We haven't covered a lot of memory recently here on bit-tech
, and in a news post last month we were pretty blunt as to why: "We'll level with you - memory reviews are... a bit dull. Running the kind of benchmarks we use for CPUs and motherboards reveals performance differences that are within an undetectable fraction of each other, prices change daily and as a result, few people read the coverage."
When you test CPUs and graphics cards, benchmarks can find real, substantial differences between products, and what's more, these results are relatively easy to understand. We're fond of using real world benchmarks such as games and open source apps, and getting results - whether it's frames-per-second or time to complete a task - that clearly show the benefit of buying one product over another.
With memory, it's trickier, because it's quantity of memory that has the biggest impact on performance, or the speed that it allows the rest of the system to run at - it's hard for us to isolate the benefits one stick of RAM gives over another. You've then also got the fact that memory customers are splitting into quite distinct groups; some just want the cheapest RAM to get the most quantity. Others want RAM based on specific ICs so as to push their overclock as far as possible. Finally the other end to this three pronged stick is that a few just want the sexiest, craziest, most aesthetically gratifying bit of kit possible.
All of which means that we were struggling to see how we should cover memory, so we turned to the community... Our original post
attracted a whopping 155 comments, and the debate was really interesting to read although centred on a few major themes.
"Stack it all up and play Jenga with it. You can then rate each type of memory based on how well it stacks, difficulty to remove from the stack, and the odds of the stack collapsing once the RAM is removed!"
suggested Bursar, which is one way to go about things, but frankly, the labs are messy enough already. While Ending_Credits suggested adding a comedy German accent, we think the Top Gear team have already got that covered and naturally we're stubborn so prefer to go our own way...
There were however, several areas that everyone seemed in agreement on: "Overclocking potential and sexiness of the heat spreaders - it's always all about the heat spreaders"
said Shepps. Multiple people suggested that we give our impressions of the RAM in terms of its heat spreaders, packaging and overall feel, as well as photos of the memory fitted to the board, along with measurements to show the space it requires.
While some mentioned compatibility, realistically it's just unrealistic to spend days and weeks seeing what RAM works in what boards with which BIOS' - we'd never manage to get any other kit tested and there's always QVL lists to refer to. While some people were interested in overclocking, most recognised that of course, we're only going to be able to deal with one sample, and that overclocking when memory is concerned is also affected by the other components, so it's not really such a valid metric for memory reviews.
Heat concerns were mentioned several times, and while we probably won't be able to get a thermal camera, we might look into finding a way to recording temperatures when we're covering memory or whole systems in general. Another area people wanted us to look at was memory technology in general, and how the stuff works, and while we've covered this previously over four large articles
, we're certainly going to be looking at updating that in the future.
And, of course, boobs
were mentioned, but we're not going there - and neither are we going to go for the suggestions that involve injuring the bit-tech
staff because our wives/girlfriends/employers health and safety department would be most displeased!
What we're likely working towards is a memory technology article - looking at whether you get better performance from low latency modules, or higher frequency modules in general - and then for product specific coverage, semi-regular memory buyer's guides where we focus on price, and what's the best buy, and then 'first look' style pieces where we give information on the latest notable memory launches - including, where possible, the IC the DIMMs are based upon - and compare the looks and stated settings they'll run at.
So, the winner? Well, lots of people put in great suggestions and the thread itself shows how knowledgeable, informed and funny the bit-tech
community can be. There can be only one winner though, so congratulations to bursar
; his suggestion was the one that made us laugh! A 6GB G.Skill Trident PC3-16000 memory kit is his prize.