Vigile writes: A big shift in the way graphics cards and gaming performance are tested has been occurring over the last few months with many review sites now using frame times rather than just average frame rates to compare products. Another unique testing methodology called Frame Rating has been started by PC Perspective that uses video capture equipment capable of recording uncompressed high resolution output direct from the graphics card, a colored bar overlay system and post-processing on that recorded video to evaluate performance as it is seen by the end user. The benefit is that there is literally no software interference between the data points and what the user sees making it is as close to an "experience metric" as any developed. Interestingly, multi-GPU solutions like SLI and CrossFire have VERY different results when viewed in this light, with AMD's offering clearly presenting a poorer, and more stuttery, animation.
Vigile writes: Details of the new NVIDIA GeForce GTX TITAN graphics card based on GK110 are already known including the 7.1 billion transistor GPU, 6GB of on-board frame buffer and full speed double precision compute power but gaming benchmarks and performance weren't revealed until today. PC Perspective has tested the TITAN up against the best graphics cards on the market and found that the new NVIDIA flagship is easily the best single-GPU solution on the market though it does fall behind the dual-GK104 based GTX 690 in most cases. Where TITAN really shines is in multi-display, 5760x1080 resolutions. Interestingly, testing of the CrossFire configurations of the Radeon HD 7970 were omitted from the article due to concerns about current FRAPS-based testing methods, and an interesting new capture solution for performance analysis is discussed.
Vigile writes: NVIDIA completes their lineup of Kepler architecture GPUs today with the release of the GeForce GTX 650 Ti but as PC Perspective's review points out, every dollar in the crowded sub-$200 space matters. The GTX 650 Ti falls between AMD's Radeon HD 7770 1GB ($119) and the HD 7850 1GB ($179) perfectly though performance leans towards the HD 7770 in most cases. NVIDIA as also removed staple features like SLI support and the new GPU Boost that was available on all previous GTX 600 cards that dynamically clocked the GPU based thermal headroom for each application, taking away some of the spark of the Kepler architecture. The saving grace for this release might be the Assassins Creed 3 bundle; getting a $50 game with a $150 graphics card is a great deal but once that promotion is over the GTX 650 Ti will need to drop in price slightly to stay relevant.
Vigile writes: Hot on the heels of the new GK104 chip inside the GeForce GTX 680 graphics card, NVIDIA used its GPU Technology Conference in San Jose to announce and discuss the upcoming GK110 GPU that will be available on Tesla cards in Q4 of this year. Sporting 7.1 billion transistors and likely a ~550 square millimeter die size, this upcoming GPGPU targeted product includes 15 of the new SMX processing divisions each containing 192 CUDA cores for a theoretical total of 2,880 cores. NVIDIA did say that shipping products would likely have only 13 or 14 SMX modules enabled but even still the company claims to have crossed well into the 1 TeraFLOPs level with FP64 double precision math. NVIDIA still has some time to bring this product out on time but there are some that worry about NVIDIA's ability to make enough of these chips for the HPC market, let alone the gaming market down the road, based on the current availability of the GTX 680 and similar consumer cards. But I guess being able to charge $8000 for a GPU helps the bottom line.
Vigile writes: NVIDIA today announced a new technology partnership with Gaikai, an on-demand gaming company that competes with OnLive, that brings GeForce GRID to the cloud gaming ecosystem. GRID aims to increase both the visual quality and user experience of cloud gaming by decreasing latencies involved in the process, the biggest hindrance to acceptance for consumers. NVIDIA claims to have decreased the time for game stream capture and decode by a factor of three by handling the process completely on the GPU while also decreasing the "game time" with the power of the Kepler GPU. NVIDIA hopes to help both gamers and clould streaming companies by offering 4x the density currently available and at just 75 watts per game stream. The question remains — will mainstream adopt the on-demand games market as they have the on-demand video market?
Vigile writes: "NVIDIA has been rolling out the new Kepler GPU in various graphics cards since March, starting with the GeForce GTX 680 and then the GTX 690 dual-GPU card with a steep $999 price tag. Today's release of the GTX 670 offers some compelling arguments for being the BEST graphics card on the market. Based on the exact same die as the GTX 680 and 690, but with a single SMX disabled bringing the core count from 1536 to 1344, the GTX 670 still includes 2GB of frame buffer running at 6 Gbps on a 256-bit memory bus. Performance of the card actually rivals the $80-100 more expensive AMD Radeon HD 7970, a card that was the fastest GPU available just a few short months ago. With a price tag of $399, the GTX 670 isn't a mid-range card by any stretch, but it may just be the best card for power and dollar efficiency. Of course, NVIDIA first needs to fight through the massive availability issues they are having with this generation."
Vigile writes: When NVIDIA launched the GTX 680 last month, it was the fastest single GPU graphics card on the market, bypassing the Radeon HD 7970 card released in January. NVIDIA was late to this generation of GPU but they are definitely targeting the high-end gamer by releasing the GeForce GTX 690 today — a dual-GPU variant based on the same GK104 chip as the GTX 680. This card features a total of 3072 shader processors, 4GB of GDDR5 memory running at 6 Gbps and a cooler made of magnesium alloy and trivalent chromium plating. While the price tag is $999, the performance of the card simply blows away anything else on the market including the dual-GPU GTX 590 and HD 6990 cards.
Vigile writes: A new Acer Ultrabook, the Aspire Timeline Ultra M3, went on sale yesterday in Asia with a little surprise: NVIDIA's unreleased Kepler GPU inside. This is NVIDIA's first 28nm architecture and will compete against AMD's Radeon HD 7000 series of graphics cards. The version of Kepler found in the Acer Ultrabook only sports 384 CUDA cores and that is a noticeable jump up from 96 cores found in the 540M, and yet the GT 640M provides gaming performance comparable to the GT 555M. Plenty of details are still needed for a firm grasp of NVIDIA's latest GPU but showing off a high-powered discrete graphics solution in an Ultrabook indicates a step forward in performance per watt.
Vigile writes: In a talk earlier this year at DICE, Epic Games' Tim Sweeney discussed the state of computing hardware as it relates to gaming. While there is a rising sentiment in the gaming world that the current generation consoles are "good enough" and that the next generation of consoles might be the last, Sweeney thinks that is way off base. He debates that claim though with some interesting numbers including the amount of processing and triangle power required to match human anatomical peaks. While we are only a factor of 50x from the necessary level of triangle processing, there is 2000x increase required to meet the 5000 TFLOPS Sweeney thinks will be need for the 8000x4000 resolution screens of the future. It would seem that the "good enough" sentiment is still a long way off for developers.
Vigile writes: AMD continues to push forward with the new 28nm process GPUs and completes the 7000-series family with the Pitcairn-based Radeon HD 7870 GHz Edition and HD 7850 cards. The HD 7870 has 1280 stream processors while the HD 7850 has 1024 SPs, both cards have 256-bit memory buses that run at 1200 MHz. As the name implies, the HD 7870 runs at 1000 MHz and is priced at $350 squaring up perfectly with the GTX 570 from NVIDIA and is able to outperform it while using 80 fewer watts of power. The HD 7850 will sell for $250 and simply blows past the GTX 560 Ti again using 50 fewer watts in the process. There are full reviews available at PC Perspective, Tech Report, HardOCP and HotHardware, but it looks like AMD might have a graphics card worth upgrading to.
Vigile writes: "The new Southern Islands GPU architecture continues to ripple through the AMD graphics card lineup and the release of the Cape Verde GPU brings it to the sub-$200 price range. The Radeon HD 7770 and HD 7750 cards will be priced at $159 and $109 respectively and offer 640 and 512 stream processors. That is signifantly less processing power than the 2,048 SPs in the HD 7970 and the new 7700 cards also drop from a 384-bit memory bus to a 128-bit one, offering quite a bit less memory bandwidth. PC Perspective did performance and power testing and found that while there are much better alternatives to the HD 7770 on the market, the HD 7750 is unique in that you can game at 1080p with high image quality settings in even the latest games without the need for any external power to the card. It would seem NVIDIA needs to get their 28nm products out the door if they want to keep pace with AMD's discrete options."
Vigile writes: AMD is discussing its future at an event in Sunnyvale today including its continued push into the world of heterogeneous computing. What was once called FSA (Fusion System Architecture) is now being labeled HSA (Heterogeneous System Architecture) and promises a combination of CPU and GPU designs with features like combined memory address spaces and instant internal context switching. In roadmaps that they unveiled today AMD shows HSA integration as quickly as 2013 with the upcoming Sea Islands GPUs and Kaveri APU. AMD also referred to themselves being flexible in architecture selection and announced the ability to combine third-party IP on its future SoCs — can an ARM + AMD combination be far behind?
Vigile writes: AMD released the world's fastest single GPU last month in the form of the Radeon HD 7970 and now they have release based on the same GPU architecture. The new HD 7950 uses the Tahiti GPU but has 1,792 stream processors rather than the 2,048 of its larger brother and still maintains the absolutely massive 3GB frame buffer. Performance testing done by the gang at PC Perspective shows that the perf gap between the two cards ranges from 15-20% depending on the game though more importantly the new HD 7950 stays ahead of NVIDIA's flagship GeForce GTX 580 while having a $50 lower price. With features like three display outputs, lower power consumption than the competition thanks to the new 28nm process and a lot of overclocking headroom (over 1100 MHz!), the $449 Radeon HD 7950 could be a great option for high-end gamers.
Vigile writes: "Last month AMD released the world's fastest single GPU in the form of the Radeon HD 7970 and now they have another product based on the same GPU technology. The new HD 7950 is based on the Tahiti GPU but has 1,792 stream processors rather than the 2,048 of its larger brother but still maintains the absolutely massive 3GB frame buffer. Performance testing done over at PC Perspective shows that the deltas between the two cards range from 15-20% depending on the game though more importantly the new HD 7950 stays ahead of NVIDIA's GeForce GTX 580 while having a $50 lower MSRP. With features like triple display outputs, lower power consumption than the competition thanks to the new 28nm process and a lot of overclocking headroom, the $449 Radeon HD 7950 could be a great option for high-end gamers if they stay in stock."
Vigile writes: It has been a few generations since AMD has had the performance leading GPU in its camp but with the new Radeon HD 7970 the tables have finally turned. Tahiti is based on the completely new Southern Islands design and is AMD's first step in its Fusion System Architecture. Built around 2048 stream processors and a 384-bit memory bus with a 3GB frame buffer, the HD 7970 is the first 28nm GPU, the first to support DirectX 11.1 and is the first to support PCI Express 3.0 bus speeds. Compared to NVIDIA's GTX 580 the new AMD graphics card runs anywhere from 13-30% faster depending on the game and resolution according to PC Perspective's testing. AMD might not have long at the top of the heap with NVIDIA's Kepler coming in early 2012, but if you want the best gaming GPU today, the Radeon HD 7970 is it.