In certain situations, when a company is using practically the same chips for two or more product generations, the need to rename a certain older-generation product is unavoidable. In order to simplify that process for the end users and avoid them drowning in a torrent of names, prefixes and suffixes, the simplest way is to rename the product with some slight changes and place it (either by name or price) into the newer category. NVIDIA is familiar with this concept, in fact, too familiar - this strategy got completely out of hand back in the era of GeForce 8, 9 and GT200. Luckily, history doesn’t have to repeat itself, and this move is entirely justified in the case of GTX560. This card is the one that should be put right behind GTX560 with the Ti(tanium) suffix, and therefore replace the very popular GTX460 with 1 GB VRAM, which was the first Fermi-based product to offer truly excellent performance at a moderate price.
 |
 |
Of course, we’re talking about the famed GF104 GPU, which was something all gamers on a budget were after. However, in order to break the pure renaming tradition, NVIDIA did their best to put GTX560 a step ahead of GTX460, which is reflected mainly in higher clocks, so even if the core of the product is the same, the new model is bound to be faster, with the only unknown thing being the percentage. We received ASUS’ representative of this card, and one from the premier DirectCU II TOP series at that.
 |
 |
The second part of this graphics card’s name refers to the cooling system that it’s using, and the solution in question is both true-and-tried and one of the best on the market. This system is famous for its heatpipe design, with the heatpipes lying directly on the GPU, in order to shorten the usual heat path “GPU - cooler base - heatpipes” by eliminating the mid-phase entirely. The rest of the system is more or less typical, with heatpipes transferring the heat evenly across the rest of the cooling body, the latter being cooled actively by two silent fans. Below the cooler, one can see the naked memory chips, which are cooled actively by the fans, but without any sort of passive cooling body. Bear in mind that DirectCU II occupies the adjacent slot, but this is something that we’ve already got used to, and the price that has to be paid to provide efficient silent cooling of ever more complex GPUs in the upper range. The digital power section’s cooling hasn’t been left unattended, luckily, and this part of the card is covered by a passive cooling body. The power section itself is very complex, with shielded coils, and has a high quality feeling to it, in accordance with the reputation of the brand ASUS. In order to provide the card with sufficient power, two 6-pin molex connectors are present. That doesn’t conclude the list of connectors on the card, though, as there’s a SLI connector as well, enabling the user to build a multi-GPU configuration.
 |
 |
Below the cooler, one can find the GF114 GPU in the hiding, which has been somewhat weakened compared to the full-blown version found on GTX560 Ti; one of its multiprocessors has been left disabled, which means that it has 336 functional stream processors out of 384 that are physically present on the card. The number of texture and ROP units present in this chip is 56 and 32, respectively. NVIDIA hasn’t proposed a default clock for this model, leaving the decision to the manufacturers, and most models are likely to be placed in the 800 MHz region, i.e. 1600 MHz for stream processors. ASUS has gone one step further, however, increasing this value to 925 MHz (1850 MHz for stream processors). This makes their TOP model one of the fastest on the market. Whichever way you look at it, the redefined clock value is significantly higher, as 125 MHz is a major difference compared to the 800 MHz that the original GTX460 1 GB was running at. This has increased the texture and pixel fillrate by 20% or so, which should definitely be felt in games. The memory chips are signed by Samsung and declared to work at up to 5 GHz effectively (1250 MHz of real clock). DirectCU II TOP’s memory clock is 1050 MHz (4.1 GHz effectively), and we have to commend NVIDIA for having finally breached the 4 GHz memory clock barrier in a mid-range graphics card.
