Apacer Giant DDR2-1066 (PC2-8500) Memory Modules
|
Test results
Testbed configuration:
- Processors: AMD Phenom 9750 (Socket AM2+), 2.4 GHz (200x12)
- Chipset: AMD 790FX
- Motherboard: ASUS M3A32-MVP Deluxe, BIOS 1001.
In these tests we also compared memory performance in ganged and unganged modes of Phenom memory controllers. In the first mode the memory controller works as a single 128-bit controller, in the second case -- as two 64-bit ones. So the ganged mode demonstrates higher results in single-core memory access mode and consequently in tasks, where the speed of a single main process is critical. Unganged mode is recommended for multitasking environments. Differences are quite big. As you can see, the effect of a chosen mode on memory bandwidth is similar to that of memory frequency. At the same time, it's sufficient to test memory in any of these modes to compare modules -- if a module demonstrates high performance in ganged mode, it will retain the advantage in unganged mode as well.
Command Rate is set to 2T. All timings, except four main values, are chosen by BIOS automatically.
| |
Apacer Giant DDR2-1066 2x2048MB |
Corsair TWIN2X4096-9136C5DF |
| Memory mode |
Ganged |
Unganged |
Ganged |
Unganged |
Ganged |
Ganged |
Memory frequency, MHz
(DDR2 MHz) |
400 (800) |
400 (800) |
533
(1066) |
533
(1066) |
587 (1174) |
575
(1150) |
Frequency of the integrated memory controller, MHz
(DDR2 MHz) |
2000 (200x10) |
2000 (200x10) |
2000 (200x10) |
2000 (200x10) |
2200 (220x10) |
2160 (216x10) |
Frequency of processor cores, MHz
(FSB clock rate x FID) |
2400
(200x12) |
2400
(200x12) |
2400
(200x12) |
2400
(200x12) |
2640
(220x12) |
2592
(216x12) |
| Memory timings, voltage |
5-5-5-15-2T,
1.8 V |
5-5-5-15-2T,
1.8 V |
5-5-5-15-2T,
2.2 V |
5-5-5-15-2T,
2.2 V |
5-5-5-15-2T,
2.3 V |
5-8-8-24-2T,
1.94 V |
| Minimum stable voltage |
(not tested) |
(not tested) |
1.8 |
1.8 |
1.88 |
1.94 |
Average memory read bandwidth (MB/sec),
1 core |
6195 |
5760 |
6728 |
6373 |
7385 |
7060 |
Average memory write bandwidth (MB/sec),
1 core |
3548 |
3588 |
3714 |
3935 |
4019 |
3957 |
Max. memory read bandwidth (MB/sec),
1 core |
7149 |
6619 |
7883 |
7428 |
8651 |
8281 |
Max. memory write bandwidth (MB/sec),
1 core |
4965 |
4983 |
4964 |
4985 |
5431 |
5342 |
Average memory read bandwidth (MB/sec),
4 cores |
10960 |
11078 |
11920 |
13110 |
13040 |
12815 |
Average memory write bandwidth (MB/sec),
4 cores |
3550 |
5104 |
3847 |
5570 |
4220 |
4182 |
Max. memory read bandwidth (w/PF, MB/sec),
4 cores |
11238 |
11105 |
11872 |
13112 |
12928 |
12647 |
Max. memory write bandwidth (NT, MB/sec),
4 cores |
6315 |
6315 |
6315 |
6315 |
6916 |
6788 |
| Minimum pseudo-random access latency, ns |
35.7 |
37.0 |
32.0 |
33.6 |
29.7 |
30.1 |
| Minimum random access latency*, ns |
88.2 |
89.5 |
75.2 |
77.6 |
69.0 |
79.8 |
*Block size: 32 MB
Conclusions
First of all we'd like to mention the excellent overclocking potential, which breaks all records among 4GB DDR2 kits, the standard timings scheme preserved. Frankly speaking, we didn't expect such results from these simple-looking modules. A much more expensive memory kit from Corsair takes the second place.
Moreover, these modules allow to reduce their voltage from the recommended level. In particular, DDR2-1066 mode is stable at 1.8 V. What concerns the overclocked mode, engineers from Apacer and chip suppliers can be really proud of 1.88 V. But this product has one great drawback: unlike other Apacer products, memory modules from this company are rare guests in retail stores. According to the management, the company is working on sales channels, and the new package for memory modules should promote the retail segment.
Excellent overclocking probably makes up for justified criticism of the strange SPD data. But if your motherboard is not as smart as ours, you will have to provide a backup memory module for the first boot, fix the startup frequency and timings in BIOS, and only then install this memory kit.
Write a comment below. No registration needed!
|
|
 |
|
|
|
