Overclock.net › Forums › AMD › AMD Motherboards › ROG Crosshair VI overclocking thread
New Posts  All Forums:Forum Nav:

ROG Crosshair VI overclocking thread - Page 813

post #8121 of 17798
Quote:
Originally Posted by y0bailey View Post

Sorry guys, just got back from a long trip so I'm WAY behind. Jjust installed the new BIOS's and I am getting WORSE results with my dram 3200mhz. .

CORSAIR LPX 8gbx2, 16-18--18-36 3200mhz that was booting with DOCP standard on 1002, now won't boot (fails over and over, defaults back to 2200) on both 0081 or 0082.

Any ideas? Anyone else done WORSE with the newest bios?

Update: 0083 is working fine (2t instead of 1t). Looking at CPUz, it now shows 2t but the tRC was super high and is now 75.


This consistent with anyone else?
post #8122 of 17798
On Bios 0082 my PC will sometimes double boot and it will automatically set the voltage to 1.5v. It has done this 3 times now, never had any issue with Bios 1002 which I am now flashing back to. This mobo is trying to kill my chip.

This is with Pstates 3.9ghz, DCOP Standard and Vcore Offset.
Edited by ShiftyJ - 4/9/17 at 8:08pm
post #8123 of 17798
Hey everyone, first time post, although I've read every one lol

I wanted to share some results I've gotten with overclocking memory, and see if any of you have seen similar results. I've gotten pretty deep into tweaking parameters, just trying to see which knobs we have available to us.

First thing was that going to the new BIOS (0081 in my case) was extremely frustrating. It seemed like I had gone backwards from where I was on 1002. However, I figured out a couple of things:

#1 The default fail count has changed from 5 to 1. This means that after 1 failed attempt, the BIOS clears all the settings and starts over. On my system, for some reason, it only trains after the 2nd or 3rd attempt, so I was never able to pass until I increased the fail count back to 5. If you get stuck in a F9 reboot loop, the red safe_boot button is your friend smile.gif

#2 For some reason, I had to LOWER the DRAM voltage to 1.32 volts from 1.35 to get my system to work at 3200. I hadn't thought of lowering the voltage before, but for some reason that worked. I wonder if this is related to the memory hole.

So, once I had gotten past these hurdles, I tried to figure out what my memory was sensitive to. I'll skip the details, but basically I put my memory on the edge of passing/failing, and looked for settings which helped. Here's the ones I have found so far:

SOC voltage - definitely impacts things, once I got above like 3400. To push from 3500 to 3600 I had to go from 1.2 to 1.25

VDDP - I also had to increase this to get higher than 3500, to 1.15 Volts. Surprisingly, it failed at 1.2 Volts, so again you can't go too high on this one

DRAM voltage - To get higher frequencies, I did have to increase this. 3600 required 1.45 Volts

Termination values - for my system it seems to like the auto value (53.3 Ohms), looks like every system might be different

Currently I'm running with Trident Z RGB F4-3200CL14D-16GTZR, 2x8MB

So, I was able to boot with the following:

BCLK=134
DRAM voltage = 1.45
SOC voltage = 1.25
VDDP = 1.15
Timings = 18:18:18:18:45
2667 strap (so 3573 DDR)
Core clock = 3.9 GHz
1700 CPU

However, this isn't entire stable running stress, so I back BCLK off to 131 (keeping core clock at 3.9 GHz)
This puts me at 3493 DDR, which I'm pretty happy with smile.gif AIDA64 is showing latency around 72 ns or so. EDIT also backed off VDDP to 1.05

If anyone has similar DIMMs, can you see if you get similar results? I'm about to try some other DIMMs next, I'll let you know what I find out. If anyone has a recipe to push even higher, I'd love to hear about it!!!
Edited by GoneToPlaid - 4/9/17 at 9:26pm
post #8124 of 17798
Quote:
Originally Posted by goncalossilva View Post


It's interesting to see how inconsistent our results are. I have a very similar but technically worse kit, the F4-3200C15D-32GTZKW.

In the new BIOS I've been able to use the 3200 strap as long as I set ProcODT to 80. In fact, out of the box, I can't do anything higher than 2666, unlike 1002 where I was happily running at 2933. For 2933 I need to set ProcODT at 68.8.

This is all at pretty much default settings (RAM at 1.35 of course) and 18-14-14-14-34 timings. Can't post with anything lower. I was using 081 (1T) but had a couple of random BSOD, so I flashed 083 (2T, which is what my RAM is rated for). Unsure if it was actually related with that or with a buggy Creator's Update though.

My 1700 isn't among the golden ones, that's for sure. The CPU isn't stable at 3.8GHz with 1.3V. I'll keep on testing next week, but I have the feeling that we may be bumping into RAM frequency/OC inconsistencies depending on the silicon lottery. And that is something that makes me feel a bit anxious: to think that the IMC of my CPU isn't up for the task of reaching 3200MHz for the memory.

Anyway, I guess we're gonna have to be really patient here, trying different settings and wait patiently for the next couple of BIOS updates. If I find out something else, I'll share it through here as I keep reading and learning from you all thumb.gif

Cheers!
post #8125 of 17798
post #8126 of 17798
This bios has about 237 extra settings that no one ever needs to change. Yet every other motherboard bios I have owned in the last 12 years has had T1/T2 dram settings, yet we need two different bioses to choose? ***?
post #8127 of 17798
Quote:
Originally Posted by hotstocks View Post

This bios has about 237 extra settings that no one ever needs to change. Yet every other motherboard bios I have owned in the last 12 years has had T1/T2 dram settings, yet we need two different bioses to choose? ***?

LOL, I was thinking the same thing, BUT I think this is an AMD limitation because of the architecture
post #8128 of 17798
Seeing you guys hitting the same CPU clock as me with 0.18 less volts makes me sad frown.gif Currently sitting on 3.85ghz with 1.38V. Don't get me wrong I'm beyond happy with the CPUs performance, just sad I cant hit that 3.9-4 range without 1.45V+ frown.gif
post #8129 of 17798
Quote:
Originally Posted by finalheaven View Post

Did you try raising your SOC? boot up issues are usually SOC related. Also not sure if its a correlation, but it appears better overclocking cores come with worse SOC? And bad SOC usually means bad memory capabilities.

My CPU doesn't overclock well... but solid with 3200mhz since day at less than 1.0v SOC. Some people need 1.2v SOC to get to 3200 and some can't get there at all.

I have set SoC Voltage up to 1.15 V.
Bit my Trident Z 3200 Mhz CL14 work fine and correctly also al lower values.
Quote:
Originally Posted by orlfman View Post

a numerous amount of issues could cause those errors. i had similar code errors when i had a corsair h110i mounted when i used it for a short while as i waited for my d15 mounting am4 kit. since it was one of the few compatible coolers out at launch. well, technically compatible. the mounting system wasn't designed for AM4, but rather AM3. but since it used the latches, it could technically work on AM4. the errors went away once i swapped it out with the d15. from what i could tell what was going on was a pressure issue. if one side was tighten more than the other, the pressure would be off, causing code 8 type errors and even inability to boot at all.

other issues could stem from utilizing the original ek rubber gasket for ek water blocks, a not so stable overclock, ram compatibility issues, motherboard issues, to simply a bad processor.

I now have the AM3+ mounting braket for my H110i GTX.
It seems perfectly sitting on the CPU, since I have never noticed strange, too high T°.
I also have checked more than once correct thermal paste spreading between AIO base and CPU HIS, it all seems to be perfectly fine and working.
At 3.85 GHz Idle (no p-states) is 30°C and maximum Full Load is 53°C.
They seems to me to be too low for thinking about a contact problem with AIO base and CPU HIS.
Quote:
Originally Posted by murrayd222 View Post

I've settled on 3.85 GHz as well, but with pstate 0 OC with a vcore offset of +o.16250. My Noctua NH-D15 hits 26C at idle and 59C at load during IBT Max. I can lower the offset by one notch, and it's still stable...but two notches down crashes IBT. For stability, I decided to keep the offset bumped up one notch than necessary though. I tried getting to 3.875 GHz with an offset up to +0.18750, but it was still a no go. For my CPU, going above 3.85 GHz results in a steeper vcore offset than I think it's worth. My 2600k lasted 5 years, and I'd like this R7 1700 to last a while as well. I've failed all attempts to get my ram to 3200 though, best I can get is 2933.
I have had for 1 month a 16 Gb Trident Z 3200 Mhz CL16 kit, which I wasn't able in any way to get work above 2666 Mhz, even with BCLK OC.
Then 1 week ago I bought a 16 Gb 3200 Mhz CL14 Trident Z kit, which instead works incredibly well.
I was able to OC it to 3600 Mhz and CL13 timings without any problem, whith a simple BCLK OC.
In this stage it all depends on compatibility, with RyZen.
Which can be something really frustrating.
Edited by Crysis90 - 4/9/17 at 10:13pm
post #8130 of 17798
Quote:
Originally Posted by GoneToPlaid View Post


SOC voltage - definitely impacts things, once I got above like 3400. To push from 3500 to 3600 I had to go from 1.2 to 1.25

That is really high... I personally wouldn't go over 1.15.... As I understand it 1.2 and above is risking your CPU.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD Motherboards
Overclock.net › Forums › AMD › AMD Motherboards › ROG Crosshair VI overclocking thread