Page 2 - NVIDIA 9600GT Architecture
Series: GeForce 9
Release Date: 2008-02-21
Interface: PCI-E 2.0 x16
Core Clock (Stock): 650 MHz
Shader Clock (Stock): 1625 MHz
Memory Clock (Stock): 900 MHz (1800 DDR Effective)
Memory Bandwidth: 57.6 GB/sec
FLOPS: 208 GFLOPS
Pixel Fill Rate: 10400 MPixels/sec
Texture Fill Rate: 20800 MTexels/sec
Max Power Draw: 95 W
Framebuffer: 512, 1024 MB
Memory Type: GDDR3
Memory Bus Type: 64x4 (256 bit)
DirectX Compliance: 10.0
OpenGL Compliance: 2.1
PS/VS Version: 4.0/4.0
Process: 65 nm
Shader Processors: 64
Pipeline Layout: Scalar MADD+MUL
Texture Units: 32
Raster Operators: 16
Data above courtesy of GPUReview.com.
Under the hood is not the earth shattering, ever so popular G92 core used in the enthusiast favorite 8800GT and 8800GTS 512MB (Heck, I have three of those cards myself) -- but its little brotha with a bigga numba, er, bigger brother with a bigger number, the NVIDIA G94 core. As we've mentioned in our 8800GT and 8800GTS review, the package NVIDIA puts forth architecturally in the G92 graphics card core is not simply a die shrink that somehow gained tens of percentages in performance gain. The G92 is really a derivative of the 90nm G80 core, rather than the 80nm based G84 used on the 8600 series. At 65nm, the new core implements a good chunk of complexity in terms of transistors -- integrating 754 million as juxtaposed to G80's 681 million transistors. The G80 based 8800GTS has 96 active stream processors, whereas the 8800GT has it increased to 112 with 56 texture address units and 56 texture units (1:1 ratio). The stream processors are also clocked high at 1.5GHz that clocks as high as the 8800 Ultra. Keep in mind, however, the 8800GTX based off the G80 core still has 128 stream processors as the 8800GTS G80 has 32 of those disabled.
When it comes to the G94, however, the G94 has only 505 million transistors -- whereas the G92 had 754 stream processors as mentioned earlier. The surface area is also reduced to 225 mm², from 324mm². That's still more than the 3870, since the AMD 3800 series are built on the 55nm fabrication process while the G94 core, like G92, is 65nm. One of the biggest changes officially is the number of stream processors: the G94 has 64, while the 8800GT has 112 and 8800GTS 512MB has 128 as mentioned in the previous paragraph -- essentially halved (Which explains the lower transistor count). To compensate, the shader clock is much higher on the G94 based 9600GT for optimization purposes, but even at that the horsepower at the wheels are still significantly lower than the G92 based 8800GT.
There are no changes in the compression algorithm of the ROPs, which has already been improved in the NVIDIA G92 revision. Additional features on the 9600GT, like the 8800GT, is the native support of dual HDCP dual link DVI output, instead of using a separate display chip. NVIDIA PureVideo 2 hardware video decoding replaces NVIDIA PureVideo on the previous generation G80 cores; which has demonstrated superior hardware decoding performance as compared -- results are seen floating around in articles with NVIDIA PureVideo 2 performance on G84 based video cards such as the 8500 and 8600 series. PCIe 2.0 support is implemented into the G94, like the G92; despite how this will have no real definition with regards to performance even on PCIe 2.0 boards that offers additional bandwidth, this evolutionary step is also fully backwards compatible. The only change is DisplayPort support --- but the connector implementation is up to the board partner and not required by NVIDIA themselves.
Like the G92 based 8800GT and 8800GTS, the 256 bit memory interface with a memory bandwidth of 57.6GB/s at stock -- which is identical to the 8800GT. Generally speaking, we can see that the 9600GT is fairly crippled when it comes to actual power -- many of which uses high clock speeds for performance compensation.
Out of the box, the Asus 9600GT TOP has it overclocked to 740MHz core (From 650MHz), 1800MHz shader (From 1625MHz), and 1000MHz memory (From 900MHz specs, but the chips used on the 9600GT TOP are stock at 1000MHz -- more on this shortly). This makes the Asus 9600GT TOP again one of the fastest factory clocked cards at press time.
1. Introduction, Specifications, Bundle
2. NVIDIA 9600GT Architecture
3. A Closer Look, Test System
4. Benchmark: FEAR
5. Benchmark: Prey
6. Benchmark: Half Life 2: Lost Coast
7. Benchmark: CS:Source HDR
8. Benchmark: 3DMark06
9. Power Usage, Overclocking
10. Noise Factor and Conclusion