Bought 1660super

Kennedy223
In a couple of years, your video card will become a pumpkin from the word at all. So don't worry too much, then buy a current one if you're still alive.

This hunix story is about 3060ti cards.

As for it will burn - not a fact, and even there, it seems, only the batch was defective, and then 3060ti. By the way, how did you know? GPU-Z shows Hynix memory?

kurskiy
No, not a party. At first, the repairmen thought that there might be a batch, but it turned out that they still put this memory on the 3060ti. And this hunix memory problem is only specifically for the 3060ti, these memory banks are not specifically used anywhere else. Its problem is rapid degradation at high temperatures, if you monitor the temperatures and do not overheat this memory (keep it in the region of 80 and below), there will be no problems with it.

Vindicar A
video card doesn't have to be fat for it to have memory problems. A lot depends on the cold and the ability to track its temperature.
I read that on a 3060Ti with hunix memory, there is a high voltage on it, and because of this it burns

Denis Kyokushin
Without this voltage, this memory cannot always take the memory frequencies declared for this video card. But the same Samsung can. And at high temperatures, the samsung does not burn like this hunix. The problem is specifically in these hunix banks, which are put specifically on the 3060ti.

Vindicare
I have a Samsung 7950 bank, the voltage of the video memory was 1.6v when it worked, but it did not give in to overclocking. Now, they say, 1.45v is an ass for the video memory

Kennedy223 wrote:
I recently bought a 1660super and found out that there is a hyunix 1660S on the vidyuhi,
this does not apply. Sit still and don't move. The rest depends on the quality of the card.


The Vindicare wrote:
Its problem is rapid degradation at high temperatures, if the temperatures are monitored and this memory is not overheated . It's a
lie. The memory on the card is new, it is not fried. One of these lies. The point is the overestimated voltage supplied to the memory, in the load there is 1.4 Volts, which is a lot.
Denis Kyokushin wrote:
I read that 3060Ti with hunix memory has high voltage on it, and because of this it burns
Stunned, really Google opened. I'm shocked. Surprised so surprised. And it does not burn, but degrades under high voltage. It works, only loses stability, like any other unit from degradation under high values.
The repairers already had tests, and they gave examples, as well as the specification, changes in voltage to the load. 1.25 volts is needed for hynix. Samsung takes more.
Vindikar wrote:
The problem is specifically in these banks of hunix, which are put specifically on the 3060ti.
Yes, because there are both hynix and samsung. And someone messed up and did not take into account that you need less for hinis. With memory, everything is in order. This is not fixed in any way at all, by no cooling and piping of the bios.

01001000 01101001 The guy
in your video says that this is a test on just one card and he does not guarantee that it will work on all 3060ti with hunix.
Here Vikon at 3:57 says that lowering the voltage does not help all cards.
Spoiler On dead hunixes, this does not help by itself, but on living ones it leads to random map freezes.
So no, there is something wrong with this hunix memory. It survives heat worse than the Samsung and, according to some reports, see the timings from the video above, lowering the voltage does not guarantee its performance.
Bios does not lower the voltage, but lowers it by replacing some elements on the board
SpoilerBut, as Vicon said from the video above, reducing the voltage for this memory can lead to card freezes.
And yes, the problem is in the memory, because when you replace it with a Samsung, the problem disappears completely, although the same voltage is applied to the memory banks that was applied to the hunix banks.
And the fact that there is some % of cards with this memory that are helped by lowering the voltage still does not allow card manufacturers to lower the voltage on all cards. All 100% of cards with this memory must take the declared frequencies and ensure stable operation, so the voltage is set at which all 100% of the cards will do this. They can't set all the cards to low voltage and get 50% of workable cards at the output, which turned out to have more successful memory.

Denis Kyokushin
Yes, here we are not talking about all the hunix memory, but specifically for the one that is placed on the 3060ti.

Vindikar
Vindikar wrote: The
person from your vidos himself says that this is a test on only one card
There, see the author's further. Everything is done and rechecked and proved, told, tables are given.
The Vindicaar wrote:
So no, the hunix's memory is not all right.
You're wrong. Watch attentively.
The Vindicare wrote:
But as the Vicon said in the vid above
It's not a vicon. Expand the horizons of repairmen. Although even that test was with high voltage.
Vindicare wrote:
And yes, the problem is in the memory, because when you replace it with a Samsung, the problem disappears completely
. Of course. Because there is 1.35 volts according to the specification. For hynix 1.25.


I thought you were smart enough to keep watching.

01001000 01101001 A person plays
the same card for 15 vids in a row and draws conclusions based on this card.
On the other hand, there is Vicon, who has more connections than all the other repairmen combined and who was informed that reducing the voltage on this memory can lead to card freezes, that is, this is roulette.
Hmm, who do you believe. I'll trust Vicon, I guess. Above, I wrote why manufacturers cannot produce cards with this hunix and reduced voltage. They need to get 100% working cards at the output, and not 50%, which got a more successful memory.

Vindicare
Your vicon came to the same voltage. there is nothing more to talk about here. He has the same video. No one has progressed further on this.

01001000 01101001
I posted above vidos and timings about possible freezes after a voltage drop. I have no reason not to trust his words.
This is indirectly confirmed by the fact that card manufacturers have been producing cards with "increased" voltage for the hunix for the second year already. This only confirms that only at this voltage 100% of the cards at the output turn out to be stably working.

Again, take the Samsung memory and this hunix. Samsung does not die at these elevated temperatures, but this hunix dies. Which also suggests that something is wrong with this hunix and it was not worth installing it on these cards en masse. Especially during the mining period, when all the shoals climb out twice.

Vindicare
This topic of dialogue is not worth it. Hynix works, and everything is fine with him. Samsung initially, according to the table and capabilities, digest large values. The whole joke. Hynix has always been a specific deshman, the worst choice and potential for overclocking and work in mining.
Here hynix is ​​covered, but not in other maps. Wonderful miracle. The temperatures there are normal, the banks are new, there are no signs of roasting, they will work, degrade and fly off for repairs to replace them with Samsung, where the factory value of 1.35 is set. So can the Samsung RAM, up to 1.5 Volts, there are sets of brackets. Other such values ​​are not allowed. Open the table and figure out the voltage, for which memory what values.
This is DDR4
And don't spread rumors and arguments. Everywhere works, and here problems. Already TWO repairmen told you this.
The Vindicaar wrote
that card manufacturers have been releasing cards with "increased" voltage for the hunix for the second year already.
Yeah, and there are Palit dual lying around - there is nothing worse than it, even giga gaming OC is even better, although giga is generally unpopular, except for the aurus in the recommendations. I also came across MSI gaming OC and gainward, the same fires. For the rest, units. Why? Not massively and not en masse, but they say it like every card. Panic spread. Personally, I silently changed mine to 3080 and 3070, a penny surcharge. The card is gone for good money.

01001000 01101001
You are turning the facts around. The repairmen just proved that there is a problem with the hunix in the 3060ti. That hynix deshman and technologically behind the Samsung, I myself know.
Is there a problem on 3060ti video cards with hunix? - there is. Is it possible to fix it by just flashing something there? No. Is the user obliged, after buying a new card, to climb into it and solder the resistors to lower the voltage, while losing the warranty? - No. There is only one conclusion from all this - if you don’t want to take risks - don’t take 3060ti or take it from those suppliers who guarantee that the card initially has a Samsung. The Chinese just put Samsung in their brands. I also know that from the middle of the 22nd year, the Gallach began to produce 3060ti at the initially reduced voltage of 1.25-1.28.

Here's some info you probably didn't know at all. What exactly did hunix provide for 3060ti cards, why do they fly, and, most interestingly, that even hunix themselves declared the operation of this memory at 1.35 voltage in their own table on their own website.
Carefully look at the timings 9:45-13:00

How do you like these RUMORS? : DDDDD
Well, yes, the problem is not general, therefore, as I wrote above - watch the temperature of the memory, do not heat it above ~ 80 - + a couple of degrees and there will be no problems.

Vindicare
Are you mocking me? Vicon sect. This video is OLD! Look new, and stop inventing. I watched them all because I owned a couple.
By the way, I was not the owner of the Colorful iGame GeForce RTX 3060 Ti Ultra W, my colleague has been in the farm for more than a year, and they work perfectly. Good bullshit toils. Here is a list of cards that showed up the least (were they at all?) in repair. These cards have proven themselves and work:


So it’s good to stir up and prove to ME, the person who has been following this since February, when there were the first patients. And I kept track of which cards were being repaired. And many were not there. Giga is the main visitor from the dual fires there. Neither one nor the other, I never put in the assembly and did not recommend to ANYONE! And now, yes, it's easier to say that 3060 ... 3070. 3060Ti is easy for people to take under warranty. But divorced here.
Vindikar wrote:
What exactly did hunix provide for 3060ti cards
? Unbelievable, these cans are only on the 3060Ti! Can't be!

01001000 01101001
You've been following badly, since you overlooked the table from the hunix itself with the specs of their memory. Or do you think she's OLD too? : D
And what did you put in the assemblies - it's purple for me. There was a problem with hunix on 3060ti. How to avoid this problem - I wrote before your appearance in this thread - do not warm the memory. The fact that for cards with good cooling the hunix heats up less and less often frays - you just discovered America right now :D

Just think, the operation of the memory at the voltages declared by the memory manufacturer in its own specs is just not guaranteed, because according to the marking this memory is not for mass production at all, but a run-in version that was released to the masses in a hurry. Buy a card - pick it up, solder the resistors to lower this voltage. Or even better - buy the top version at a price of 3070, so that there are no memory problems. In general, a problem-free card and this memory from hunix is ​​excellent, I recommend it to everyone: D

So I repeat once again, everything that is needed, I wrote here in this thread even before you appeared. Do not warm the memory, period. It's you who came running here to flood.

Kennedy223 3 years flight normal card good don't worry

01001000 01101001
Enough of the gag, hunix on their website indicated the voltage for their memory is 1.35v, card manufacturers put this voltage on the cards. No need to shift the blame from hunix to card manufacturers.
https://product.skhynix.com/products/dram/gddr/gddr6.go?appTypCdAPX06&treeNo=1052
Back in January, overclockers wrote that chips with CS status are not a product for mass production https://overclock3d.net/news /memory/sk_hynix_s_newest_gddr6_modules_will_power_future_gpus_with_20_gbps_speeds/1 , but hynix didn't give a damn about that either, and now we have cards with these H56G32CS4DX005 chips. Is it also the fault of the card manufacturers that hynix put unfinished and undertested probes into mass production?
Vikon has already told everything in detail on this topic, why are you again trying to pour your water here, idle talk and drag in unrelated things? I will ignore all this. There are dry facts, and there is nonsense from a huckster-collector who is uncomfortable when someone raises a topic with a 3060ti and hunix. No need to try to chat me up, when I see a similar flow of water, I skip it. Fuck someone else's head.

01001000 01101001
Who is confused about what? Hynix released samplers, the voltage indicated in the passport was 1.35v. Card manufacturers have done everything according to the passport. Then it turned out that if you don’t have a top-cooled card, this hunix can fall off, because it’s a shitty sampler, stuck together in a hurry on the knee, which does not hold the elevated temperature that memory from other manufacturers normally holds.

And in the end, in order to insure against the PROBLEM THAT HUNIX brewed, you either need to solder the card to lower the voltage, or buy a version with top cooling for an overprice.
No one got confused in anything - the hunix crap, those who want to buy a 3060ti on the hunix or have already bought it, the advice is not to warm the memory above 80. This does not apply to cardholders on the Samsung. What's confusing here? everything is extremely simple, it seems like.

And by the way, yes, Hynix greatly overestimated his memory. Until now, banks above 7 GHz do not have the status of mass production, although probes have already been riveted to 10 GHz.