ADATA DDR4-2400 RAM: Is there life in the hi-end segment?
The question of the expediency of transition to a new generation of RAM has been repeatedly considered. To the great joy of most users, DDR4 was implemented only on the Intel Socket 2011-3 platform, so only real iron fans and enthusiasts were puzzled by the right choice of RAM of this standard. What to buy from what is on the market? Maybe choose the nominal frequency less, but disperse more? I propose to look at the ADATA memory kit - the standard frequencies are not transcendental - DDR4-2400, but is this the limit?
I will not bother with the description of the appearance, as usual, I will not, here are a couple of pictures - profile, full face, packaging.
')
I note the average height of the radiators on the RAM, which brings some inconvenience in terms of compatibility with processor coolers. Although, in an amicable way, when buying such an expensive and productive platform, many probably were puzzled by the purchase of a liquid cooling system, and in this situation, the problem of the height of radiators is irrelevant.
Under the radiators, chips from the MFR series from SKHynix were found. Statistics on these chips allows you to count on quite decent results of overclocking and optimization of timings.
Let's get to the essence of the question:
Memory in comparison will be two sets: already described ADATA XPG Z1 and Corsair Vengeance DDR4-2800 . Why such a choice? It's very simple - memory bars are made on the same chips, and they cost more than different money (2400 are not available for sale on the Market yet, but there are 2133 MHz , and judging by newegg.com their price is approximately equal =) This raises the question - and so expensive , can it save even a little?
Total output should be 3 graphics with 4 columns each - ADATA nominal, ADATA overclocking, ADATA timings lowering and Corsair nominal.
I'll start overclocking the memory. It was not possible to take a frequency of 2800 MHz in spite of the fact that the ASUS Rampage V Extreme motherboard has a ready-made profile of settings for plates with such chips.
“Well, you have to increase the voltage ... Danger? Of course, dangerous. But, the risk, as they say, ... ”(C) Ivan Vasilyevich changes his profession.
As a result, the memory started working at 2800 MHz, but the supply voltage was 1.425V instead of the standard 1.2V. IMHO, it is quite acceptable, only 19 percent excess. The theoretical heating of the memory chips is quite compensated by the installed radiators. Memory, according to tactile sensations, did not become more heated.
With a fixed frequency of 2400 MHz and a power of 1.4125V, it was possible to lower the timings to 13-16-15-39-2T.
Well, let's bring the picture together
The performance in the nominal differs from the overclocked state by about 2 percent ... Is the game worth the candle? Decide for yourself. The result in acceleration and the face value of the Corsair diverge by less than half a percent. Lowering the timings allows you to win another half percent. At the output we get the difference between the maximum and minimum result - 4 percent .... not enough is obtained.
There is a bonus in the other: of course it is necessary to see in toys. Paradoxically, World Of Tanks responds adequately to everything related to an increase in frequency. No wonder the ideal processor for this game called the Intel Pentium G3258 with an unlocked multiplier. Inexpensive and you can significantly raise the multiplier during acceleration. Here and here the situation is similar. I replaced the video card for the test with the AMD Radeon R9 290X - the difference between the two DDR4-2400 and DDR4-2800 modes was 10 percent, and if in absolute values - 92 and 101 frames per second, respectively. I agree right away with the speakers, who will tell me - they say, and so it is more than 90 fps, then why invent? Yes, indeed, but the fact remains.
First of all, DDR4 only appears on the market and is installed on only one platform, so now the choice is very meticulous.
Secondly, if there is no difference, why pay more? Modules assembled on the same chips work about the same, despite the different markings. Therefore, it is enough to study the specialized forums and sites in order to understand what memory is best suited for the characteristics and will not be space for the price. Here ADATA XPG Z1 manifests itself from the advantageous side.
I will not bother with the description of the appearance, as usual, I will not, here are a couple of pictures - profile, full face, packaging.
')
I note the average height of the radiators on the RAM, which brings some inconvenience in terms of compatibility with processor coolers. Although, in an amicable way, when buying such an expensive and productive platform, many probably were puzzled by the purchase of a liquid cooling system, and in this situation, the problem of the height of radiators is irrelevant.
Under the radiators, chips from the MFR series from SKHynix were found. Statistics on these chips allows you to count on quite decent results of overclocking and optimization of timings.
Let's get to the essence of the question:
Test bench:
- Intel Core i7-5960X
- ASUS Rampage V Extreme, Intel X99 Socket
- OCZ Vector SSD 240 Gb
- NVIDIA GeForce 7600GT
- Corsair AX1200i
Memory in comparison will be two sets: already described ADATA XPG Z1 and Corsair Vengeance DDR4-2800 . Why such a choice? It's very simple - memory bars are made on the same chips, and they cost more than different money (2400 are not available for sale on the Market yet, but there are 2133 MHz , and judging by newegg.com their price is approximately equal =) This raises the question - and so expensive , can it save even a little?
The task is as follows:
- Check whether ADATA XPG Z1 DDR4-2400 CL16 can operate at 2800 MHz with timings 16-18-18-36 (this is the nominal value of the more expensive Corsair kit)
- Check how low timings can be for operation at 2400 MHz
What I use to test stability and performance:
- Stability - the HyperPi 0.9.9b test - a multi-threaded calculation of the Pi number up to 32 million decimal places - catches errors in the memory “one-two-three”.
- Performance - Cinebench R15 and AIDA64 bandwidth test - tests rendering and memory bandwidth.
Total output should be 3 graphics with 4 columns each - ADATA nominal, ADATA overclocking, ADATA timings lowering and Corsair nominal.
I'll start overclocking the memory. It was not possible to take a frequency of 2800 MHz in spite of the fact that the ASUS Rampage V Extreme motherboard has a ready-made profile of settings for plates with such chips.
“Well, you have to increase the voltage ... Danger? Of course, dangerous. But, the risk, as they say, ... ”(C) Ivan Vasilyevich changes his profession.
As a result, the memory started working at 2800 MHz, but the supply voltage was 1.425V instead of the standard 1.2V. IMHO, it is quite acceptable, only 19 percent excess. The theoretical heating of the memory chips is quite compensated by the installed radiators. Memory, according to tactile sensations, did not become more heated.
With a fixed frequency of 2400 MHz and a power of 1.4125V, it was possible to lower the timings to 13-16-15-39-2T.
Well, let's bring the picture together
The performance in the nominal differs from the overclocked state by about 2 percent ... Is the game worth the candle? Decide for yourself. The result in acceleration and the face value of the Corsair diverge by less than half a percent. Lowering the timings allows you to win another half percent. At the output we get the difference between the maximum and minimum result - 4 percent .... not enough is obtained.
There is a bonus in the other: of course it is necessary to see in toys. Paradoxically, World Of Tanks responds adequately to everything related to an increase in frequency. No wonder the ideal processor for this game called the Intel Pentium G3258 with an unlocked multiplier. Inexpensive and you can significantly raise the multiplier during acceleration. Here and here the situation is similar. I replaced the video card for the test with the AMD Radeon R9 290X - the difference between the two DDR4-2400 and DDR4-2800 modes was 10 percent, and if in absolute values - 92 and 101 frames per second, respectively. I agree right away with the speakers, who will tell me - they say, and so it is more than 90 fps, then why invent? Yes, indeed, but the fact remains.
Findings:
First of all, DDR4 only appears on the market and is installed on only one platform, so now the choice is very meticulous.
Secondly, if there is no difference, why pay more? Modules assembled on the same chips work about the same, despite the different markings. Therefore, it is enough to study the specialized forums and sites in order to understand what memory is best suited for the characteristics and will not be space for the price. Here ADATA XPG Z1 manifests itself from the advantageous side.
Source: https://habr.com/ru/post/376453/
All Articles