Assembling your own PC can be a confusing venture, especially if you’re no hardware geek and have to resort to ‘expert’ opinions when selecting hardware components. This isn’t so much of a problem when faced with selecting a video card or processor, because there are plenty of benchmarks for these components to be found on major hardware sites such as Anandtech and Tom’s Hardware.
It’s different when it comes to selecting memory modules, or R(andom)A(ccess)M(emory). There are practically no real-world performance benchmarks to be found for memory modules, so newcomers are left to ask around on hardware forums. This invariably leads to a wealth of hardware geeks responding with the most expensive and ‘high-performance’ DDR3 modules, when it’s not necessary.
As a newcomer you’ll be slapped in the face with terms such as CAS latency and clock speeds, but no one mentions why these things matter for a gaming PC. This is because such terms only matter in theoretical benchmarks, not real-world performance. In other words: there is no tangible benefit to shelling out for better memory modules.
The theory is that lower latency, higher clock memory modules result in better performance than higher latency, lower clock modules. A variety of memory and PC benchmarks support this theory, with better memory modules resulting in significantly higher scores in such tools as PC Vantage and SiSoftware Sandra. However, hardware benchmarks do not equate to real-world performance.
It’s telling that most major hardware sites, such as AnandTech and Tom’s Hardware, rarely include real world performance indications for memory modules. This is because there are no actual performance gains to be had by shelling out for better memory. It would simply result in a collection of near-identical values, which isn’t what the target audience (overclockers, not regular PC users or gamers) is interested in.
So if you simply want a great gaming or media PC, then you might as well save some money on RAM. Get the least expensive DDR3 4GB dual channel memory modules you can find, but make sure it’s from a reputable brand such as Corsair, G Skill, Kingston or OCZ. The money you save on memory can be put to better use in purchasing a better video card, processor or even a solid state disk.
By now you’re probably wondering whether I’m just pulling this out of my ass, so it’s time to give you some real world performance statistics. I’ll start with a fairly recent (september 2009) article by AnandTech that compares DDR3 modules ranging from DDR3-1333 to DDR3-2133 that are mostly low-latency modules: CAS-5 to CAS-8. The price difference between the top and bottom modules is roughly 70 euro, so we’d expect to see some decent performance differences here.
The theoretical benchmarks geared towards overclockers demonstrate huge performance differences. Then we move on over to gaming performance with a Far Cry 2 benchmark and see that that difference between top (DDR3-1600 CAS-6) and bottom (DDR3-1333 CAS-5) is a whopping 2 FPS average. This on top of a base 65 FPS, so it’s not noticeable at all. It’s different with transcoding in Sony Vegas Pro, which grants a 50 second improvement for the top contenders, but that’s not something the average PC user or gamer is likely to do.
More real world performance figures can be found over at Tom’s Hardware, who resorted to cheating on the gaming benchmarks in order to give the illusion of a greater impact from memory modules than you’d normally see as a gamer. They’ve turned the detail down to medium with all frills off and the video resolution is set to a measly 1280×800. This is because modern games are heavily bottlenecked by the video card and both memory modules and processors have very little to no impact on performance at higher detail levels and video resolutions.
Still, even with Tom’s Hardware’s exaggerated results, it’s painfully obvious how little impact memory modules have on gaming performance. In Far Cry 2 and GTA IV the lowest framerate comes in at 100 FPS, which makes the resulting difference between top and bottom of respectively 4 and 2.5 FPS absolutely worthless. Left 4 Dead on first glance seemed to be more heavily impacted, but the lowest framerate in that benchmark is 146 FPS, which means that you’ll never notice the 14 FPS increase that the most expensive memory modules provide.
Keep in mind that the performance difference would be even smaller in high detail and high resolution gaming, because even the most powerful video cards on the market bottleneck performance to such a degree that memory modules have exactly no noticeable impact on game performance.
Tom’s Hardware also included a few application benchmarks, which detail how little impact memory modules tend to have on demanding software applications. These are applications that do not benefit from better RAM at all:
- Video transcoding in XviD, DivX or Mainconcept
- Audio encoding in Lame and audio transcoding in iTunes
- 3DS Max 2009 image rendering
- AVG Anti-Virus
- Adobe Acrobat
- Adobe PhotoShop Image Processing
- WinZIP File Compression
There were only two exceptions to this list, namely Adobe Premier video rendering and WinRAR file compression. Both of these applications actually benefit more from a simple processor or hard disk upgrade than they do from better memory modules, which again makes it a questionable expenditure.
And that is why I claim that you might as well avoid anything but the least expensive memory modules of a decent brand. It’s also why my quarterly budget gaming PC articles always feature cheap DDR3 4GB RAM: perfect gaming performance at a low price.
There are exceptions to the rule of course. A prolific video-editor with a healthy budget could benefit from upgrading the memory modules in his/her PC, but even then it’s still not the key component to upgrade. A video-editor would do better in sticking with a heavy-duty processor and high quality hard disk drive, before considering a potential memory upgrade.
All of this is not going to stop overclockers from buying and recommending these products, because they are key to gaining the best possible theoretical performance – and that’s what being an overclocker is all about. However, it’s still something you’d best keep in mind when you next find yourself shopping for a new PC or hardware upgrade.