Debut of AMD AM2: the long-awaited DDR2 on AMD Athlon X2
Date: 23.05.2006
|
|
Synthetic benchmarks
At the very start, we found out the following: since no
modifications to the processor core which would affect the performance
of new AM2 chips have been introduced, we should look for dependence of
performance variation on the mode of memory controller operation. Since
in the new AMD processors (like in "old" Ê8) all depends on the
HyperTransport and there is no FSB at all, the memory controller
integrated into the chip does not match any memory type starting with
DDR2-400 and DDR2-533 up to DDR2-667, DDR2-800, and even DDR2-1066.
That is why the major focus in our tests was made on selecting
the most varied memory operation modes, i.e. we were after an in-depth
investigation of variations to the system performance versus the memory
frequency and latency timings. Of course, for curiosity we might have
added results of tests for other AMD chips or other processor
architecture into the comparison tables. But there was no sense in
that, because found it more visual to approximate the difference in
performance of Socket AM2 and Socket 939 systems to other reviews
describing Socket 939 chips instead of depleting our results through
bringing in fewer tests of the Socket AM2 system with various memory
settings.
On this page, we present the results of test in synthetic
benchmarks. For convenience, we highlighted the DDR2 memory operation
mode in yellow tints, and in green - for systems with DDR400. After the
memory type, there go the memory latency timings at which we ran the
tests.
Let's start with Everest Ultimate ver. 2.80.534.
The very first results vividly prove the thesis that AMD
processors with support for DDR2 memory would have had nothing to do in
the epoch of DDR2-533 occurrence, let alone DDR2-400. As is seen in
practice, the more or less impressive "gap" is seen just in a system
with DDR2-800 that offers quite decent latency. Even DDR2-667 sometimes
loses to the regular not overclocked DDR-400.
Now let's see what PCMark'05 will show.
On the whole, the above stated is proved well enough by the
results of tests run in PCMark'05.
And this is the final verdict by PCMark'05:
Summing up the results of testing the new system in synthetic
benchmarks, we note that in general the latency and performance of
DDR400 memory is in fact comparable to a system equipped with DDR2-667
4-4-3.
That is, being guided by the test results, if I, say, were
after replacing a system based on Socket 939 Athlon 64 X2 and DDR400
memory and if the retail didn't offer anything better than a
combination of the same Athlon 64 X2 for Socket AM2 and DDR2-667
memory, then such a replacement would hardly be reasonable. Quite a
different thing would be DDR2-800, but in any case the buyer should
carefully examine the price tags for this memory before the purchase.
We now will see how reasonable that will be when tested in real
applications...
 |
Top Stories: |
 |
 |
|
MoBo:
| |
|
|
VGA Card:
|
|
|
|
CPU & Memory:
|
|
