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

ROG Crosshair VI overclocking thread - Page 915

post #9141 of 17390
Quote:
Originally Posted by madweazl View Post

Yea, there is a substantial amount of vdroop at load; even at LLC 2 I get 69mv but I think your vcore is likely fine for 3.8. If you havent done anything wild on the RAM side, .950 is likely fine there as well.
LLC compensates for the vdroop, yes. It can help with an overclock as well; if you're at 1.45 vcore (hypothetical) and didnt want to go any higher, you could use LLC to counter the vdroop (and should in this case in my opinion). At 3.8, 1.35 should be enough under load for the vast majority of CPUs (likely somewhere around 1.42 in BIOS with LLC 2). As posted above, somewhere around .950 on the SOC seems to work well for me.

except using LLC is highly ill advised by the powers above us.
post #9142 of 17390
Quote:
Originally Posted by hughjazz44 View Post

The whole point of CPU boosting is to provide brief moments of higher performance to aid in everyday tasks, like opening an application. Since opening an application only takes a short while, the CPU can boost a single thread temporarily to get the job done quicker, then go back to normal. This makes the end-user experience "feel" better. The CPU wasn't meant to sustain boost. If you want permanent higher performance, you need to overclock. Or buy a faster CPU.

This is literally something that can be done with Pstate useage, except with every core for better sooper vroom power. I've noticed if it doesnt need all cores to do something, Pstate wont rev up all cores anyway so its practically the same thing except you can get all-core boosting when you need all cores.

Its why I don't understand why they couldn't have at least gotten a CCX worth of cores boost. I think it would put the architecture to its full potential if the boost would be on core 0-1 and 4-5. Even 0-3 simultaneous boost would be more effective at stock. But, that will probably make the TDP go up and they cant have that frown.gif.

I'm no engineer tho, it just seems like it would function better that way.

oh and XFR is pretty useless to overclockers anyway.
post #9143 of 17390
LLC is not ill advised if you know what you are doing. I've used it on countless systems and if it was that dangerous it wouldn't be in the bios. Sure if you are setting 1.5v with LLC 5 you are a moron, but running anything 1.425v or under with a little LLC is fine, just watch in hwinfo, if your maximum volts ever goes higher than you are comfortable with, decrease volt or LLC. I am on 1.425v LLC 3 and I never even see it get above 1.425v, most of the time it is at about 1.39v
post #9144 of 17390
Quote:
Originally Posted by hotstocks View Post

LLC is not ill advised if you know what you are doing. I've used it on countless systems and if it was that dangerous it wouldn't be in the bios. Sure if you are setting 1.5v with LLC 5 you are a moron, but running anything 1.425v or under with a little LLC is fine, just watch in hwinfo, if your maximum volts ever goes higher than you are comfortable with, decrease volt or LLC. I am on 1.425v LLC 3 and I never even see it get above 1.425v, most of the time it is at about 1.39v

nonono, Raja, elmor, and the stilit all each advise against using LLC. Its more about things you cant see happening when you use LLC, such as higher strain on the FETs, the spikes after the load is removed that can go well above what you set it to be, and the 'ringing'. It was some odd pages back where there was another hot discussion about LLC usage and why we should stick to just raising the voltage to where we need to counter Vdroop.

Not telling you not to do it, even i'm still at LLC 2. I'm just saying its ill advised.
Edited by Reikoji - 4/13/17 at 5:30pm
post #9145 of 17390
Quote:
Originally Posted by hotstocks View Post

LLC is not ill advised if you know what you are doing. I've used it on countless systems and if it was that dangerous it wouldn't be in the bios. Sure if you are setting 1.5v with LLC 5 you are a moron, but running anything 1.425v or under with a little LLC is fine, just watch in hwinfo, if your maximum volts ever goes higher than you are comfortable with, decrease volt or LLC. I am on 1.425v LLC 3 and I never even see it get above 1.425v, most of the time it is at about 1.39v

Is 1.425V a 24/7 OC?
New and Shiny
(18 items)
 
  
CPUMotherboardGraphicsRAM
Rynze 7 1700 ASUS ROG Crosshair VI Hero Sapphire HD7950 G.SKILL TridentZ F4-3200C14D 
Hard DriveHard DriveHard DriveHard Drive
Kingston HyperX 3K Crucial MX300 Western Digital Black Western Digital Green 
Hard DriveCoolingOSMonitor
Western Digital Red Noctua NH-U14S Windows 10 Pro Dell U2414H 
MonitorKeyboardPowerCase
Dell P2414H Ducky One Corsair RM650x NZXT H440 White 
MouseAudio
Logitech G502 Proteus Spectrum Xonar DX 
  hide details  
Reply
New and Shiny
(18 items)
 
  
CPUMotherboardGraphicsRAM
Rynze 7 1700 ASUS ROG Crosshair VI Hero Sapphire HD7950 G.SKILL TridentZ F4-3200C14D 
Hard DriveHard DriveHard DriveHard Drive
Kingston HyperX 3K Crucial MX300 Western Digital Black Western Digital Green 
Hard DriveCoolingOSMonitor
Western Digital Red Noctua NH-U14S Windows 10 Pro Dell U2414H 
MonitorKeyboardPowerCase
Dell P2414H Ducky One Corsair RM650x NZXT H440 White 
MouseAudio
Logitech G502 Proteus Spectrum Xonar DX 
  hide details  
Reply
post #9146 of 17390
Quote:
Originally Posted by Reikoji View Post

This is literally something that can be done with Pstate useage, except with every core for better sooper vroom power. I've noticed if it doesnt need all cores to do something, Pstate wont rev up all cores anyway so its practically the same thing except you can get all-core boosting when you need all cores.

Its why I don't understand why they couldn't have at least gotten a CCX worth of cores boost. I think it would put the architecture to its full potential if the boost would be on core 0-1 and 4-5. Even 0-3 simultaneous boost would be more effective at stock. But, that will probably make the TDP go up and they cant have that frown.gif.

I'm no engineer tho, it just seems like it would function better that way.

oh and XFR is pretty useless to overclockers anyway.

Because it's not about "sooper vroom power". It's about completing a single task in a short time. If you open Chrome, and it XFR's one thread to 4.1GHz to open Chrome faster, the end-user will think "Hey! That opened fast!". They WON'T think, "Why did only 1 thread go to 4.1GHz?! Why didn't they all?!" You need to think like a typical computer user, and not like an overclocking obsessed techie.

Your CPU spends the VAST majority of it's life waiting on you to do something. Opening apps and stuff create spikes in usage. Then while you read the website, it sits and waits again. It's those spikes that benefit GREATLY from core boosting and and XFR.
post #9147 of 17390
Quote:
Originally Posted by hughjazz44 View Post

Because it's not about "sooper vroom power". It's about completing a single task in a short time. If you open Chrome, and it XFR's one thread to 4.1GHz to open Chrome faster, the end-user will think "Hey! That opened fast!". They WON'T think, "Why did only 1 thread go to 4.1GHz?! Why didn't they all?!" You need to think like a typical computer user, and not like an overclocking obsessed techie.

Your CPU spends the VAST majority of it's life waiting on you to do something. Opening apps and stuff create spikes in usage. Then while you read the website, it sits and waits again. It's those spikes that benefit GREATLY from core boosting and and XFR.

The typical computer user isn't using monitoring software or overclocking in the first place. They probably aren't even building their PC's themselves either, goin streight to Ibuypower or something and never looking at a bios screen. And when you make use of P-states to change your clock, you get that down-clock and down-volt when your computer is waiting for you to do something as well. That is why it is literally the same thing and XFR/boost could be better than what it is.
post #9148 of 17390
Quote:
Originally Posted by Reikoji View Post

The typical computer user isn't using monitoring software or overclocking in the first place. Those people all buy Dells. And when you make use of P-states to change your clock, you get that down-clock and down-volt when your computer is waiting for you to do something as well. That is why it is literally the same thing and XFR/boost could be better than what it is.

You don't think Ryzen will be put in Dells? And it's not "literally the same thing". P-states change based on load to mitigate power usage. XFR throws all that to the wind to get a brief, controlled, overclock. And XFR only happens when a single thread is being loaded. P-states can change regardless of thread counts.
post #9149 of 17390
Quote:
Originally Posted by hughjazz44 View Post

You don't think Ryzen will be put in Dells? And it's not "literally the same thing". P-states change based on load to mitigate power usage. XFR throws all that to the wind to get a brief, controlled, overclock. And XFR only happens when a single thread is being loaded. P-states can change regardless of thread counts.

Of course, and that is usually better.

At stock XFR and boost give ME, emphasis on me, somewhat over 4100mhz boost. Cool. But in my observation that boost doesn't always get applied to everything single core, just little mundane things like as you say opening a web page. its never anything constant, and that is why it lacks. Take CPU-Z for instance. Single core score won't be the score 4100+mhz will give me, it will be the score 3700mhz non-boost will give me. But its supposed to boost single core right? what happened there? The boost is lacking in such reguards. Will it boost 4100mhz for cinnebench single core? probably not, thats not mundane enough.

Pstate OCing to 4100mhz, I get that 4100mhz score and still only one core is actively boosted to 4100mhz while that is happening, the rest stay at the 2400mhz downclock. Then multi-core starts happening and I get the much more desired all boost.

And I do understand that they do it the way they do it to keep that TDP at or very close to the advertised 95w, which is another thing out the window to overclockers anyway.
post #9150 of 17390
Quote:
Originally Posted by Reikoji View Post

Of course, and that is usually better.

At stock XFR and boost give ME, emphasis on me, somewhat over 4100mhz boost. Cool. But in my observation that boost doesn't always get applied to everything single core, just little mundane things like as you say opening a web page. its never anything constant, and that is why it lacks. Take CPU-Z for instance. Single core score won't be the score 4100+mhz will give me, it will be the score 3700mhz non-boost will give me. But its supposed to boost single core right? what happened there? Another reason the boost is lacking.

Pstate OCing to 4100mhz, I get that 4100mhz score and still only one core is actively boosted to 4100mhz while that is happening, the rest stay at the 2400mhz downclock. Then multi-core starts happening and I get the much more desired all boost.

And I do understand that they do it the way they do it to keep that TDP at or very close to the advertised 95w, which is another thing out the window to overclockers anyway.

You're thinking about this totally wrong. It's not designed to sustain the XFR frequency. AMD isn't selling you a 4.1GHz CPU, then purposely gimping it to 3.7GHz to save watts. They're selling you a 3.7GHz CPU, and giving you the added benefit of brief bursts of speed. They could've just as easily said - 3.7GHz. Period. No boost. No XFR. No nothing. If you want 4.1GHz, pay us $200 more for the hyper-binned 4.1Ghz variant.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD Motherboards
Overclock.net › Forums › AMD › AMD Motherboards › ROG Crosshair VI overclocking thread