GPU Wars: Nvidia vs. AMD

Nvidia preparing GTX 680M for Computex

noreply@blogger.com (Jun) — Sat, 19 May 2012 15:10:00 +0000

According to a post over at Videocardz.com, it appears that Nvidia is preparing a new Kepler-based mobile chip. It will be based on the GK104 GPU and will be known as the GTX 680M. According to early performance figures, the GTX 680M will not be able to dethrone AMD's current mobile flagship, the HD 7970M.

Although precise clocks are still missing, the early specs for the GTX 680M suggest that we are looking at a GK104-based chip that will have either 744 or 768 CUDA cores and feature up to 4GB of GDDR5 memory paired up with a 256-bit memory interface. The rumoured TDP of this chip is up to 100W.

The leaked benchmarks shows that the card could end up around 37 percent faster than current Nvidia flagship, the GTX 670M, which is based on Fermi architecture. It scores 4905 points in 3DMark 11 performance preset when paired up with Core i7-3720QM Ivy Bridge CPU.

Still slower than HD 7970M

According to a post over at Videocardz.com, it appears that Nvidia is preparing a new Kepler-based mobile chip. It will be based on the GK104 GPU and will be known as the GTX 680M. According to early performance figures, the GTX 680M will not be able to dethrone AMD's current mobile flagship, the HD 7970M.

Although precise clocks are still missing, the early specs for the GTX 680M suggest that we are looking at a GK104-based chip that will have either 744 or 768 CUDA cores and feature up to 4GB of GDDR5 memory paired up with a 256-bit memory interface. The rumoured TDP of this chip is up to 100W.

The leaked benchmarks shows that the card could end up around 37 percent faster than current Nvidia flagship, the GTX 670M, which is based on Fermi architecture. It scores 4905 points in 3DMark 11 performance preset when paired up with Core i7-3720QM Ivy Bridge CPU.

As noted, the 4905 score, if proven to be accurate, is not enough to dethrone AMD's current flagship, the HD 7970M. According to previous benchmarks of the HD 7970M, it scores 6125 points in the same 3DMark 11 when paired up with Intel's Core i7-2820.

Of course, we are still looking at early reports and rumours, but it is nice to know that Nvidia is at least considering such a chip and we are surelly looking forward to see some more details at Computex event when it kicks of on June 5th.

Source: fudzilla.com

Nvidia Responds to GeForce 600 Series V-Sync Stuttering Issue

noreply@blogger.com (Jun) — Sat, 19 May 2012 14:53:00 +0000

Tom's Hardware has received several requests to look into a v-sync issue that a number of owners are seeing with Nvidia's new GeForce GTX 600-series graphics cards. A number of readers have pointed us in the direction of a very long forum thread where multiple customers express concern over stuttering problems with their Kepler-based graphics cards.

Our editorial team played through a number of games with GeForce GTX 600-series board and was unsuccessful in reproducing the issue. As such, we're fairly confident that the issue does not apply to every configuration based on Nvidia's newest cards; some folks are experiencing it, while others aren't.

Nvidia provides an official response to the reported v-sync stuttering issue with its GeForce GTX 600 Series

Tom's Hardware has received several requests to look into a v-sync issue that a number of owners are seeing with Nvidia's new GeForce GTX 600-series graphics cards. A number of readers have pointed us in the direction of a very long forum thread where multiple customers express concern over stuttering problems with their Kepler-based graphics cards.

Our editorial team played through a number of games with GeForce GTX 600-series board and was unsuccessful in reproducing the issue. As such, we're fairly confident that the issue does not apply to every configuration based on Nvidia's newest cards; some folks are experiencing it, while others aren't.

Eager for an explanation, Tom's Hardware reached out to Nvidia for its response, understanding that owners of $400, $500, and $1000 graphics cards want to know why their boards are misbehaving. The company sent over an official response to the GeForce GTX 600-series stuttering issue, which reads:

We have received reports of an intermittent v-sync stuttering issue from some of our customers. We’ve root caused the issue to a driver bug and identified a fix for it. The fix requires extensive testing though, and will not be available until our next major driver release targeted for June (post-R300). For users experiencing this issue, the interim workaround is to disable v-sync via the Nvidia Control Panel or in-game graphics settings menu.

Although that's not necessarily the most reassuring message, we're at least glad to see that Nvidia has identified the issue and is working on a fix. We'll update you if we hear any more on this, so stay tuned!

Source: tomshardware.com

Kepler GK110 detailed by Nvidia

noreply@blogger.com (Jun) — Sat, 19 May 2012 14:35:00 +0000

Nvidia has officially published the Kepler GK110 whitepaper detailing all the specs that that had us wondering for months. The Kepler based GK110 will first show up as a Tesla K20 graphics card is built for intense computing applications that include data analytics, weather modeling,, computational chemistry and pshysics and aimed at both server and workstation systems.

As far as the specs are concerned, the GK110 features 7.1 billion transistors and promises up to three times the performance per watt when compared to previous Fermi architecture. It packs a total of 2880 cores organized in 15 SMX modules. Each SMX feature 192 single-precision CUDA cores, 64 double-precision units, 32 special function units (SFU) and 32 load/store units. The GK110 GPU features six 64-bit memory controllers, which adds up to a 384-bit memory interface.

Total of 2880 cores in 15 SMX modules

Nvidia has officially published the Kepler GK110 whitepaper detailing all the specs that that had us wondering for months. The Kepler based GK110 will first show up as a Tesla K20 graphics card is built for intense computing applications that include data analytics, weather modeling,, computational chemistry and pshysics and aimed at both server and workstation systems.

As far as the specs are concerned, the GK110 features 7.1 billion transistors and promises up to three times the performance per watt when compared to previous Fermi architecture. It packs a total of 2880 cores organized in 15 SMX modules. Each SMX feature 192 single-precision CUDA cores, 64 double-precision units, 32 special function units (SFU) and 32 load/store units. The GK110 GPU features six 64-bit memory controllers, which adds up to a 384-bit memory interface.

The memory subsystem inside the Kepler based GK110 includes 64KB of on-chip memory for each SMX that can be allocated with a bit more flexibility when compared to the previous Fermi architecture, enabling 32/32KB split between shared memory and L1 cache. In addition to L1 cache, the GK110 Kepler also has 48KB of Read-Only Data cache. In case you lost the number, this adds up to 960KB of shared memory.

As far as the L2 cache is concerned, the Kepler GK110 packs 1536KB of L2 cache, double the amount of L2 cache found in the Fermi architecture and offers up to twice the bandwidth per clock. As expected the GK110 also has ECC memory protection support for all register files, shared memories, L1 and L2 cache and DRAM memory.

In addition to these specs, Nvidia also included a couple of Kepler features that will show up in the GK110 and that include Dynamic Parallelism, Hyper-Q, Grid Management and Nvidia GPUDirect. We already wrote about these features and if you are looking for more details, you can always check out the Nvidia GK110 whitepaper located here.

All in all, the GK110 looks like a pretty impressive GPU, but we honestly doubt that we'll see a Geforce graphics card based on the GK110 anytime soon.

Source: fudzilla.com

nVidia frightened Chinese chips could take over by 2017

noreply@blogger.com (Jun) — Sat, 19 May 2012 14:30:00 +0000

Roll the clock back just 10 years and, somewhere inside China’s massive bureaucracy, a decision was made to no longer rely on Western companies designing the core technology on which China’s future would be built.

A project was kicked off at the Institute of Computing Technology under the watchful eye of Professor Hu Weiwu.

“Professor Hu?”, we hear you ask – and the answer is “Yes – you cheeky sod”.

The first chip was a 266MHz, 32-bit processor fabricated on a ‘nice and chunky’ 0.18 micron process. In modern parlance, that’s 180 nanometres. China had kicked off its first salvo into the CPU market with a very basic chip that was aimed purely at products like the tills you find in shops.

Intel, you might remember, was pushing out 3.06GHz Pentium 4 chips and AMD was ruling the processor performance roost with its Athlon 64 FX range about to hit the market.

Roll the clock forward 10 years and we have a very interesting quote from nVidia’s Chief Scientist, Bill Dally, circulating the web. He has been quoted as saying that China’s efforts at making a processor were “Laughable”, but that the rate of development and licensing was “Frightening” to the point where nVidia is predicting that China will be matching the West very soon and pulling ahead by 2017.

In a double thumbs up with big grin “You like?” moment, China’s CPU programme (Godson/Loongson) has decided that X86 and Windows should not really form any part of its future – so the main thrust is on Linux. That’s not to say that other operating systems cannot be ported to the Chinese processor – just that it is more ‘for show’ then a serious push.

Source: kitguru.net

Nvidia to make a mess with rebranded GT 620

noreply@blogger.com (Jun) — Sat, 19 May 2012 13:50:00 +0000

We found out a rather interesting piece of information regarding the rebranded entry-level GT 620 graphics cards and it appears that not all cards on the market will be the same.

We started asking questions about it when we saw the specification list on Nvidia's website for the GT 620, we realized they are not exactly the same as the GT 430 and some GT 620 graphics cards that were already launched by some Nvidia partners. We thought that it is a simple mistake but the situation seems much worse.

Some will be 64-bit, some 128-bit

We found out a rather interesting piece of information regarding the rebranded entry-level GT 620 graphics cards and it appears that not all cards on the market will be the same.

We started asking questions about it when we saw the specification list on Nvidia's website for the GT 620, we realized they are not exactly the same as the GT 430 and some GT 620 graphics cards that were already launched by some Nvidia partners. We thought that it is a simple mistake but the situation seems much worse.

Apparently, some GT 430 graphics cards that partners currently have on stock will get a new shiny box and a BIOS flash. The original GT 430 was based on the GF108-300 chip with a 128-bit memory interface but the new chip will be GF108-100 and will feature a 64-bit interface. To make things worse, both should hit retail/e-tail at similar date so it will be hard to differentiate which chip you get unless you take a closer look at the box.

The similar situation is with the GT 630 that will get a new chip as well. The old GT 440 was based on the GF108-400 with DDR3 and GDDR5 support paired up with a 128-bit memory interface, while the new one will be GF108-401 that will only feature DDR3 support. However, this is not a big deal since you are practically getting almost same specs, unless you are going for GT 630 with GDDR5 memory.

The GT 610 will get a plain old BIOS flash, new box and will use the same GF119-300 GPU found on all GT 520 graphics cards.

Source: fudzilla.com

Nvidia Blames TSMC's 28nm Process Technology for Slow Sales

noreply@blogger.com (Jun) — Mon, 14 May 2012 03:30:00 +0000

Although the yields of semiconductors made using 28nm fabrication process at Taiwan Semiconductor Manufacturing Company seem to be improving, their undersupplies is just becoming more evident. For Nvidia Corp., it just hurts.

Nvidia on Friday said that the problems with supply of chips made using 28nm process technology at TSMC persisted throughout its first quarter of fiscal 2013. Even though the company's better-than-expected margins signal that the costs are getting down, which is an indicator of improving yields for leading-edge products, the firm does not expect its supply issues to get resolved shortly.

Nvidia Does Not Expect 28nm Issues to Be Resolved Soon

Although the yields of semiconductors made using 28nm fabrication process at Taiwan Semiconductor Manufacturing Company seem to be improving, their undersupplies is just becoming more evident. For Nvidia Corp., it just hurts.

Nvidia on Friday said that the problems with supply of chips made using 28nm process technology at TSMC persisted throughout its first quarter of fiscal 2013. Even though the company's better-than-expected margins signal that the costs are getting down, which is an indicator of improving yields for leading-edge products, the firm does not expect its supply issues to get resolved shortly.

"Demand is high for Kepler and although supply will continue to improve, we are not able to meet all our OEM and channel demand in Q2 FY2013. We do not expect the 28nm supply situation to resolve itself until later this year," said Chris Evenden, director of investor relations at Nvidia, during quarterly conference call with financial analysts.

For the Q1 FY2013 that ended on April, 29, 2012, Nvidia reported revenue of $924.9 million, net income of $60.4 and gross margin of 50.1%. The results were generally better than expected by many, primarily thanks to shipments increase of Tegra 3 system-on-chip.

Nvidia's operating expenses (to which the company usually attributes GPU/GPGPU implementation costs) were even higher than projected by around $7 million at $390.5 million. Nonetheless, the firm's gross margins was 50.1%, higher than predicted 49.2%. This happened primarily because of sound launches of the GeForce 6-series GTX "Kepler" products that are now sold in price ranges between $399 and $999.

Due to seasonally weak calendar first quarter, sales of Nvidia were down across both consumer and professional lines.

Source: xbitlabs.com

NVIDIA’s 28nm chips get priority over AMD’s at TSMC

noreply@blogger.com (Jun) — Mon, 14 May 2012 02:56:00 +0000

If you are in the hunt for NVIDIA’s latest and greatest Gefore GTX 680, chances are you either managed to grab one out of pure luck or have given up completely. As of writing, Newegg does not have a single GTX 680 in stock. Even one of the largest PC stores down under only have a few in stock.

NVIDIA’s relationship with their primary foundry, TSMC, has no doubt been unpredictable in recent times. They blamed TSMC for denting NVIDIA’s competitive edge, then later credited the foundry for the efficiency of the latest Kepler cards.

At one point it got to the point where NVIDIA threatened to find another foundry, and reports arose that Samsung was preparing qualification samples. In response to this, TSMC is now giving NVIDIA priority over others such as AMD and Qualcomm for the manufacture of 28nm chips. Something that is sure to enrage AMD and only add fuel to the rumour of Qualcomm Snapdragon S4 delays.

Source: kitguru.net

AMD preparing HD 7970 GHz Edition

noreply@blogger.com (Jun) — Mon, 14 May 2012 02:24:00 +0000

According to a post over at Atomicmps site, AMD has had its yields improve significantly and the company will launch the HD 7970 GHz Edition - a reference Tahiti XT-based graphics card that will run at beyond 1000MHz for the GPU.

Of course, some partners have jumped the gun and we have already seen some HD 7970 that are actually clocked higher than 1000MHz for the GPU, including the 1100MHz clocked Powercolor PCS+ HD 7970 Vortex II and Sapphire's HD 7970 Toxic, already seen during Cebit.

Something that partners already did

According to a post over at Atomicmps site, AMD has had its yields improve significantly and the company will launch the HD 7970 GHz Edition - a reference Tahiti XT-based graphics card that will run at beyond 1000MHz for the GPU.

Of course, some partners have jumped the gun and we have already seen some HD 7970 that are actually clocked higher than 1000MHz for the GPU, including the 1100MHz clocked Powercolor PCS+ HD 7970 Vortex II and Sapphire's HD 7970 Toxic, already seen during Cebit.

Of course, those were probably cherry picked samples but the current crop of AMD's Tahiti ES chips easily hit 1250MHz GPU clocks. The memory on HD 7970 GHz Edition cards should remain unchanged - 384-bit memory interface and 3GB of GDDR5 memory gives it a nice edge over the GTX 680.

The HD 7970 GHz Edition will of course go against Nvidia's GTX 680 and while it is still not clear if AMD plans to go over 1000MHz on reference cards, we are sure that partners will certainly follow with their own versions at some point in time.

Source: fudzilla.com

POLL #2 Result: Do you overclock your graphics card?

noreply@blogger.com (Jun) — Sun, 06 May 2012 14:15:00 +0000

POLL: Do you overclock your graphics card?

RESULT: Majority says No...

Nvidia GeForce GTX 660 Ti Not Coming Until Q3 '12?

noreply@blogger.com (Jun) — Thu, 03 May 2012 14:49:00 +0000

With the release of the GeForce GTX 690 announced this past weekend and the GeForce GTX 670 expected to hit the selves on May 10th, Nvidia will not be releasing the GeForce GTX 660 Ti, as previously expected at the same time. Information coming out of Overclockers.co.uk has Andrew Gibson (Overclockers.co.uk) quoted as "660Ti, you have zero chance, its approx 6 months away........" There is no official word from Nvidia on the release of the GTX 660 Ti.

What information we do have on the GTX 660 Ti (though at this point still considered a rumor) is that it will feature 6 SMX clusters (1152 CUDA cores). This is two fewer than the GTX 680 and one fewer than the expected GTX 670. The GTX 660 Ti is expected to have an estimated price of around $249 at release, which would put it against the AMD Radeon HD 7800 series.

Nvidia's GeForce GTX 660 Ti is now expected to hit the selves in Q3 '12 and not the previously expected May 10th, along with the GeForce GTX 670, according to reports.

With the release of the GeForce GTX 690 announced this past weekend and the GeForce GTX 670 expected to hit the selves on May 10th, Nvidia will not be releasing the GeForce GTX 660 Ti, as previously expected at the same time. Information coming out of Overclockers.co.uk has Andrew Gibson (Overclockers.co.uk) quoted as "660Ti, you have zero chance, its approx 6 months away........" There is no official word from Nvidia on the release of the GTX 660 Ti.

What information we do have on the GTX 660 Ti (though at this point still considered a rumor) is that it will feature 6 SMX clusters (1152 CUDA cores). This is two fewer than the GTX 680 and one fewer than the expected GTX 670. The GTX 660 Ti is expected to have an estimated price of around $249 at release, which would put it against the AMD Radeon HD 7800 series.

Rumored specifications for the GeForce GTX 660 Ti are:

• Transistors Count: 3.5 billions
• Process: 28 nm
• SMXs: 6
• CUDA Cores: 1152
• TMUs: 96
• ROPs: 24
• Base Clock: 1006 MHz
• Memory Clock: 1502 MHz
• Memory: 1536 MB GDDR5 192-bit
• Bandwidth: 144 GB/s
• Single-Precision Computing power: 2.35 TFLOPS
• TDP: 150 W

Source: tomshardware.com

Nvida patents parallelisation of GPU data

noreply@blogger.com (Jun) — Thu, 03 May 2012 14:35:00 +0000

Graphics dalek Nvidia has filed for an extension to its patents for a hierarchical processor array.

It looks like Nvidian thinks that Patent 7,634,637 can be extended to cover some new ideas it has to fix a core design problem hat results in wide and ineffective graphics rendering pipelines. Nvidia has applied for a patent that describes the idea is that there are two or three tiers of processing cores with dedicated functions. The pipelines would include different shaders, such as a vertex shader unit, a geometry shader, a pixel shader and some others.

Keep all your processors hierarchical

Graphics dalek Nvidia has filed for an extension to its patents for a hierarchical processor array.

It looks like Nvidian thinks that Patent 7,634,637 can be extended to cover some new ideas it has to fix a core design problem hat results in wide and ineffective graphics rendering pipelines. Nvidia has applied for a patent that describes the idea is that there are two or three tiers of processing cores with dedicated functions. The pipelines would include different shaders, such as a vertex shader unit, a geometry shader, a pixel shader and some others.

In the patent Nvidia says that "each massively parallel stage in a stage-by-stage pipeline tends to provide little granularity of control of portions of each parallel stage. Each "massively parallel stage becomes unwieldy and prohibitively time-consuming to design". As the massively parallel stage struggles during operation to find sufficiently wide units of work to fully occupy the data path its usage is cut back.

Nvida's cunning plan is to keep parallelisation efficient, by using multiple levels of processing hierarchies with multiple classes of graphics operations being associated with a different stage of graphics processing. Each level would include a module that is capable of processing all graphics functions. There would also be one top-level component that is able to distribute certain classes of work to lower level classes of processors.

It also comes out with a third-level class in the processor hierarchy that would be reserved for general purpose computations, and a specialised graphics function module that can perform graphics operations carried out based on frame buffer data.

The result is a design which is configured to execute a large number of threads in parallel.

Source: fudzilla.com

Quick Note: NVIDIA Serves Up $999, Dual-GPU GeForce GTX 690

noreply@blogger.com (Jun) — Thu, 03 May 2012 14:24:00 +0000

NVIDIA today unleashed a graphics card that is sure to whet the appetites of hardcore PC gamers around the world. Unfortunately, the price of entry is so high that only those with massive amounts of disposable income will be likely to take the plunge.

The new GeForce GTX 690 uses dual Kepler GPUs on a single board. Compared to the single-GPU GTX 680, NVIDIA says that performance nearly doubles in most gaming situations. The GTX 690 is of course built on a 28nm process and brings with it 3,072 CUDA cores.

For the truly insane gamers, two GTX 690s can be paired in SLI mode for some quad-core graphics goodness.

Price of entry is high, but so is the payoff

NVIDIA today unleashed a graphics card that is sure to whet the appetites of hardcore PC gamers around the world. Unfortunately, the price of entry is so high that only those with massive amounts of disposable income will be likely to take the plunge.

The new GeForce GTX 690 uses dual Kepler GPUs on a single board. Compared to the single-GPU GTX 680, NVIDIA says that performance nearly doubles in most gaming situations. The GTX 690 is of course built on a 28nm process and brings with it 3,072 CUDA cores.

For the truly insane gamers, two GTX 690s can be paired in SLI mode for some quad-core graphics goodness.

"The GTX 690 is truly a work of art -- gorgeous on the outside with amazing performance on the inside," doted Brian Kelleher, senior vice president of GPU engineering at NVIDIA. "Gamers will love playing on multiple screens at high resolutions with all the eye candy turned on. And they'll relish showing their friends how beautiful the cards look inside their systems."

All of this performance comes at a cost, however. The GTX 690 will have an MSRP of $999 when it launches in limited quantities on May 3 -- wider availability will come on May 7.

Source: dailytech.com

Nvidia teases new product

noreply@blogger.com (Jun) — Thu, 19 Apr 2012 01:51:00 +0000

Nvidia has released a teaser picture on its Geforce Facebook page hinting that we'll see a new announcement pretty soon.

The teaser picture doesn't reveal much and shows black block with etched Nvidia logo and "it's coming" title written below it. It does appear to be a graphics card, at least by the looks of it, but your guess is good as ours. It is still a neat teaser picture. The new card could either be a dual-GPU Geforce GTX 690 (some call it the GTX 685), or can be something that we simply didn't expect.

"It's coming"

Nvidia has released a teaser picture on its Geforce Facebook page hinting that we'll see a new announcement pretty soon.

The teaser picture doesn't reveal much and shows black block with etched Nvidia logo and "it's coming" title written below it. It does appear to be a graphics card, at least by the looks of it, but your guess is good as ours. It is still a neat teaser picture. The new card could either be a dual-GPU Geforce GTX 690 (some call it the GTX 685), or can be something that we simply didn't expect.

In any case, we'll know soon enough and we'll keep digging to see what we can find out.

Source: fudzilla.com

GPUs Now a Tool for Accelerating Cancer Research

noreply@blogger.com (Jun) — Thu, 19 Apr 2012 01:49:00 +0000

As many of us know, cancer is currently treated with a highly potent beam of radiation. This works very well to kill the cancer cells and hopefully destroy the tumor, but radiation does not discriminate; it kills all cells in its path – cancer or otherwise.

Doctors avoid the healthy cells by taking accurate MRI scans. The problem with these 3-D models is that most are out of date – sometimes by months. Inaccuracies in the scans lead to questions such as: How large has the tumor grown? Has the tumor recessed somewhere else? The answers to these questions are critical when high power beams of radiation are being directed into a patient. An updated MRI scan every day is quite inconvenient for most patients, making that solution unfeasible and too expensive.

GPUs are predicting the action of cancer inside the body.

As many of us know, cancer is currently treated with a highly potent beam of radiation. This works very well to kill the cancer cells and hopefully destroy the tumor, but radiation does not discriminate; it kills all cells in its path – cancer or otherwise.

Doctors avoid the healthy cells by taking accurate MRI scans. The problem with these 3-D models is that most are out of date – sometimes by months. Inaccuracies in the scans lead to questions such as: How large has the tumor grown? Has the tumor recessed somewhere else? The answers to these questions are critical when high power beams of radiation are being directed into a patient. An updated MRI scan every day is quite inconvenient for most patients, making that solution unfeasible and too expensive.

With more and more developers and enterprises utilizing parallel computing to run their supercomputers, more are using clusters of Nvidia GPUs, or to be more specific, Tesla GPUs. Tesla GPUs have been seen in clusters driving the most powerful supercomputers and allowing future moon craft to act more intelligently. This innovation in the use of Tesla GPUs may be transferred further into the medical field and may even be used to diagnose and treat cancer.

Many of the above questions can be answered by using Nvidia Tesla accelerated computers. These computers will have the capability, with extreme computing performance, to predict the movement and change of the patient’s tumor. All this can be done with one outdated MRI. This technology, called ART (Adaptive Radiotherapy), will be used to more accurately focus treatment on the tumor and not live flesh.

Source: tomshardware.com

Report: Nvidia Stopping Production of GeForce GTX 580, 590

noreply@blogger.com (Jun) — Thu, 19 Apr 2012 01:34:00 +0000

Information coming out of SweClockers states that Nvidia has reportedly stopped production of GeForce GTX 580 GPUs. Launched in November, 2010, the GeForce GTX 580 had been the fastest single-GPU graphics card until AMD responded with Radeon HD 7900 series over a year later. With the successful launch of the GeForce GTX 680 and more Kepler based GPUs slated in May, the GeForce GTX 580 has served its purpose to users around the world. The cards will remain in inventory until fully excused out in the market. Nvidia has already stopped production of GeForce GTX 590, with the rumored dual Kepler-based card already in the works.

With the success of the Kepler-based GTX 680 and more units slated for May and beyond, Nvidia's partner manufacturers have stopped production of the GTX 580 & GTX 590

Information coming out of SweClockers states that Nvidia has reportedly stopped production of GeForce GTX 580 GPUs. Launched in November, 2010, the GeForce GTX 580 had been the fastest single-GPU graphics card until AMD responded with Radeon HD 7900 series over a year later. With the successful launch of the GeForce GTX 680 and more Kepler based GPUs slated in May, the GeForce GTX 580 has served its purpose to users around the world. The cards will remain in inventory until fully excused out in the market. Nvidia has already stopped production of GeForce GTX 590, with the rumored dual Kepler-based card already in the works.

Source: tomshardware.com

AMD confirms Radeon HD 7000 Series price cuts

noreply@blogger.com (Jun) — Thu, 19 Apr 2012 01:32:00 +0000

Ever since Nvidia introduced its 28nm Kepler-based GeForce GTX 680, we have been hearing rumours of imminent AMD price cuts on its own 28nm Southern Islands-based GPUs and now they're official.

AMD has cut the MSRP for the Radeon HD 7970 to $479, while the Radeon HD 7950 now comes in at $399. It will take some time for the new prices to filter down to retailers, but we are already seeing some cuts in Europe. The cheapest Radeon HD 7970 now sells for €404 and you can pick up a Radeon HD 7950 for €350+.

Sweetens the deal with 3 free games

Ever since Nvidia introduced its 28nm Kepler-based GeForce GTX 680, we have been hearing rumours of imminent AMD price cuts on its own 28nm Southern Islands-based GPUs and now they're official.

AMD has cut the MSRP for the Radeon HD 7970 to $479, while the Radeon HD 7950 now comes in at $399. It will take some time for the new prices to filter down to retailers, but we are already seeing some cuts in Europe. The cheapest Radeon HD 7970 now sells for €404 and you can pick up a Radeon HD 7950 for €350+.

AMD also shaved off a few bucks from the Radeon HD 7770, which is now priced at $139, but sadly the cuts did not extend to Radeon HD 7800 Series cards.

In addition to new prices, AMD is also bundling three games with the cards: Deus Ex: Human Revolution, Dirt: Showdown and Nexuiz, whatever that is.

Source: fudzilla.com

More Nvidia rebrands on the way

noreply@blogger.com (Jun) — Thu, 12 Apr 2012 05:42:00 +0000

After our initial article that rebranded Nvidia Geforce GT 620 will indeed hit retail, we have been asking around and managed to dig out a bit more info regarding it. According to our sources, it appears that the entire entry-level lineup will be rebranded and we'll be able to eventually see GT 620, GT 630 and Geforce 610 on retail/e-tail shelves.

Unfortunately, we were a bit off with our initial article as it now appears that retail and OEM rebrands of the entry-level 6xx lineup will be different. While the GT 620 OEM and Geforce 605 OEM are rebranded GT 520 and Geforce 510, the retail versions will be quite a bit different. According to our current info, the retail version of the GT 620 will be a rebranded GT 430. The GT 430 and thus the GT 620 is based on the Fermi 40nm GF108 GPU with 96 shaders, 4 ROPs, 16 TMUs and a 128-bit memory interface with GDDR3 memory.

Entire entry-level 6xx lineup

After our initial article that rebranded Nvidia Geforce GT 620 will indeed hit retail, we have been asking around and managed to dig out a bit more info regarding it. According to our sources, it appears that the entire entry-level lineup will be rebranded and we'll be able to eventually see GT 620, GT 630 and Geforce 610 on retail/e-tail shelves.

Unfortunately, we were a bit off with our initial article as it now appears that retail and OEM rebrands of the entry-level 6xx lineup will be different. While the GT 620 OEM and Geforce 605 OEM are rebranded GT 520 and Geforce 510, the retail versions will be quite a bit different. According to our current info, the retail version of the GT 620 will be a rebranded GT 430. The GT 430 and thus the GT 620 is based on the Fermi 40nm GF108 GPU with 96 shaders, 4 ROPs, 16 TMUs and a 128-bit memory interface with GDDR3 memory.

According to our info, Nvidia also plans to rebrands the GT 440, which will magically (with BIOS flash and a new shiny box) become the GT 630. The Geforce GT 440 (non OEM one) is based on the same GF108 40nm GPU but has slightly higher clocks and is paired up with GDDR5 memory. The last but not least is the GT 610 which will in retail/e-tail world is simply the GT 520, based on the GF119 GPU with 48 shaders, 4 ROPs and 64-bit memory interface.

The rebranding really doesn't come as a surprise. According to our info, partners expect that the upcoming AMD's Trinity and even Intel's Ivy Bridge will devastate the entry-level GPU market. Intel's Ivy Bridge could end up with a performance similar to the GT 430 and Trinity will even go higher. With this in mind, Nvidia would rather simply rebrand existing cards. As noted, OEM rebranding will be a bit different and OEM 6xx should not be the same as entry-level retail/e-tail 6xx cards.

Source: fudzilla.com

More info leaks out on NVIDIA’s dual-GPU GTX 690

noreply@blogger.com (Jun) — Thu, 12 Apr 2012 05:32:00 +0000

A leak through Chinese channels not only further engrains previous confirmation that NVIDIA has a dual-GPU based GeForce GTX 690 in the works, but also lets a few more details out the bag on the specs front.

This upcoming dual-GPU wielding weapon will be based on the 28nm GK104 silicon and is said to require juice from two 8-pin PCIe power connectors. The card will harness a PCI-E 3.0-compliant bridge chip and its display output configuration will comprise three DVI and one DisplayPort.

Certainly, the biggest concern for many will be the power draw, but given the kinds of people that are serious about laying down big dollars for the top tier models like these, it won’t strike potential buyers as anything outrageous.

Apparently NVIDIA is setting the PSU requirement for the GeForce GTX 690 to be a minimum of 650W. If anything, seems a little lowly? Odds are most people buying one of these cards will already have the funds (if not already own) a high quality 750W+ power supply to ensure stability won’t be an issue.

It can also be expected that the GTX 680′s complete feature set will be in place with the GTX 690.

Source: kitguru.net

Nvidia releases GeForce 301.24 beta with FXAA, Adaptive VSync, more

noreply@blogger.com (Jun) — Thu, 12 Apr 2012 05:29:00 +0000

Making up for March's relatively dull update, Nvidia has announced what it deems an "essential upgrade for all GeForce users." Released yesterday, the 301.24 beta drivers carry many of the goodies introduced with the GTX 680 a few weeks ago, as well as new performance improvements, bug fixes and SLI/3D Vision profiles.

Among the new features is FXAA, an anti-aliasing technology that is supposedly up to 60% faster than 4xMSAA and offers similar, if not superior results to such "ageing" solutions. Nvidia offered a detailed comparison in Batman: Arkham City last December, showing how FXAA fares against varying degrees of MSAA and CSAA.

This week's beta drivers also pack Adaptive VSync, which automatically adjusts vertical sync depending on your game's frame rate to prevent screen-tearing and other hiccups. Vertical sync is enabled if your game reaches the ideal frame rate (i.e. 60fps) and it's disabled when speeds dips below the target performance.

Download GeForce 301.24 beta (release notes)
Desktop: Windows XP 32-bit | Windows XP 64-bit | Windows Vista/7 32-bit | Windows Vista/7 64-bit
Mobile: Windows Vista/7 32-bit | Windows Vista/7 64-bit

GeForce 301.24 has a slew of Nvidia Surround enhancements, including the ability to enable a fourth "accessory display" for basic tasks like checking email and Web browsing, desktop management options, "bezel peeking" to reveal hidden game information, custom resolution management, and central display acceleration.

Nvidia touts another ~20% boost in Skyrim's indoor scenes (compared to last month's 296.10 WHQL release). Depending on your setup, you might also feel gains in Batman: Arkham City, Bulletstorm, Civilization V, Deus Ex: Human Revolution, Dragon Age 2, Far Cry 2, Just Cause 2, Metro 2033, Total War: Shogun 2, and StarCraft 2.

We've included a list of SLI and 3D Vision profile updates below and you'll find various bug fixes on page nine of the release notes. Considering Microsoft hasn't approved the update yet, you might want to wait a week or two until the WHQL build surfaces. If not, get downloading and share your findings in the comments.

New & Updated SLI Profiles
Alan Wake
Call of Juarez: The Cartel
Counter-Strike: Global Offensive
Deus Ex: Human Revolution - The Missing Link
Left 4 Dead
Orcs Must Die!
Portal 2
Risen 2: Dark Waters
The Darkness II

New & Updated 3D Vision Profiles
All Zombies Must Die! – Rated Fair
Ghosts 'n Goblins Online – Rated Good
Krater – Rated Poor
Oil Rush – Rated 3D Vision Ready
Postal III – Rated Good
Rayman Origins – Rated Good
SevenCore – Rated Fair
Stacking – Rated Good
Unigine Heaven Benchmark v3.0 – Rated 3D Vision Ready
Wargame: European Escalation – Rated Good
Warp – Rated Good
Wings of Prey – Rated Fair

Source: techspot.com

POLL #1 Result: Are you impressed by Nvidia Kepler?

noreply@blogger.com (Jun) — Thu, 12 Apr 2012 04:50:00 +0000

POLL: Are you impressed by Nvidia Kepler??

RESULT: Majority says No...

Details on Nvidia Kepler GK104-based GTX 670 Ti Surface

noreply@blogger.com (Jun) — Sun, 08 Apr 2012 18:19:00 +0000

The next two SKUs based on the GK104 Kepler look to be set to release in May under the names GTX 670 and GTX 670 Ti. According to German publication 3DCenter.org, these GK104s will take the same route of the GF100: GK104 on the new cards will have one fewer Streaming Multiprocessors (SMX). These are basically GTX 680s that didn't make the cut during manufacturing with one of the SMX units disabled. The SMX units holds 192 CUDA cores, so with seven out of eight SMX units enabled, you end up with a 1344 CUDA core count. This allows Nvidia to salvage the graphics card and provide end-users with a card a tier or two down the performance charts.

When something goes wrong during the manufacturing process, what are manufacturers to do? Do what many other have done, you turn the part into a new item with the defective part disabled!

The next two SKUs based on the GK104 Kepler look to be set to release in May under the names GTX 670 and GTX 670 Ti. According to German publication 3DCenter.org, these GK104s will take the same route of the GF100: GK104 on the new cards will have one fewer Streaming Multiprocessors (SMX). These are basically GTX 680s that didn't make the cut during manufacturing with one of the SMX units disabled. The SMX units holds 192 CUDA cores, so with seven out of eight SMX units enabled, you end up with a 1344 CUDA core count. This allows Nvidia to salvage the graphics card and provide end-users with a card a tier or two down the performance charts.

Rumored Specifications for the GeForce GTX 670 Ti:

• 4 Graphics Processing Clusters (GPC)
• 7 SMX Units (192 CUDA cores per units)
• 1344 CUDA Cores
• 112 Texture Units (TU)
• 32 Raster Units (ROP)
• 256-bit Memory Controller
• 2 GB GDDR5 Memory
• Estimated base core clock of 900 MHz (boost clock not known)
• Estimated 1250 MHz (5.00 GHz GDDR5 effective) memory clock, with around 160 GB/s memory bandwidth
• Estimated price $349-399 dollars

The GTX 670 Ti is expected to see a 20 percent drop in performance when compared to the GTX 680. With its estimated performance and price point, the GTX 670 Ti is set to do battle against AMD's Radeon HD 7950 and previous generation GTX 580.

Source: tomshardware.com

Rebadged GT 620 not for OEM only

noreply@blogger.com (Jun) — Sun, 08 Apr 2012 18:13:00 +0000

After doing a piece on Nvidia's GT 620 and Geforce 605 OEM-only graphics cards, which are actually rebranded GT 520 and Geforce 510, we got a nice tip that the GT 620 is not actually OEM-only card and it will also hit retail/e-tail.

In case you missed it, the GT 620 features the same GF119 40nm GPU with 48 stream processors found on the Geforce GT 520. The only difference between the GT 520 and the "new" GT 620 is the amount of memory, or rather half of it on the GT 620, but we are pretty sure that this won't happen in retail/e-tail, since it is a simple matter of box change. The card will still work at 810MHz for the GPU, 1620MHz for shaders and 898MHz for up to 2GB of DDR3 memory paired up with a 64-bit memory interface.

It will hit retail

After doing a piece on Nvidia's GT 620 and Geforce 605 OEM-only graphics cards, which are actually rebranded GT 520 and Geforce 510, we got a nice tip that the GT 620 is not actually OEM-only card and it will also hit retail/e-tail.

In case you missed it, the GT 620 features the same GF119 40nm GPU with 48 stream processors found on the Geforce GT 520. The only difference between the GT 520 and the "new" GT 620 is the amount of memory, or rather half of it on the GT 620, but we are pretty sure that this won't happen in retail/e-tail, since it is a simple matter of box change. The card will still work at 810MHz for the GPU, 1620MHz for shaders and 898MHz for up to 2GB of DDR3 memory paired up with a 64-bit memory interface.

According to our sources, partners have already ordered new boxes and it shouldn't be long before we see a 520 to 620 switch in retail/e-tail. This does not come as a surprise since we are close to Ivy Bridge launch and with its "improved" graphics it simply makes the entry-level GPUs obsolete and useless.

We will certainly keep an eye out for more rebadging.

Source: http://www.fudzilla.com/home/item/26665-rebadged-gt-620-not-for-oem-only

Nvidia GeForce GTX 680 Game Benchmarks

noreply@blogger.com (Jun) — Sun, 08 Apr 2012 04:00:00 +0000

A collections of real-world gaming tests using Nvidia GeForce GTX 680, listed alphabetically.

"When NVIDIA briefed us on Kepler and the GeForce GTX 680, they said their goals with this new architecture were to produce a product that was faster, smoother, and richer than the previous generation. We think they pulled it off." -hothardware.com

GTX680 used to crowbar in less desirable low end cards

noreply@blogger.com (Jun) — Thu, 29 Mar 2012 17:34:00 +0000

Consumers may well think that when businesses buy from their suppliers, that it works in much the same way as buying normal products from a normal shot. For example a bottle of wine from supermarket.

No chance.

Imagine asking for a good bottle of wine and being told, “Sure, as long as you take 3 cases of house lager, 12 bottles of Château No-Hope and a crate of wonky bottle openers”.

Surprising?

Maybe – but this is how the channel operates when a desirable product hits town. And it’s supported by vendors like nVidia. No let off from AMD either, with the big-calibre-Radeons having demanded on more than one occasion in the past, that several lower spec cards accompany them on any delivery shipment.

One UK distributor had 10 of the ‘chickens teeth’ GTX680 cards arrive and immediately explained to its customers, “Order now, because when they are gone – they are gone. Oh. By the way. You’ll need to order some mainstream product as well”.

Source: kitguru.net

Catalyst 12.3 supports HD 7000 series, squashes various bugs

noreply@blogger.com (Jun) — Thu, 29 Mar 2012 17:32:00 +0000

Radeon owners have a fresh driver update today, courtesy of Catalyst Software Suite 12.3. As usual, the release covers cards spanning back as far as the HD 2000 series and introduces full support for the latest HD 7000 products (except on Windows XP). Besides supporting Southern Islands GPUs, Catalyst 12.3's release notes don't mention any new features or game frame rate optimizations. However, a plethora of bugs have been fixed.

Most of the fixes are for Windows 7, including one that caused Tom Clancy's H.A.W.X to crash after task switching with anti-aliasing and morphological anti-aliasing (MLAA) enabled. Another caused Furmark Benchmark to crash when launched in high performance mode, while Alan Wake had issues when launching in DirectX 9 mode with multiple cards in Crossfire, and Far Cry 2 caused a system hang when running the game on high.

Image flickering and other assorted graphical glitches have been solved in Quake 4, XPlane, and TES V: Skyrim on Windows 7, while AMD has solved corruption in Enemy Territory Quake Wars when playing on Windows Vista. AMD notes that it's still working on other artifacting and crashing problems in Enemy Territory Quake Wars, Rage, and Crysis 2 on Windows 7, as well as Dirt 3, Crysis Warhead, and Dragon Age 2 on Windows XP.

Download Catalyst 12.3 WHQL (release notes)
Desktop: Windows XP 32-bit | Windows XP 64-bit | Windows Vista/7 32-bit | Windows Vista/7 64-bit
Mobile: Windows Vista/7 32-bit | Windows Vista/7 64-bit

Folks running Windows 8 are still best off with the month-old 8.93.7 RC10 driver, which doesn't cover HD 7000 products. Assuming you have an HD 5000 or 6000 series desktop card, the driver adds support for Windows 8's WDDM (Windows Display Driver Model) 1.2, including native Stereo 3D support for full screen and windowed gaming, optimized screen rotation and power consumption, as well as improved sleep and resume performance.

Download Driver 8.93.7 RC10 (release notes)
Desktop/Mobile: Windows 8 32/64-bit

Source: techspot.com