Ram speed doesn't matt-

Yeah, it matters in broken as fuck games like Fallout and ARMA

>>59990643
No you dumb indian nigger. Imagine your game textures are stored in ram. They need to be sent back to the cpu or gpu so they can get reprocessed again and then displayed to your monitor. If your ram isn't fast enough, the cpu and gpu will fuck around doing nothing, waiting for information.

>>59990613
>being bottlenecked by ram in 2017

Take me back when basic AF ram got the job done and was cheap

Wait for DDR5

>>59990712
Or oc some gskill to 5ghz alongiside with 5ghz 7700k lmao

>>59990613

brb ordering a 7700k and 3600mhz ram

>>59990613
>muh poorly optimised gaem

>>59990728
Nice, Noctua mounts for RAM.
Don't forget the steel DIMM slot

>>59990643
Fallout 4 (and all Bethesda games DESU) is probably the worst optimized games I've seen in years. It seems to run fine in outdoors areas but as soon as you step into the city it turns into a slideshow, unless I drop shadow distance.

>>59990671
the solution is to keep the texures in the damn graphics memory

>test high speed memory in 4 games out of 200 that actually benefit from RAM
>WOW LOOK BUY FASTER RAM OKAY

That's what we call cherrypicking

>>59990779
>I buyed slow ram and regret it: the post

>>59990808
>I buyed
you what

>>59990774

this

/thread

>>59990774
>>59990830
You can't /thread over your own reply faggot

>>59990754
That's because the renderer is single threaded and you're getting CPU bottlenecks due to a stupid amount of drawcalls because Bethesda is shit and doesn't optimize

>>59990779
Did you actually watch the video?
He specifically says the extra performance doesn't warrant the much higher cost of higher speed ram unless you have money to burn.
Even makes a comparison of the cost of high speed ram over the sweet spot speed and how you could use that money for better processors.

There's a point of diminishing returns shown in the video.
OP cherrypicked that screenshot to mislead /g/ if anything.

>>59990868
Dude if you look up from the botton there's an 80fps difference. EIGHT TEEN fucking 80, it's not a big difference, it's insanely huge.

Ok. So how long until RAM becomes another component which must be cooled actively? If we had 6 Ghz RAM Ryzen would be even better.

>>59990842
i didn't but surely can

/thread
faggot

>>59990613
Another 2013 meme like muh i5 .Or muh 970/1060

>>59990868
Anyway, RAM speed does matter on "esports" games, which covers a big chunk of gamers.

Does higher ram MHz decrease the latency or just increase bandwith?

>>59990975
It increases both, CAS timings aren't real DRAM latency.

>>59990975
Both
>>59991005
You are retarded

>>59990613
Fast RAM is like a $20 premium over the low tier shit so there's little reason not to but it instead. It's the 4GHz stuff that has garbage value.

>>59990975
Both. Higher frequency means there is less between cycles of new data being transferred, and it also means you're delivering data more frequency so in a given unit of time your total volume of data transferred went up.

>>59991005
CAS timings refers to the latency between the memory getting a command from the controller and the memory actually doing it. And the command can be various things. It's still real latency

>>59990868
>>59990912
Yeah, and we shouldn't have to pay more to get the same performance of other games because of the shit faggot developers that probably have a deal with ram makers.

Fuck it, we're all faggots for playing meme games anyways

>>59990613
intel btfo, just wait till amd gets DDR4 9999 mhz and ryzen will have all the fps

>>59990747
DIMM sag when?

>>59991162
Lmfaoooo youre stupid ask fuck all you basically said is like yeah we should get the same performance from a rx480 same as from a 1080ti it's just the developers lmao. Idk how someone can be so stupid kys lmao

>>59991185
No, not at all. That's a staggering fucking difference between the FPS and the RAM speed when such variables as that should be marginal at best.

Don't forget to get raped and kill yourself, you retarded kike loving fucking faggot sack of ugly nigger shit with down syndrome on the way out

We need something better, DRAM latency is still around 5-6 times slower than the L3 caches, the bandwidth isn't even comparable.

File: 1472854824054.png (60KB, 738x566px) Image search: [Google]

60KB, 738x566px

>graph
1336 FAGS BTFO
HOW WILL THEY EVER RECOVER
KINGSTONIGGERS ON SUICIDE WATCH

File: c3-speed-vs-latency-table.png (279KB, 1200x575px) Image search: [Google]

279KB, 1200x575px

>>59991148
>It's still real latency

But it's completely irrelevant information without also considering the clock speed. People herp and derp about DDR4 having higher latency because of its looser timings, when that is in fact bullshit. A stick of 3200MHz DDR4 with 18-18-18-48 timings would have identical real latency to a stick of 1600MHz DDR3 with 9-9-9-24 timings.

RAM latency hasn't increased over the years, even though timings have, because the higher clock speeds completely offset the looser timings. You get the same latency along with much higher bandwidth.

>>59990712
Is DDR5 even drafted yet?

File: 1385420441583.jpg (251KB, 852x939px) Image search: [Google]

251KB, 852x939px

>>59990643
>>59990754
>>59990740
???
These games are just using more CPU power compared to GPU power

The backend and AI is much more complex than a game like BF1

Am I a meme? I have 2800 and never see it on these charts. It's either vengeance or ripjaws. Should I just OC it to 3000? I have like 5 case fans too so passive cooling shouldn't be an issue.

But after my first OC (which I reverted from after this bullshit) my fucking ports don't turn on half the time when I start my PC. Thanks Intel and Asus.

>>59991618
No 2800 is ok

I got 3200 MHz RAM and it cannot go over 2133 MHz or my PC gets stuck in boot loop. That's what I get for rushing into Ryzen so early.

>>59991725
It sounds like you could have a shit motherboard desu

>>59991725
more like not checking compatibility
cannot believe how many times i've read about b dies or whatever the fugg. out of stock motherboards can't come back in stock soon enough

>>59991725
Sucks man I have a watercooled 5.1ghz 7700k and 4266mhz gskill trident z. Never had a problem

>>59991725
Don't worry, there's still BDW-E motherboards that can't go over 3000MHz 2 years later ;)

Fallout 4 runs at constant 60fps on my ddr3 i7 setup

File: 1353170076811.png (151KB, 274x238px) Image search: [Google]

151KB, 274x238px

>>59991808
The compability list doesn't mean shit on Ryzen boards, my friends b350 tomahawk can only handle 2133mhz while mine can get to 3000mhz, we use the same Corsair vengeance lpx 3000mhz and the same bios, it's compatible according to the compability list. Will probably get another mobo like the ASRock b350 or go over to Intel where it just works.

>>59990613
I've always been told it doesn't matter up to a certain point, not that it doesn't matter at all. This graph seems to support that. For example, the difference between 7700K with 3200Mhz RAM is negligible with 3600Mhz. The difference between 3200Mhz and 2666Mhz is not negligble.

>>59991931
Even the difference between 7700K 3600Mhz and 7700K 2666Mhz is only 10 frames.

>>59991909
Memory controller is all on the Ryzen SoC m8. The motherboard has nothing to do with it. Also did you update BIOS and try different SoC voltages?

>>59990613
>1080p

>>59991969
>having a life, friends, and a girlfriend

>>59991954
The motherboard has everything to do with it, else you wouldn't see people only getting 2600 and some getting 3600 on Ryzen

And you sure as fuck wouldn't be seeing "memory compat" EFI updates coming every 3 days

>>59991993
The BIOS updates improving memory compatibility aren't for the motherboards, they're new AMD AGESA microcode updates for the SoCs. You can look at AMD's Ryzen community blog, it's AMD doing the work to improve RAM compatibility.

https://community.amd.com/community/gaming/blog/2017/03/30/amd-ryzen-community-update-2


>We have reduced DRAM latency by approximately 6ns. This can result in higher performance for latency-sensitive applications.
>We resolved a condition where an unusual FMA3 code sequence could cause a system hang.
>We resolved the “overclock sleep bug” where an incorrect CPU frequency could be reported after resuming from S3 sleep.
>AMD Ryzen™ Master no longer requires the High-Precision Event Timer (HPET).

>We will continue to update you on future AGESA releases when they’re complete, and we’re already working hard to bring you a May release that focuses on overclocked DDR4 memory.

We already have reports from the manufacturers that early versions of the May AGESA based BIOS will have most memory kits hitting 3200 at least.

>>59992064
You're confusing AMD's AGESA update with EFI updates from vendors, both improve memory compatibility.

>>59991909
>we use the same Corsair vengeance lpx 3000mhz

The brand logo on your sticks means absolutely nothing. All that matters is who made the chips underneath, and Corsair use multiple suppliers. Some of their stuff is Samsung, some is Hynix, even on the exact same parts. Ryzen hates Hynix stuff.