greenmaji said:
Any chance of an artical were overclocking, 1:1 vs downward dividers and real world gaming performance would be doable?
Some testbed graphs to look at would be extreamly helpfull.
TIA jmke
we no longer have Core 2 test setup in-house at the moment; but's in our wishlist of things to work on once we do get our hands on Core 2 again.
Dragonprince said:
I don't understand where this mentality has come from
from the world of facts
the game you mention (Oblivion) is one of the FEW exceptions on the rule. The FACT is that 99% of new games are very very very very GPU dependant, sure you need a speedy enough CPU, but we're talking 2ghz A64 or 3Ghz P4 here, as being "speedy enough".
Whether you have 3800+ or Core 2 X6800 in your gaming system next to 7900GT SLI, you won't notice the difference. We're not talking about benchmark numbers here, we're talking about NOTICING the difference;
if you run at lower resolutions with that SLI system you'll go past playable 60FPS point immediately, and anything over that you simply won't notice (and if you're running on LCD/TFT you better have vsync on to prevent image tearing and thereby locking FPS to 60). Now if you start turning up the detail in-game/resolutions/AA/AF which drops performance, you still won't notice UNLESS the game is CPU dependant (I can name 2 now which will benefit Core 2, Oblivion, Rise of Legends) which is highly unlikely... so you're increasing the load on the GPU and thus it becomes the bottleneck.
No matter how fast your CPU gets, your game FPS won't increase, so whether there's a 3800+ inside or X6800, you would not be able to tell the difference
when playing GPU dependant games.
now you go on about CPU.. my statement is about MEMORY timings; which, even in the BEST CASE SCENARIO with specific benchmarks shows a tiny difference; NO WAY will it impact real world gaming, even Oblivion won't show a thing I'm quite convinced.
MEMORY SPEEDS however CAN show a difference, but again, in RW... very very very small, since it's not the memory subsystem which is at it's operating limit, but the GPU
MadMan007 said:
You forget people might buy things based on faulty assumptions, I think that's what this article is trying to show especially for stock speeds.
100% correct, my idea for the article was too evaluate if more expensive memory will give you better performance , comparable to the extra $$ you are paying for it. OCZ's PC7200 DDR2 sticks allowed us to run at a wide variety of speeds/timings and thus the outcome was known after a 14+ hours benchmark session.
if you run your system at stock speeds and don't intend to overclock (I run my
work-station at stock speeds FYI) you're best of with PC4200, unless you can get PC6400 for the same price.
now if you are going to overclock, all bets are off, since you're no longer running a default system speeds, keeping memory 1:1 with FSB will give you that tiny bit of extra performance, and in this case higher rated memory will give you more headroom.
the article is not meant to tell people at OVERCLOCKERS FORUMS new things, if you are overclocking, you already knew most of what the article is saying. It's meant to tell people who DONT overclock and would rather pay less than more, that the Core 2 CPU is happy with cheaper PC4200 and buying higher rated memory won't give you a noticeable performance boost.
edit: pratical example from
http://forumz.tomshardware.com/hardware/GB-DDR2-RAM-speed-ftopict197700.html
I'm a bit of a unexperanced computer builder, and I'm planning on building a gaming PC. I was going to get 2GB of RAM, DDR2 800. My friend told me that for today's games, 2 gigs is fast no matter what, and DDR2 557 would be a better buy, since its cheaper, and at 2 gigs the speed doesn't matter. Is he right? Should I get DDR2 557 instead?
(I didn't post the link the madshrimps article there, somebody else did
)
thank you all for the feedback!