re: PC Discussion - Gaming, Performance and Enthusiasts

970 Ti maybe, but I would assume the 980 Ti would be beyond what I want to spend on cards--$550/card is bad enough for me, and I may only get one anyway (I say that, but I know my case is going to look too empty and my loop too extravagant without a second card). My main goal for a long time has been to justify switching to NVIDIA for a change, not necessarily to upgrade but just to get some more hands-on advanced tweaking/OCing experience with team green. That goal was delayed by the brief surge in mining profitability, followed by not being able to resist getting two 290s for only $200 each.

And when you go all-in with water cooling, GPU upgrades get much more expensive as it is because the block/backplate is $150+, proprietary, and difficult to sell.

This fricking guy

The 970s are pretty cheap for the considerable upgrade, $339 isn't bad at all when I can sell the 270x for 110-130.

Or wait a month or two and grab one for around 280-300 used and call it good.

Thinking about upgrading my RAM, I went with the G.Skill X Series 2x4GB (fine with 8GB don't need more) at 1866MHz.

Would it be worth it to get the TridentX Series 2400MHz on sale right now? Mobo can only go up to 2133, but the 2400 TridentX Series is currently cheaper than Sniper Series 2133.

I'm a G.Skill homer so don't both with any other brand.

9, and the TridentX is 10 but the Sniper Series is 9.. But 2133 vs 1866 as well, but that's out of my range of knowledge about just how much better 2133 is vs 1866.

This is what I have in it now.

This is the TridentX

This is the G.Skill Sniper Series 2133

Overclock your ram

1. You'll notice no difference in speeds
2. Its easy to overclock

Here's some info for you, lifted from my guide:



The other timings besides CAS latency also matter, but are really only important when comparing two very similarly performing kits. They all represent some sort of delay in moving data in and out of the memory.

So using that formula

DDR3-1866 CL9
(9/1866) * 2000 = 9.65ns

DDR3-2400 CL10
(10/2400) * 2000 = 8.33ns

Lower is better of course, so the DDR3 2400 kit would have better access time. If you bump the speed down to 2133, you can tighten the timings as well, but with both kits running at their rated speeds, the 2400MHz kit will give about a 10% theoretical performance increase when you account for all the other timings as well. I doubt you'll notice anything beyond a placebo effect.

If you lower the clock to 2133 and only lower the CAS latency to 9 and change no other timings, performance comes out to be close to identical to your current kit. If you can get it from 10-12-12-31 to 9-10-10-28, you're back to about a 10-11% performance increase over your current kit. If you could somehow get it to 2133 MHz at 9-9-9-24 (highly unlikely), you could achieve about 20% better performance over your current kit. But here's the thing. That 2400 kit is running at 1.65v, and your kit is running at 1.5v. I would be shocked if you couldn't just bump your 1866 to 2133 @ 1.6-1.65v without changing any timings. And then you've accomplished the same thing without having to buy new RAM and sell the old, and it only took a minute.

This RAM performance difference will only kind of sort of matter when you are playing games that are highly CPU-bound (and still we are looking at a few FPS difference at best), or if using an APU that relies on system memory instead of dedicated vram.

Yep, just see what it does. If it doesn't like that, try 9-10-10-28, which is what I'd predict that 2400 kit could do at best if lowered to 2133.

yea for 120 bucks it seemed pretty legit