News Crash Tested: Nvidia Improves RTX 30-Series Stability With New Driver

JarredWaltonGPU

Senior GPU Editor
Editor
Tom's, could you please list which cards from which manufacturers do and don't have the MLCC blocks?
My understanding is:

Asus: 6xMLCC
EVGA: Varies, 6xMLCC on top models? 4xSP-Cap/2xMLCC on others? (I don't have a card yet)
Gigabyte: 6xSP-Cap on some models, not sure on others
MSI: 5xSP-Cap/1xMLCC on some, 4xSP-Cap/2xMLCC on others
Nvidia: 6xMLCC
PNY: I don't know
Zotac: I don't know
 
  • Like
Reactions: vinay2070

spongiemaster

Admirable
Dec 12, 2019
2,273
1,277
7,560
I'm suing over your comment for defamation, and implying no one considers my 0.2% testing results difference meaningful! sniff!:'(
You don't need to sue me. Don't you have three 3080's? That's like $45,000 on Ebay.

Jokes aside, the article was very useful. Basically shows that if you want a 3080, and you can find one at a reasonable price, to go for it. The problems are getting fixed, and the performance isn't affected.
 

Shadowclash10

Prominent
May 3, 2020
184
46
610
I'm suing over your comment for defamation, and implying no one considers my 0.2% testing results difference meaningful! sniff!:'(
So, aren't the results here well within the margin of error? Something as samll as, oh I don't know, the weather being 10 degrees hotter during one test, or say the internals being warmer after a previous test? Not criticizing, simply asking. Regardless, I think that if the difference is so small it is basically inconsequential?
 
I was reading elsewhere that SP-Caps have a better life expectancy with less chance of cracking. Therefore the decision between the 2 types was more about life expectancy vs a few % higher performance. Does anyone know if their is any truth to this? Definitely would put a different spin on this whole capacitor discussion.
 

Blacksad999

Reputable
Jun 28, 2020
70
48
4,570
Considering the cards technically only need hit their advertised boost clock, anything even near 2k is pretty good. So, the Tuf boost advertises as: 1740MHz, the EVGA FTW 1800MHz, etc.
 
Does this mean that the next cards, from the assembly line, will include a 'hardware' fix?
The more I read on this the more I come to the conclusion this isn’t a hardware problem. Expert reviews say the caps can only make <=50Mhz difference. I saw a video where 2 caps were replaced and it made a 30mhz difference. The other factor is the boost in question only operates for a few seconds, I would guess that over 99% of the time the gpu isn’t using this max boost anyway.
 

JarredWaltonGPU

Senior GPU Editor
Editor
When I went through the article, I was like, he has a lot of patience!!! :)
It's absolutely required for testing hardware. I have 100+ hours logged in multiple games on Steam, for games that I basically never played but only benchmarked. Strange Brigade is a good example of this, or Shadow of the Tomb Raider -- I've played that for maybe 10-15 hours, but certainly not the 225 hours I have on record! And the great thing is that about once a quarter, or at least semi-annually, I need to go and retest every GPU to keep the GPU Benchmarks Hierarchy up to date. But as far as jobs go, writing about new hardware is pretty fun for me. :)
 

bigdragon

Distinguished
Oct 19, 2011
1,106
546
20,160
My takeaway is that gaming performance increased slightly, synthetic benchmark performance dropped significantly, and temps improved slightly. I use GPUs for gaming and game dev so the improvements are more appealing than losing synthetic numbers.

Nvidia should have delayed Ampere, however. There is no excuse for rushing untested or barely-tested designs to market like this. I am enjoying seeing scalpers panic due to early-adopter customers trying to return scalper cards! LOL!