quite some headroom there if it's a good set, what are you running it at currently?

Ok so 1 dimm per channel, 4/8 slots used each with b-dies.

Don't have any experience with quad channel myself, but maybe your imc/board is running into limitations on oc after 3200?
But if you were able to hit 3700 with mixed sticks that wouldn't make much sense then.

You can change the order of your sticks too, sometimes that can make a change.
How about giving 3600 MHz 16-18-16-38-1T a try? Leave others auto first.
If issues maybe assign some subs:
tWR 12
tRFC 300
tRRD_L 6
tRRD 4
tWTR_L 10
tWTR 6
tRTP 6
tFAW 20
tCWL 12

Good luck.

that's no quad channel...

haha, new toy incoming @owikh84?

So the two of you already seem to be quite committed. Wow.

Motherboards seem to be really lovely this time, lots of tweaking options n very good components.

Regarding DJR I haven't read anything about it yet, but looking at it, it seems to me that they might not be able to tighten timings on lower frequencies. But for Intel it seems to be a nice alternative. What's the availability & price on these locally?

Short story right there 😉
Yeah I'm staying on Zen2 or at least AM4 myself until Meteor Lake comes to light which will be when I reevaluate.

Short addition to your words on IF. Zen3 should get latency further down as there will be 8 cores per ccx eliminating cross talk via IO. So yeah, steady improvement. 3100 cs 3300 is a great preview on this. Shared cache and IO interference vs exclusivity and no rerouting. Now AMD are mostly still compensating with caches.

The intel boards (not looking at the price) are however quite nice. But saying that up there I wasn't implying I was positively taken by the CPU's.

The most exciting thing probably is ASRock making a run again and allowing TDP to be adjusted for 65W chips on non z boards to 125W and by that raising base frequency (hence BFB). Lets see if Intel is allowing it this time. I have a feeling given the competition they might.

This post has been edited by nrw: May 12 2020, 12:20 AM

in one of the those yeah... indeed not ethical.
however it seems that he has been using a board with 4-DIMM layout, too.

Is the silicon prediction even remotely accurate?

I can't imagine how it would be able the reliably predict silicon quality on first boot. (Unless they changed this)

Next time I use a few emojis just to make sure.

Waiting some some deeper testing content now, enjoy tinkering.

This post has been edited by nrw: Jun 6 2020, 04:33 PM

Wouldn't be the first time for ASUS. Perhaps it can be narrowed down to a specific memory pcb type or have you ruled that out, already?

Great testing, you've got some great dedication towards Intel if you spent your own money on this gen with couple boards.
Don't bother about tFAW anything less than x4 is subbed anyways as you surely know.

I remember ASUS having issues on 2-DIMM boards with a2 pcb layout some time back.
But I kinda doubt this would happen to them across 'board' again on an entire platform.

Ye I noticed these sticks, saw one kit up for sale where you said you were going for the new 17-18-18-38 4400s Royals.
I also wanna play with one more kit before DDR5 times but Teamgroup is pretty much not approachable in Malaysia, neither directly nor though dealers.
Guess I'll have to order that one overseas, as well.

oh nah, I should have clarified.
Boards with 2 DIMM-Slots (the old APEX line) not 4 DIMM Boards with 2 DIMM configuration.

I'll drop you a PM about the latter, thanks & appreciated!

lowyat.co.kr

132ns, crazy.

You're attracting some big names right here cstkl & owihk

damn owi, ur 8 threads are strong. wth happened there

not much of a difference between these 2 right now then.

I just ran the bench twice earlier after seeing this.
https://www.3dmark.com/3dm/63181883 this is the better run

I don't really think with it's length (too short) and the cpu utilization (not even 100% when running full threads despite power limit not hit and cpu well below 90C) this benchmark is any good to be fair.

Nah not all, I'm running by core (+specific) as daily. 54, 54, 53, 53, 53, 52, 52(51), 52(51). Thermal throttling >=90°C. Cache stock.
The one in brackets is my daily, the upped 52 was for the run above. Both should be stable albeit having just tested 51 extensively.