Beeswitt

ℹ️ Please read this post before creating a thread on GPU Performance Issues, as it will help you get faster, better, stronger help from the community. (Assuming the steps below don't empower you to fix it yourself.)
 
02/07/2024 📰 Nvidia GPU and affecting web browser scrolling? Check out this driver: https://nvidia.custhelp.com/app/answers/detail/a_id/5519/~/geforce-hotfix-driver-version-551.46
 
ASSUMPTIONS MADE IN THIS GUIDE:
 
That your issue  isn't a KNOWN ISSUE. To check, search the AMD, Nvidia and Intel Knowledge bases for the word "stutter" and see if they list any known issues that describe your issue.
The PC boots and runs with some modicum of stability The GPU(s) display to your monitor without issues (i.e., this is not a guide for blank screen issues during boot!) If it's issues with games, ideally it's not game-specific
(You're welcome to try this guide, but it's less-likely to work if the issues only happen in a specific game) The operating system is Windows 10 or Windows 11
HOUSE KEEPING:
🛑 Before asking for help (do not reply here), do all of these (house keeping) things first:
 
🙋‍♀️ ASKING FOR HELP 🙋‍♂️
🔬 COLLECT DATA FOR TROUBLESHOOTING FIRST 🧪
 
✏️ MAKING YOUR NEW TOPIC/POST ✏️
ℹ️  If you want fast, expedient help from the volunteers, make a new topic in the Troubleshooting forum (not here in the Guides)
ℹ️ Already have a post in Troubleshooting? Don't make a new topic! Just edit the original post and provide the information below:
🛑🛑🫸 Replying to this post to get help❓ We see you! 👁️ Start a new topic please! 🛑🛑
 
📝 If you have questions about the guide itself, found a mistake, have recommendations, etc. reply below and let me know!

Hello!
 
I have microstutters and freezes in almost all games, regardless of graphics settings. For example: The Stanley Parable, The Beginner's Guide, The invincible, GTA IV and especially Fortnite (so many microstutters, but fps is just fine). All of these games are not very demanding on hardware, and judging by the load displayed in MSI Afterburner, my pc is not even close to overload.

Basically, these freezes occur when changing scenes or when changing lines of dialogue. For example in The Stanley Parable, the game runs smoothly if you walk around one room, but as soon as I move to the next location and the narrator starts saying a line of dialogue, I experience one microstutter. Then again the next location, a new line and again a single stutter. An interesting fact is that in the case of The Stanley Parable, if I hear the same dialog line in a second time, the freeze may not appear, as if it only works on new lines of dialogue. But after restarting the PC, again every dialogue causes a freeze. Maybe dialog line is sitting in something like cache, and pops out of there with no problem? But when it needs to be loaded directly from game directory, it lags?
Also, sometimes, after restarting Windows, freezes are completely gone. Then another restart and they appear again.

It is very exhausting trouble, that i just cant figure out. Can anyone help me with that?
 
My system:
CPU - i7 4960x
Motherboard - Asus Rampage IV Extreme (bios modified for nvme support)
RAM - DDR3 HyperX Beast 2400MHz 4x8Gb (HX324C11T3K4/32)
GPU - MSI RTX 2060 Super Armor OC
Cooler - DeepCool AK620
SSD (Windows) - Samsung 970 Evo Plus 250Gb (via m.2 to pcie adapter)
SSD (Games 1) - OCZ Revodrive 3 x2 240Gb
SSD (Games 2) - Intel Sata SSD 480Gb
HDD (Media files) - Seagate 4Tb
OS - Windows 10 Pro
 
What I tried to do:
- Clean reinstallation of Windows and drivers
- Reinstalling games
- Update video card drivers
- Disable and enable overclocking

What i think is nice to try (i will do this in next couple days):
- Install games on another ssd
- Take out all drives and check with just one ssd
- Testing with another Display
- Trying another HDMI cable

If you have any thoughts about what i can try to do, please, inform me

Focus on cpu oc for now and we can oc the rams and gpu later so for now just turn xmp off for the rams and revert gpu back to stock (save oc settings first tho)
 
Afaik x79/99 seems to see some beneifits to cpu oc (apparently lower vcore requirement) if you set bclk strap to 125 so id set bclk strap to 125
 
Vcore can be pushed as high as your cooling allows, for now set 1.35v vcore with llc off, can push this higher later on
 
Seems like ivy e is kinda shit at clocking so start at 4.37ghz with 35 multi, then try 4.5ghz with 36 multi and so on till unstable, its helpful to figure out what your max bclk with 125 strap is for overclcocking in smaller increments so id reccomend checking max bclk first by upping bclk by 1 increment, start at 128, once it stops posting drop it into decimals till you find the highest bclk thatll post, as an example 131.4bclk, then you subtract 0.5 from that so 130.9bclk just to ensure stability
 
As for trying even higher clocks with more volt when 1.35v isnt enough and say 4.6ghz with 36 multi at 127.8bclk wont be stable, you can raise your volt to whatever your cooling can handle be it 1.4v or 1.5v (llc off so itll probs droop by ~0.05v), if efficiency is a concern then try 1.3v 1.35v 1.4v and 1.5v then draw a curve to see where diminishing returns hits and run your chip at that point just before diminishing returns, if efficiency is not a concern you can fuck all that and run the highest volt your cooling can handle with the conditions of <1.35v can go up to 100c tjmax, <1.45v can go up to 85/90c, and >1.45v youll wanna keep under 80c due to degradation and voltage rollover (too high temp and the cpu wont clock higher with more volt) under prime95 smallest ffts or aida64 stresstest
 
 
Once you get past cpu oc and youve gotten a setting you are happy with you can move on to ram oc
 
This thread seems to have some pretty helpful info regarding volts cause unfortunately i only have real ddr3 oc experience with x58 and partial am3+ experience (yes have already passed the 3000 ddr3 mark atleast on x58), as for the rams themseves check thaiphoon burner and cpuz just to make sure of the ics, though i bet theyre hynixes anyways since all the high speed 8gb sticks use hynix 4gbit ics like mfr bfr afr etc. And all of them do 3000+ with ease, just a matter of the imc being able to keep up or not, think a universal setting thatd work for those hynixes would be 12-16-16 at 1.65-1.8v, wont need to push volt higher as thats pretty pointless (youd only need to push volt for real high freq or useless cl and i doubt youll get much over 2800 ddr3)
 
 
As for gpu oc just a matter of playing with sliders, have furmark in the background to check for artifacting and crashing, find the max gpu and vram oc without artifacting and lower by 10-20mhz for stability
 
 
And all of this is perfectly safe for the hardware btw as these are rather mild voltages were looking at
 
just be aware that particularly unstable ram or bclk oc can corrupt your os if you are particularly unlucky (it does happen but its quite rare) so id suggest using a diff windows install if possible, doesnt have to be a diff drive if you dont have one, you can just make a partition with diskpart or disk manager whatevers the name with your current drive like 50gb ish or so then you just install another windows on that partition, and once you are done you can simply delete that partition with your windows bootable usb (the one you use to actually install windows in the first place) and re integrate it into your main partition again

So basically just a stopgap pc
 
2400 ddr3 is pretty slow though so if possible try cranking it up abit further, hynix mfr/bfr do well over 3000 so ics wont be a limiter
 
Ivy e controller should be pretty close to gulftown in terms of max ram oc and there is a ddr3 4000mhz sub for the 4960x (24 multi, 160bclk strap) so id atleast hope for 2800, it is dual rank though so you may only get around 2800 ish
 
This is 2800 stable on 2 double sided cfr sticks with an i7 930, board has a trash 3rd channel so ill need to get a better board (has trouble posting even a slow 2500 with 3 sticks), pretty easy to get this running since its just a matter of set vdimm set vtt and set primaries
 
you problably wont need much voltage as long as you dont dump all your voltage into cl and tcwl, problably ~1.8v instead of the 2.1v im running here cause cfrs are pretty mid clockers
 
As for the cpu just run it as fast as the cooler allows, id consider ~1.52v safe but you problably wont be running over 1.45v anyways
 
Strap problabaly 125 since the 2400 ram multi works, unsure about 2666 and 2800 ram multis but 3000 is problabaly pushing it for a dual rank 4x8 config
 
 
And right i forgot to check the damn bios overview so might aswell give some oc settings while i check for pcie lane settings, make sure you save your current profile before attempting oc just so you can fallback onto your current oc

Oops my bad read "4970" instead of "4960X" 😛 
Intel has such bad naming habits, with the 4960X  numbered between the 4770 and 4970, while being something totally different !
 

try removing the extra pcie cards and see if that fixes your issue
 
You can also try putting your gpu in a diff slot cause im pretty sure you can control the bandwidth of the pcie x16 slots so you can put your gpu in whatever x16 slot and still have it be running in x16, atleast thats what i see on my p6t deluxe v2 and p6x58d-e where i can choose pcie slot configs and have the second x16 slot have full x16 bandwidth
 
 
Btw thats a pretty interesting build
Is this a retro build or something? Cause the case and cooler screams modern white theme build but the components scream 2013 especially the gpu, not like getting a better one is that expensive either (used 8gb rx 580s going for like 60$ or less)

total PCIE bandwidth is far from a problem on HEDT platforms, you may have a problematic GPU, but agian it doesn't make much of a difference.

That is lga 2011 not 1150 so lots of pcie lanes available

Hello!
 
Faced with the problem of RAM. My friend gave the computer for sale, I had to clean it, replace the thermal grease and prepare it (I did it a lot of times, with different computers and never encountered such a problem). After I returned everything to its place and turned it on, a cyclic reboot began. After some time, I realized that the problem is in the RAM (or with its connection with the processor). I have two 4Gb sticks, and 4 slots on the motherboard. The first two (closest to the processor) do not work. And the other two are working, the computer is working properly, but only in single-channel mode. The first two slots (DDR3_4 and DDR3_2) are channel A. After, I pulled the motherboard out of the case, removed the cooler from the processor and tried again. All slots are OK, the computer turns on in dual channel mode! But putting everything back, installing the cooler on the processor and installing the motherboard in the case, it all happened again!
What is the problem? Did someone manage to solve this problem?

PC:
CPU - i5 3570
Motherboard - Gigabyte GA-Z77-DS3H (Rev. 1.1) (BIOS - F8)
RAM - Corsair Vengeance DDR3 2x4Gb (1600Mhz)
GPU - Palit Geforce GTX 650 Ti
PSU - Thermaltake Litepower 700w
OS - Windows 10 Pro (x64)

Thank you in advance!
Без названия.mp4

Yes, they pushed all the way in.
 
I got it from the fifth attempt
 
Thank you!

Ensure that the cooler pins are properly pushed in, from my experience they you have to push quite far.

Thank you ?

And another small tip, buy a face mask to protect your lungs from dust.
Without it, I will not even open the case?

Yup, I think that doing that would be a good idea.  With "are the pins pushed in all the way", I meant the CPU cooler pins, but checking the RAM is indeed also a good idea.  From my own experience I know that these stock Intel coolers are easy to not have been connected properly - sometimes we are just too careful.
 
I should look some more into compressed air, brushes, alcohol, and wipes, thanks for the tips! Just cleaned my dust box of a PC yesterday, 5 years of dust in a dusty room is a fair amount... ?

Possibly an unintentionally badly seated CPU, possibly because there is more pressure on one side than on the other, causing some pins not to make proper contact. Are the pins pushed in all the way? Does it work when you lay the case horizontally? Another, careful, reseat of the CPU and cooler might help.
 
(that's really well cleaned! How did you do that? )

Hmm, this is a good guess, I'll try to install a cooler more carefully tomorrow morning. Maybe the problem occurs after installing the cooler.

Yes, RAM is inserted correctly.
From the horizontal location of the PC, nothing changes.
 
Compressed air, brushes, alcohol, lint-free wipes, vacuum cleaner with nozzle for hard-to-reach places and a little love for my work ?

SODIMM is the form factor (Small Outline Dual Inline Memory Module), all RAM now is SDRAM, the S meaning it is synchromised to the CPU clock

SODIMM is the socket/connector type, SDRAM is the type of RAM.  Its not really discussed much these days but yes its correct.

Nice to know that
 
The reason why I said “radiator” is that in my native language, there is no word that means “heat sink”.
“Heat sink” is translated as "radiator"

No worries.. EVERYONE knew what you meant anyway.. That's just people busting your chops for using the wrong term, because everyone one here is perfect and their feces doesn't stink... lol

Do you mean heat sink?
Many cheap motherboards have no heat sink for the VRMs, so the quad core Ryzen 3 lineup isn't much trouble. It would be an issue if you had a 6 or 8 core.

According to their Spec page: https://www.gigabyte.com/Motherboard/GA-A320M-H-rev-11#support-cpu
 
It lists all CPU's as supported since Bios Version "F1" which appears to be the initial Version so i would assume it supports it out of the Box.
 
You are however correct in that many 300 series Boards need a Bios Update to support em (especially if they have been on a store Shelf for a year in some brick and mortar Store)