Intel's Hexa-Core Core i7-5820K was released in August 2014 for a reasonable $ 390 – the cheapest Haswell e-processor currently available – and has become popular with enthusiastic builders over the years.
Today, the chip can be bought for almost half the original price, and we managed to find one online for around $ 200, which is roughly the price.
The 5820K works with a basic frequency of 3.3 GHz as standard, but can increase up to 3.6 GHz depending on the workload (admittedly a slight increase) and 15 MB L3 cache in addition to 1.5 MB L2 cache pack a significant increase in 2014 compared to the mainstream Core i7.
Although the chip is pin compatible with previous generation Ivy Bridge-E and Sandy Bridge-E cards, we have an updated socket, LGA2011v3. The socket change was somewhat justified this time because Haswell-E saw the introduction of DDR4 memory on Intel's high-end desktop platform. As with previous generations, support for quad-channel memory remained, and the official specification required DDR4-2133.
Since the release of the 5820K, we've seen the price of the 6800K jack in this category rise to well over $ 400, and last year the 7800X dropped back to $ 390. Of course, 2017 also brought the Ryzen 5 1600 for $ 200 and the 8-core / 16-thread R7 1700 for $ 290. Intel's high-end desktop platform actually faced real competition in 2017 that you never really saw.
Our focus will be on playing for this re-test. The goal is to accomplish two things: 1) to show those of you who are currently using a 5820K the benefits of upgrading, and 2) to give used buyers information about whether the 5820K and an associated X99 motherboard are Worth a visit – $ 200 for the 5820K plus $ 80 to $ 100 for a supporting motherboard seems to be a great value on paper.
What we can say in advance is that it is probably not worth jumping for the 8700K if you are mainly playing on the 5820K. It makes more sense to wait considering what we saw in our last 4770K test in Haswell.
The 4770K at 4.8 GHz at 1080p at medium quality was only 15% slower with a GeForce GTX 1080 Ti. At maximum quality, this margin was reduced to only 10%, which further decreased to 4% when the GPU workload was increased to 1440p.
While we don't expect the 5820K to make a big difference, you still love benchmark results, and so do we, so we don't let this little fact stop us. We tested the 5820K with the four-channel DDR4-2666 memory. This is the highest memory speed that our processor would work with, at least with the available sticks. The 4770K was tested with DDR3-2400 memory and the 8700K with DDR4-3200 memory.
Before we get to the game results, we have some Cinebench R15 results as well as some corona benchmarks and power consumption figures.
First, a quick look at Cinebench R15, the most popular rendering benchmark of all. Here, the standard 5820K scored 6% fewer points than the R5 1600, and the Ryzen CPU also achieved a better single-thread result. Overall, however, they weren't drastically different. After overclocking, the 5820K reached 1305 points, which was on the level of the 7800X and a whopping 21% behind the overclocked 8700K.
If you want to increase productivity, the 8700K offers noticeable benefits, although you're better off with a Skylake-X processor like the 7820X or a Ryzen 7 CPU.
Here's a look at the Corona results and we see that the 5820K can only keep up with the 7800X once it is overclocked, although in this test it is way ahead of the Ryzen 5 1600 and almost catches the R7 1700. Still overclocking the 5820K was 10% slower than the standard 8700K and 24% slower once the 8th generation processor was overclocked.
Before jumping to the game results, take a quick look at the power consumption. In stock, the 5820K consumes slightly more power than the 8700K and 12-core / 24-thread 1920X threadripper, but less than the 7800X. After overclocking, the aging chip consumed the same 270 watts as the 5.2 GHz 8700K, so nothing too extreme, but at 4.6 GHz the 5820K gets quite power hungry.