Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

We have an issue with RTX 3060Ti and we need to talk about it

MaximuT
 Share

So.... My friends work in repair service. An they are having a lot of 3060 Ti with issues. Most of them show that there is some strange relationship with Hynix memory. We tried to Google it, but it seems that this issue either being removed or not officially posted. The main issue that received 3060Ti cards come with Hynix memory (did not found with other memory), but according Manufacturing documents we can install Samsung memory. So, if we want to fix this issue , we need to change all Hynix memory, since time passes and if we leave any of these memory, card will return with same memory issue. We cannot identify which of the eHynix memories on card is broken, since one test they pass, othere test - Black screen. Logs dont help. The only solution that we have at this moment - to change all memory on card to Samsung.

And most of these observations are not official. We collected this data from those 3060TI that we receive from clients (miners and ordinary users) and statistic from our colleagues from other repair services. When one of repair partners tried to official announce it (say not to but 3060Ti) sicne they have issues), one of our local distributors contacted with us “showed us “BanHammer”

People come to us saying that they (cards):

- crash

- lagg

- BSOD

- dont work properly 

Most of them are PALIT and Gigabyte ones

Several month we have this strange issues with Hynix memory. for the last month we received 30 RTX 3060ti Cards. And mostly due to these memories cards die but they are repairable. In worst cases - Graphic chip dies.

Each month the quantity of GPU with Hynix memory issues that we receive increase exponentially.

 

Is it due to cooling issues? Perhaps not... BUT thesee Hynix memory extrimely sensitive to temperature changes

So, i googled the issue for friends, and most of results are provided by Miners but issue was about hash speed. They compare the same card with Samsung and Hynix... and in most of cases Hynix ones work not well. Riggs with Samsung memory can work well, but when Hynix card is added - All rig starts to work not so optimal

 

we are trying to find any other reason why cards die... but all our stats are collected in one country... so... we are looking for additonal info in the world. 

 

Link to comment
Share on other sites

Link to post
Share on other sites

5 minutes ago, MaximuT said:

We tried to Google it, but it seems that this issue either being removed or not officially posted

Shit, let me put my tinfoil hat on first.

*hat is on*

 

 

This is the first time I am seeing someone report an alleged widespread issue with 3060tis.

8 minutes ago, MaximuT said:

most of results are provided by Miners but issue was about hash speed

Because daggerhashimoto/ethash benefits from higher graphics card bandwidth, higher clocking vram = better hash rates. This is not a point that supports your claim.

 

11 minutes ago, MaximuT said:

Hynix memory extrimely sensitive to temperature changes

As you predicted, I suspect this has something to do with cooling. Stock fan profiles on these cards are very focused on silence, vrams might be subjected to temperatures they were not designed to quite easily. I dont think issue is temperature changes though, if that was the case you simply can test it by applying moderate heat to one of these vram chips (less than solder melting temperatures) and see if it kills a card that was previously working fine.

 

mY s YsTeM iS Not pErfoRmInG aS gOOd As I sAW oN yOuTuBe. WhA t IS a GoOd FaN CuRVe??!!? wHat aRe tEh GoOd OvERclok SeTTinGS FoR My CaRd??
 HoW CaN I foRcE my GpU to uSe 1o0%? BuT WiLL i HaVE Bo0tllEnEcKs? RyZEN dOeS NoT peRfORm BetTer wItH HiGhER sPEED RaM!!dId i WiN teH SiLiCON LotTerrYyOu ShoUlD dEsHrOuD uR GPUmy SYstEm iS UNDerPerforMiNg iN WarzONEcan mY Pc Run WiNdOwS 11 ?woUld BaKInG MY GRaPHics card fIX it?
 MultimETeR TeSTiNG!! aMd'S GpU DrIvErS aRe as goOD aS NviDia's YOU SHoUlD oVERCloCk yOUR ramS To 5000C18

 

Link to comment
Share on other sites

Link to post
Share on other sites

19 minutes ago, Levent said:

Shit, let me put my tinfoil hat on first.

*hat is on*

xD 
yeah... we also had such hats... but when started to receive more and more 3060Ti card -we start digging. 

 

20 minutes ago, Levent said:

Because daggerhashimoto/ethash benefits from higher graphics card bandwidth, higher clocking vram = better hash rates. This is not a point that supports your claim.

We would agree with this comment... but we also have cards from ordinary users that did not mined (According to their comments)

23 minutes ago, Levent said:

As you predicted

also was one of the first reasons why they might be dieing... but guys checked them and fans works are on Best level. and also... if was, lets say, 5-10 cards per month we would not accent on this issue... but we guys constantly receive 3060TI cards with Hynix 

PERHAPS, since we record these issues only on local level, on European and US market People also might think like you and they throw the blame on the users, but we would like to get more global statistics, if there are such... After all, the easiest way is not always the right way.

30 minutes ago, Levent said:

This is the first time I am seeing someone report an alleged widespread issue with 3060tis.

That is why we need to talk about it. 

 

We had other issues with 2GB hynix memory that were installed on AMD Radeon cards, but since these cards are not so popular on local market, we didnt take them into account. 

Link to comment
Share on other sites

Link to post
Share on other sites

So, it seems that there are not so many people here that can assist with these issues.

Link to comment
Share on other sites

Link to post
Share on other sites

  • 5 weeks later...

Just want to update the status of this issue that here in CIS (mostly Russia) is being accured. 

"According to last information, the video memory malfunction mimics the video chip malfunction. The memory even passes special technological tests. The repair consists in replacing the faulty "cans" of memory, but unfortunately such repair is not enough for a long time, and after a while another chip fails. A radical solution to the problem is to replace all Hynix memory with memory from another manufacturer, such as Samsung. Signs of memory chips malfunction and their mass nature indicate a defect in production or miscalculations in chip design, perhaps one large batch of problematic chips has come. Previously, defective memory was encountered by owners of graphics cards with Micron memory on board and now the turn has come to Hynix.”

 

 

Perhaps there are someone who faced with the same issue outside CIS? 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×