The AMD Ryzen All In One Tread /Overclocking/Memory Speeds & Timings/Tweaking/Cooling

So aida64 still shows 74ns and its not Aida64 flaw then?

 

it looks like it , in the download instructions it says this:

( Posted | March 28, 2017 )
Today FinalWire released a new stable update to the desktop editions of its popular system information software. The new AIDA64 release implements optimized benchmarks for AMD Ryzen “Summit Ridge” and Intel “Apollo Lake” processors, supports the upcoming Microsoft Windows 10 Creators Update as well as the latest graphics and GPGPU computing technologies by both AMD and nVIDIA.
AVX2 and FMA accelerated 64-bit benchmarks for AMD Ryzen processors
Microsoft Windows 10 Creators Update support
Optimized 64-bit benchmarks for Intel Apollo Lake SoC
Socket AM4 motherboards support
Improved support for Samsung NVMe SSDs
NZXT Kraken X52 sensor support
GPU details for nVIDIA GeForce GTX 1080 Ti, Quadro GP100, Tesla P6

 

It's only the Trial Version, but here's my AIDA64 Memory Benchmark. I have not attempted to tighten the timings any on my RAM after getting it stable at 2933mhz.... I was tired of dragging my case out to reset the cmos so I have left it at the rather odd 14-14-21 timing until there is a newer bios available.




img photos

 

Well it seems as per this htread http://www.overclockers.com/forums/...e-LED-2x8GB-DDR4-3200-CL16-CMU16GX4M2C3200C16

Mine are v4.24 E-die so that means Dual rank.
Time to shop for v4.31's, the only Corsair B-die kits...

Now if I get B-die but in 2x16GB format I would still be with a dual rank config as opposed to 2x8GB? Just want to make sure I got this right...

 

well the 1.1 BIOS is quite raw. using corsair dominator platinums @ 2400, cas14 @1.22v resulted in a system that ran prime95 blend for 30mins no problemo, handled light gaming for an hour perfectly fine...then the system went to sleep & the OS was irreparably corrupted. couldnt repair install, even from a disk (!)



i was surprised to see micron in corsairs hand-picked IC model. figured it would be samsung for sure...as yall can see, its dual rank, so basically im going to be ironing out stability problems until its addressed @BIOS level. i think its worth it to downclock them temporarily, as they tend to OC pretty well & zen likes mem mhz quite a bit.

anyway increased vdimm to 1.25v, vpp to 2.6v, & vref to 0.63v. been a bit gunshy about pushing 2666, but i might increase the voltages+nb & give it a shot shortly.

still waiting on AM4 bracket (preordered it) for corsairs h115i. corsair said they started shipping yesterday. ill give yall update on the 1700 mhz ceiling as soon as i can start overvolting it

 

Goddamn memory management blue screens. Need to run memtest. Also it could be any driver pretty much. Started after I got ipvanish and new geforce drivers. Ah annoying

 

Here is mine (I was right it is Samsung - D-Die)

 

Some benchies for AoS new Build (more than +20% for ZEN k17)

Spoiler

 

I'm guessing stock 1700 there?

 

@Chipsy, here cookie for You:

The L3 latency is now displayed correctly ~10-11ns (with Your RAM ~8-9ns)
For Fabric RAM latencies at ~50-60ns we need to wait for new ASUS BIOS
It will be this week IMO :nerd:

Spoiler

 

Mine... -_-"

 

Anyone using MSI boards managed to get "close" to their target vcore?

I'm @ 1.376Mhz - LLC mode 5, which results in 1.384Mhz idle and 1.368 load - If I compensate with a higher LLC then the idle overshoot goes up

VRM's also seem to get quite toasty under load.

Currently struggling to get a 1800X and MSI x370 Pro carbon stable @ 4Ghz with the above, wondering if anyone can give me some pointers?

---

On another subject, the ram's running @ 3200 C16 which is good

NB voltage @ 1.05 - not sure if this has helped. Not had AMD is years so, safe to say not 100% sure on a few things e.g. this NB voltage

(spec in my profile is my other "non-work"station)

 

RAM being used Samsung B-die

Spoiler

 

high levels of LLC can & will do that. its meant to offset vdroop @load, & yeah it definitely ends up messing with idle voltages because of how its designed. which board do you have? i havent overvolted the 1700 on that pro carbon yet since its on the stock cooler right now. as such, cant comment on VRM temps or LLC behavior...but my initial advice would be to turn off LLC & find out exactly wat voltage your chip needs to stabilize under full load, ie prime or OCCT 4.42...then lower vcore a notch or two & slowly increase the LLC to a moderate level at most.

its not a good idea to run high levels 24/7 unless you have to - a bit of vdroop from the board helps prevent voltage spikes while a bit of LLC prevents voltage dips. the best situation for your hardwares stability & longevity is a happy medium

as an aside, ive heard a few people say bumping the nb voltage above stock helped stabilize higher mem frequencies. i havent experimented with that either because of the memory corruption i ran into earlier...lol. n1 getting yours @3200!

 

If i'm not mistaken, isn't there a new bios for the Crosshair VI? 1002?

 

Been out for a while. But they did add it to their website yes. Not the AGESA one tho.

 

i assume that's the one to wait for? i'm on 902

 

Managed to get the rig booted @ 3200 but was not stable. Did not bump up the NB voltge, just the dram. It took 1.395 before the bios would even post. 1.3900 still would error out stating that the overclock failed.

 

Let me guess, your ram says ver 3.31 on the sticks?