Tests of low-end AMD Athlon 64 X2 for Socket AM2
Author: Date: 26.09.2006
|
|
Benchmarking
For running the tests, we selected a rather extensive kit of processors, so now we can answer not only the question 'Which one is faster?' but a few at a time:
- How slower has AMD Athlon 64 X2 3600+ turned relative to AMD Athlon 64 X2 3800+? Is it because of cutting down the L2 cache size half as much?
- Has the DDR2 brought a performance boost to AMD Athlon 64 X2?
- Which is the value dual-core processors is more powerful - AMD Athlon 64 X2 3600+ or Intel Pentium D 915?
- What if we compare the performance of all of them versus Intel Core 2 Duo?
To answer these questions, we assembled three test configurations.
Test configuration for AMD Socket 939:
Motherboard
|
|
Memory
|
|
HDD
|
Samsung HD080HJ (80 GB, 7200 rpm, 8 MB, SATA-300)
|
Video Card
|
|
Power supply unit
|
SuperPower 480X, 480 W
|
Operating system
|
Microsoft Windows XP SP2
|
Drivers
|
NVIDIA ForceWare 91.31 WHQL
|
Test configuration for AMD Socket AM2:
Motherboard
|
ASUS M2N-SLI Deluxe (nForce 570 SLI )
|
Memory
|
2х DDR2-800 512 MB GEIL PC6400 (5-5-5-15)
|
HDD
|
Samsung HD080HJ (80 GB, 7200 rpm, 8 MB, SATA-300)
|
Video Card
|
|
Power supply unit
|
SuperPower 480X, 480 W
|
Operating system
|
Microsoft Windows XP SP2
|
Drivers
|
NVIDIA ForceWare 91.31 WHQL
|
Test configuration for Intel LGA775:
Motherboard
|
ASUS P5B Deluxe/Wi-Fi-AP (Intel P965)
|
Memory
|
2х DDR2-800 512 MB GEIL PC6400 (5-5-5-15)
|
HDD
|
Samsung HD080HJ (80 GB, 7200 rpm, 8 MB, SATA-300)
|
Video Card
|
|
Power supply unit
|
SuperPower 480X, 480 W
|
Operating system
|
Microsoft Windows XP SP2
|
Drivers
|
NVIDIA ForceWare 91.31 WHQL
|
To start with, let's see how the processors respond to the change of operating speed of the memory subsystem. We start with AMD Athlon 64 X2 3600+ in combination with DDR2-400/533/667/800.
Despite that Futuremark PCMark’05 has proved practically indifferent to memory performance , the other benchmarking suites showed a performance boost as high as 35% due to transition from DDR2-400 to DDR2-800, that is, in most ways it doesn't make sense saving on memory modules.
And what about Intel Pentium D 915? Unfortunately, ASUS P5B Deluxe/Wi-Fi-AP does not allow for operating with DDR2-400 if the system bus is higher than 533 MHz. We also had no DDR2-1066 available, but the missing results can be calculated approximately using those already produced.
Based on the produced results, we can conclude that for Intel Pentium D fast memory modules is of less importance than for AMD Athlon 64 X2. Of course, in this case much depends on the chipset, motherboard and BIOS, as well as memory latency timings, but we believe it makes no sense to buy expensive DDR2-800/1066 memory with low timings for these purposes. The bottleneck of these systems is the bus running at 800 MHz.
Having sorted out with the memory requirements of systems being tested, we now move on to comparing the CPU performances in various tasks. We start with the synthetic benchmarks..
SiSoftware Sandra 2007 SP1 equates the performance of Athlon 64 X2 3600+ and Athlon 64 X2 3800+, and despite the different cache memory sizes the benchmarking suite shows a minor lag of Athlon 64 X2 3800+ Socket AM2 behind the Socket 939 at arithmetics with higher memory operating speed; it shows excellent performance of Athlon 64 X2 3600+, especially in the overclocked state, in comparison with both Intel Pentium D 915 and even Intel Core 2 Duo E6300.
Futuremark PCMark’05 has long been a matter of complaints that it is biased towards processors of higher clock speeds – which we could observe in the CPU test. Athlon 64 X2 3600+ and Athlon 64 X2 3800+ are everywhere on par, except the graphics tests. Intel Pentium D 915 rehabilitates mainly due to the encoding tasks, but anyway lags well behind at graphics.
We now move on to practical tests of data compression (archiving a folder containing programs and documents of 1 GB total size), as well as audio encoding (540 MB *.wav into *.mp3) and vide encoding (a 74 MB *.mpeg movie into *.avi). Here the faster is the better.
WinRar 3.60, unlike 7-ZIP 4.42, offers better optimization for multithreaded compression on dual-core processors, which is well seen by the results. On the whole, we finally managed to find tasks where Athlon 64 X2 3600+ lags behind because of the reduced cache, and Athlon 64 X2 3800+ demonstrates a small performance boost due to the faster memory operation. Media encoding is still a "soap-box" for Intel processors, especially while compressing audio with LAME codec, but overclocking Athlon 64 X2 3600+ changes the situation and it is ready to compete even against Intel Core 2 Duo E6300, whereas Intel Pentium D 915 is no match for it at that.
The memory speed is important for fast operation in CAD/CAM suites, so Socket AM2 processors win at that over Socket 939 even if the L2 cache .size is smaller. But here Intel Pentium D 915 is already a rather serious competitor which after overclocking catches up with almost Intel Core 2 Duo E6300.
Judging by the results of gaming benchmarks by Futuremark, we can see some dependence of AMD CPU performance on the L2 cache size - there is almost no lag. But the DDR2 memory has shown not as much use as we expected. That does not prevent Athlon 64 X2 3600+ from being faster than Intel Pentium D 915 in new versions of the package.
Quake 3, sensitive to the operation speed of the memory controller, shows interesting results. All the Intel processors have proved faster at that. The DDR2 with у AMD did not hold water. But modern and resource-intensive games will show a bit different result…
Games demonstrate some minor lag of Athlon 64 X2 3600+ behind Athlon 64 X2 3800+, but on the other hand we see some small performance boost for Socket AM2 processors. Intel Pentium D 915 is not as good a performer as in multimedia. And only the much greater cache size saved it from lagging well behind Athlon 64 X2 3600+ .
Final Words
Let's try summarizing all the answers to the questions we posed in the review. AMD has done a pretty successful job of the dual-core Athlon 64 X2 3600+ which almost doesn't lag behind Athlon 64 X2 3800+ and in almost all the tasks runs faster than Intel Pentium D 915 despite the reduced L2 cache size and with no practical use from DDR2 memory. The most affordable dual-core Intel processor proved faster in only a few tasks, which is also good, so it makes no sense to disregard it. In the long run, it is the price and priority of tasks to be done will matter most to the thrifty user, unless the user is not into overclocking. After all, we note that even Athlon 64 X2 3600+ if overclocked can yield a performance comparable to Intel Core 2 Duo E6300, and the costs may prove even smaller in terms of the price of processors and motherboards. Those who are not short for funds will not find anything topical from the tests we ran – with a good motherboard they will overclock Intel Core 2 Duo to a performance level unattainable to competitors, but the cheap-rate processors are not meant for such buyers.
We appreciate LLC PF "Servis" (Dnepropetrovsk, Ukraine) for the processors and other hardware presented for tests.
|
Content: |
|
|
|
Top Stories: |
|
|
|
MoBo:
|
|
|
|
VGA Card:
|
|
|
|
CPU & Memory:
|
|
|