FX-8370, Looking for some Technical help :D.

[kaitheus]

Ok so recently I installed the FX-8370, How ever I'd like to know a bit more about its technical layout such as it Stock Vcore etc, I am asking for this primarily because at one point I was getting a stock reading of v1.35 and now I am getting a reading of v1.275. Both readings the CPU was running at 4Ghz. How ever when it was reading 1.35 the multi was at 20.5 for some reason, so technically I was at 4.1GHz at that point, yet I never set that up that way in the bios :/.

Note that turbo boost is off as well as all the power saving states etc and cool & quiet is disabled etc, and the OC section of my mobo is set to auto atm which means essentially that I'm not OCing the CPU atm.

Also I'm not new to PCs just that some times its better to just ask for help rather then go head 1st into the vulcano lolz.

Any help would be appreciated as I use this PC for heavy video encoding so I need it Stable lolz.

 

[Darkbreeze]

If Catalyst control center is installed you need to go into the power settings there and slide the performance slider so that min and max are both 100% or the power features will still sometimes operate as will the multiplier fluctuation regardless of BIOS settings.

 

[Darkbreeze]

Like this:

 

[kaitheus]

That's odd that is the Only option in the CCC I don't see and I am using the latest drivers o.o.

 

[Darkbreeze]

Try going into the advanced power options in control panel. Deep in the options is the same setting for cpu performance. Set min and max at 100%.

 

[kaitheus]

Ok so I just went in to the performance power plan and set that up for max performance/no sleep etc through the control panel, then I restarted the PC just cause lolz.

Also I was initially asking about what is the actual Stock Vcore of a FX-8370, cause I for some reason find 4GHz @ v1.275 a bit low ? How ever I just played a bit of PlanetSide 2, Warframe and Star Citizen with no issues. Also I think that power option in the CCC only appears if I am Actually OCing the CPU at that time, cause I remember seeing it with my 955BE when I had it OCed to 3.8GHz.

Going to Encode a video in Sony Vegas before I finally head off to bed as that'll really stress the CPU, not like Prime95, but I do that another time lolz.

 

[slyu9213]

Power options should be there no matter what. It's basically the same thing as using the High Performance power option in Windows that sets CPU to 100% min and max.

 

[Darkbreeze]

There is no "Stock" voltage for it if the system is constanstly shifting it about, but an average voltage for that chip with no turbo and no power reductions is around 1.275 to 1.3v. My 8320, which is the same chip just a lower bin, is generally around 1.35 with a 4.5Ghz overclock. Even with EVERY possible option that could cause automatic adjustment of the voltage disabled it still varies a slight amount depending on system load. Enabling load line calibration in the BIOS is helpful for stability as well.

In the power options, you need to click on "change plan settings" next to the Performance plan, and then on Advanced settings. In advanced settings, open Processor power management and set the minimum processor state to 100% if you wish. Make sure the cooling policy there is also set to Active.

 

[kaitheus]

There is no "Stock" voltage for it if the system is constanstly shifting it about, but an average voltage for that chip with no turbo and no power reductions is around 1.275 to 1.3v. My 8320, which is the same chip just a lower bin, is generally around 1.35 with a 4.5Ghz overclock. Even with EVERY possible option that could cause automatic adjustment of the voltage disabled it still varies a slight amount depending on system load. Enabling load line calibration in the BIOS is helpful for stability as well.

In the power options, you need to click on "change plan settings" next to the Performance plan, and then on Advanced settings. In advanced settings, open Processor power management and set the minimum processor state to 100% if you wish. Make sure the cooling policy there is also set to Active.

This was actually what I needed, so between v1.275 and v1.35 is normal for this CPU. Also the only option I see in the Processor Power Management area is for the Cooling Policy, which is set to active, everything else I aready had set to max performance.

I think once I have let this new CPU go through a burn in period I'll actually OC it to 4.4GHz or so. Which should be fine on Air I assume with a push/pull config.

My Setup atm, Quite Please with it so far 😀.
http://kaitheus.deviantart.com/gallery/49861010/Carbide-Air-540-WIP

 

[Darkbreeze]

I have a standard Noctua NH-U14S in a default single fan push configuration and come nowhere near thermal limits running Prime95 small FFT's on version 26.6, which is the version you want to test with. Your motherboard is a major factor as well since it's often the VRM thermals that create issues when using a tower cooler that doesn't afford much if any airflow across the board itself. Case cooling, and as much of it as possible, can be become the larger factor in some overclocking situations.

What is your motherboard model?

 

[kaitheus]

Asrock 990FX Extreme 4, which has heatsinks/pipes from the NB to the VRMs, which also has a tini fan on it, You can actually see just about everything from this shot, http://kaitheus.deviantart.com/art/My-New-Corsair-Carbide-Air-540-8-Nice-An-Clean-461390670 .

 

[Darkbreeze]

8+2 power phase on that board so it's good enough. Overclock to your hearts content and thermal limits. Keep in mind, once you establish a permanent stable overclock, you need to run Prime as determined above for a full 24hrs, with nothing else running, to verify stability. I've seen rigs fine up fifteen hours and then throw errors or halt workers between fifteen and 24hrs. After 24hrs the chances of micro-errors being introduced into the system if there has been no failure during testing is infinitely reduced.

 

[kaitheus]

8+2 power phase on that board so it's good enough. Overclock to your hearts content and thermal limits. Keep in mind, once you establish a permanent stable overclock, you need to run Prime as determined above for a full 24hrs, with nothing else running, to verify stability. I've seen rigs fine up fifteen hours and then throw errors or halt workers between fifteen and 24hrs. After 24hrs the chances of micro-errors being introduced into the system if there has been no failure during testing is infinitely reduced.

Ok good to know, I usually ran Prime95 for 8 hours or so, but I think I can manage 24 hours, just as long as I have something to keep me busy in the mean time lolz. Also why do they call it an 8+2 power phase, and not just a 10 ? lolz.

 

[Darkbreeze]

With 8+2 power phase, 8 phases will be dedicated to the CPU, 1 phase to the RAM, and 1 phase to the Hypertransport.

 

[kaitheus]

With 8+2 power phase, 8 phases will be dedicated to the CPU, 1 phase to the RAM, and 1 phase to the Hypertransport.

Ok that Actually makes sense now that I think about it, it means like you said that 8 are completely dedicated to Just the CPU so that OCs are far more stable etc, which is why I got this board in the 1st place lolz.

Thanks for all the help, I think I am running stable at v1.275 @ 4GHz no OCing atm, will wait to actually OC this thing probably in a week or 2, then I'll push it to 4.4-4.5GHz, run Prime95 for 24 hours and go from there. Shouldn't have any issues with pushing 4.5 or even a bit higher on this chip, but I think I'll keep it modest at 4.5ish lolz.

 

[Darkbreeze]

Beyond 4.5Ghz doesn't really show a lot of improvement in gaming with most current cards anyhow except at very high resolutions or high resolutions with multiple monitors, but 4.5Ghz does well with FX8.

 

[kaitheus]

Beyond 4.5Ghz doesn't really show a lot of improvement in gaming with most current cards anyhow except at very high resolutions or high resolutions with multiple monitors, but 4.5Ghz does well with FX8.

I was kind of wondering about that actually, wheather at some point there was deminishing returns or not, and since I only play at 1080p 4.4-4.5 should be fine once its all set in stone lolz.

 

[Darkbreeze]

Yep. You should be good there.

 

[mdocod]

With 8+2 power phase, 8 phases will be dedicated to the CPU, 1 phase to the RAM, and 1 phase to the Hypertransport.

Actually, 8 "phases" for the CPU, 2 phases for the CPU-NB (split power plane).

The memory VRM is separate and can be found over by the DIMM slots.
The chipset VRM is separate and can be found over by the chipset.
The HT bus runs from the CPU-NB to the chipset, it is most likely "powered" by both the +2 phase part of the CPU VRM config and the chipset VRM.

There is no "Stock" voltage for it if the system is constanstly shifting it about, but an average voltage for that chip with no turbo and no power reductions is around 1.275 to 1.3v. My 8320, which is the same chip just a lower bin, is generally around 1.35 with a 4.5Ghz overclock.

Actually, there IS a "stock" voltage. It is known as VID and is stored on the chip along with other chip-specific identifiers and data. In most cases, if we disable APM and turbo, with voltage controls set to auto the motherboard will configure the CPU voltage to match the VID. Vishera 4000/6000/8000 series CPUs can have "stock" VID's anywhere from ~1.2-1.4V.

Beyond 4.5Ghz doesn't really show a lot of improvement in gaming with most current cards anyhow except at very high resolutions or high resolutions with multiple monitors, but 4.5Ghz does well with FX8.

Resolution has very nearly nothing to do with CPU workload. A given CPU at a given speed will have about the same performance whether running a 720P or 4K monitor. Resolution effects the render workload, which is pretty much entirely the responsibility of the GPU.


---------

Reading this thread has just been painful.. it's like a continuous stream of misinformation.

 

[Darkbreeze]

To start with, it wasn't my quote, I simply cut and pasted a quick reference from here:

http://www.overclock.net/t/891696/a-short-power-phase-design-explanation

Guess I could have read it better, as I was just looking for the quick description, and now that I went back and read the actual thread it clearly is incorrect, but it's irrelevant. The specific engineering details are not what he was looking for anyhow. He got the gist and your critique could really have been more complimentarily constructive rather than simply being intended to tear down what's been said.

And if you want to argue about whether or not one of the phases is for voltage, you can argue with Gabriel at Hardwaresecrets as he clearly indicates that's the case in this article. His words, not mine.:

http://www.hardwaresecrets.com/article/Everything-You-Need-to-Know-About-The-Motherboard-Voltage-Regulator-Circuit/616/4



And all of these speak to your statement that CPU workload is unrelated to high resolution gaming:

http://www.tomshardware.com/answers/id-2335204/effect-resolution-cpu-performance-gaming.html

http://www.techpowerup.com/forums/threads/low-res-gaming-needs-a-faster-cpu-nuh-uh.122199/

And if you look at almost ANY set of gaming benchmarks using the same resolution and GPU, the faster CPU will post higher frame rates, so how can you infer that there is no relationship between CPU performance and high resolution gaming. I realize that an increased resolution will shift much of the load to the GPU and result in higher frame rates and overall performance, but I'm talking about different CPUs with the same resolution, title and GPU model.

 

[mdocod]

To start with, it wasn't my quote, I simply cut and pasted a quick reference from here:

http://www.overclock.net/t/891696/a-short-power-phase-design-explanation

Guess I could have read it better, as I was just looking for the quick description, and now that I went back and read the actual thread it clearly is incorrect, but it's irrelevant. The specific engineering details are not what he was looking for anyhow. He got the gist and your critique could really have been more complimentarily constructive rather than simply being intended to tear down what's been said.

Actually, "tearing down" the misinformation here is sort of the point. Those details are actually important in the greater scheme of things... Kaitheus has reached out to the community in hopes of acquiring technical information to help him on his path to achieving a stable performance tuned system for video encoding.

Consider for a moment, someone performance tuning their vishera CPU on a 4+1 phase or 4+2 phase. We can infer something important about how the board will handle performance tuning different aspects of the CPU based on consideration of that VRM configuration. On a board with only 1 phase for the CPU-NB, we can expect less useful performance tuning headroom on the CPU-NB (including the memory controller, and what sort of memory speeds it might support).

And if you want to argue about whether or not one of the phases is for voltage, you can argue with Gabriel at Hardwaresecrets as he clearly indicates that's the case in this article. His words, not mine.:

I don't even understand the context of what you are implying here. I have no interest in participating in an argument that doesn't exist anywhere but an imaginary place in your head. I have not spoken to the contrary of or the favor of the debate you are implying I am supposed to be carrying out, and will not entertain your attempted deception and distraction here.

And all of these speak to your statement that CPU workload is unrelated to high resolution gaming:

http://www.tomshardware.com/answers/id-2335204/effect-resolution-cpu-performance-gaming.html

http://www.techpowerup.com/forums/threads/low-res-gaming-needs-a-faster-cpu-nuh-uh.122199/

And if you look at almost ANY set of gaming benchmarks using the same resolution and GPU, the faster CPU will post higher frame rates, so how can you infer that there is no relationship between CPU performance and high resolution gaming. I realize that an increased resolution will shift much of the load to the GPU and result in higher frame rates and overall performance, but I'm talking about different CPUs with the same resolution, title and GPU model.

You implied that higher resolution gaming came with increasing compute requirements from the CPU. That was false.
You seem to be attempting to "change your story" in the middle of making your argument here. This doesn't surprise me as you've been making it up as you go here through the whole thread.

The hard limits of frame rate imposed by the CPU are NOT significantly influenced by resolution. If you cap at 76FPS on a CPU bottleneck at 1024*768 resolution, then that's going to be the CPU limited cap at any resolution, doesn't matter. When the CPU limited cap gets "covered up" by a different bottleneck (GPU), then the FPS changes but the hard cap is still there. When we change the compute performance, the hard cap changes.

 

[Darkbreeze]

And if you want to argue about whether or not one of the phases is for voltage, you can argue with Gabriel at Hardwaresecrets as he clearly indicates that's the case in this article. His words, not mine.:
I don't even understand the context of what you are implying here. I have no interest in participating in an argument that doesn't exist anywhere but an imaginary place in your head. I have not spoken to the contrary of or the favor of the debate you are implying I am supposed to be carrying out, and will not entertain your attempted deception and distraction here.

Sorry, typo, I meant Memory, not voltage. You DID indicate none of the phases are for memory. You said

"Actually, 8 "phases" for the CPU, 2 phases for the CPU-NB (split power plane)."

My understanding is that one of the phases IS for memory, as is indicated from many technical sources, including this one:

But there is one caveat. On some motherboards the
phase that controls the memory or the chipset voltage is
located close to the other phases, making you to have a
wrong phase count if you simply count the number of
chokes present near the CPU socket. We show this case in
Figure 12: even though the portrayed motherboard has
four chokes, it is a three-phase motherboard, as only
three of the phases are used to generate the CPU main
voltage (Vcore); on this motherboard the fourth phase is
used to generate the memory voltage.


As far as the CPU and it's relationship with a higher frame rate, basically regardless of resolution, whatever.

A higher frame rate is directly tied to an increase in CPU requirements, e.g. 60 fps will require twice as much CPU power as 30 fps. However, 60 fps may not require twice the bitrate, since the improvement in fluidity from 30 fps to 60 fps means there is less time for the eye to notice a quality loss.

Therefore, you can either increase the capability of the CPU to accommodate the decrease in frame rates you may experience at a given resolution, or increase the resolution which results in shifting more of the process to the GPU. Any way you look at it, there is NO situation where having a slower or less capable CPU is beneficial, so I don't even know why you would remark that comment.

And this is not an appropriate venue anyhow, so I'll leave you to it rather than argue on the OP's thread.

 

[mdocod]

Darkbreeze,

I said exactly what I meant, and what I said is correct.

The "+2" or "+1" that we list with AM2/+AM3/+ socket board specifications is the power regulation for the CPU-NB, NOT MEMORY. You can keep misinterpreting "sources" to attempt to finagle your way out of this, but that's the way it is and you're going to have to figure out a way to live with it. If I have to make a freaking diagram imposed over a picture of an AM3+ board pointing out exactly where the VRMs are for the chipset, CPU, CPU-NB, memory etc I would be happy to do that for you.

----------------

The cpu hard capped performance in a real-time workload like gaming DOES SCALE WITH CHANGES TO CPU PERFORMANCE. I have never claimed anything to the contrary. You're arguing a point that is totally out of context from your original false claims.

 

[Darkbreeze]

Ok bud. Whatever. Happy trails. I guess EVERY technical source on the internet, EXCEPT for you, must be wrong huh?

Phase power design is often listed as 4+1, 6+2, 8+3, etc. The number preceding the plus sign (4, 6, and 8 in those examples) is indicative of the number of phases dedicated to the CPU, thus a 6+2 phase power design will dedicate 6 phases to cleaning CPU power. The number after the plus is for other components, often RAM or HT (HyperTransport) for AMD.

 

[mdocod]

Darkbreeze,

Did you even read those sources you are citing?

From the "next page" of your original source:
http://www.hardwaresecrets.com/article/Everything-You-Need-to-Know-About-The-Motherboard-Voltage-Regulator-Circuit/616/5

FYI: "Integrated Memory Controller" is the CPU NorthBridge.
The article clearly explains exactly the point I am making here, that the +1 or +2 on AM3 is the VRM for the CPU-NB (memory controller).

In the article you are referencing as your source, there's a diagram showing the position of the SEPARATE VRMs for the chipset and memory, exactly where I told you they would be.

--------

I'm sorry you are finding other sources written by rank amateurs. I'm sorry you failed to interpret the only one of your sources that has the correct information. This isn't "me vs the internet." This is "you are wrong today, learn to accept that and move on."