AMD Phenom X4 945 vs 955 Black Edition

After the presentation of the new socket AM3 in February was not the fastest processors from AMD on the stage, it is the end of April, the so-called "Dragon Update". This provides two main new processors with 3.0 and 3.2 GHz, AMD Phenom X4 names of the 945 and 955 Black Edition. In addition, the AMD OverDrive software on the 3.0 version status brought and fitted with many new features. For while the clock is apparently increasing at 3.2 GHz on the paper to pull any trees, should the software with special sections for the main memory and a type of Turbo-mode for the Phenom additional space required for additional performance. The Phenom X4 945 and 955 area is in the technology in January presented model Phenom X4 940 BE almost all the same. Two MB of L2 cache are also one of the party, an additional L3 cache with a size of six MB of course. The only difference is in the socket AM3, which brings the two new models with DDR3 support, while the first model still had to resort to DDR2. Since we have already the first AM3 model have shown, however, that the space on a nearly identical motherboard to the Phenom X4 Sun care is pretty , are also with the new models in this area not expect miracles. Rather, it is directed to the 200 MHz to 3.2 GHz consequently increased stroke of the new flagship Phenom X4 955 its free multiplier.

Since we are using the new socket AM3 , the different storage modes and the benefits of DDR3 memory and all other conditions already at the start of the AM3 platform with the Phenom X4 810 have explained, we now want the two new subjects speak for themselves only. We had our existing test system to the well-deserved retirement sent, including AMD, the new test system will also benefit. Both have seen in the last year a total of 36 processors, 21 of them from Intel, 15 of AMD. Since there have been over the months in which we undergo such a test system does not change, many changes in the market, it was not only given the new processor generation time for a change. Of course, an entire generation provides for the replacement of the entire system to always best, so we postponed the plans from the former early summer to autumn. We have resorted to many suggestions from readers, we have received in recent months. At this point we want to make again some explanations of the new test system and explain why we have chosen for these components. One of the largest and most often attached to criticism was certainly the older Antec PSU with 480 watts. This is sent in a well-earned retirement and replaced with a brand new power supply from Cooler Master. For our test, we use the smallest version with 700 watts, as stronger power supplies fall, especially under light load efficiency in some cases significantly. If the 20 percent threshold in the 700-watt subjects at 140 watts, it is screwed up at the 1,100-watt model is already at 220 watts - a value with the new test system for smaller processors, not even under full load prime achieved.

Another important point is the new graphics card. After we complete last year not just low-power ATI Radeon HD 2900 XT in action had it, we rely on Nvidia's first (relatively low-power) 55-nm solution for high-end segment, the GeForce 9800 GTX +. In particular, the significantly lower absorption line ( sh. in the article by HT4U ), even from competitors in the form of the ATI HD 4850 and the consistently faster performance were important reasons for the selection. All other eligible models (HD 4870, GTX 260, GTX 280) certainly provide a higher performance, but only at disproportionate increase in power consumption. To convene the innovative triple channel RAM just to be on the software side was approximately essential of transitioning to Windows Vista 64-bit version. Even this step we took the new test system, so that hereafter, the three slots on the X58 boards with two gigabytes are equipped - all the additional Intel boards rely on the dual-channel mode on two of these modules. The disparity of 4 to 6 GB of course on paper there is, in benchmarks but not realizing quasi. We also consider that such situations should be in the near future, the most ordinary variants or in the form of DDR2 memory with the old-established Core 2 Duo and Core 2 Quad previously are. Therefore, all (older) AMD processors also come with a potent DDR2 memory, therefore, the maximum 1,066 MHz makes. Since till now only the Phenom and Athlon benefit from this memory, the older models work with a lower clock, but with improved timing. For Socket AM3 us is analogous to a circuit board from Asus is available, which is based on the 790FX and the SB750. Furthermore, it is also a quasi-identical spin for the socket AM2, which is logical to DDR2.

In totaling to the recently premeditated from the ground test system, we have also carried the benchmark course an update. Still, some hypothetical tests of the party, as they habitually are one step further than any real application. There are new features that have originated a place in processors, first tested in their performance. Since these theoretical tests make use of the user on the PC but not much, we are also numerous sensible applications for help. These are now the most important selection of multimedia applications, but also everyday things like packaging of files. Rounding out the test a hardly any games where we were looking for a selection of games with built-in benchmark capability, and play out, which means Fraps not only the average frames per second, but the minima and maxima were indomitable.

As is habitual in the past, each benchmark is executed five times to convene with any outliers in performance can. This happens in almost all applications, regrettably, more often than anticipated, both the negative and the positive direction. The average of the five tests (excluding extreme outliers) is then relocated to the individual diagrams. In games will make use of the resolution of 1,280 x 1,024 pixels, since at higher resolutions roughly only the graphics card plays a role in inferior resolutions but have no relation to certainty. In general, we test it, if not further illustrated, with the highest detail. The values of AA / AF can vary from game to game; though, the dissimilar setups are a small simulation of what may now vary with each new game. All tested processors have to confirm with the graphics card these tests that they can manage well with a few quality-enhancing features as well as a maximum.