Chaintech Apogee GT DDR3 1600

We have got a kit of DDR3 memory by Chaintech and will introduce you to our impressions. Chaintech, formerly for its motherboards and video cards, decided to refresh its assortment, which gave birth to the participant of today's review - Apogee GT. Let's take a look at it.

Whatever you take, the setting follows the same trends. For instance, in the box that we got we found merely two memory modules, 1GB each, with two much ado on the box cover! To the left, the label says "Ultimate Memory Technology", to the right - "Deluxe Edition", and under the "GT" label there is the inscription "Blazer Series". Box_rear.jpg

On the back of the package, there is a small annotation on the contents, i.e. the memory modules. According to the manufacturer, the efficiency of the new cooling system displayed near the annotation has been increased by 50%. But there is no indication as to versus what and under which conditions it has been increased.

The memory modules are placed on the velvet pad, as if there were luxurious perfumes or jewels.

Apart from the two memory modules, the kit includes a radiator with screws, washers and springs, as well as a brief user's instructions brochure.

The cooling radiator is fasted to the modules with screws at the top. To compensate for the various distance between the memory slots on various motherboards, the slot for screws has been done for the whole width of the radiator, due to which the radiator can be moved along the memory modules.

On each of the memory modules, there is a sticker displaying the specifications. We can see that the memory modules by Chaintech are able running at the effective 1600 MHz frequency with the latency timings set to 7-7-7-18 at the supply voltage 1.6 V. Not bad at all, indeed.

The radiator is fastened to the memory module with two screws. Not only is it quite out of the ordinary, it is easy to dismantle, which we are doing right now.

The memory modules are single-sided, so on one part of the radiator there is a heat-conductive grey pad to contact the memory chips, on the other side - there is a pad of porous material to press the memory module to the send part of the radiator.

The memory chip are made by Elpida. According to the description of these chips on Elpida's web site, their nominal effective clock speed is 1066 with the timings set to 7-7-7. Therefore, during the manufacture of Apogee GT memory modules there were used selected chips because Chaintech guarantees their operability at 1600 MHz with the latency timings set to 7-7-7-18 and slightly increased supply voltage equal to 1.6 V. Recall that as per the DDR3 standard the supply voltage equals 1.5V. We'll find out if Chaintech keeps its promise from further tests.

Benchmarking results and conclusions

We'll be running today's tests using Everest Ultimate 4.0 and SiSoft Sandra 1424. We already conducted a similar test, so as not to repeat we'll be testing the memory on a motherboard based on the Intel X38 chipset.

Since the chipset Intel X38 sets the system bus speeds and the memory bus speed synchronously, that is, the latter depends on the former, so we had to set the system bus speed to 400 MHz. The side effect was the slightly increased CP clock speed 200 MHz above the nominal, however, that should not affect the operating speed of the memory.

In the end, we were able to set the memory frequency to 1600 MHz. Note that at the so high clock speed the memory started up trouble-free in the Command Rate "1T" mode, so we decided to know that gain from that. So, at 1600 MHz we ran the tests with the latency timngs set to 7-7-7-21 CR=1T and 7-7-7-21 CR=2T. But the promised configuration of latency timings 7-7-7-18 proved to be inoperative with our motherboard ASUS P5E3, so we raised the latter timing value to 21. Don't worry - that is a mere trifle which almost did not affect the memory performance.

We also tested the memory at 1333 MHz, but this time with the latency timings set to 6-6-6-18-1T. The reduction if the memory frequency allowed to reduce the latency timings. To find out the effect of the system bus speed on the memory subsystem speed, we ran tests at the system bus speeds set to 333 and 400 MHz, with the memory frequency equal to 1333 MHz.

Today, we'll know the effect of the Command Rate "1T/2T" parameter and the system bus speed on the memory subsystem speed for the Intel x38 chipset. Perhaps the situation with other chipsets may be different, and we'll find that out in another article. OK then. Off we go!

The memory at 1600 MHz with "1T" won, and slightly overtook the results produced with the parameter "2T". The difference in the results for memory running at 1333 MHz proved to be minor, albeit there is one: at the system bus speed 400 MHz, the results proved to be higher.

The results of memory write at the system bys speed 400 MHz are almost equal, and the frequency of the memory itself and the "1T/2T" almost did not affect. But the gain from using the 400 MHz system bus proved to be quite weighty and amounted to almost 20% as compared to the 333 MHz bus.

At the test of copying speed in the memory, the results ranked in the expected sequence - 1600 MHz 1T takes a lead, followed by 1600 MHz 2T, then 1333 MHz at the system bus speed 400 MHz. Then, the memory at the system bus speed 333 MHz leaves the the others at a 20% gap. The second time, it appears that the system bus speed 400 MHz is highly favorable to the memory subsystem. We also note that the difference between the memory with "1T" and "2T" proved to be minor as is between the memory at 1600 MHz and 1333 MHz at the system bus speed 400 MHz.

The memory latency with "1T" proved to be less, which is better. But the latency slightly increased as compared to the 333 MHz bus due to use of the 400 MHz bus. On the other hand, the difference is so minor that it is not worth mentioning it.

The memory bandwidth test in SiSoft Sandra issued quite predictable results. The gain from using the "1T" latency timing amounted to 2%, but the gain from using the 400 MHz bus the gain proved to be substantial and amounted to almost 10%.

As is the case with Everest Ultimate 4, in the latency test during random access SiSoft Sandra the memory gained from using the Command Rate 1T parameter, but lost from using the 400 MHz system bus. The difference is negligible, as you can see.

Final Words

Today, we have not only tried a new kit of high-speed DDR3 memory but also found out quite interesting traits of the Intel X38 chipset, and presumably, similar results would have been produced with other chipsets by the company. We'd also like to note that at 1333 MHz the memory kit by Chaintech started up trouble-free at reduced timings, which may be quite useful if it is impossible to set the 400 MHz bus. On the other hand, the gain will be much smaller than from using the 400 MHz system bus speed and the memory frequency 1600 MHz. So, owners of motherboards based on the Intel x38 chipset with support for DDR3 should think about that.