The new NVIDIA's strategy of aggressive promotion of the chipsets into the market, adjustment of the price policy to the market realities and no more focusing on integrated graphics started yielding the fruit. At least, the launch of the first nForce wasn't followed by such a massive attack as bringing into the market mainboards from six brandname manufacturers. Note that none of the boards is equipped with integrated graphics (!), i.e. the success of the chipset depends on the market reaction to a classical, nonintegrated, solution. Sooner or later we will see boards with a north bridge coming as an IGP with a GeForce4 MX based graphics core, but the first impression is the most important, and it will be formed now, when the boards without video are the only available on the market. 

The nForce2 chipset was already examined some time ago. But I'd like to attract you attention to the companies chosen by NVIDIA as partners for launching the new chipset. Here they are: ABIT, ASUSTeK, Chaintech, EPoX, Leadtek, MSI. What do they have in common? 

Notability. All of them are famous players on the market and belong to the league ranking the top leading board makers. All of them? And Leadtek? Reportedly, mainboards for this company, which is known mostly for its video cards, are produced by EPoX. Thus, NVIDIA gives one more chance to its favorite to make more money with its chips. :) 

Experience in production of "cool" boards. Although not everyone stuffed its solution with as many features as possible, there is no free space on textolite of all the boards. And if we compare characteristics of the budget nForce2 based boards and expensive models on other chipsets, it will be hard to say who is going to take the palm. 

Orientation toward enthusiasts and overclockers. MSI is more universal in this respect, but it's also the first who releases products for overclocking. The other names (except maybe Leadtek) are well known to any enthusiastic assembler and overclocker. 

So, the image of the nForce2 projected by NVIDIA is notability, a feature-laden design, enthusiasm and overclocking. The chipset is not for everybody - it's for those who know what's what in a computer case, loves everything modern, fast and advanced. Well, if the image is backed by the objective test data, it will be the most perfect outcome. 

Test conditions

Testbed: 

• Processor: AMD Athlon XP 2700+ (13x166=2167 MHz), Socket 462(Socket A) 
• Mainboards on the NVIDIA nForce2: 
• ABIT NF7-S (BIOS 10) 
• ASUS A7N8X (BIOS 1001C) 
• Chaintech 7NJS (BIOS of 01.11.2002) 
• EPoX 8RDA+ (BIOS of 28.10.2002) 
• Leadtek K7NCR18D-Pro (BIOS of 06.11.2002) 
• MSI K7N2 (BIOS 3.0B2) 
• Mainboard on the VIA KT400: 
• Iwill KK400 (BIOS of 15.10.2002) 
• Memory: 2x256 MB PC3200(DDR400) DDR SDRAM DIMM Kingmax, CL 2.5 
• Video card: Palit Daytona GeForce4 Ti 4600 
• Hard drive: IBM IC35L040AVER07-0, 7200 rpm 

Software: 

• OS and drivers: 
• Windows XP Professional SP1 
• DirectX 8.1b 
• NVIDIA nForce UDP 2.0 
• VIA 4-in-1 4.43 
• NVIDIA Detonator v40.72 WHQL (VSync=Off) 
• Test applications: 
• Cachemem 2.4 MMX 
• BAPCo & MadOnion SYSmark 2002 (Internet Content Creation + Office Productivity) 
• MadOnion 3DMark 2001 SE build 330 
• Gray Matter Studios & Nerve Software Return to Castle Wolfenstein v1.1 

Would you now take a look at the characteristics of the today's contestants:

Test results

To get rid of useless information you will be given only the most peculiar, interesting and popular results: games, business, graphics and multimedia software and a low-levl memory test. All the other results do not contradict to what you can see on the diagrams. If you consider some test data left overboard to be very important, just contact us and we will take that into account next time. 

Games

So, in the low resolution the memory controller of the nForce2 is ahead, as well as in the high one, though it doesn't make much sense because the chipset and processor get less and less work to do, while a video card receives greater loads. The two tests match well: in 3DMark 2001 the nForce2 outscores the KT400 by 9%, and in Return to Castle Wolfenstein (800x600x16) by 13%, but remember that in this case the test resolution is lower. Well, when there is something that depends on a chipset, the new NVIDIA's solution performs excellently. 

Business applications, graphics and sound

...Well, this is a good old SYSmark 2002. There is not much that needs comments: like before, the leading position of the nForce2 is not doubtful. If we compare the best and worse results (the latter belongs to the KT400) we will see that the nForce2 goes ahead by 10% in the business software and by 12% in the programs working with multimedia content and graphics. It's interesting that the gain of the new model is hovering around 10%. The nForce2 based boards demonstrate almost equal performance (the deviation of 3% is on account of the system error of the SYSmark). 

Low-level memory test

Another good old lad is Cachemem. That's right that this test runs under the virgin DOS - it's impossible to eliminate influence of the multitask and very complicated operating system in the low-level extreme tests. All the boards on the nForce2 show very stable operation. As a rule, Cachemem shows difference of about 3-4% even if it's very smooth in the high-level software; in this case it's also very smooth even at such level. It means that the efficiency of all the boards is almost identical. The unsophisticated algorithm of the test let the legendary DASP show its might and it demonstrated the stunning excellence of the nForce2 in the memory read and write speeds. In such simple but theoretical algorithms the read speed can be better by 34% and the record one by 26%, but remember that in the real software the figures are much more modest. 

Summary

So, the most popular tests are completed. But what for did we do such a routine work? 

First of all, contrary to the last article on nForce2, today we are dealing with 6 models instead of one, and all of them are produced by the best manufacturers. Besides, since the above mentioned review we have gotten new drivers and updated BIOSes, collected statistical data etc. Maybe it makes sense to think the performance of the NVIDIA nForce2 over? Let's try. 

Secondly, it's interesting to compare the products. The boards are different, someone could have bungled and released a very slow solution. Sometimes it happens. Well, let's start with the first item. 

Performance of the NVIDIA nForce2 chipset

Let me remind you what we said about the nForce2 last time "...The NVIDIA nForce2 performed so well that together with its functional characteristics it becomes the most appropriate thing to buy, especially for newly announced and future AMD processors (such a blessed picture can be spoiled only by a too high price)". 

Taking the today's scores into account I have to add the following: performance of the nForce2 is stably higher compared to the flagship of VIA Technologies, sometimes even by over 10%. Impressive. What about the prices? High performance is not free, and the nForce2 based boards are more expensive than those based on its competitor, and they are pricier by more than 10%; i.e. the price difference is greater than the performance gap. But it was always so! In the upper market sector each added percent in performance costs sometimes dozens of percents of the price increase. That is why, the conclusion is that the nForce2 is a quite attractive solution in terms of functionality, efficiency and price for those who consider the first two parameters much more important than the third one. 

Comparative performance of the NVIDIA nForce2 based boards

Contrary to usual, one of the diagrams is taken not into the section of the comments, but in the completing part. The performance of the VIA KT400 (represented by the Iwill KK400 board) is taken as 100%, the other columns express the difference in performance of the other boards regarding this etalon. First of all, let me explain how we obtained the results. If we have to compare the number of crocodiles in the Nile with the distance to Alpha of Centaurus :) - the quantities are different, including the order (some can be measured in units, other in millions), - there is only the geometrical mean that can help, - this is a root of the Nth degree of a product of N selected results. This is how we obtained the figures for the general performance - this is a geometrical mean of all the scores from the above diagrams expressed in percents relative to the reference system. 

The summary diagram shows that all the manufacturers managed to build good boards on the nForce2. Or, they all made the same number of the same errors :). The best score differs from the worst one by 1.25%. That is, they don't differ at all. That is why you should take a close look at the characteristics of the boards in the table above, read the brief descriptions of the models you liked and select the one according to its functions, design, accessories and price, rather than a performance level. 
 
 
 

Write a comment below. No registration needed!