Performances: Synthetic
After all these presentations, let us pass to most interesting: tests! We chose here new protocol of new tests including some benchmarks that we will present to you a little later. In order to have an objective vision of the situation and a better visibility on the scores, we segmented the results in three parts: Synthetic, Practical and Plays. Let us re-examine last once the configurations used:
Platform Athlon 64 : Athlon 64 3200+ (2.0 Ghz) on Asus K8V (K8T800) - Radeon 9700 Pro
Athlon Platform 64 FX : Athlon 64 Fx-51 (2.2 Ghz) on Asus SK8N (nForce3 Pro 150) - Radeon 9700 Pro
Opteron Platform : Opteron 244 and 246 (1.8 and 2.0 Ghz) on Asus SK8N (nForce3) - Radeon 9700 Pro
Platform Athlon XP : Athlon XP 3000+ and 3200+ (2.1 and 2.2 Ghz) on Asus A7N8X (Ultra nForce2) - R9700 Pro
Duron Platform : Duron 1.6 Ghz @ 2.2 Ghz on Asus A7N8X (Ultra nForce2) - Radeon 9700 Pro
Platform Pentium 4 : Pentium 4 3.0 and 3.2 ' It Ghz on Asus P4C800 (i875P) - Radeon 9700 Pro
Pentium Platform 4 EE : Pentium 4 3.2 Ghz ' Extreme Edition'/W 2M L3 hides on Asus P4C800 (i875P) - R9700 Pro
Xeon Platform : Dual-Xeon 3.06 Ghz 512k and 1M L3 on Asus PC-dl (i875P) - Radeon 9700 Pro
In these configurations, only Duron is overclocké to 2.2 Ghz. All the configurations function with 512 Mo of memory DDR400 Cas2.5-2-2 (except Opteron and Athlon 64 FX in DDR333). In the case or alone of the modules of 512 Mo was available, the memory usable by Windows was attached via the parameter "/MAXMEM=512 ". Thus let us begin our synthetic tests with a good old man Sandra Max3! in mode CPU Benchmark:
ha TÅLAMOD om det inte funkar direkt så låt sidan vänta i någon minut lr låt sidan "vila" i 5 min ... av någon konstig anledning fungerar dett lysnade för mig
SORRY men ni MÅSTEHÖGERKLICKA OCH KLICKA PÅ "VISA BILDEN" på menyn som poppar upp
http://www.x86-secret.com/pics/cpu/k8-2/bench-sandra.gif
Indexed quasi-uniquement on the frequency, Sandra gives here a disappointing result for range AMD64. Whereas Xeon are largely at the head of with the optimization of the code of Sandra for the SMP, Athlon 64 Fx-51 obtains here a result close to Athlon XP 3200+. Very logical since the two processors are given rhythm to 2.2 Ghz and have a core of identical execution. On this benchmark which calls neither upon the memory, nor with the mask, the scores are thus logically very near. With the identical one, one notes that Pentium 4 and Pentium 4 EE and its 2 Mo of L3 obtain a result strictly identical for the same reason. Now let us pass to the Multimedia test same benchmark:
SORRY men ni MÅSTEHÖGERKLICKA OCH KLICKA PÅ "VISA BILDEN" på menyn som poppar upp ---------ha TÅLAMOD om det inte funkar direkt så låt sidan vänta i någon minut lr låt sidan "vila" i 5 min ... av någon konstig anledning fungerar dett lysnade för mig
http://www.x86-secret.com/pics/cpu/k8-2/bench-sandra2.gif
Even classification here where it is noted that all CPUs based on the core of execution of K7 finish with the drag. The conclusions are here strictly identical to benchmark CPU Although interesting to test the performances of the pure core of execution, this test does not remain about it less one synthetic test which does not reflect the true total performances of the processor. Let us finish with Sandra by the traditional test report which, even if we left it in Buffered, gives all the same an interesting indication on the performances memories:
SORRY men ni MÅSTEHÖGERKLICKA OCH KLICKA PÅ "VISA BILD" på menyn som poppar upp-----ha TÅLAMOD om det inte funkar direkt så låt sidan vänta i någon minut lr låt sidan "vila" i 5 min ... av någon konstig anledning fungerar dett lysnade för mig
http://www.x86-secret.com/pics/cpu/k8-2/bench-sandra3.gif
One notes here the profit gotten by an architecture Dual Channel compared to Single channel. If Pentium 4 and Athlon 64 FX are in a handkerchief of pocket with Opteron in light withdrawal, Athlon 64 remains confined with the band-width memory of its single channel of DDR400: 3.2 Go/s . Obviously of a very great effectiveness, the internal controller report of Athlon 64 3200+ offer a band-width near to the theoretical maximum. Xeons just like Athlon XP are attached by their FSB although they lay out that a largely oversize memory for their needs. Let us pass now to more interesting, a benchmark recognized that the architecture of Athlon 64 affectionne particularly: SuperPI:
SORRY men ni MÅSTEHÖGERKLICKA OCH KLICKA PÅ "VISA BILD" på menyn som poppar upp---------ha TÅLAMOD om det inte funkar direkt så låt sidan vänta i någon minut lr låt sidan "vila" i 5 min ... av någon konstig anledning fungerar dett lysnade för mig
http://www.x86-secret.com/pics/cpu/k8-2/bench-superpi.gif
In spite of the 1000 MHz of delay, Athlon Fx-51 carries out here the dance, follow-up of very ready by Pentium 4 3.2 Ghz ' EE' then by its traditional version. Hardly one second behind, one finds here Athlon 64 3200+ which offers performances finally worthy of its P-Rating. If Athlons XP remain with the drag, it is necessary here to greet the performances of Athlon 64 3200+ which arrives almost has to rise on the level of Pentium 4 3.2 Ghz. In the same way, the excellent performances of Athlon 64 Fx-51 and Pentium 4 3.2 Ghz ' EE' have in competition only their respective prices. Before passing to the serious things with the practical tests, let us finish with PCMark 2002 which summarizes well the general state of the synthetic benchmarks:
SORRY men ni MÅSTEHÖGERKLICKA OCH KLICKA PÅ "VISA BILD" på menyn som poppar upp---------ha TÅLAMOD om det inte funkar direkt så låt sidan vänta i någon minut lr låt sidan "vila" i 5 min ... av någon konstig anledning fungerar dett lysnade för mig
http://www.x86-secret.com/pics/cpu/k8-2/bench-pcmark.gif
f Athlon 64 3200+ remainder handicapped by its memory on only one channel, the performances in pure calculation are very correct, although a little in on this side Pentium 4 3.2 Ghz. The palm in the very top-of-the-range cost alternatively in Athlon 64 FX or Pentium 4 3.2 Ghz ' Extreme Edition '. In PCMark 2002, one can notice the bug on the test report probably due to the 2 Mo of additional mask.
Översättningen är från franska till egnelska och är gjort av altavistas script babelfish så klaga inte på mig.
Om någon har en server därt jag kan llägga upp bilderna så är jag tacksam då den nuvarande sajten troligen snart blir nerstängd ....
[http://www.x86-secret.com/articles/cpu/k8-2/a64-6.htm]