|
Memory Speed
How well does the Athlon 64 perform with slower memory?
UT2004
| Memory Speed |
1024 x 768 |
1280 x 1024 |
1200 x 1600 |
Difference |
| 200MHz |
20.79 |
15.09 |
11.45 |
-15.4% |
| 266MHz |
22.73 |
16.33 |
12.48 |
-7.9% |
| 333MHz |
24.07 |
17.22 |
12.92 |
-3.4% |
| 400MHz |
24.85 |
17.82 |
13.41 |
0.0% (Base) |
Aquamark
| Memory Speed |
Default Settings |
Difference |
| 200MHz |
40,029 |
-14.0% |
| 266MHz |
43,222 |
-7.1% |
| 333MHz |
45,211 |
-2.8% |
| 400MHz |
46,536 |
0.0% (Base) |
Here we are seeing more performance losses in the lower resolutions, suggesting
that the CPU is a limiting factor in the high resolutions.
The biggest performance loss between 400MHz and 200MHz is around 15%. We found
that unlike the Athlon XP platform, the system had no difficulties with running
the memory at a slower speed, probably because the memory interface is entirely
between the CPU and the memory modules, without the involvement of the north
bridge at all.
These figures do tell us one important fact: if you wish to upgrade to Athlon
64, you may not need to buy DDR400 if you already have plenty of slower memory.
Even antiquated DDR200 would be an admirable stop gap measure until memory prices
fall again and there is little reason to upgrade your DDR333 to DDR400. This
could save you £100 or more on your upgrade costs! Remember the Athlon
64 platform does not use dual channel memory, so there is no disadvantage to
using a collection of mismatched modules either.
<< Back | 1 | | 2 | | 3 | | 4 | | 5 | | 6 | Next >>
|
