August 9th, 2011
Radeon HD 7000 Series GPU Architecture detailed
We got the information about the code names of graphics cards AMD Radeon HD 7000, which, according to unofficial information will be released in September this year.
Some time ago we reported that AMD has already started to produce the first test chips for graphics cards with the family of Southern Islands gpus.
However, it appears that work on new designs has been even more advanced than you might think. It is noteworthy that all of the following described GPU will be made in a 28-nanometer process.
AMD Radeon HD 7000 Series to be PCI-Express 3.0 Compliant
AMD’s next generation of graphics processors (GPUs) that will be branded under the HD 7000 series, are reported to be PCI-Express Generation 3 compliant. The desktop discrete graphics cards will feature PCI-Express 3.0 x16 bus interfaces, and will be fully backwards-compatible with older versions of the bus, including Gen 1 and Gen 2. Motherboards sold today feature Gen 2 PCI-E slots, although some of the very latest motherboards launched by major vendors feature PCI-Express 3.0 slots.
The new bus doubles the bandwidth over PCI-E 2.0, with 1 GB/s of bandwidth per lane, per direction. PCI-Express 3.0 x16 would have 32 GB/s (256 Gbps) of bandwidth at its disposal, 16 GB/s per direction. AMD’s next generation of GPUs, codenamed “Southern Islands” will be built on the new 28 nm process at TSMC, and will upscale VLIW4 stream processors. Some of the first PC platforms to fully support PCI-Express 3.0 will be Intel’s Sandy Bridge-E. Whether AMD’s GPUs have hit a bandwidth bottleneck with PCI-E Gen 2, or is AMD trying to just be standards-compliant, is a different question altogether.
An early version of the AMD Catalyst drivers 11.7 gave us some basic overview about upcoming models.
Low-end graphics market will be represented by the Radeon HD 7570 and HD 7670. The first one will use GPU codenamed Pro Lombok and the second Lombok XT. Both chips will be compatible with VLIW4 technology, which means that the four processor cores will be grouped in one cluster shading. The mid-end market will include three graphics – Radeon HD 7790 graphics processor with Thames LE gpu, Radeon HD 7850 graphics processor with Thames Pro gpu and Radeon HD 7870 XT with Thamse.
AMD Details Next Gen. GPU Architecture
Some major design changes are afoot for AMD’s fusion platform and for future GPUs. AMD is calling is next GPU architecture GCN or “graphics next core.” At the first AMD developer conference, titled AFDS, AMD showcased some new features for it’s upcoming Llano GPUs. The major changes are as follows:
VLIW4 to SIMD – This change allows for better GPGPU performance. VLIW5 and VLIW4 were efficient for graphics calculations, but not for general purpose (varying) CPU computations. SIMD brings the best of both worlds. In fact, NVIDIA has been using this for quite some time within their Fermi architecture.
Unified Memory – Both CPU and GPU can utilize the same address spaces. The GPU now has built in address translation hardware just for this specific design implementation.
Improved C++ GPGPU programming and debugging support – Previously developers had to utilize assembly and some C to achieve the desired implementation. Now programmers can take full advantage of a high level language with AMD GPUs.
Compute Units – Instead of the traditional steam processor, AMD will utilized its new SIMD architecute (which is comprised of multiple ALUs making up a 16 bit wide vector SIMD block). Each compute unit has 4 SIMD blocks with an L1 cache. Overall, each CU with have its own L1 cache and all CU’s will share a L2 cache (which all NVIDIA and AMD GPU’s currently employ).
Partially Resident Textures (PRT) – As with John Carmack’s MegaTexture technology, textures can now be partially loaded in memory for quicker access. This allows for the use of larger textures to be more efficiently used on a much larger scale than previous implementations. The main differnece is of course that insead of being implemented via software (id tech 5) it will now be implemented via hardware.
Thus, AMD’s fusion GPU code named Llano, will bring some major changes to the table that will dramatically improve on-die GPU performance when calculating and handling general purpose computations. In terms of graphics improvements the main difference revealed thus far is in terms of textures and memory usage. More information will be established as time progresses. Whether or not some of these changes will go into affect for AMD’s upcoming Southern Island GPU’s (Radeon HD 7000 sereies) has yet to be seen.
- READ MORE (Source): AMD Radeon HD 7000 Series to be PCI-Express 3.0 Compliant | techPowerUp
- READ MORE (Source): AMD Details Next Gen. GPU Architecture