3DMark 2005Pro: Testing 40 cards.

Starting with Times 3DMark 2001, Futuremark tests have become a standard video card performance tool for all manufacturers of similar products. At least, only these tests can be seen on the screens of test laboratories of companies, and only on these tests is performed on the finished products on the factories. Once I was very surprised to see SpecViewPerf on one of the LEDTEK test laboratory screens – it turned out that they were tested by a new quadro 🙂

You can scandal, indignant, ignore and curse, but the fact remains a fact – it is the numbers received on Futuremark test packages, such as 3DMark2001Se and 3DMark2003, until recently served as a guide for manufacturers and necessarily appear in the documentation and even often applied to the scheduled graphics cards. This industrial standard and it is impossible to foreign. Will the current novelty be?

Of course, yes, moreover, most likely, he will replace the morally outdated 3DMark2001se.

2001Se was a benchmark, measuring the performance of the entire system as a whole. So, for example, the result of the first game test (Car Hi + Car Low) – very little depended on the overclocking of one video card, and the serious increase in it could only be obtained along with the CPU acceleration. Dragothic and Nature – tests, indeed, more dependent on the video card, and LOBBY (High + Low) – depended mainly from the subsystem percent + memory.

The reisal in them was directly depended on the frequency of the project, from whether in Synchron or in Asynchron, the memory, and from timings. This explains a rather small increase in 2001SE when overclocking only VGA, without CPU, unlike 2003. That is, summarizing, it can be said that, unlike the 2001Se, 2003, almost completely depends solely from the video card (with the exception of the first test, where the processor-dependence Manifested much more than three others).

And in this 2005 – rather heir 2003 than 2001Se.


As in the previous innovations, 3DMark 2005 consists of a test set – familiar Game Tests, CPU TEST, Feature Test and the new Batch Size Test. What they imagine and what is the alignment of modern video cards in this test we will try to tell in this review. Immediately, it is necessary to note the fact that absolutely all tests from the set require support for PS 2.0, or 3.0. Those.

All huge variety of VGA starting with the GeForce 4 times and below are not supported.

Gaming test. processor

The final result in traditional “brands” is formed from the results of three game tests. If you need to remove this number, everything else can not be labeled.

The formula for which the final result is calculated:

In the previous 3DMark 2001/2003 there were 4 game tests. Now only three, and all of them are implemented on a single engine, only the sets of shaders used, the conditions of illumination and textures of the materials are distinguished.

The first game test Return to Proxycon (Return to Proxikon) offers to see the continuation of the battle on the spacecraft from the 2003 version (Battle of Proxycon). The cargo ship is attacked by nonlabo-equipped fraternity and with honor tolerate defeat. The spectacle is impressive and the first time the 20th looks without disgust.

The internal space of the ship is an extremely detailed demoscene with eight sources of lighting.

Second game test Firefly Forest Direct Heir to Mother Nature. The action takes place in the night magic forest and has two lighting sources – moonlight in combination with a soaring firefly. Picture of the maximum photorealist.

The most difficult test of all three.

And the last test – Canyon Flight. The flying ship passes through the aqueous canyon and enters the unequal battle with the monstrous monster. By complexity, the test is inferior to Firefly Forest, which will be clearly visible in testing (just one source of lighting – the sun), but the realization of the water space is one of the best in the graphics industry from all created.

Nevertheless, in the implementation of the physical model of water there are bugs. For example, as when the monster is emerging and falling, there are beautiful splashes and water dust, but absolutely there is no excitement of waves, which is simply impossible when falling like a similar body into water. Waves should be such that they would cover the flying ship along with Botatman and Captain.

Well, yes forgive the developers 😉

CPU test.

For the first time the CPU test in the 3DMARK package appeared in the previous version – 3DMark2003. Now testing is carried out on the basis of the first and third gaming demoscene in the resolution of 640×480. The developers came simply and original – put the execution of vertex shaders to the central processor which is absolutely not ready to perform such tasks, thereby loading it so much that the influence of any video card in such conditions is really minimal.

The formula for which is calculated the final result of the CPU TEST:

Availability and accounting of Hyperthrea

Leave a Reply

Your email address will not be published.