Now that DDR4 matters why would you buy anything else?
http://www.techspot.com/article/1171-ddr4-4000-mhz-performance/page3.html
I only use computers for spreadsheets, autocad and audio/video so they don't really matter.
>>54702754
If you read the reviews, it shows that it does indeed matter for some of those. Video/audio conversions are much faster on higher DDR4.
You gain about 10% performance on Excel going from ddr4 2133 to 4000. I'd imagine the gap is even larger from DDr 1333 to DDR4 4000 mhz
Handbrake and photoshop absolutely loves those higher bandwidth ram. Whooping 60% performance from 2133 to 4000 Mhz.
>>54702754
ECC ddr4 is still quite expensive so in that case, your right. You could stick with and old K7 and probably be ok though. No need to make things faster like boot times, that shit's just inconvenient right.
>>54702689
I'll care when I see a chart showing its performance in a task that isn't useless
>>54702890
...such as?
>>54702857
I meant watching videos and listening to audio, not editing.
>>54702861
I don't really care, I turn on pc when I open the office and go make a coffee meanwhile.
>>54702934
>>>54702857
>I don't really care, I turn on pc when I open the office and go make a coffee meanwhile.
Oh ok, I see. I used to make a sandwich and wait for my 1200 -baud acoustic coupler to load my favorite bulletin board as well but I really didn't care because I like sandwiches.
>>54702689
>Fallout 4
That game is one of the very few whose performance scales almost linearly with memory speed. Most other games will not show such gains even with such high memory frequencies.
>>54703821
But they do show gains, showing bandwidth does actually matter, and ddr4 is "better."
>>54703821
It's happening in more and more games, it's not JUST fallout.
Expect DX12 and Vulcan to make even greater use of memory bandwidth.
>>54702857
>You gain about 10% performance on Excel
wew, I'll be able to work my way through those spreadsheets in no time at all!
>>54702689
Assuming the speed scales, this is where you get eventually.
>>54704617
10% is a lot if your server uses excel data for some reason, those 10% adds up over time. If you're upgrading over 1333 mhz for example, the scaling would probably be around 20%, thats quite a bit of efficiency.
>More and more modern games benefit a lot from RAM speed
>DDR4 prices are identical to DDR3 ones now
>/g/ cucks would still have you buy obsolete DDR3 shit
Why should anyone hear this shitty board again?
>>54702689
It only helps CPU bottleneck situations, and only on skylake with DDR4, still doesn't matter for 99% of people
>>54705269
>and only on skylake with DDR4
works the same with haswell-e too.
>>54705283
even less people would have that
final page, single 980ti, effectively no reason difference
still really doesn't matter for RAM speed, if it's faster at the same price it'd be worth it, but otherwise, it's price climbs really quickly
>>54705334
That doesn't show me anything really, is it skylake or haswell-e? is it dual or quad channel memory? What is the density of the memory? CAS timings?
>>54705437
It's the same CPU as the other tests a 6700 at 4.5ghz
>>54705334
>it's price climbs really quickly
price difference between 2133Mhz and 3200Mhz is little enough that you might as well get the faster RAM if you can afford the extra $20-30.
>proof fallout4 is a POS depending on your hardware
thx OP
>>54705334
Those results directly contradict Digital Foundry's, who found that not only does The Witcher 3 scale extremely well with faster RAM (up to around 3000MHz), but that faster RAM actually makes a bigger difference than a CPU overclock.
https://www.youtube.com/watch?v=Er_Fuz54U0Y
>>54705510
You need a $100+ motherboard rather than a $40 board if you want to run RAM at 3200MHz speed.
>>54706259
If you have a K series CPU you'll have that better motherboard anyway so your argument is a bit pointless for most people.
Anyone building a PC for gaming would be stupid not to get a K series CPU, it easily adds another 2-3 years of useful life span to a CPU.
>>54706301
K CPUs are overpriced. It's cheaper to just upgrade more often.
>Non-K
http://pcpartpicker.com/p/vCNMf7
>K
http://pcpartpicker.com/p/9KzR3C
The K variant costs almost TWICE as much for maybe those 2-3 extra years.
>>54706437
Wow you're a fucking retard. You think an $80 cooler and a $150 motherboard is required?
You can get a $20 heatsink and a $100 motherboard.
That makes it cost about 1/4 more but lasts 2-3 years longer.
Dont be retarded.
>>54704745
wtf if FO4 doing to cause this?
The last time I bothered to check if RAM makes a difference in gaming performance, this was happening
What the hell changed? Is this a Fallout thing?
>>54706437
>its cheaper to just upgrade more often
>he never bought a 2500/2600k
Hahahahahahahha
>>54706605
>technology has moved on since 2010
Wow, breaking news!
>>54706605
>Is this a Fallout thing?
I think so. I guess Fallout is just written like shit or something.
>>54706741
I'm still using my Q6600.
>>54706927
For some reason Fallout is streaming a lot of data constantly in and out of RAM rather than holding stuff cached. Thats the reason why memory bandwidth has such an impact.
The game isn't even using a lot of texture data, and the geometry is nothing special either so I can't figure out what they're doing it for.
>>54706927
You think wrong, dumbass. >>54706231
>>54704745
Is that with APU/iGPU?
>>54706990
Nope.
>>54706990
Pretty sure that's only with 2 980tis, one 980ti here is nothing
>>54705334
>>54704745
>double the fps with higher speed RAM
I don't believe this at all
>>54707045
nah dude it's magic
Basically a cheap tier upgrade for your gpu.
OP is troll. Bethesda are known for very shitty game optimizations. The system RAM in this case is literally being used as vRAM for some retarded reason.
This is why you can only get about 30-40 FPS on max 4k settings with an i5 and a Fury X GPU. Whereas you can get about 80-90 FPS on max 4k settings with an i5 and a Fury X GPU playing BF4
The kicker? Battlefield 4 looks better and mote realistic.
Is it only Fallout that runs faster with more RAM speed?
>>54702689
I'm curious what a newer CPU with eDRAM L4 would do with these benchmarks.
Most games don't have huge working sets, are there are other ways of getting big bandwidth bumps than just cranking DRAM clocks through the roof.
>>54702861
>boot times
why the fuck is this memed about so much on /g/
who the fuck restarts that regularly that it matters
and who the fuck just sits there and stares at it while it boots?
it's not like you're not going to waste far more "productive" time on your system playing shitty games or doing absolutely nothing anyway
http://www.anandtech.com/show/8959/ddr4-haswell-e-scaling-review-2133-to-3200-with-gskill-corsair-adata-and-crucial/7
>>54707256
Pretty much. System RAM is NEVER supposed to be used for situations where high bandwidth affects overall performance.
>>54707256
No. Lots of games see improvement, but mostly only up to 3000Mhz, beyond that you don't see any real improvements.
>>54708212
Also though, the difference is not as big as what is in the FO4 benchmarks here. On average going from 2133Mhz to 3000Mhz will only get you maybe 3-4 FPS in your typical title. It's not groundbreaking but it is there. If you're running some slow ass 1333Mhz DDR3 you can see real performance increase from overclocking it.
>>54707341
>I'm poor I can't afford newer faster stuff.
Seriously dude boot times used to be really slow, like 5 minutes slow. Progressively components got faster and boot times got quicker, and quicker. This is just another step.
