so wtf is with this new AM4 motherboards only supporting up to

>>58584143
That sure is some high quality and up-to-date source you got there. Let's just wait until they finally release the dared thing along with its specs, okay?

>>58584143
>high end computer
>AMD

>>58584154
>supporting a company that has 1% performance increase between generations

AM3+ boards mostly came with a 1600mhz max with 1866 considered an OC luck of the draw thing. Later high end boards supported 2100+, can't be bothered to check exact speed.

I think those used a memory controller on the north bridge though (right? Maybe I'm wrong) while Zen has the controller on die.

Is it possible do something externally to increase bandwidth if the memory controller really does turn out to suck? Or would it just add too much latency to be worth it?

>>58584214
AMD hasn't had the memory controller on the northbridge for 15 years. You are definitely wrong.

There is a difference between official and unofficial speeds.

>>58584790
So what made later boards able to support such higher speeds? I'm genuinely curious.

>>58584143
Wait till the ryzen is actually released

>>58584198
>and 40% better performance

>>58584143
Jedec standard goes up to 2400mhz. Memory above that is out of spec.

>>58584214
>>58584790
>>58584865
AM3+ all supported 1866mhz officially
FM2 depends on the processor, top end officially supported 2133mhz

>>58584143
oh wow
that's pretty amazing

meanwhile I'm running 3GHz ram cuz why the fuck not

>>58584865
They didn't do any different to support those speeds, later boards advertised it because the IMC was easily capable of those speeds.

You can use 2400mhz ddr3 on an i5 2500k that only officially supports 1600mhz without much issue as well.

>>58584143
That is the official supported maximum speed, just like how Kaby Lake supports an official maximum of 2133/2400mhz (https://ark.intel.com/products/97129/Intel-Core-i7-7700K-Processor-8M-Cache-up-to-4_50-GHz under memory specification), just like how the first gen i7s officially only supported DDR3-1066.

Its the highest speed from both manufacturers that they certify that their memory controller is guaranteed to hit. As anyone running these chips knows, its not the maximum possible speed. A very good 7700k could hit 4ghz+ on the memory controller, and I personally dont know anyone who ran their Nehalem i7 ram at 1066mhz for any extended period as most people had their ram running at 1600mhz or faster.

>>58584143
Most boards shown at CES were 2667+, 3000+, or not yet stated because they can reach higher. One vendor said 4000mhz kits were booting with no trouble.

You really have to be grasping at straws to be posting some obscure Chink shit. You're a low energy shitposter. Sad.

>>58585445
The supported frequency listed by the CPU spec, and what a board lists as supported are entirely different things.

>>58585089
/thread

>>58584198
Not realising that their grip on market share means they've shelved their actual innovation as a contingency

Go study more and shitpost less

>>58585470
>The supported frequency listed by the CPU spec, and what a board lists as supported are entirely different things.
And since the IMC is on the CPU in the first place, its the quality of the CPU IMC that ultimately determines what the whole smash will do.

Besides, as >>58585089 said, DDR4 JEDEC standard caps at 2400mhz. Anything past that is considered out of spec.

My hope is that AMD has managed to sort their shit with their memory and cache subsystem. IIRC it wasnt all that good with the construction cores and for the post-piledriver stuff they had to get a third-party designed IMC in for them, and the less said about their Construction Cores cache the better.

File: 1484861475586.png (26KB, 602x500px) Image search: [Google]

26KB, 602x500px

>>58584143
>he thinks3200mhz is always better than 2400mhz
>he doesnt know what CL14 CL9... is

go back to /v/ kid this board is for adults