nVIDIA uses MCM technology to package the Geforce4 410 Go core and 2 or 4 memory particles on a substrate. At that time, the MCM packaged display chip was aimed at mobile devices.
And Quadro4 700 XGL uses P80 form factor, Quadro4 900 XGL uses P83 form factor, Quadro4 550 XGL uses similar form factor to P71, but uses different video memory. AGP 4X.
4.21 Quadro4 380 XGL
Model: P112
PCB Code:180-10112-0000-A01
Chip model:NV18GL-A1
Released on November 12, 2002. Chip date is 0227 weeks, PCB date is 0229 weeks.
After repeated delays, nVIDIA released the GeForce FX5800/5800 Ultra graphics cards on November 18, 2002, and the actual release date was in March of the following year. The NV30 chip is made with 130 nm transistors, operates at a frequency between 400 and 500 MHz, and can access 128MB GDDR2 via a 128-bit bus running at 800 MHz or 1000 MHz.
One of the reasons why GeForce FX uses "FX" instead of the number "5" is that many former engineers of 3dfx have participated in the development of GeForce FX. Due to the incomplete realization of the innovative design, and the imperfect GDDR2 memory. This causes the GeForce FX5800/FX5800 Ultra graphics cards to generate a lot of heat. The performance barely surpasses the Radeon 9700 Pro (R300 series) that made the limelight in 2002.
On March 6, 2003, ATi released the Radeon 9800 Pro. It is a version that optimizes the Radeon 9700 Pro, increases the frequency, and can well support GDDR2. Its performance exceeds the GeForce FX5800 Ultra graphics card.
Radeon 9800 Pro caught nVIDIA by surprise. I had to give up the immature GDDR2 video memory for the time being. On May 12, 2003, it released the GeForce FX5900 Ultra, which uses GDDR1 video memory, and has improved graphics quality, heat dissipation, and power consumption. This enables GeForce FX5900 Ultra/FX5950 Ultra to compete with Radeon 9800 Pro/9800 XT graphics cards.
On June 13, 2003, Xabre Graphics (XGI), a newly established subsidiary of SiS, acquired Trident’s graphics chip development business at an undisclosed price. On September 15, 2003, XGI announced the Volari V3/V5/V8 chip graphics card and the Volari Duo V8 Ultra graphics card with dual display chips. Among them, the Volari V3 graphics card is the Trident XP5 graphics card, and the Mobile Edition is called Volari XP5. The performance of Volari Duo V8 Ultra graphics card is slightly higher than that of GeForce FX 5700 Ultra, and sales are less. Because of its powerful appearance created by its dual display core, it is now the darling of collectors.
The second product in VIA’s Chrome product line is DeltaChrome S8, which was not developed smoothly. S3 designed the DeltaChrome S8 evaluation board in August 2003, and it was not officially released until July 28, 2004. It supports DirectX 9a, and its 3D processing capability is close to that of the mid-level graphics card FX 5700Ultra at that time.
On July 23, 2002, Intel supported AWG (Arapahoe Work Group) officially released the PCI Express computer bus, referred to as PCI-E, and officially referred to as PCIe. Due to the close cooperation between ATi and Intel, it was the first to showcase the industry's first graphics card using PCI Express at IDF2003 on September 19, 2003 (week 0338). The specific model of this card was not announced. The chip shown in the picture is labeled "215S8KAGA11F", which has integrated PCI-E. It is an engineering sample chip of RV380 (X600). It can be seen that the first graphics card in the industry that uses PCI Express is Radeon X600, and the PCB date is 0330 weeks.
Since the publication of the PCI Express interface standard, nVIDIA's R&D pace has lagged. The solution is to design the AGP-to-PCI Express bridge chip BR02. BR02 stands for Bridge Revision 02, also known as High Speed Interconnect (HSI). It can convert AGP signals to PCI-E signals, and it can also convert PCI-E signals to AGP signals. On February 17, 2004 (0408 week), nVIDIA released the GeForce PCX series of graphics cards using the BR02 chip. The BR02-A1 ES chip in the picture is dated 0345 weeks. It seems to feel the huge pressure of ATi and accelerate the finalization. The BR02-A2 ES chip has a date of 0407 weeks and is finally finalized in A2.
GeForce PCX is mainly based on the unsuccessful old product GeForce FX series, and its conversion performance will be slightly lost, so the bridge chip has been criticized. ATi also said: "Where is there a way to build a bridge". In the period of GeForce 6 and GeForce 7, the advantage of the BR02 chip finally emerged, that is, only one interface chip needs to be developed. ATi chips are all native, that is, a single chip launches native PCI-E and native AGP versions at the same time. Finally, due to cost pressures, the Rialto unidirectional bridge chip that converts PCI-E signals to AGP was also developed.
The embarrassing situation of nVIDIA's lagging behind PCI-E interface graphics cards in desktop computers prompted it to cooperate with many notebook computer manufacturers to announce the mobile PCI Express Module (MXM) specifications for notebook computers on May 17, 2004. The full name of MXM technology is Mobile PCI eXpress Module.
nVIDIA also released the entry-level and mid-range versions of GeForce FX: mid-range FX5600 (NV31), FX5700 (NV36) and entry-level FX5200/5500/5100 (NV34), as well as the corresponding mobile version.
There are two types of Rev A00 from GeForce FX, T01 using GDDR1 video memory and P126 using GDDR2. There is no difference between T01 and P126 from the front, except that the capacitor used by T01 is slightly smaller.
5.1.1.1 T01
PCB Code:180-10T01-0000-A00
At the beginning of August 2002, the first NV30 silicon taped-out. It takes about 4 weeks (0236 weeks) to package the chips, and 4 weeks (0240 weeks) to test the chips. The PCB date of this T01 is 0240 weeks, which should be the earliest version. The NV30 chip on the PCB is the first packaged chip, and the chip is marked as follows:
Ibiden Daisy Chain（Ibiden Co Ltd ,Daisy Chain Topology）
T01 is a chip evaluation board for NV30 chip, using mature and reliable MBGA package GDDR1 video memory. After the verification is feasible, directly copy the P126 product evaluation board/reference board using GDDR2 video memory.
The cooling system is the prototype of GeForce FX5800 (QUADRO 2000). The fan power cable is in the lower left corner, and the length is not enough to plug into the upper right socket. The lower left socket is blocked by the heat sink. It was pasted with glue on the lower left fan socket 3 cm to the left. Suspected exhibit samples.
Samsung developed the industry's first 1Gbps 128Mb GDDR2 SDRAM in July 2002. nVIDIA GeForce FX is the first graphics card to use GDDR2. On November 18, 2002 (0247 week), COMDEX2002 released GeForce FX. The promotional picture of GeForce FX at that time is the following picture. The Samsung graphics memory particles on the engineering sample card of the picture are the ES version and the date is 0238 weeks. The PCB date should be 0240 or 0241 weeks.
The video memory particle date is 0241 weeks, which is the official version. The PCB date is 0242 weeks, the NV30 chip date is 0243 weeks, and the version is A1. Although the chip is marked as NV30GL, it is not distinguished as QUADRO and GeForce FX. When the NV30 chip is the A2 version, it is marked as GeForce FX to indicate formal mass production, and marked as QUADRO 2000 is still in the engineering sample stage.
This heat dissipation system is optional for GeForce FX. Due to its low heat dissipation capacity, it is mainly used for the heat dissipation of low-end graphics cards.
The chip date is 0244 weeks, and the PCB date is 0244 weeks.
This slightly lower cost turbo radiator (nVIDIA's FX Flow cooling technology) was selected and finally adopted.
This revised version has been revised and officially finalized. This card is still undergoing important tests on GDDR2 video memory and even the chip. It can be seen that until the final GeForce FX5800 Ultra is finalized, there are still many problems in the hardware, but time is not waiting, so we can only bet on the driver. The final engineering sample PCB date is 0301 weeks.
On February 17, 2004, nVIDIA released GeForce PCX series graphics cards: GeForce PCX5950 (NV35), GeForce PCX5750 (NV36), GeForce PCX5300 (NV34) and GeForce PCX4300 (GeForce4 MX 4000), which are essentially AGPs with bridge chip BR02 Card.
nVIDIA later changed the name of GeForce PCX5950 to Geforce PCX5900, but there are still manufacturers using this form factor to make the so-called GeForce PCX5950 graphics card.
MCM (Multi-Chip Module) is a packaging and packaging technology for bare dies, chips, and integrated circuits. This packaging technology can encapsulate two or more dies on one substrate. Usually, only one display core die is packaged in the display chip. As mentioned earlier, nVIDIA GeForce4 Go once packaged the display core and 2 or 4 video memories on a base to save space and reduce energy consumption. It is mostly used on mobile chips. Entering the PCI-E era, nVIDIA’s BR02 bridge chip is packaged on the same base with the display chip and video memory through MCM packaging.
5.12.2 E233
PCB Code:180-10233-0000-A00
PCB date is 0352 weeks. The BR02 chip is packaged on the same base as the graphics chip and video memory.
