Back in 2008, when it looked like NVIDIA owned the GPU market, and AMD seemed lagging behind on the performance and efficiency game, the company sprung a surprise. The company's RV770 silicon, the first GPU to implement GDDR5 memory, trounced NVIDIA's big and inefficient GeForce GTX 200 series, and threw AMD back in the game. GDDR5 helped the company double the memory bandwidth, with lower pin- and memory-chip counts, letting the company and its partners build graphics cards with fewer components, and earn great margins, which the company invested in development of its even better HD 5000 series, that pushed NVIDIA with its comical GeForce GTX 480, to hit its lowest ever in market-share. Could AMD be looking at a similar turnaround this summer?

Since the introduction of its Graphics CoreNext architecture in 2012, AMD has been rather laxed in its product development cycle. The company has come out with a new high-end silicon every 18-24 months, and adopted a strategy of cascading re-branding. The introduction of each new high-end silicon would relegate the existing high-end silicon to the performance segment re-branded, and the existing performance-segment silicon to mid-range, re-branded. While the company could lay out its upcoming Radeon R9 series much in the same way, with the introduction of essentially just one new silicon, "Fiji," it could just prove enough for the company. Much like RV770, "Fiji" is about to bring something that could prove to be a very big feature to the consumer graphics market, stacked high-bandwidth memory (HBM).

Sapphire today announced its Radeon HD 4730 accelerator. The new GPU surfaced along with Radeon HD 4790. Together these GPUs fill the Radeon HD 4700 series, bringing relief to the demand Radeon HD 4770 created, but is apparently unable to meet due to technical problems with the 40 nm manufacturing node of foundry partner TSMC.

The Sapphire Radeon HD 4730 uses a PCB that is pin-compatible with Radeon HD 4870, as it is based on the same RV770 core with a different configuration. The company used a simple cooler with radially-projecting aluminum fins, which it has been using on inexpensive models of the Radeon HD 4850, and HD 4830. The memory is subjected to its air-flow, while the VRM area has a heatsink of its own. The GPU has 640 stream processors enabled, with a 128-bit wide GDDR5 memory interface, holding 512 MB of memory. Sapphire cut down on the number of memory chips used, thanks to the narrower memory bus, while maintaining the memory bandwidth of similarly clocked 256-bit GDDR3. The core is clocked at 750 MHz, while the memory is at 900 MHz (3.6 GT/s effective). The outputs provided on the card are DVI-D, D-Sub, and HDMI. The card takes its power input from two 6-pin power connectors. The card uses a 3+2 phase power circuit. The Sapphire HD 4730 is expected to be priced at 80€.

AMD is preparing yet another performance-mainstream Radeon HD 4000 series SKU. The Radeon HD 4790 finds lineage from the Radeon HD 4890, currently AMD's fastest GPU. The RV790 GPU will be given a new set of specifications and memory configuration, to yield an SKU that performs better than the Radeon HD 4770, and slightly better than HD 4850. It beats us as to why it is positioned in the HD 4700 series, and not say "Radeon HD 4860", but we are too late to comment on that.

Specifications-wise, the RV790 core runs at 600 MHz, slightly lower than the RV770 in Radeon HD 4850 (625 MHz), but its performance increment over it comes from the use of GDDR5 memory. The GPU makes use of a 256-bit wide GDDR5 memory interface. It handles 512 MB of memory clocked at 800 MHz (3200 MHz effective). Being based on the RV790, it is pin-compatible with any existing HD 4890 PCB. Price-wise, it is expected to sit between the HD 4850 and HD 4870.

AMD is dressing up a new ATI Radeon SKU to go to office soon. The Radeon HD 4730 beats our guesswork, and wits both in one shot. The new SKU is not based on a scaled-down 40 nm RV740 GPU the HD 4770 is based on, but rather on a new RV770 variant, dubbed the RV770CE. This GPU is physically identical to the RV770, and has part of its configuration identical to the RV770LE (Radeon HD 4830), while its memory interface is 128-bit GDDR5 instead of 256-bit GDDR3.

The only explanation we have in store is that perhaps AMD is looking to save on using lesser number of memory chips per board (while not crippling the memory interface), and that RV770 is in healthier stock than the RV740. Pictured below is the PowerColor Radeon HD 4730 PCS, one of the first cards based on the GPU. Other important specifications, such as stream processor count, and clock-speeds are yet to surface. Ideally it should cash-in on a price-segment below that of the Radeon HD 4770.

A relatively unknown company, Triplex, has designed a Radeon HD 4830 graphics accelerator that does not require the 6-pin PCI-Express power connector. This is especially interesting for two reasons: that it is RV770, and that it runs at reference clock speeds despite shedding its traditional power design. Spotted on XtremeSystems, this engineering sample PCB features a 2+1 phase power design, that draws all its power from the PCI-Express slot.

Also featured are 512 MB of 256-bit GDDR3 memory, DVI-D, HDMI and D-Sub outputs, and a seemingly two-slot cooler design that is yet to be pictured. The card lacks CrossFireX fingers. The GPU has 640 stream processors, DirectX 10.1 compliance, and a 256-bit memory interface. It has AMD reference clock speeds of 575/900 MHz (core/memory). For reference the third picture shows a Radeon HD 4670 accelerator of the same make, and PCB length.

AMD clinched the performance crown from NVIDIA, and retained for a significantly long amount of time with the dual-RV770 based Radeon HD 4870 X2 graphics accelerator. The company recently introduced the RV790 GPU, which went into making the Radeon HD 4890 graphics accelerator. Our reviews of this card, especially the AMD Radeon HD 4890 CrossFire review, brought forth some interesting findings with regards to how the accelerator works in tandem with another of its kind.

Holding significantly higher clock-speeds than the Radeon HD 4870, the accelerator managed to consume lesser amount of power in a pair than a single Radeon HD 4870 X2 accelerator. As a solution, it emerged faster than the NVIDIA GeForce GTX 295. The findings of several such reviews sparked of fresh speculations surrounding AMD planning a dual-RV790 accelerator with which it could potentially regain the performance crown. All such speculations were laid to rest by David Cummings, AMD's director of product marketing of discrete desktop graphics, under the graphics product group, in a recent interview with X-bit Labs (read here). Cummings claims that the company has no plans to create a "Radeon HD 4890 X2", at least not yet. The Radeon HD 4890 fills the gap between the HD 4870 and HD 4850 X2, while the HD 4870 X2 holds post at $399. A dual-RV790 card would not be feasible keeping the global economic climate in mind. The HD 4890 accelerators though, are capable of running in tandem with up to four cards of its kind, provided the system supports such a configuration.

An unexpected visitor to Guru3D.com's offices was a pre-release sample of a yet to be released 40 nm RV740-based Radeon HD 4750 graphics accelerator. Not bound by any NDAs with AMD, the website went ahead with a little (p)review of the card. The HD 4750 is the RV740XT model, and features GDDR5 memory. The name contradicts an earlier report suggesting HD 4770 to be the shelf-name for the RV740XT, and HD 4750 for the GDDR3-based RV740Pro. It features 640 stream processors, core clock speeds between 650~700 MHz and GDDR5 memory clocked at 800 MHz (3.20 GHz effective), across a 128-bit wide memory bus. The RV740XT comes with a rated shader compute power of 900 GFLOPs, as against 740 GFLOPs the RV770LE-based HD 4830 is rated at, while having similar specifications. It comes with 32 TMUs and 16 ROPs.

The findings of the preview show it to be somewhere between the performance levels of the Radeon HD 4830 and HD 4850. Interestingly, Guru3D omitted GeForce 8800 GT/9800 GT from the comparison, though GeForce 9600 GT was left to face the onslaught from stronger ATI GPUs. The Radeon HD 4750 is expected to be priced below the $100 mark and is expected to outperform most competitive accelerators in its price-range. To read the review, head over to Guru3D here.

It is clear now, that AMD's next performance GPU, the RV790 will be built on the 55 nm manufacturing process. To build on this information, sources tell German website Hardware-Infos that one of the factors that will help the new GPU perform better and run at higher clock-speeds than its predecessor, the RV770 does, is the newer improved 55 nm manufacturing process it will be built on. Going by the foundry-partner codename "55GT", the RV790 may get a little help from its superior silicon fabrication, which invariably makes it more expensive to manufacture.

One of the reasons behind why AMD is starting its 40 nm GPU lineup with a mainstream GPU such as RV740, is that the 40 nm process needs further development by foundry companies. It hasn't developed to the level that safely permits manufacturing high-end GPUs with stellar transistor-counts. For the same reason, NVIDIA's 40 nm conquest will be flagged off by the entry-level GT218 GPU. High-density circuits built on the current 40 nm process are known to be very prone to electrical leakage.

Conceived to be a successor to the RV770 graphics processor, the RV790 emerged out of various media-leaks as the next big thing for ATI, in its never-ending pursuit for market leadership. Information on this mysterious new GPU has been trickling in at leisure, and it has been shaping up to be a current-generation GPU built using newer manufacturing technologies that could make it cheaper to produce. What we do know about RV790XT based engineering samples so far, is that they carry faster memory chips that are rated to operate a higher frequencies, 1 GHz to be precise, but specified by AMD to run at 975 MHz.

Fresh details sourced by Hardware-Infos from internal sources point out to the possibility of AMD stepping up the core clock speeds by as much as 100 MHz, taking it to 850 MHz. The ramp in clock speeds however, come at the expense of increasing core voltage to 1.3 Volts. There have been several reports speculating the core voltage so far, on of which points at 1.23V for 750 MHz. Another grey-area for now is the RV790's shader domain, with a certain level of ambiguity surrounding stream processor count. All of this leads us to believe that AMD hasn't finalized the specifications of RV790 and could be racking its brains on this one. AMD is expected to make critical announcements on its new GPUs in March and might choose the CeBIT event as the ideal platform.

Having taped out late last year, samples based on the new RV790 graphics processor have been doing rounds in the industry. A few more details have surfaced about it. Earlier noted to have identical clock speeds to that of the RV770XT (Radeon HD 4870), the RV790 samples are now known to have higher memory clock speeds. While the Radeon HD 4870 has its memory frequency at 900 MHz GDDR5 (effective 3.60 GHz), the samples carry memory clocked at 975 MHz (effective 3.90 GHz). Interestingly the memory chips on the sample, labeled IDGV1G-05A1F1C-40X, made by Qimonda, are specified to run at 1.00 GHz, reaching the 4 GHz effective memory speed mark. The samples feature 1 GB of memory. The RV790 is AMD's new current-generation graphics processor built on the newer 40nm silicon fabrication process. The new process is expected to reduce the GPU's power consumption and thermal footprint. The RV790 is conceived to be an immediate successor to the RV770 GPU.

Some of the first batches, rather iterations, of RV790 and RV740 engineering samples have begun surfacing. Sources reveal bits and pieces of the new GPUs' specifications to Hardware-Infos. Being some of the first samples, these are merely iterations en route the development of the final product, though trend has it that preliminary information about AMD GPUs have a tendency of turning out true. We will exempt the RV770's final stream processor count from these.

The RV740, a mainstream GPU from AMD, is on course of becoming the first GPU in production, to be built on the 40nm manufacturing node. It carries 640 stream processors and a core clock speed of 700 MHz. It features a 128-bit wide GDDR5 memory bus, churning out bandwidth that rivals equally clocked GDDR3 with double the bus width. The memory bus will be clocked at speeds between 800 and 900 MHz (3.2 GT/s and 3.6 GT/s). Products will carry 512 to 1024 MB of memory. The GPU houses 32 texture memory units (TMUs) and 8 raster operations pipelines (ROPs).

The RV740 graphics processor by AMD could well be the first GPU in production, to be built on the newer 40nm manufacturing process, giving its makers something to brag about. The RV740 design and specifications are largely derived from the RV770LE (covered here), while bringing the GDDR5 memory standard to the mainstream.

Chinese industrial journals tell DigiTimes that the company has completed taping-out the RV740, and that the GPU is awaiting mass-production by AMD's foundry partner, Taiwan Semiconductor Manufacturing Company (TSMC). Mass production of products based on this GPU could commence by the end of Q1 2009.

Sources tell Expreview that AMD is working on a new graphics processor (apart from the mysterious RV790), as an immediate reaction to the 55nm high-end GPUs NVIDIA is set to release. NVIDIA on its part is working on GeForce GTX 260 (55nm), GeForce GTX 285 (successor to GeForce GTX 280) and the dual-GPU monstrosity, GeForce GTX 295. To rival this, according to the source, AMD is working on the RV775 and derivatives based on its yields' performance.

These GPUs are slated for January thru March 2009. It has some very interesting specifications. To begin with, 840 stream processors. While stream processors on a current generation RV770 cannot be added/subtracted in quantities of 40 stream processors, perhaps, the increment of 40 SPs is derived from an extra ALU cluster the RV770 comes with by default (which helps in die harvesting, where the spare ALU cluster makes up for any damaged cluster, thereby improving yields). This would require near-perfect GPUs in the making (if re-engineering of the RV770 on a large scale is to be avoided). Another interesting point is regarding the smaller die. At this point we are clueless as to how that is possible without re-engineering the RV770 at a large scale. Treat yourselves to the specifications meanwhile:

In an exhibition of pure technological flexibility by AMD, the RV770 graphics processor is going to be used along with GDDR4 memory, in PowerColor's newest Radeon HD 4850 PCS+ GDDR4 graphics card. The inclusion of GDDR4 memory is to boost the memory clock speeds and to provide room for a little more memory overclocking.

The PCS+ GDDR4 features a GPU cooler designed by ZEROtherm, which is advertised to keep the GPU up to 10 °C cooler than what the reference cooler would. It features heatsinks for the memory and the VRM area. It features 512 MB of GDDR4 memory clocked at 2.20 GHz, with the core clocked at 675 MHz. The card supports CrossFireX, and sports dual-DVI connectors. There is no word on its availability pricing as of now.

The RV770 graphics processor changed AMD's fortunes in the graphics processor industry and put it back in the race for supremacy over the larger rival NVIDIA. The introduction of RV770-based products had a huge impact on the mid-range and high-end graphics card markets, which took NVIDIA by surprise. Jen-Hsun Huang, the CEO of NVIDIA has been quoted saying that they had underestimated their competitor's latest GPU, referring to RV770. While the Radeon HD 4870 graphics accelerator provided direct competition to the 192 shader-laden GeForce GTX 260, the subsequent introduction of a 216 shader variant saw it lose ground, leaving doubling of memory size to carve out the newer SKU, the Radeon HD 4870 1GB. Performance benchmarks of this card from all over the media have been mixed, but show that AMD isn't giving up this chance for gaining technological supremacy.

In Q4 2008, NVIDIA is expected to release three new graphics cards: GeForce GTX 270 and GeForce GTX 290. The cards are based on NVIDIA's G200 refresh, the G200b, which incorporates a new manufacturing technology to facilitate higher clock-speeds, stepping up performance. This looks to threaten the market position of AMD's RV770, since it's already established that G200 when overclocked to its stable limits, achieves more performance than RV770 pushed to its limits. This leaves AMD with some worries, since it cannot afford to lose the wonderful market-position its cash-cow, the RV770 is currently in, to an NVIDIA product that outperforms it by a significant margin, in its price-domain. The company's next generation graphics processor would be the RV870, which still has some time left before it could be rushed in, since its introduction is tied to the constraints of foundry companies such as TSMC, and the availability of the required manufacturing process (40nm silicon lithography) by them. While TSMC takes its time working on that, there's a fair bit of time left, for RV770 to face NVIDIA, which given the circumstances, looks a lost battle. Is AMD going to do something about its flagship GPU? Will AMD make an effort to maintain its competitiveness before the next round of the battle for technological supremacy begins? The answer is tilting in favour of yes.

ASUS has decided to run around and release a new MATRIX card, the EAH4850 MATRIX/HTDI/512M. The particular card is powered by a 55nm RV770 GPU with 800 Stream Processors, has 512MB of GDDR3 memory and the proprietary dual-slot Hybrid Cooler. While the EAH4850 MATRIX might sound like it's going to be an overclocked card thanks to its bigger cooler, it's not. It carries the stock 625MHz core and 1986MHz DDR memory clock. Instead the card comes with an integrated hardware and software set of controls that makes adjustments to the GPU and memory voltages depending on the card load and mode. ASUS software is also monitoring the card's GPU, memory, Power IC, and ambient temperatures. The MATRIX Radeon HD 4850 is expected to hit the stores in November.

After last week's brief introduction of the first dual RV770 Radeon HD4850 spotted in the wild, we've now managed to find some benchmark scores for you thanks to the guys over at VR-Zone. Apparently they have the Sapphire ATI Radeon HD 4850 X2 card in their lab and have decided to pair it with an Core i7/X58 platform to deliver the maximum out of it. This card is clocked at 625MHz for the core, 1000MHz for the 1GB GDDR3 memory and has 1600 stream processors (800 for each of the two RV770 GPUs). More pictures of the card and results from different widescreen resolutions can be seen here.

The ATI Radeon HD 4830, the newest foot-soldier from the red-camp, is AMD's answer to the GeForce 9800 GT, a gap left by the company, that may have been eating into its profits for months now. As usual, TechPowerUp received its review samples from TUL (PowerColor), AMD's long-standing partner, and AMD itself. We reviewed both of them, as soon as the product became official today. During the course of reviewing them, with the card from AMD in particular, our reviewer, W1zzard noticed an anomaly: the sample from AMD was showing an abnormal stream processor count of 560.

W1zzard also authors the GPU-Z diagnostic utility, and it is his routine chore to program the utility to detect a new GPU. The newest build of GPU-Z detected the card from AMD to have as many as 80 stream processors disabled from the original specifications for the Radeon HD 4830. In his article, W1zzard attempts to explain this anomaly.

Earlier this year, the unthinkable happened when ATI brought in stiff competition with NVIDIA in almost all segments to snatch the performance crown. All NVIDIA could do as a retaliation was to hurry up with a transition to the 55nm silicon fabrication technology and price cuts among all its segments. The latter is of more importance since price cuts for NVIDIA's products have been huge, in many cases, as much as 40%. While Radeon HD 4850 in its 512 MB and 1 GB avatars took on GeForce 9800 GTX/GTX+, the GeForce 9800 GT slipped in from nowhere. Technically at launch the GeForce 9800 GT isn't much more than 8800 GT except a few features that only nForce chipset users could benefit from, among other changes. Indications are that this product isn't faring bad in the market and to retaliate ATI is devising a new graphics core, the RV770LE.

Simply put, the RV770LE is a toned down variant of the RV770 core that will position below the RV770 Pro (Radeon HD 4850). It features lower clock speeds than the RV770 Pro and more importantly, a 192 bit memory bus and 480 Stream Processors. This also affects the TMU and ROP counts with the new core having 12 ROPs and 24 TMUs. Memory configurations would be affected as well. While some sources indicate that it would be branded as Radeon HD 4830, others say it's likely to be branded as the Radeon HD 4750 and is slated for release in October.

Within this month, AMD will release a FireStream GPGPU card based on the successful RV770 graphics processor. This card provides computing parallelism derived from 800 stream processors, that churn out a peak collective computational power of 1000 GFlops. That's a 100% improvement over the FireStream 9170 in terms of computational power and a 60% improvement in terms of GFlops per Watt. That said, the power draw of this card has gone up to 150W compared to its 100W predecessor. This card is the fastest in the industry in terms of double-precision floating point performance. It packs 1 GB of GDDR3 memory on a 256-bit wide memory bus.

FireStream series of cards to catered towards delivering blistering performance of AMD's GPUs to business users in high performance computing situations such as financial analysis, seismic processing, simulations, rendering, climate research, protein folding etc. Apparently the card retains the display logic with a single DVI connector though normally GPGPU devices lack it. This card will be out within this month and will carry a price-tag of US $999.

Info-Tek GECUBE, the renowned Taiwanese graphics card maker, released the HD4850 Overlocking series with a Zalman fan onboard in July. This proved a great success on the market thanks to its spectacular gaming and video playback performance. To capitalize on its popularity among enthusiasts and consumers, The GECUBE team has now developed a more advanced offering in the form of the HD4850 1.0GB Overclocking series. The new HD4850 1.0GB Overclocking series includes an exclusive ceramic fan that delivers not only an ultra-quiet working environment, but also extreme performance. The new card also comes with the RV770PRO 2nd generation 55nm chipset, the PCI-Express 2.0 interface, full support for DirectX10.1/SM4.1 and ATI CrossFireX technologies but also built-in HDCP, HDMI, Blu-Ray and HDVD decoder support.

NVIDIA suffered its first red-quarter in five years. There are several contributors to this, namely an up to US $200M write-off to cover expenses in recalling and restoring faulty mobile graphics processors.

Another factor has been a replenished product lineup from competitor AMD/ATI that is taking on NVIDIA products at mid thru high and enthusiast segments of the market, in essence ATI now has products to counter NVIDIA at every possible segment, with more dressing up to go to office.

Seeking Alpha spoke with CEO Jen-Hsun Huang, he was quoted saying:

We underestimated the price performance of our competitor's most recent GPU, which led us to mis-position our fall lineup. The first step of our response was to reset our price to reflect competitive realities. Our action put us again in a strong competitive position but we took hard hits with respect to our overall GPU ASPs and ultimately to our gross margins. The price action was particularly difficult since we are just ramping 55-nanometer and the weak market resulted in taking longer than expected to work through our 65-nanometer inventory.

Huang says that with their transit to the 55nm silicon fabrication process, they hope to do better.

PCPop has access to an upcoming mobile (notebook) graphics processor (mGPU) from ATI, codenamed the M98. This graphics board makes use of the RV770 core and will be accompanied by 512 MB of GDDR3 memory, up to 1 GB in some models. It could be branded under the Radeon HD 4000 series. It can be dubbed the fastest till date if it's pitted against NVIDIA's current high performance mGPU, the GeForce 9800M GTX which makes use of the G92 GPU with a similar configuration to that of the desktop 8800 GT / 9800 GT cards (112 shader units). It then becomes comparable to GeForce 8800 GT versus Radeon HD4850 where the latter obtains 15% to 20% performance gains.

Pictures show the M98 Axiom Package featuring the the RV770 GPU next to four GDDR3 memory chips made by Qimonda. The GPU is marked "M98 XT", it is an engineering sample.

The success of the RV770 graphics processor (GPU) meant that partners of ATI could experiment with the design and come up with several products based on the GPU. The likes of Sapphire, HIS, PowerCOlor, Palit, HIS, etc., have already come up with numerrous non-reference designs, though all those designs had a common ATI designed PCB, with a primary PCB design with common layout of the various components, where the memory configurations and coolers used made all the difference. Sapphire has come up with a new non-reference PCB design for the Radeon HD 4850 that places components in a different way, perhaps it aides manufacturing and keeps the memory chips at a distance from the VRM circuitry enough to keep it cool enough to be passively cooled by a central cooler's air-flow? Whatever be the design methodology, Expreview pictured an upcoming Sapphire product based on this new PCB.

PowerColor is preparing an overclocked Radeon HD4870, the PCS (Professional Cooling System) model. This model uses 1 GB of GDDR5 memory and uses overclocked parameters, such as 800 MHz core. This card comes with a cooler made by Zerotherm comprising of heatpipes conducting to aluminum fins on either sides of a fan. According to TG Daily, this card could be based on the SuperRV770 concept, where high-performing parts are used along with 1 GB of GDDR5 memory. According to PowerColor, this cooler could keep the temperatures down by up to 20° C versus reference ATI cooler even at the overclocked 800 MHz core setting.