, to output to multiple displays
Eyespeed refers to GPU compute and to AMD’s “open initiative” approach to (everything and especially) to OpenCL, in contrast to Nvidia’s use of their own proprietary GPU language, CUDA. We see AMD partnering with Cyberlink, Arcsoft, Viewdle, Adobe, Microsoft and more companies (some of which are also Nvidia’s partners) to bring you, the end consumer, quality video processing and playback; and of course, UVD 3 accelerated decoding for 3D BluRay playback.
We are especially going to be evaluating performance and AMD’s claim of 35% better performance per mm over HD 58×0. That means that the HD 6870 should be about equal in performance to HD 5850 overall. Not really too much has changed from Cypress but we understand that Barts has up to 2x the performance of the tessellator in the HD 58×0 GPU. And of course we shall watch for this in our performance testing. Here is the Barts GPU from AMD’s own presentation slide.
To see what it brings new, we note that the UVD engine has been updated; HDMI 1.4a is available for 3D Blu-ray and we see an improved Tesselator Engine. Although there are fewer SIMD Engines than the Radeon HD 5800 series, AMD now uses a second Ultra Threaded Dispatch Processor and an improved engine logic. We have noted in previous reviews, that Nvidia’s Fermi GPUs are faster in heavily tessellated scenes than competitive AMD Cypress GPUs. Well, now AMD claims a solid tessellation improvement over Cypress and HD 58×0 series and calls their method “tessellating the right way”.
Tessellating the Right Way
- Focus on most efficient tessellation usage models
- Ideally want ~16 pixels per polygon
- Best balance of image quality and performance
- Adaptive tessellation helps achieve this
- Use high levels only for objects close to viewer, on silhouette edges, or in areas with fine detail
- Use low levels for distant or simple objects to improve performance and avoid geometry aliasing problems
Of course, Nvidia will take issue with the above and argue that high levels of AA require per pixel application of tessellation beyond what AMD’s solution is tuned for. We shall see what happens in the long run with game developers. Nvidia’s argument will probably be that AMD is focusing on triangle size and “ideals” to cover that their tessellator is a bottleneck for the rest of their GPU, compared to Fermi. That is why Nvidia chose to do a fully parallelized implementation in Fermi. In contrast, AMD has decided to stick with improving their original design and counting on the assumption that games will not be so heavily tessellated so as to make any practical difference to the gaming experience for the next two or three years – and certainly not before they move to their own more completely redesigned Northern Islands GPU architecture on the 28 nm process which is due next year.
Morphological Adaptive AA
AMD’s new morphological anti-aliasing technique works as a post process effect. In other words, the GPU finishes rendering each frame as usual – but before presenting it to the display, it runs it through another shader pass to perform the filtering. This differs from traditional multi-sample and super-sample AA techniques where the filtering occurs during the rendering of each frame. In fact, this technique can eliminate aliasing for still images, though it’s intended to work better when in motion.
The filter works by first detecting high contrast edges with various pixel-sized patterns that are normally associated with aliasing, and assumes they should actually be straight lines that are not aligned to pixel edges. It then estimates the length and angle of the ideal line for each edge, and determines the proportional coverage by the lighter and darker color for each pixel along the edge. Finally it uses this coverage information to blend the colors for each pixel. All of this is actually being accomplished by the Catalyst drivers through a DirectCompute shader while the Local Data Share is used to keep adjacent pixels in memory for a low overall overhead. It will be interesting to see if AMD chooses to extend this morphological adaptive AA to the 5000 series as there is no reason it cannot be done, except perhaps to differentiate HD 6000 series from the current one.
AMD’s diagrams (below) should help to illustrate how this is accomplished.
Since the edge detection step requires frequent sampling and re-sampling of adjacent pixel colors, it offers a lot of opportunities for data re-use by using the LDS (Local Data Share) hardware to avoid redundant data fetches and to significantly improve performance. AMD sent us a driver very late in our testing and we are unable to evaluate it as yet. We simply cannot comment on what we have not yet evaluated.
Anisotropic Filtering (AF)
With the HD 5000 series, AMD brought genuine angle-independent filtering to gaming by putting an end to angle-dependent deficiencies. However, our own Senior Editor BFG10K pointed out the flaws in AMD’s Anisotropic filtering with the transitions, here, here and here. AMD listened to us and the enthusiast community and they have improved the transitions between filter levels. Well, examining these improvements are beyond the scope of this performance evaluation, but rest assured that BFG10K will again provide the definitive answers in a future review right here at ABT.
Like Cypress, all Barts GPUs are produced with the 40 nm process. AMD’s new reference Radeon HD 6870 has 1120 Stream Processors with its core operating at 900 MHz with 1GB of GDDR5 at 4.2 GHz on a 256-bit bus. There are 32 ROPs and 56 Texture units on a smaller die than that of the Radeon HD 5850 GPU and there are also fewer transistors. Yet, even with this reduction in die size and transistor count, the Radeon HD 6870 should perform similarly to the Radeon HD 5850 due to architectural refinements and fine-tuning. The HD 6870’s maximum load board power is 151 watts and its idle is 19 watts which is an improvement in idle wattage compared to the HD 5850, and also matching the HD 5850’s load wattage. The HD 6870 requires two 6-pin PCIe power connectors (bottom card). The reference HD 6870 is 9.5 inches long.
The new AMD Radeon HD 6850 is clocked at 775 MHz with 1GB of GDDR5 at 4 GHz on a 256-bit bus and features 960 Stream Processors. There are also 32 ROPs. Maximum load board power is 127 watts and the idle is at 19 watts, just like the 6870. The reference HD 6850 is 9 inches long and only requires one single 6-pin PCIe power connector (top card, above).
To properly bring you this review, we are using our reference Diamond HD 5870 (850/1200 MHz) as our standard and we put all of our Radeon cards through their paces this week with the very latest performance drivers – Catalyst 10-10. AMD should be quite proud of this driver set as it brings sold performance increases over Catalyst 10-9 which we analyzed here recently. We are comparing the HD 5870, HD 6870 and HD 6850 to the EVGA (super-overclocked) and Galaxy GTX 460-1GB (at stock clocks), plus the (stock-clocked) Galaxy GTX 460-768MB version. Remember that we are comparing a HD 5870 with a SEP of $400 with video cards about half its price. We are using Nvidia’s very latest (“detonator”) WHQL GeForce 260.89 drivers that also give solid performance increases over the previous version, so both vendors are giving us their very best for this review.
Specifically, you will see us pit our HD 5870 against the new HD 6870 and HD 6850 and also against the reference GTX 460-1GB, the EVGA (highly) overclocked version, as well as the GTX 460-768 MB version in 22 modern games and 2 synthetic benchmarks using 1680×1050, 1920×1200 and 2560×1600 resolutions. Since we are using the fastest of the fast upper-midrange video cards, it makes sense to test at the highest resolutions that they can handle and with the most demanding settings. We shall also overclock our HD 68×0 cards to give you an idea of what AMD’s board partners will soon be bringing you. In the above picture, the GTX 480 (top) and the non-reference design PowerColor HD 5870 PCS+ (left), are included for a visual size comparison.
Is AMD’s HD 68×0 worth the premium over its rival, Nvidia’s overclocked GTX 460-1GB?
Since Nvidia has signaled their willingness to engage in a price war by dropping their GTX 460’s pricing to a bit below AMD’s new cards, we naturally want to know if the new AMD HD 68×0 cards are worth their new price premium. They are certainly priced very competitively and we will show you where it comes down to performance and features. Surprisingly, we received our Galaxy GTX 460-768MB video card on Thursday morning – the same day that our review was due to be published! Well, we delayed our review so that you could more easily make up your mind by having this card to compare directly to the HD 6850 just as the HD 6870 will compare to the stock and super-overclocked GTX 460.
We had quite a bit of a mix going on in our test bed. We tested our GTX 460s at their reference clocks and also with the vendor set factory overclocks – in EVGA’s case, it is pretty extreme – from the GTX 460’s stock clocks of 675/1800 MHz to 850/2000 MHz! Our HD 6850 overclocked like a champion – from 775/1000 MHz to 900/1100 Mhz. Unfortunately, the overclock on our HD 6870 can best be described as “mild” – from 900/1050 MHz to 950/1075 MHz. Although the HD 6870’s overclock is perceived as disappointing, it is a blessing in disguise as it will give you an idea of scaling while still keeping your own overclocking expectations realistic. We only used CCC to set our Radeon overclocks and we did not increase the core voltage nor change the fan profile.
Can you CrossFire-X your HD 6870 with the HD 6850?
Normally, we would not expect a person buying a HD 6850 to consider CrossFiring it with a HD 6870. Quite often a gamer will buy one card and perhaps later on, add a second more powerful one; something you cannot do with SLI. Using CrossFire-X, each dissimilar card can keep its own unique clocks and there is some load-balancing as the weaker card adds to the overall experience. Well, we wanted to see if it would work; something only a real hardware geek would do. Read on!
Because of severe time constraints on this article, HD 68×0 CrossFire will be examined in depth in a further article as well as 3-panel Eyefinity (which can be driven off of a single Radeon) verses Nvidia’s competing 2D Surround which requires SLI to make it work. We used our Intel Core i7-920 at 3.8 GHz for this evaluation so there was no chance of any CPU bottlenecking. Read on to see our test bed and the games we used.