ASUS GTX 560 Ti Graphics Card Review

• GF114 chip
• 1.95 Billion transistors
• 900MHz Core Clock Speed
• 1800MHz Processor clock
• 4-phase power circuitry
• 4200MHz Memory clock
• 256-bit memory interface
• 1024MB GDDR5 memory
• Power Monitoring hardware
• CUDA
• PhysX
• 3D Vision
• NVIDIA Surround
• 2 Graphics Processing Clusters
• 8 Streaming Multiprocessors
• 384 CUDA Cores
• 64 Texture units
• 32 ROP units
• 512KB L2 Cache
• 134.4 GB/s memory bandwidth
• 52.6 Gigatexels/second bilinear texture fill rate
• 40nm fabrication process
• 2x Dual-Link DVI-I, 1x Mini-HDMI
• 2-slot form factor
• 2x 6-pin power connectors
• 170W TDP
• 100C Thermal Threshold

The GeForce GTX 560 is based upon a new chip called the GF114. This chip has 1.95 Billion transistors, and is built on TSMC’s 40nm fabrication process. If you might remember the GeForce GTX 480 was released early in 2010 with 480 CUDA cores and was the fastest single chip card on the planet until the release of the GeForce GTX 580 late last year. The GeForce GTX 580 was a “Full GTX 480” and that is basically what a GTX 560 Ti is, a “Full GTX 460”.

The chip is composed of two Graphics Processing Clusters each consisting of four Streaming Multiprocessors with each having 48 CUDA cores. This is exactly the same configuration of the GTX 460 i.e. 48 SPs per SM instead of 32 on the 480/580 and 8 SMs, but the 460 had 7 SMs enabled for 336 CUDA cores while this new graphics chip has all eight enabled.  NVIDIA redesigned the chip at the transistor level much like they did on the 580. There are 64 texture units and 32 ROPs. The TDP (Thermal Design Power) limit of the 560 is 170W which is slightly higher than on the 460 at 160W.

The GTX 460 had 336 CUDA cores with a clock speed of 675MHz/1350MHz/3600MHz (Core/Shader/Memory). With the ENGTX 560 Ti DCII TOP/2DI/1GD5, ASUS has increased the clock speed to 900/1800/4200MHz, on the full 384 CUDA core GTX 560 Ti part, yet they have only increased the TDP by 10W. So while the TDP is higher than the 460, the increase in clock speeds of the reference card and the enabling of the shaders is logical. The thermal threshold on the GTX 560 Ti is 100C and board partners will release overclocked cards some of whom I will review.

The GeForce GTX 560 Ti has eight PolyMorph engines which are NVIDIA’s term for their Tessellation engines on the card along with two raster engines. NVIDIA’s performance in Tessellation is generally higher than their competition from AMD due in large part to the number of PolyMorph engines. Tessellation is one of the major features of DirectX 11 which all GF1xx parts support as well as multithreaded rendering and Compute shaders. NVIDIA says their cards do DirectX 11 right and games like HAWX2 which uses tessellation extensively generally perform better on their cards.

As you might know a company called 3DFX introduced multi-GPU gaming in the 1990s with the advent of the Voodoo 2 SLI boards that required a 2D card to operate. Sadly 3DFX went out of business in 2000 and NVIDIA introduced their version of SLI in 2004. Today virtually every Intel motherboard with two graphics slots can support two or more video cards working together. The GeForce GTX 560 Ti card supports up to 2-way SLI allowing for scaling of up to 90% in games. I will cover SLI more in the SLI 560 article.

The memory interface on the GTX 560 Ti matches the interface on the GTX 460 1024MB card. Each set of 8 ROPs on the GF114 has a 64-bit memory controller. As the GTX 560 Ti has 32 ROPs that means the card has 4 64-bit memory controllers or a 256-bit memory interface.  The memory speed has also increased from the 4008MHz on the reference 560 to 4200MHz on the ASUS ENGTX 560 DCII TOP/2DI/1GD5 card. This means the ASUS card has a memory bandwidth of 134.4GB/second an increase over the reference card that has 128.3GB/second bandwidth.

NVIDIA specific features include PhysX, 3D Vision, NVIDIA Surround and SLI. Physics equations have traditionally been done on the graphics cards. A small company called Ageia developed the PhysX software and hardware to bring physics in games. NVIDIA bought Ageia a few years back and has now implemented support for the PhysX into their GeForce line of products allowing graphics cards such as the GTX 560 to do the physics effects on the graphics card instead of the CPU, allowing for better effects such as realistic clothing, dust, fog, destructible persistent damage and much more to be done with better performance than on the CPU. 3D Vision is NVIDIA’s technology to play stereoscopic 3D games and movies using a 3D Vision kit, a 120Hz monitor, and a GeForce card that supports it.  NVIDIA Surround allows up to three monitors at 7680x1600 resolutions (3x2560x1600) to be used to play games in surround view. Note, that to use NVIDIA Surround two cards in SLI mode are required.

ASUS Specific features on the ENGTX 560 TI DCII TOP/2DI/1GD5 card

• DirectCU II
• Dual Fans
• Super Alloy Power Design
• Super Alloy Choke
• Super Alloy Capacitor
• Super Hybrid Engine
• Super Alloy MOS
• SAP CAP
• Voltage Tweak
• Xtreme Design VGA
• Dust-proof fan
• GPU Guard
• Fuse Protection
• EMI Shield

If you might remember, ASUS introduced their DirectCU feature in 2009 with the launch of the EAH5850 DirectCU card and has used it on many cards from both NVIDIA and AMD. CU is the scientific symbol for copper and the DirectCU cooler had a copper plate and heatpipes in direct contact with the graphics chip.

The second generation DirectCU II cooler has replaced the single fan with a dual-fan cooler while still keeping direct contact with the GPU. The fans contribute double the airflow over the reference design and have a 30dB noise level, which is 11dB less than the reference cooler on the GeForce GTX 560 Ti. The direct contact of the flattened copper heatpipes to the GPU result in a up to 20% improvement in heat versus the reference design which uses a metal plate that hinders heat dissipation. Note that the GTX 560 in general operates at low temperatures compared to its bigger brothers.

The next feature is something called Super Alloy Power or SAP. SAP consists of Super Alloy Chokes, Super Alloy MOS, Super Hybrid Engine, and Super Alloy Capacitors. Super Alloy Chokes are made out of a special alloy formula and manufactured under high pressure and temperature. They have a concrete core which delivers buzz-free operation compared to chokes with hollow casing that can create buzzing vibration during operation. Super Allow Capacitors increase the maximum voltage threshold by 30% and the card lifespan to 150,000 hours in 75C ambient, which is 2.5 xs longer than traditional cards. The Super Hybrid Engine is a chip that intelligently switches between high and low-intensity power profiles in real-time offering up to 15% performance boost without the chip. The Super Alloy MOS is smaller, cooler, and more efficient than generic MOSFETs, increasing the maximum voltage threshold by 30% over traditional MOSFETs. SAP CAPs are on the back of the GPU providing 30% more efficient power to the GPU.

ASUS introduced their Xtreme Design for their motherboards in 2009 with the launch of their X58 motherboards like the P6X58D Premium motherboard.  Last year they brought many of these features into the graphics card space including   Dust-proof fan, GPU Guard, Fuse Protection, and EMI shield. The Dust-proof fan is sealed twice over to make it virtually dust-proof increasing the lifespan of the fan and preventing nasty dust build-up. As video cards get heavier and bigger they sometimes warp the PCB of the graphics card or even damage the components. GPU Guard strengthens the structural reinforcement on the card helping the card stay crack-free. ASUS Fuse protection provides doubled protection against electric current overload. The Electro Magnetic Interference shield on the card blocks radiation and improves image quality. It effectively blocks 66% radiation over traditional graphics cards.