You're right... Thanks
You're right... Thanks
CPU is also important for gaming, maybe GPU does most of the job, CPU is responsible for PhysX if you're running NV card and only have one card in your system.
Overclocking Ram helps CPU work faster, hence, more efficiency for PhysX, hence better gaming experience.
You literally have no idea what you're talking about. The whole point of PhysX is that it offloads the physics load to the GPU from the CPU, it doesn't require a second card. PhsyX only even exists in a tiny fraction of games. With the exception of APU using the integrated graphics improved ram performance will have a negligible impact. Whilst games may well be CPU bound, its unlikely they'll be memory bandwidth bound.
Which 3dmark and have you got a link? A quick look seems to suggest completely the opposite to what you are saying:
from Anand's sandy bridge memory scaling article: http://www.anandtech.com/show/4503/s...-the-best-ddr3
I'm talking about optimized DRAM timing with extremely tighten first, second and third timings.
This picture from online review website is not what I'm talking about. There is no difference if you leave all DRAM timing setting in BIOS to AUTO, which is the case for most online review website.
I don't have any link as it;s my own experience. Besides, the sweet spot for DDR3 is 2800C9 with extremely tighten first, second and third timings, which is not considered in this review.
Do Not trust these online reviews, run some benchmark and compare the results, what kind of review this is. Everybody is qualified to do this.
That review clearly covers the impact of different timings at a variety of speeds, if there was any difference at all, you'd expect it to be more pronounced at the lower speeds. Just to give you the benefit of the doubt, greybeard posted this the other day: http://www.corsair.com/en-us/blog/20...swellrealworld which shows the difference in performance from using higher bus speeds with an overclocked i7. You'll note that there is a difference in heavily CPU bound applications, (which I suspect was why starcraft was chosen for the gaming benchmark) but, given the comparative difference in price, I maintain that its still not worth it until you've hit the top end of every other part in the PC.
I think I have made myself very clear in previous reply, the key is not only first timing shown on these pictures and reviews. When the DRAM freq is getting higher and higher, second and third timing become equally important nowadays. The problem is second and third timing are not considered thoroughly in these reviews. So these reviews don;t fully address the problem.
We understand exactly what you're saying, what I'm saying, and what the results posted by kalneil quite amply demonstrate, is that you're wrong. With the exception of some very niche use cases where memory performance is key latency has next to no impact on performance, nor does RCD or Precharge. You'll note that the graph shows different timings not just for CL, but also TRCD etc.
Noones trying to say they don't matter at all, just pointing out that your assertion that they have a noticeable impact on gaming performance is incorrect.
OK, can you show me the second and third timing used for this reviews? How many times do I have to repeat myself that when the DRAM freq is getting higher and higher, second and third timing become equally important, even more important than first timing in some cases. How much improvement is 'noticeable' may I ask?
Is 2-3 FPS noticeable in gaming?
I think perhaps you're talking about something different, the graph kalneil posted lists Cas Latency, Row To Column delay, RAS Precharge & Row Active time. The only other significant timing not stated is command rate, which may have an impact, but certainly not a significant one.
There are currently 1 users browsing this thread. (0 members and 1 guests)