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

ROG Crosshair VI overclocking thread - Page 1010

post #10091 of 16945
Quote:
Originally Posted by bluej511 View Post

in games and benches the difference between cl 14, cl 16 and cl 18 is barely there, with cl 16 even beating out cl 14 in a few cases.
https://www.techpowerup.com/reviews/AMD/Ryzen_Memory_Analysis/

Thank you for the link. This post supports my point that 2933-14 is identical to 3200-16 in real performance. I used to get 2933 and it ran fine at 14 CL but wanted 3200. Now it runs 3200-16 but the end result is the same. Granted, I have not yet done real app comparisons. Am working on these, will post when ready.

My guess is that 2933-12 is same or maybe slightly better than 3200-14 for memory throughput.
12/2.933=4.091 ns
14/3.200=4.375 ns
Ryzen data fabric may add its nuances.

Would be wonderful if someone could post a table of scores such as
2666-16 CB15=1xxx
2666-14 CB15=...
2933-16 CB15=1yyy
2933-14 CB=...
2933-12 CB=...
...
3200-14 CB15=1zzz
Preferably all the way to 3600 or 3733 if you can get to that speed. Ryzen speed may be fixed, say at 3.8 or 4.0. As opposed to particular apps or games CB is accessible to everyone and is a reasonably harsh test. Without the BIOS setting for CB15 (as not all reading this forum may have that setting).
Edited by 1TM1 - 4/19/17 at 9:21am
post #10092 of 16945
Thread Starter 
Quote:
Originally Posted by Ubardog View Post

Im sorry this is gonna be rude but im trying to be polite. So sorry. This will properly damage our chips in short term use with SAFE SET voltages. for a £270 board this is a major biggy that should have NEVER seen further than the R&D room. The 1.5v on boot loop has fueled this even more for me. Bugs with Vid and offset have been present from bios one and you focus on ram speeds rather than chip killers and Voltages. I feel you tryna please the masses and not focus on what really important. I bought a bespoke mobo I expected a few mishaps on new chipset but this board is winding me the hell up. When else is this happening behind the sences Because i get some weird readings every now and again?

You done well with the bios's and help on this page. there major problems lurking that have been masked in last bios only. Gupsterg and I only use modest volts but if it was someone a little more adventurous then they would have been hitting higher Volts than 1.5v and continued to try and get stable causes god knows how many loops. I have spent so much time in boot loots playing with bkcl and ram timings, I have even got distracted by RL and left pc in loop for about 5 mins. mad.gif Next time I tinker i will check my probit Vcore as well during fails If its showing high then *** !!!

I contacted support and got fobb'ed off with in short "contact the seller for advice" . My replacement back plate come this week if It dose not clear 90% of "MY" problems then im pushing support more.
My volts are higher than most's reported here. I see regular 200A spike on CPU /170-0.4v on SB/ In the past seen reports of 2v on chip from SVI.
Ram Is set 1.35 show 1.413-1.439/ Vcore is offset +0.11875 voltage shows 1.373 in bios But up to 1.46 in HW64/
Just seems like the read outs from board or the voltage controllers are knackered Why should a £270 board show different voltages to another. surely all boards of this standard should be 12v as 12v and 3.3 as 3.3. I have always seen it JUST under but never 12.361 and 3.401 its bs really


I know some of these are reading error. BUT ARE THEY ? i know they will not be what they are stated but still. I have looked though everyone's HW64 SS's and nothing this far out of wack.

I gonna stop my self here biggrin.gif

Angry vent over.
Quote:
Originally Posted by Ubardog View Post

The thing is I contacted Asus support. Got told contact seller. At least Asus would understand it where as laptop direct probably will not.If i got this from amazon It would be exchanged by now. I expected more from asus really.

As for elmor say they have know about this issue again angered me over the whole thing. Why even put 0.1% effot in to anything else if a bug like this existed. Always safety 1st. If this was in a building trade and a potentially safe Voltage was KNOW to give a lethal voltage and repeated ignored IE 110v giving out 440v (1.35v giving 1.5v+) If reported to proper people the whole site would be shut down and there would a HSE investigation and heads would roll.

I know all my volts show high, I can work around this but when there is massive drops and peaks of voltages and amp's I can not.

On a positive note Today my official Nxzt am4 bracket arrived today. Its Fitted and im running IBT max as I type
Completely free Shipped from USA to Uk

Testing beginning again biggrin.gif

Hoping it clears some readings up.

It's an issue with AMD CBS along with a SMU firmware bug, it's the same on any other board which enables this menu. The alternative is that we remove the option. I'm sure you would not prefer that? I'm looking into possible fixes but this is the status at the moment. If you want to avoid this bug, don't change VID or use Manual mode.

Unreliable readouts you'll have to take up with Martin.
post #10093 of 16945
Quote:
Originally Posted by 1TM1 View Post

Thank you for the link. This post supports my point that 2933-14 is identical to 3200-16 in real performance. I used to get 2933 and it ran fine at 14 CL but wanted 3200. Now it runs 3200-16 but the end result is the same. Granted, I have not yet done real app comparisons. Am working on these, will post when ready.

My guess is that 2933-12 is same or maybe slightly better than 3200-14 for memory throughput.
12/2.933=4.091 ns
14/3.200=4.375 ns
Ryzen data fabric may add its nuances.

Would be wonderful if someone could post a table of scores such as
2666-16 CB15=1xxx
2933-16 CB15=1yyy
...
3200-14 CB15=1zzz
Preferably all the way to 3600 or 3733 if you can get to that speed. Ryzen speed may be fixed, say at 3.8 or 4.0.

Any comparison also requires a full snapshot of all associated memory timings (beyond primary memory timings) to make sure you know where the gains are coming from. When/if AMD open things up, the testing will be more valid.
post #10094 of 16945
Quote:
Originally Posted by crossbone View Post

The question though is - are you also on the CU update and have you tried the RealBench Stresstest, to confirm?

Yes; standard balanced plan. I've run every benchmark/stress test out there the past couple weeks on maximum settings.
Quote:
Originally Posted by ggdfdgd3 View Post

Are you using the windows default nvme driver or samsung nvme driver?

Btw I can also confirm that the freezing only happens with memory intensive stress tests, OCCT standard or Prime95 Small FFT for example works fine.

Possibly something to do with paging so maybe disabling the pagefile could be something to try?

Using the W10 driver.
Edited by madweazl - 4/19/17 at 10:22am
Inferior FTW!
(22 items)
 
FrankenRyzen
(13 items)
 
 
CPUMotherboardGraphicsRAM
i5 6600K 4.5@1.320v Gigabyte Z170X-Gaming 7 Founder's GTX 1070 200/800 Corsair Vengeance LED 3200@4000 
Hard DriveCoolingCoolingCooling
MyDigitalSSD BPX 120 (RAID 0) EK-FB GA Z170X Monoblock XSPC RX360 (White) XSPC RX240 (White) 
CoolingCoolingCoolingCooling
EK Vardar F2s (x5) EK-FC1070 GTX1070 (nickel/plexi)  EK GTX1070 FE backplate EK Double Bay PWM D5 MX 
CoolingCoolingOSMonitor
Bitspower 250 Z-Multi Res EK HDC fittings (12mm) Windows 10 Samsung C32F391 
KeyboardPowerCaseMouse
Razer Blackwidow Chroma EVGA Supernova G1 1000w Phanteks Enthoo Primo Razer Naga Epic 
Mouse PadAudio
Razer Vespula Sennheiser HD598 
CPUMotherboardGraphicsRAM
Ryzen R7 1700 @3.9 Asus Crosshair VI Hero Radeon HD 6950 G.skill RGB 3200 @3500 C16 
Hard DriveCoolingOSMonitor
Samsung 960 Evo CM 212 Evo (push/pull) Windows 10 Pro 46" Sharp Aquos 
KeyboardPowerCaseMouse
Razer Anansi 750w Keian CM ATCS 840 Logitech G300 
Mouse Pad
My desk... 
  hide details  
Reply
Inferior FTW!
(22 items)
 
FrankenRyzen
(13 items)
 
 
CPUMotherboardGraphicsRAM
i5 6600K 4.5@1.320v Gigabyte Z170X-Gaming 7 Founder's GTX 1070 200/800 Corsair Vengeance LED 3200@4000 
Hard DriveCoolingCoolingCooling
MyDigitalSSD BPX 120 (RAID 0) EK-FB GA Z170X Monoblock XSPC RX360 (White) XSPC RX240 (White) 
CoolingCoolingCoolingCooling
EK Vardar F2s (x5) EK-FC1070 GTX1070 (nickel/plexi)  EK GTX1070 FE backplate EK Double Bay PWM D5 MX 
CoolingCoolingOSMonitor
Bitspower 250 Z-Multi Res EK HDC fittings (12mm) Windows 10 Samsung C32F391 
KeyboardPowerCaseMouse
Razer Blackwidow Chroma EVGA Supernova G1 1000w Phanteks Enthoo Primo Razer Naga Epic 
Mouse PadAudio
Razer Vespula Sennheiser HD598 
CPUMotherboardGraphicsRAM
Ryzen R7 1700 @3.9 Asus Crosshair VI Hero Radeon HD 6950 G.skill RGB 3200 @3500 C16 
Hard DriveCoolingOSMonitor
Samsung 960 Evo CM 212 Evo (push/pull) Windows 10 Pro 46" Sharp Aquos 
KeyboardPowerCaseMouse
Razer Anansi 750w Keian CM ATCS 840 Logitech G300 
Mouse Pad
My desk... 
  hide details  
Reply
post #10095 of 16945
Higher ram speed is better even if same latency as lower tighter timings because the infinity fabric runs at 1/2 ram speed. I am running at 3430mhz C16 ram and it is faster than 3260mhz C14 ram by a little in benchmarks, but more importantly a lot more in games where Ryzen is lacking because the 2 ccx cores need to talk to each other as fast as possible and that is where the real penalty is. I've seen benchmarks of gaming Ryzens with 3600 C16 ram that are beating Intel 7700 overclocked. That is how important fast ram/infinity fabric speed is. Latency is not very important. Sure hope 3600 mhz ram speed is an option in May
post #10096 of 16945
Quote:
Originally Posted by badhairguy View Post

I'm sure this has been mentioned several times already, but if you are having issues with improper temps and/or fan curves because of the +20 offset, play around with the SenseMI skew number in the bios. My 1700X at stock clocks and voltage with an H110i was idling at 60C which I knew couldn't be right, so I set senseMI skew to 288 from 272 and now I'm getting 35-38C at idle, 60C under load.

Having said that, you are never really going to know what your temperatures are, but with the offset we are already in that boat only with senseMI skew adjusted, now you won't get throttling or the crazy fan oscillations.

Ok, thx

Does SenseMI Skew change tctl then? Would make sense since you say you don't know you exact temps?

The optimal solution would be:

Teh reported CPU temp by the BIOS is obviously a mix of temperature readings including tctl - so if this mix would be calculated via tctl - 20c offset things should be a lot better.
This would make the fans quieter and also still show the correct tctl values - so you always know the correct temps
post #10097 of 16945
Quote:
Originally Posted by newguyagain View Post

Ok, thx

Does SenseMI Skew change tctl then? Would make sense since you say you don't know you exact temps?

The optimal solution would be:

Teh reported CPU temp by the BIOS is obviously a mix of temperature readings including tctl - so if this mix would be calculated via tctl - 20c offset things should be a lot better.
This would make the fans quieter and also still show the correct tctl values - so you always know the correct temps

Yeah it changes tctl
The Beast
(15 items)
 
  
CPUMotherboardGraphicsGraphics
AMD Ryzen 1700X @ 4GHz (32.25X120.4) ASUS Crosshair VI Hero (0082) MSI Nvidia GeForce GTX 1070 Gaming Edition MSI Nvidia GeForce GTX 1070 Gaming Edition 
RAMHard DriveCoolingOS
G-Skill TridentZ F4-3000C14D-32GTZ @ 3210MHz (2... Samsung 960 EVO NVMe m.2 500GB Corsair H110i Windows 10 64bit 
MonitorKeyboardPowerCase
LG 34uc87 ultrawide 3440x1440 Corsair Strafe Corsair AX760i Platinum Corsair 460X RGB 
Mouse
Logitech G602 
  hide details  
Reply
The Beast
(15 items)
 
  
CPUMotherboardGraphicsGraphics
AMD Ryzen 1700X @ 4GHz (32.25X120.4) ASUS Crosshair VI Hero (0082) MSI Nvidia GeForce GTX 1070 Gaming Edition MSI Nvidia GeForce GTX 1070 Gaming Edition 
RAMHard DriveCoolingOS
G-Skill TridentZ F4-3000C14D-32GTZ @ 3210MHz (2... Samsung 960 EVO NVMe m.2 500GB Corsair H110i Windows 10 64bit 
MonitorKeyboardPowerCase
LG 34uc87 ultrawide 3440x1440 Corsair Strafe Corsair AX760i Platinum Corsair 460X RGB 
Mouse
Logitech G602 
  hide details  
Reply
post #10098 of 16945
Regarding the memory latency, and performance, some things are not easily explainable with Ryzen. I see no performance difference between 1T and 2T Command Rate for example, using 2x16 GB Dual Ranked at 3200-18-16-16-36. Also several tests by now proof you get a better performance using Dual Ranked modules, compared with Single Ranked modules using the exact same settings (because of rank interleaving), also the Data Fabric seems to mask some improvements from lower memory latency settings, which is why the higher memory frequency seems to be more important.
post #10099 of 16945
Quote:
Originally Posted by badhairguy View Post

Yeah it changes tctl

Ok then its not an option.

There has to be a way to see the unchanged values as AMD wants them to be displayed and at the same time achieve a quite fan profile.

Another option which would work is that Asus could change its FanXpert to allow a custom fan profile curve of up to 95-100C for Ryzen "X" CPUs - which then would lead to a quite PC and alos knowing all the "real values".

@elmor
Would that be an Option?
post #10100 of 16945
Quote:
Originally Posted by newguyagain View Post

Ok then its not an option.

There has to be a way to see the unchanged values as AMD wants them to be displayed and at the same time achieve a quite fan profile.

Another option which would work is that Asus could change its FanXpert to allow a custom fan profile curve of up to 95-100C for Ryzen "X" CPUs - which then would lead to a quite PC and alos knowing all the "real values".

@elmor
Would that be an Option?

If you want to know the real value then change sensemi skew so it reads 20C less. AMD added +20C offset as we all know.
The Beast
(15 items)
 
  
CPUMotherboardGraphicsGraphics
AMD Ryzen 1700X @ 4GHz (32.25X120.4) ASUS Crosshair VI Hero (0082) MSI Nvidia GeForce GTX 1070 Gaming Edition MSI Nvidia GeForce GTX 1070 Gaming Edition 
RAMHard DriveCoolingOS
G-Skill TridentZ F4-3000C14D-32GTZ @ 3210MHz (2... Samsung 960 EVO NVMe m.2 500GB Corsair H110i Windows 10 64bit 
MonitorKeyboardPowerCase
LG 34uc87 ultrawide 3440x1440 Corsair Strafe Corsair AX760i Platinum Corsair 460X RGB 
Mouse
Logitech G602 
  hide details  
Reply
The Beast
(15 items)
 
  
CPUMotherboardGraphicsGraphics
AMD Ryzen 1700X @ 4GHz (32.25X120.4) ASUS Crosshair VI Hero (0082) MSI Nvidia GeForce GTX 1070 Gaming Edition MSI Nvidia GeForce GTX 1070 Gaming Edition 
RAMHard DriveCoolingOS
G-Skill TridentZ F4-3000C14D-32GTZ @ 3210MHz (2... Samsung 960 EVO NVMe m.2 500GB Corsair H110i Windows 10 64bit 
MonitorKeyboardPowerCase
LG 34uc87 ultrawide 3440x1440 Corsair Strafe Corsair AX760i Platinum Corsair 460X RGB 
Mouse
Logitech G602 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD Motherboards
Overclock.net › Forums › AMD › AMD Motherboards › ROG Crosshair VI overclocking thread