Tech-Hounds.com

Because gamers play games, not benchmarks





Quake 4 - Data Processing Terminal, 1024 x 768
CL 2 1T 512 MB
CL 3 1T 2 GB
CL 3 2T 4 GB
67
124.538
173
65
123.55
167
61
123.319
168
.
70
128.396
175
70
127.168
172
70
127.73
184
.
67
121.201
169
67
122.818
166
67
122.121
172

Lower memory timings and Command Rate is still the main factor influencing performance in this game. However, what the graphs doesn't show is that with 2 GB and 4 GB, there's no stutter at all on the first run. That's good, since in reality you don't really play the same level over and over again to avoid stuttering.

SC: Chaos Theory - Lighthouse, 1024 x 768
CL 2 1T 512 MB
CL 3 1T 2 GB
CL 3 2T 4 GB
62
98.496
160
62
98.506
160
62
98.485
160
.
62
98.572
160
62
98.564
160
62
98.549
160
.
62
98.268
160
62
98.313
160
62
98.2905
160

Despite being a quite 'new' game, Splinter Cell: Chaos Theory behaves pretty much the same whether you have 512 MB, 2 GB or 4 GB with any kind of timing.

Serious Sam II - Greendale, 1024 x 768
CL 2 1T 512 MB
CL 3 1T 2 GB
CL 3 2T 4 GB
16
61.034
89
30
61.214
88
29
61.035
88
.
28
63.403
91
26
63.079
91
34
63.316
93
.
27
62.614
89
34
62.626
91
34
62.528
90

Performance graphs are poor indicators of just how smooth a game runs with large amounts of memory. But then again, they're meant to show performance over a span of time (average fps) and of course, can't capture per second frame rates (or with less than a second time measurements). We could display an fps progress graph, however, some stutter occur at random points (Battlefield 2) or occur in less than a second. So, even an fps progress graph can't really capture it. In this situation, a subjective evaluation is more appropriate - we can tell you that from experience during testing and gameplay, 2 GB and 4 GB plays exactly the same with no stuttering as opposed to 512 MB. At least in games that do need the extra memory such as Battlefield 2, F.E.A.R and Quake 4.

Let's look at the load times.


In msec

In percent

CL 2 1T 256 MB
CL 3 1T 2 GB
CL 3 2T 4 GB
CL 2 1T 256 MB to CL 3 1T 2 GB
CL 2 1T 256 MB to CL 3 2T 4 GB
Battlefield 2 152692 44108 44903 -71.11% -70.59%
Call of Duty 5061 4910.33 4987 -2.98% -1.46%
FEAR 33988.33 3594 3778 -89.43% -88.88%
Quake 4 256521 19782 19767.33 -92.29% -92.29%
Serious Sam II 9960.33 8270.33 8198 -16.97% -17.69%
Average


-54.55% -54.19%

Well, the most obvious differences will be in Battlefield 2, F.E.A.R and Quake 4. 70 to 90 % less time to load is quite something - from basically 4 minutes to just about 20 seconds in Quake 4. Of course, those savings are actually the same regardless. of whether you have 2 GB or 4 GB. The games (or all applications for that matter) only care about memory larger than what it needs, not how much larger. So, why choose 4 GB? Well, despite what the developers and everyone suggest, some people do like to run other programs on the background when they're playing, particularly if you're online. More programs running at the same time means you have to have more resources, so if you're one of those people, you might want to consider 4 GB instead of 2 GB. After all, the performance difference if you're using an Athlon 64 system is minimal and memory modules are cheap these days. Those who might need that extra bit of performance could opt for 2 GB, probably with 4 single banks 512 MB modules capable of lower latencies (2.5 or maybe 2). They are available, but you have to be willing to pay more money for them than the standard, value oriented modules.

Conclusion:

Having more memory won't provide you with higher performance (lag free gaming), but it will help stuttering issues with some memory hungry games. Only timings does, but your mileage may vary with your processor. However, not all stutter will disappear, only those related to memory. Stuttering caused by loading a map / level or saving a savegame will probably not disappear with having more memory. These are storage related stuttering, and we'll look into solving those issues in another article. As we said before, if you're experience lag in gameplay and you've updated your drivers and apply the latest game patches, you will likely have to use a faster processor and / or graphics card or turn down some details or the resolution you're using to play the game.

With an integrated memory controller, the Athlon 64 is less prone to memory tweaks. Processors with less bandwidth potential such as a smaller cache processor (like the Sempron) or a bandwidth starved one (higher multipliers) will probably benefit more than our Athlon 64 3500+. However, the impact will likely be at most 10%, which is consequently the same performance difference between the Athlon 64 3500+ and it's higher clocked sibling (the Athlon 64 3800+).

Even with lower timings and Command Rate, you will likely experience an increase in performance no more than 3 to 7 percent, with the most gain (overall) is achieved by changing Command Rate from 2T to 1T. Using low latency modules does provide some performance increases, but they either don't have as big of an impact or you have to pay more for these low latency modules.

That said, there are other benefits to having more memory - if your applications and games are able to make use of it. Much less stuttering (what we're focusing on this article) and far shorter load times are what we've shown here today. Newer games are making use of higher resolution textures and more texture layers (texture, specular, normal maps, etc). If you're planning to run games at their highest setting, preferably with a fast, high end graphics card, your PC must have the extra memory to store all those data. 2 GB should be enough for these new games and for most users, but if you can tolerate the slightly lower performance of using Command Rate 2T, you can upgrade to 4 GB. For lag free gaming, we still recommend 2 GB over 4 GB, since games requiring more than that is unlikely to appear within 1 or 2 years down the road.

[Previous Page]
[Go to top]
[Home]
Disclaimer and Privacy policy.