AMD Radeon R7 (Bristol Ridge) vs Intel HD Graphics (Bay Trail) vs Intel HD Graphics 5000

AMD Radeon R7 (Bristol Ridge)

The AMD Radeon R7 (Bristol Ridge) is an integrated graphics card for the fastest mobile AMD Carrizo APUs. At its launch, the R7 is available on the top model FX9830P with 512 GCN shader units and a core clock of 900 MHz. The technical specifications do not differ from the AMD Radeon R7 in Carrizo, but the clock is up to 100 MHz higher. Depending on the cooling and the configurable TDP (12-45 Watts), the performance can differ quite a lot. The performance also depends on the memory (best performance with dual-channel DDR4).

Architecture and Features

According to AMD, the Radeon R7 consists of eight cores based on the third GCN generation (often called GCN 1.2 or 2.0 in the press). It is therefore very similar to the Tonga desktop chip and supports DirectX 12 (FL 12_0). Furthermore, its shaders can be used with Mantle, OpenGL and OpenCL. HSA 1.0 is supported for running code on both CPU and GPU cores simultaneously.

Compared to the Kaveri generation (Kaveri R7 with GCN 1.1), the new color compression in particular results in better memory utilization and improves the gaming performance by 5-7 %. 

The improved video decoder of Carrizo/Bristol Ridge now includes UVD 6 (Unified Video Decoder), which can decode HVEC / H.265 up to 4K in hardware. Furthermore, the VCE (Video Codec Engine) has been improved for faster transcoding of H.264. More information on architecture and features of the Bristol Ridge architecture can be found in our launch article.

Performance

Depending on the configurable TDP of the Bristol Ridge chips between 12 up to 45 Watts, there can be significant differences in games. Overall, the results are slightly better compared to the Carrizo Radeon R7 due to the higher clocks for comparable model. You can still use the Kaveri R7 benchmarks to check the gaming performance.

Power Consumption

Depending on the configurable TDP (cTDP), the 28 nm chip is specified at 12-45 Watts. It can therefore be used for thin and light notebooks as well as medium-sized devices or All-in-Ones (AiO).

Intel HD Graphics (Bay Trail)

The Intel HD Graphics (Bay Trail) is a low-end integrated Bay Trail graphics card found in certain Atom models (Z3770), nettops (J2850) and notebook SoCs (N3510). It supports DirectX 11 and is based on the Ivy Bridge GPU. Clock rates and shader cores, however, are considerably lower.

Compared to Ivy Bridge, which offers either 6 or 16 Execution Units, the HD Graphics (Bay Trail) comes with only 4 EUs. The core clock is significantly lower as well. Depending on the specific model, the maximum Turbo Boost is 896 MHz or less and the memory controller can support DDR3(L), DDR3L-RS or LPDDR3 in single- or dual-channel operation.

The fastest notebook models equipped with this GPU nearly match the performance of the HD Graphics (Sandy Bridge) and the AMD Radeon HD 6310. Therefore, older and less demanding Windows games like World of Warcraft or Half-Life 2 can be played fluently in very low settings. Performance is not sufficient for most modern titles.

Compared to competing ARM SoCs, Bay Trail outperforms the Adreno 320 (which can be found in different Qualcomm SoCs like the Snapdragon 600) and offers a performance similar to the Tegra 4 GPU. That is enough power to run even the most demanding Android games smoothly in very high resolutions as of 2013.

The integrated video decoder supports all popular codecs such as MPEG2, H.264, VC1, VP8 and MVC and is suitable for resolutions up to 4K up to 100 Mbit/s. The user can connect up to two displays via HDMI 1.4 (max. 1920 x 1080) or DisplayPort 1.2 (max. 2560 x 1600). Another new feature is the support for Wireless Display and Quick Sync, Intel's fast and power efficient H-264 hardware encoder. Some of these features are not available on all models.

Depending on the model, the power consumption of the entire SoC is somewhere between 2 and 10 W. The most efficient versions are therefore suitable for passively cooled tablets, whereas faster variants are used in larger subnotebooks with active cooling.

Intel HD Graphics 5000

The Intel HD Graphics 5000 (GT3) is a processor graphics card included in some of the ULV Haswell processors of 2013 (e.g. Core i5-4250U). It represents the highest configuration of the integrated graphics unit (40 execution units), but lacks the additional eDRAM cache of the Iris Pro Graphics 5200. The relatively low base clock can be automatically overclocked using Turbo Boost technology. Depending on the processor model, the base and turbo clock rates may differ, resulting in varying graphics performance between CPU models.

In comparison to the HD 4000, the graphics core has been modified extensively. The GPU supports DirectX 11.1, OpenCL 1.2 and OpenGL 4.0. It also features an improved decoder for 4K videos and the fast Quick Sync encoder.

According to benchmarks in 3DMark 11, the HD Graphics 5000 is up to 50 percent faster than the previous HD 4000. In games, however, the performance advantage is significantly lower. With simultaneous load on the CPU, the low TDP of the ULV models (15 W) limits the Turbo Boost of the GPU. Although the HD 5000 features 40 Execution Units, the graphics performance is still somewhat below the HD Graphics 4600 with only 20 EUs (HD 4000: 16 EUs). Overall, the HD 5000 is thus just behind the AMD Radeon HD 7660G and at the level of a dedicated Radeon HD 7650M. Current games (as of 2013) will run fluently in low to medium-low settings.

Due to the 22nm 3D Tri-Gate production process, power consumption is relatively low. The HD Graphics 5000 is available on certain ULV dual-core Haswell models in the 15 Watt TDP range.