Page 2 - NVIDIA 8800GTS Architecture
The shiny G92 on the 8800GTS that reflects my camera... yes, just like last time. Anyone want to get me a dSLR?
Manufacturer: NVIDIA
Series: GeForce 8
GPU Core, Process: G92, 65nm
Interface: PCI-E 2.0 (Backwards compatible)
Core Clock (Stock): 650 MHz
Shader Clock (Stock): 1625 MHz
Memory Clock (Stock): 970 MHz (1940 DDR)
Memory Bandwidth: 62.08 GB/sec
Pixel Fill Rate: 10400 MPixels/sec
Texture Fill Rate: 41600 MTexels/sec
Memory Type: GDDR3
Memory Bus Type: 64x4 (256 bit)
DirectX Compliance: 10.0
OpenGL Compliance: 2.1
PS/VS Version: 4.0/4.0
Shader Processors: 128
Pipeline Layout: Scalar MADD+MUL
Texture Units: 64
Raster Operators: 16
Data above courtesy of GPUReview.com.
The heart of the new 8800GTS is NVIDIA's new G92 core. This is where the truth comes in -- as we wrote in our previous 8800GT review, it doesn't hurt to 'borrow' some of the same set of information, right? I mean, I wrote it, and the same information applies... there's little change between the 8800GTS G92 and 8800GT in terms of architectural specifications.
Anyways, with a funky naming scheme for revised G92 cards, the whole 8800GTS-8800GT-8800GTS models would instantly confuse users who are not watching this industry closely. The G92 die shrink is quite a bit faster -- and more power efficient -- than NVIDIA's 90nm G80 used in the previous generation 8800 series cards. Relatively speaking, the G92 based 8800GT is designed to outperform the previous generation 8800GTS G80; in which the G92 based 8800GTS (Typically configured with 512MB RAM) is literally a better performer than the G80 8800GTX in controlled situations. Of course, NVIDIA doesn't want the 8800GTS G92 to completely outperform their once flagship card at stock levels; no core revisions are planned for the 8800GTX and 8800 Ultra. We'll examine this section in our benchmarking section shortly.
Architecturally speaking, the package NVIDIA puts forth in the 8800GTS graphics card is not simply a die shrink that somehow gained tens of percents in performance gain. Being that, 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 8800GTS has it increased to 128 with 64 texture address units and 64 texture units (1:1 ratio). The stream processors are also clocked high at 1625MHz on the 8800GTS G92 (The Asus 8800GTS TOP's shader is at 1.8GHz, but most users report that the shaders on the 8800GTS G92 overclocks really well) that challenges the speed of the 8800 Ultra in this respect. 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.
Thanks to the 65nm process, the due surface shrinks down approximately 33% to 324 mm² from 484 mm² even with much more transistors on board. With that taken into account, there's still a reduction in SPs, texturing power, and not as many ROPs as its higher G80 relatives, but still more shader and texture horsepower than the 8800GTS G80. Anyway, the ROPs are cut down to 16 like the 8800GT as the previous generation 8800GTS actually had 20 on board. The NVIDIA 8800GTS has a texture fill rate of 41600 MTexels/sec.
On the other hand, compression algorithm of the ROPs has been improved for insanely high resolutions with the new G92. This may not mean much in terms of 3D graphics acceleration for gaming purposes, simply due to the fact that the 8800GTS is not powerful enough to run most games at 30" LCD monitor resolutions anyway. The 8800GTS G92 in SLI may benefit, but other than that -- this architectural enhancement may mean less to the end user.
Additional features added back onto the 8800GTS 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 articles with NVIDIA PureVideo 2 performance on G84 based video cards such as the 8500 and 8600 series. PCIe 2.0 support is implemented as well; despite that 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.
One step back from the previous generation cards is its 256 bit memory interface with a memory bandwidth of 62.08 GB/s as compared to the 8800GTS G80's 320 bit memory interface offering up to 64GB/s. It's still higher than the 8800GT at 57.6 GB/s.
The Asus 8800GTS TOP clocks the core at 740MHz from stock 650MHz; an increase of nearly 100MHz. The memory runs at 1035MHz actual clock -- while it shows that the Qimonda memory used on the card runs at stock 1000MHz effective according to the memory specifications (Discussed on the next page), it's quite impressive that the core runs at 740MHz out of the box. Being that, the Asus 8800GTS TOP is literally one of the fastest variants of the 8800GTS on the market today -- there's only one card at press time that has higher clocked memory at 1050MHz, but with a slower core!
Page Index
1. Introduction, Specifications, Bundle
2. NVIDIA 8800GTS 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
