DdR4
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
>DdR4 is just a meme
thats a difference of 3 frames
its a meme
>>53385092
Alright then you stick with ddr3 and enjoy not having your better performance
>>53385092
People can see the difference of 3 frames, just because you can't doesn't automaically make it what you 4channers call a "meme".
>>53385092
Why are you using ddr3 then?
Ddr3 is just a meme too.
>>53385157
>see the difference of 3 frames
most certainly, when talking about 1 frame vs 4 frames
>>53385064
what makes this benchmark useless is that even though it's the same chip, it's being used on different boards. Obviously there is going to be a difference, regardless of the fact that one uses faster ram.
>>53385064
skylake supported ddr3?
>>53385430
It still supports it.
>>53385413
Motherboards don't affect performance much. Definitely not enough for 3.5fps in game.
>>53385430
Not officially. But with motherboard manufacturers support.
>>53385464
You're missing the point. A benchmark testing a video card should be done with the same board, ram, and the same chip to ensure no extra variables enter the equation.
The fact that they used different boards with different ram adds that extra variable, thus making the equation (i.e, the entire benchmark) null and void.
What should have been done to test RAM performance (and solely RAM performance) would have been a gamut of game benchmarks for DDR3 and DDR4, then get an average out of them.
All this is screen cap is showing now is how one card used on a board compared to a board with different ram with one game. Nowhere near benchmark tier.
>>53385589
no you fucking idiot they're testing the variable that is ram.
>>53385609
ON TWO DIFFERENT BOARDS
FOR ONE GAME
It's like you've never done a benchmark before.
https://www.youtube.com/watch?v=utWnjA4NzSA
>>53385636
much like how all benchmarks are invalid because they can't test an fc 8350 on the same board as an i5 3470
>>53385674
fx 8350*
>>53385695
>benchmark meant for comparing two chips
and thats why you're not getting it.
See, they go through a gamut of game testing for chips too.
Not one game and suddenly
>ddr4 is just a meme
>>53385064
That's approximately 1-2 generations of CPU performance, 7% overall increase.
I guess that's fine for a few more bucks.
>>53385791
i don't know about America but in aus ddr4 and ddr3 are exactly same price last time I checked.
Also ddr4 is still in its early days. Ddr3 started at 1066/1333, and has gone up to 2133 being available at only a $2 difference to 1600.
Ddr4 should follow suit, since the standard is ddr3 1600, we should see ddr4 3000mhz start taking off at decent prices in 2017, which will be amazing for APUs.
>>53385866
CL14 3200mhz DDR4 kits have been a thing since the beginning of the year. 16GB is $130 or less right now depending on where you look.
www.newegg.com/Product/Product.aspx?Item=N82E16820232205
>>53385172
>Framerates well beyond even 120 Hz
>Still use low-as-shit resolutions
Why do benchmark websites do this?
I get that they're trying to better highlight the differences between RAM types, but that seems a bit stupid to me. If you need to use nonstandard settings for memory to make a difference, then clearly memory does not make a difference. Don't falsify your benchmarks for the sake of making a difference show up.
>>53385866
Nobody fucking cares what costs what in the land of spiders and bugaboos.
>>53385959
You cared enough to reply
>>53385923
Price difference with ddr4 2400mhz?
>>53385944
>far cry
>the resolutions they have listed
>ddr2
this is most likely benchmark results from 2009 m8
>>53385064
Don't game fps generally depend on the gpu and *video* memory?
>>53385172
this!
#nomemesinmymachine
>>53385944
it was undoubtedly taken around 2009. the resolutions are fine.
>>53385064
It would be nice if they mentioned the MHz of each.
I think I might try downclocking my skymeme's DDR4 from 3000MHz to 2133MHz and see what the performance difference is on GTA V.
>>53385157
Only when it is proportionally significant.
"Are there two lions or three?" Is more proportionally significant than "are there 96 lions or 97?"
Why is everything "meme this meme that this is a meme that's a meme" all of a sudden? Have we been invaded?
>>53385992
It depends on everything in your system, but the usual major bottleneck is your graphics card, yes.
>>53386028
"<something> is a meme" is a meme
>>53386011
You got that from vsauce didn't you?
Fucking Micheal hasn't released a new video in months
>>53385064
>3 FPS
>Doesn't show the actual RAM speed
Running 2133 MHz DDR3 and don't think I need to upgrade until DDR4-6000
>>53386028
Jealous, Milhouse?
new mobos *need* DDR4, you don't have a choice. also come back when everybody has 128 GB of RAM and we can store whole databases in memory.
>>53385979
>>53386003
Are you trying to imply it's impossible to set up a proper benchmark in 2009?
Or are you just saying that back then benchmark websites knew even less about technology than today?
>>53386159
in 2009 1920x1080 was high resolution.
>>53386183
And in 2009, 153 fps was nowhere near a realistic scenario for a benchmark.
>>53386183
4K monitors have been around since 2001
>>53386213
what point are you trying to make? that you would like there to be less info in the graph?
>>53386241
My point is rather simple:
Benchmarks are only useful to consumers if they represent real-world scenarios.
Synthetic benchmarks only serve the developers of a product. And I'm pretty sure those don't need some amateur benchmarking website to feed them their statistics.
>>53385157
>4channers
>>53386241
>>53386298
Also, I actually take _offense_ to the fact that they're presenting me with these garbage statistics. It's like saying: “You came here to find out if DDR2 vs DDR3 made a difference. Well guess what, you're getting this fat load of shit instead”
If I look up DDR2 vs DDR3 benchmarks, I would expect to see statistics using _realistic_ parameters. Like say I run a game at 1440p on my GTX 970 and high/very high settings.
The question I need to ask myself is: “How much of an FPS difference would I notice in the real world, _if anything_?” Not “How much of a number change would I see in this meaningless synthetic benchmark?”
Also, a statistic that shows all memory models being tested at the exact same FPS contains just as much useful information as a statistic that shows them being different. It contains data about them being equal, which is just as useful to know as them not being equal.
>benchmarking RAM modules with games
wtf is wrong with these faggots, there are whole programs dedicated to measure cache/RAM speed
>>53386345
A: I am a gamer. I am interested in playing games. I buy expensive hardware for the purpose of playing games. I want to know whether I will notice a difference in games from this hardware, or not.
B: Here you go, see how much difference it makes in actual games and decide for yourself.
C: Why use games to benchmark? What is wrong with you faggot. There are whole programs dedicated to measure cache/RAM speed
See how out of place your argumentation is?
>>53386364
dumbass, you don't need a game to benchmark a game, and a specialized program would obviously be a better option, since it would give you data that's much more reliable than some shitty game
it's like saying "hey, let's compare these two electrical motors in a heavy load truck". do you think that that would make a lot of sense? sure you'd get some results, but... there are tools that measure power, torque, etc. that can give you the actual technical specs.
btw, the pic in the op doesn't even mention the speeds in the RAM modules, or the number of modules, and, you know, DDR3 also supports quad channel.
>>53386437
>you don't need a game to benchmark a RAM module
fixed
>>53386298
is far cry not a real world scenario?
>>53386454
Yes and no.
Yes in the sense that it's a realistic scenario to be playing Far Cry. But no in the sense that it's not a realistic scenario to be interested in performance metrics for.
Far Cry doesn't come even remotely close to taxing the system they're testing on. So, you have a situation where either
1. You significantly overpaid for the GPU in the first place. (If your goal was to run games on the order of Far Cry)
or
2. You don't care about performance benchmarks in Far Cry because your PC is so powerful that it runs at 150 fps either way.
It's like looking up benchmarks for pong. Who in their right mind would do that? There's no useful data to be derived from figuring out which GPU renders at 5000 fps and which renders at 5001 fps.
A game that runs at 150 fps on _every_ tested system is never going to be a realistic scenario for a benchmark, PERIOD.
>>53386437
>it's like saying "hey, let's compare these two electrical motors in a heavy load truck". do you think that that would make a lot of sense? sure you'd get some results, but... there are tools that measure power, torque, etc. that can give you the actual technical specs.
You're optimizing for the wrong questions here.
Using a synthetic benchmark to measure _exactly_ how many kilowatts the engine outputs under nominal peak performance is useless to a consumer. You've gained no information over just looking up that figure in the spec sheet.
The difficulty in sizing and buying the right hardware comes from understanding how technical specifications transfer to real world performance - especially as the number of configuration parameters reach the hundreds.
*Anybody* purchasing hardware has to construct roughly an inverse of this function - by starting with their real world requirements and then projecting backwards to the required hardware.
So data that incorporates this projection (i.e. hardware -> real world performance) are inherently much more valuable than the specs themselves - unless you have a perfect mathematical model of all the physics involved and a supercomputer to simulate it on.
>>53386512
well that was just one of the ones tested. i reverse image searched it and here's the next page.
full article: http://www.legionhardware.com/articles_pages/why_is_ddr3_no_faster,1.html
>>53385064
What about APUs?
>>53386569
So to follow up: To somebody designing a heavy load truck, your measurements of synthetic power and torque will be inherently useless.
Of course, a properly educated engineer will be able to take these figures and approximate the mapping himself in order to draw conclusions about how which motor will perform in the actual truck; but that education requires an understanding of how these parameters transfer to begin with.
The most important point of benchmarks is to establish this understanding; especially because game rendering engines are such incredibly complex systems sensitive to so many parameters that you cannot possibly figure out on paper. You get surprised over and over again.
>>53386581
Thanks, that's a much more interesting benchmark.
>>53385157
>4channers
Get out
>>53386590
Bristol Ridge hasn't been released yet.
There are not APUs using DDR4.
>IT DOESN'T GIVE MORE MORE FPS IN MY VIDYA, MUST BE A MEME
I'm not even shitting on video games here.
using ddr4 right now is some gay shit for real.
It literally provides more of a benefit to The Witcher 3 than CPU overclocking. Tops out around the 3000MHz mark though.
https://www.youtube.com/watch?v=Er_Fuz54U0Y
>>53385944
They didn't have 4K monitors 9 years ago, numb nuts.
>>53386602
Put down the gun you flat-chested piece of shit, we both know you won't do it.
>>53386773
my CRT does 1536x2048. I'm sure there are better models out there that can handle higher resolutions
>>53386618
But theoretically speaking APUs should benefit a lot more from DDR4 than just CPU+GPU configurations right?
>>53387109
Of course, the IGP uses system memory as its buffer.
Though no speed of DDR4 is enough to provide adequate bandwidth.
>>53385064
DDR4 is basically the same price as DDR3, I'll take 3 more frames with on extra money out of pocket. My CPU requires it anyway.
>>53387129
We're currently seeing r7 250s in APUs, hopefully ddr4 can push it up to r7 260x performance.
>>53387721
No. Just no.
Kaveri's 8 CU are disgustingly bottlenecked. Two channels of DDR4 at 4000mhz still wouldn't be enough to deliver adequate bandwidth to 8CU at 800mhz.
>>53386773
>he's too young to know about https://en.wikipedia.org/wiki/IBM_T220
>>53388047
You're an idiot. If it's that much of a bottleneck you wouldn't see an improvement going from an a8 7600 to an a10 7800.
Truth is an apu with ddr3 matches it's dedicated gpu equivalent in performance so well you can crossfire them.
Ddr4 will only make things better.
>>53388534
>>53387721
this.
thanks anons
>>53388826
i can't wait for zen laptops.
Hopefully hybrid crossfire is better then too
>>53388865
>https://en.wikipedia.org/wiki/IBM_T220
>tfw man wasn't that like 8000Nz
>>53388865
me too anon
meanwhile ddr4 wouldn't last.
Intel released specs for Xpoint the next ram type
DIMM fags BTFO
DDR4 is K.I.A. IN ONE GENERATION
hahahahahahahahahahaha
historically this (DDR4) would be the shortest lived ram type ever
>>53389280
shorter than rdram?
![SC[1].png](https://i.imgur.com/LM8fAeq.png "SC[1].png")