KarathKasun

1400 scores better than 1300x in newer games because moar threads.

You need to manually set pretty much all timings on Ryzen to get much above 3200.

3 hours ago, dave_k said:

Strix board has the same amount of capacitors like the carbon. ..  

But the Strix has more SoC VRM phases.  The split of caps for Core/SoC is likely weighted more toward the SoC.

3 minutes ago, dave_k said:

But Vcore VRM on both Strix and Tomahawk is 4 phase for VCore

Not quite true.  Strix has doubled low-side FETs.  Its somewhere between a 4 phase and fully doubled 4 phase, all its missing compared to the latter is extra caps and chokes.

So it should be better than the Tomahawk but worse than the Gaming Pro Carbon or Krait boards.

When Ryzen APU's finally hit the market, the Strix will have an advantage.  It should be able to max out a quad core with no problem and it will be better equipped to handle iGPU overclocking.

Just now, dave_k said:

well i have the ir guy and my friend is getting B350 tomahawk next week or so so i'll try to measure

B350 Tomahawk is only 4+2, its VRM is half of the B350 Gaming Pro Carbon/Krait which is 8(4x2)+2.  So its not apples to apples.

Never trust a software sensor.  They can be in different places on different boards, and that will skew temperatures.  They can also be biased or calibrated differently.  You need to use probes or an IR thermometer.

Just now, dave_k said:

tell me why B350 Carbon VRM ran 30°C higher during gaming?

Based on what kind of test?  Measured how?

Thats where you are wrong.  At 25c PK616BA can push 50A per FET and the PK632BA can push 78A, and they can maintain nearly triple the current of the OnSemi parts at 100c.  They just are not as efficient, which is less important when you double up the FETs in each phase because you halve the on/active resistance.

Also, the MSI boards have double the chokes/inductors and caps as well.

You can push a board much further under LN2 because the CPU is so cold that it cools the area around the socket, this bypasses a number of issues present with low end VRMs.

Also, under those conditions you do not care if it will only last 30m or so.  You just need a bench or two and validation shots.  The same style VRM (B350 Prime) can hit 150c under stress with an R7 @ 4ghz, I don't think anyone believes that is a good situation.

They are 11A to 20A each on the low side.

123-224w maximum wattage @ 25c. not Counting the high side FETs or efficiency problems.

Not a great VRM for anything but a quad core honestly.  After the Zen APU's (near 50/50 core/SoC power split) come out though, it will likely find its place.  It should be fine for mild R7 or moderate R5 hex overclocks, but that does not appear to be its intended use.

The VRM quality on the mATX & ITX boards is questionable at best.  Its likely that your board is not able to supply clean enough power to warrant pushing over ~1.3v.  I would set voltage to 1.3v and see where your maximum stable OC ends up.

FWIW, on a mITX board (similar VRM quality compared to the ITX boards) my R5 1400 can just barely do 3.8ghz @ 1.4v.  An R7 on the same board would likely clock even lower because the power draw through the VRM is MUCH higher.

5 minutes ago, BuckGup said:

If you know the cause why don't you know a fix?

Because its likely different for every game.

Automatic updates that exist in all of the things.  Settings people are changing that they don't mentally connect with graphics quality or causing graphics quality issues.  Screen overlays.  Mods.  GPU acceleration in web browsers...

Any number of reasons.

Scaling can make things worse unless you use a very high smoothing factor.  In MC using DSR introduces its own shimmering into planar textures (think ocean) because its filtering on top of filtering.

Downscaling can only fix so much high frequency noise.  If anything reduce your texture LOD, things will be 'smoother' at the cost of reducing detail somewhat.  MSAA is dead mostly due to the direction rendering has developed over the last 5-10 years, and at this point post process AA is not great in many games.  You could try SweetFX or some other shader code injector if you want good post AA.

Its hard to read a whole thread and take all of it serious when the posts in it sound like paranoia.

L.A. Noire has aliasing pretty bad with "AA" turned on in game, but that's what I remember from any GTA based game from Rockstar.  Mostly comes down to AA not working right with deferred shading, you just get post processing AA.

Not trying to be a jerk or anything, but disable all user override profiles for the game and test again.  Overrides can totally wreck in game LOD algos.

6 minutes ago, BuckGup said:

Disabling has no effect

What iGPU?  Some of the AMD ones have a 'texture alignment fix' option in the driver (old VLIW ones especially).

I can break out my cheap Haswell laptop and replicate if that is similar hardware.  Screen cap below is from GTX 670 with forced AF, that's the only time the problem surfaces.

I would assume that game optimizer software (GeForce Experience comes to mind for many users) could also flip registry bits to force AF as well.

2 minutes ago, BuckGup said:

Except the problem happen on iGPUs too.

You can force AF on iGPUs.  There is an option in the intel iGPU control panel for it if you dig around enough, and I believe it is defaulted to 4X or 8X if you choose the 'quality' preset.  The quick fix is to disable AF in game.

2 minutes ago, lordante said:

Ok so:

Windows media creation tool has a CRC check. So impossible the win10 reinstal drive is corrupted.

If you don't trust it you can CRC check it yourself and compare to another users CRC check.  Both users would need to be using the same language for the media AFAIK.

You can CRC check an install media to see if it has been altered, its not that difficult.

Windows media creation tool downloads from MS.

For ESO try disabling all driver forced quality settings.  Forced AF/AA can break quite a few games.

On 7/15/2017 at 10:23 AM, BuckGup said:

Yeah thats why some games aren't the best to test in. Minecraft is great due to it's block style.

Here are too screenshots. One has the problem one doesn't.

That is what happens when you force 16X AF in Minecraft (mentioned in my other post).  The only way to make it not look like that is to disable in game AF if you force it with the GPU control panel.  Actually, it looks like a bunch of these problems could be coming from GPU driver forced quality settings.

Getting L.A. Noire setup again ATM. 

Don't have GTA V either, may have a copy of GTA IV hanging around though.

Quick rundown of what I have on the system...

Antichamber

ARK

Arma 2

Arma 3

Assasins Creed 1 & 2

All Batman games

All Borderlands games

Dirt 3

Day of Defeat: Source

Doom

Devil May Cry

Dota 2

Elder Scrolls Online

Fallout 3

Fallout: New Vegas

Fallout 4

Grid Autosport

Killing Floor

Metro 2033 (OG)

No Mans Sky

Portal 1 & 2

Rust

Serious Sam Classic

Space Engineers

Strike Suit Infinity & Zero

System Shock 1 & 2

Team Fortress 2

Wolfenstein: The old Blood

Mass Effect: Andromeda

Sleeping Dogs

Watch Dogs

L.A. Noire

BioShock 2 & Inf

2 hours ago, lordante said:

Your screen just confirm you didn't read anything on this thread.
If you had take a look at it, you will know this moutain is "screened" from the other side.

So you try to play the super hero, but finaly, you missed the bug on the two screens about the same moutain. You did not read the thread, how can you confirm or cancel anything on it?

ANYWAY congratulation, you are not "infected" or "affected", now, read the thread, stop thinking you are the messia, and see what is wrong on the moutain image given earlier.

BTW, it is simple, i up to date my computer's drivers every week (since years now). I listed all my Hardware (and changed all parts too). The only factors staying: Intel, windows (but i changed from 7 to 10).
The only one who get ride of this did something "new", because Ryzen is "new".
The only one making the test is me (even if it will be a fail, we can remove Intel suspicion).

But if you want us to make "pro" test, OK, just give me the money!!!
I sold all my Hardware to get to Ryzen (and i made a smart move, selling my old hardwares make me win near 200€). I will not test product one by one, because if i do and the corruption is the one passing between Cmos/GPU... I'm fu..d!

.

Did you even look at the Dropbox links?  Of course not, because you are too worried about trolling to even think.

For those too worried about conspiracy theories, the links in my post have 20-30 screen caps of the area in game.

1 hour ago, BuckGup said:

Alright thanks for the picture. Now can you open Pay Day 2 and play the first mall heist and simply don't move.

Don't have Pay Day 2.

I can tell you right now that Minecraft does NOT work with AA at all.  DSR crashes when you go fullscreen. FSAA was forced on in the driver for these, it does nothing.  It has done nothing for a long time now.

Folder with caps from R5 1400 with GTX 670.

Folder with caps from IvyBridge using iGPU in linux.

I remember why I stopped playing MC now, its a game built solely from bugs and programming glitches.  The game resolution scales with Windows UI scaling... WUT?  That's one mystery solved.  Also, do NOT force AF unless you disable it in game, it breaks the graphics pretty bad if you have it enabled in both places.  16X AF creates odd shimmering in some places due to the design of the textures, 4X AF is plenty for MC and causes the least issues.

Attached image is about what MC looks like using DSR.  Getting DSR working is very jankey but looks pretty good.

Steps to get MC working with DSR...

disable DPI scaling for the Minecraft shortcut

make sure full screen is disabled in game

set DSR factors in NV Control Panel

set desktop resolution to your desired DSR factor (4k on 1080p gives 4x AA)

start game

maximize the game window

NEVER hit F11, game will crash.

Just now, jjohnthedon1 said:

Iv seen it like that Incpuz too 

Its a platform thing, not a software thing.