Need help bad

[jasonalwaysready]

so:

these super extreme 2666 sticks cost $188 and on extreme settings gained 2-3 fps.
average schmoe ram from name brand manufacturer costs $60

$190 - $60 = $130

that difference applied to the gpu will get you a 7970 or a 670

how many more fps can you get out of a 670 than a 660 ti?


so for $10 more get a better set? sure, why not? it's just $10. maybe they look better but i just cant see anything noticeable in performance out of them.

 

[PP Mguire]

10-15% gains give or take.

That's ok, the 600 dollar 2800MHz Corsair Plat kit he did gave it *drum roll* 3fps gain. Vs the same 1333 kit on a 4.8GHz 3770k. On a stock CPU there will be no gains.

 

[Warmonger]

Adding 60-80 bucks replacing RAM and CPU won't give him any cash for shipping costs. Those XMS kits will run 9's at 1.65v easy.

Which is absolutely horrible.

 

[PP Mguire]

Unnoticeable.

 

[Warmonger]

Unnoticeable.

My 1600MHz kit runs at 8-8-8-20 on 1.30v. That's how "absolutely horrible" that XMS3 kit is.

 

[PP Mguire]

There is absolutely no real word benefit between 1600 8's and 1600 11's. You either paid a nice price for your kit, or you have nicely binned IC's. Either way, it doesn't matter. DDR3 prices are slowly rising again, the cheapest decent kit when linked was the XMS3. If he wants to spend extra money on some lower numbers, then that's his prerogative. His money is better spent elsewhere.

 

[Warmonger]

There is absolutely no real word benefit between 1600 8's and 1600 11's. You either paid a nice price for your kit, or you have nicely binned IC's. Either way, it doesn't matter. DDR3 prices are slowly rising again, the cheapest decent kit when linked was the XMS3. If he wants to spend extra money on some lower numbers, then that's his prerogative. His money is better spent elsewhere.

That XMS3 kit costs exactly the the same price as the kit i'm using, for the record. And timings that different will make a real world difference in memory bandwidth. The sticks I have come factory at 11-11-11-28 @ 1.35v, once you play with the timings in the bios I was able to achieve 8-8-8-20 @ 1.30v. Which is way tighter timings using less volts than the kit ran at stock. Or you could go the exact opposite and crank up the frequency (Intel prefers frequency over timings). This same kit will do 2133MHz @ 9-9-9-24 on 1.50v, you can't tell me that's not impressive.

 

[PP Mguire]

What kit, and lets see them run at 2133. For all we know, the XMS kit could do the same. Have you tried them? Does it matter to the OP who wont be tweaking like an enthusiast? That's probably going to be a no.

We aren't memory bottlenecked. Still moot point. There are no performance gains from faster frequencies, or tighter timings. If there was, I would run my kit at full blown 2133.

Edit: You still running the Samsungs?

 

[Warmonger]

What kit, and lets see them run at 2133. For all we know, the XMS kit could do the same. Have you tried them? Does it matter to the OP who wont be tweaking like an enthusiast? That's probably going to be a no.

We aren't memory bottlenecked. Still moot point. There are no performance gains from faster frequencies, or tighter timings. If there was, I would run my kit at full blown 2133.

Edit: You still running the Samsungs?

Still? Their 30NM, which is newer memory than every single kit you can buy today. Samsung's 20NM line is coming out soon. And excuse me for being such a hardware/enthusiast nut. Being an IT Expert is a hobby, I find computer electronics/programming fascinating.

P.S. There is no such thing as a "memory bottleneck", you can max any game on the market on 200MHz 184pin DDR1. The only time you will see a "bottleneck" with memory is when doing extreme amount of multi-tasking or really heavy payloads. This is where higher frequency comes into advantage, higher frequency ram in a machine can handle a large payload better than slightly slower memory with tighter timings (means more responsiveness from the system under load).

 

[PP Mguire]

I saw over on OC you were running Samsung RAM. Those look to be the sticks that a ton of TPU guys gobbled up to try and get ridiculous clocks on and it was a 50/50 chance whether or not it would happen. Back then, I just bought the HyperX T1s and ran them at 2000 with ease LOL.
No need to apologize, I love hardware and tweaking as well. Just the majority of people who post in these areas are here for a budgeted basic build (usually for gaming) and wont ever tweak to get maximum performance. That's why I was saying (a lot lately it seems) these 1600MHz kits are quite fine for anybody. You kinda proved that too. Memory bandwidth isn't an issue and hasn't been for quite a while. I could run my quad kit at 1066 and I wouldn't see an FPS difference. Hence why I've been saying, 1600 8's, 9's, 11's whatever, these people wont notice the difference because it isn't there.

If they want to benchmark and overclock, sure, I'll point them in the right direction.

I even linked to my review coworker running a game with 2800 Plats. 3FPS difference on a 600 dollar kit of RAM, with a CPU clocked to handle the higher bandwidth. That was also on the IGP, which will deliver better performance with higher system RAM frequency.