Kingston HyperX Beast DDR3 2400MHz 8GB Memory Kit Review
Andy Ruffell / 12 years ago
Once we were out of the BIOS and in Windows, we opened up CPU-Z to double check what settings we had, and to make sure that the relevant options we had set in the BIOS had been applied. They had and CPU-Z confirmed a DRAM frequency of 1200MHz (2400MHz Effective) with timings of 11-13-13-30-2T and that we had 8GB in total.
After CPU-Z had confirmed our settings had been applied, we fired up AIDA64 to check the stock performance of the memory on our Z77 motherboard.
At stock, we saw some very nice strong numbers with 21193MB/s read, 18812MB/s write, 23143MB/s copy and a latency of 35.6ns. When looking at previous reviews, these numbers were very good and showed good performance at stock.
After we had the stock performance, it was time to see how far the memory could be pushed on stock timings so off we went back into the BIOS to see how far things could be pushed. The first step was to overclock our i7 3770k to 4.5GHz to eliminate any bottleneck from the memory controller. Once this was done, we tried to raise the memory divider from 2400MHz to 2600MHz, but didn’t hold out much hope and rightly so as the system failed to post. Instead we knew that we’d have to increase using the BCLK and slowly increased the BCLK until we hit our limit of105.5MHz.
This gave us a final memory speed of 2532MHz with stock timings of 11-13-13-30-2T at 1.65V.
We took to opening up AIDA64 again to see if the performance had changed and actually found a mixed bag of results. While the read speed had decreased from 21193MB/s to 20071MB/s and the copy speed had decreased from 23143MB/s to 20815MB/s but the write speed had actually increased by a dramatic amount from 18812MB/s to 22197MB/s and the latency had decreased from 35.6ns to 32.2ns.
The next step for us was to see how far we could push the memory with auto timings. The board used will automatically set timings on the memory depending on its speed.
We stayed at 105.5MHz and started to increase from that point and found that the modules had a bit more room to play with now that the timings had automatically slackened to 11-13-13-35-2T.
We tried once again on the 2600MHz divider but this was still not possible, so going back to the BCLK way of overclocking, we found we were able to push a little bit further to 106MHz giving us a final speed of 2544MHz with timings of 11-13-13-35-2T at 1.65V.
Once we had booted back into Windows and saw 100% stability through a variety of different tests, we opened up AIDA64 one last time to check the performance again. The performance had increased in certain areas (write and copy) though the copy speed was still lower than at stock. We ended up with speeds of 20234MB/s read, 22256MB/s write, 20988MB/s copy and a latency of 32.3ns.
Though we tried to push the memory further, we wasn’t prepared to increase the voltage further than the stock rated 1.65V to do so as this is classed as unsafe on the Z77 platform without extreme cooling.