• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA RTX 50 SERIES - Technical/General Discussion

Regards the 4090 connectors, unless I'm mistaken I don't believe I've seen any conclusive evidence of any instances of this that weren't down to user error.
It's not user error. The only reason the 4090 melts less is because it draws less power. It has the exact same power deliver setup. It is every bit as flawed as a 5090.
 
Just because your personal "experience" is better with a 5090 doesn't mean the situation as a whole isn't a total cluster-****.

The 5090 has zero price/performance improvements and only one highly contentious exclusive feature. Add in the melting connectors, lacklustre FE cooler that sacrifices heat & noise for the sake of a largely pointless 2-slot form factor, missing ROPs, missing PhsyX support and virtually zero availability and it's a total mess.

Regards the 4090 connectors, unless I'm mistaken I don't believe I've seen any conclusive evidence of any instances of this that weren't down to user error. This clearly isn't the case with the 5090 and they're melting even when fully seated plus we have the weird 'balancing' issues with individual cables as shown by Derbauer.

If you can get one at MSRP and don't have any problems then great, enjoy it and the extra performance but this doesn't mean the launch as a whole hasn't been a disaster.

The 4090 was 100% melting, it happened to mine and I can absolutely assure you it wasn't down to user error , I know of others that have melted also again cable fully seated and secure yet melted ... it's a **** design and should not have been used on a gpu period, it's far to fussy.
 
Last edited:
So it's ok for PSUs to be designed in a way which can push way too much power down individual cables?
But it's not ok for anything else to be designed that way?
Every single battery is designed that way. Short the terminals and it will give you all the current it can.

The PSU is limited. It's single rail 12v and limited to whatever that big number written on the side of it is. If it says 1000w that means it will give 83A. It cannot know what loads or what connectors you're going to use so the standards all push that responsibility to the device(s) drawing current.

The problem is that Nvidia's design does not respect that responsibility and can allow inappropriate current down too few conductors.
 
Last edited:
Just looked at the invoice from when I built a whole new PC with my 980Ti upgrade back in 2016. 8Pack overclocked i7-6700K bundle, HoF 980Ti, Phanteks case, PSU (still in use until this afternoon), HDD, SSD and DVD drive (why? lol) for a grand total of £1675.93. At the time, on my helpdesk salary, it was a tough purchase to justify. I tried getting my wife to talk me out of it, but as usual that didn't work "you work hard for your money, treat yourself".
Now the invoice for my Zotac 5090 Solid... £2239.99 and nothing else with it (hopefully a tiny pack of Haribo when it arrives in a few hours). But this time I'm no longer on that helpdesk salary, I'm now a senior dev and was lucky enough to be able to impulse change from a 5080 to a 5090 and not worry too much about it. Still not really happy with the price, and wasn't happy with the wait (that is over today, unless it is missing ROPs), but I'm looking forward to (hopefully) not upgrading GPU again for a long time now.


My Zotac 5090 Solid is due to arrive this afternoon, I'll post some photos when it gets here.
Oh awesome! Please share some benchmarks and temps (core and memory) when you have a moment. I am really curious as to have it performs given it is one of the smaller 5090s.
 
Just looked at the invoice from when I built a whole new PC with my 980Ti upgrade back in 2016. 8Pack overclocked i7-6700K bundle, HoF 980Ti, Phanteks case, PSU (still in use until this afternoon), HDD, SSD and DVD drive (why? lol) for a grand total of £1675.93. At the time, on my helpdesk salary, it was a tough purchase to justify. I tried getting my wife to talk me out of it, but as usual that didn't work "you work hard for your money, treat yourself".
Now the invoice for my Zotac 5090 Solid... £2239.99 and nothing else with it (hopefully a tiny pack of Haribo when it arrives in a few hours). But this time I'm no longer on that helpdesk salary, I'm now a senior dev and was lucky enough to be able to impulse change from a 5080 to a 5090 and not worry too much about it. Still not really happy with the price, and wasn't happy with the wait (that is over today, unless it is missing ROPs), but I'm looking forward to (hopefully) not upgrading GPU again for a long time now.

This is a good example. People doing well in their respective fields that would have spent extra disposable income on golf clubs or some other snore fest.

Now Nvidia are pricing to tap into that money. It's exploitative but they know the market is there.
 
I do actually think that a 5080FE for under £1k is ‘fair’ - OK, it’s disappointing from the perspective of anticipated gen on gen increase, but it’s not wildly bad price to performance.
It’s not just gen on gen which is bad though, go back 4.5 years and you’re looking at +66% performance for 43% more money, so just a 23% increase in performance per dollar since 2020.

2016 - 2020 would have got you 126% performance for 16% more money or a 110% increase in performance per dollar which is over 4x higher than today.
 
Every single battery is designed that way. Short the terminals and it will give you all the current it can.

The PSU is limited. It's single rail 12v and limited to whatever that big number written on the side of it is. If it says 1000w that means it will give 83A. It cannot know what loads or what connectors you're going to use so the standards all push that responsibility to the device drawing current.

The problem is that Nvidia's design does not respect that responsibility and can allow inappropriate current down too few conductors.
I reckon this is why Nvidia might get away with all this. They don't supply the end-to-end chain. Most electrical devices come with a power cable that plugs into the wall and it's clear you are at fault for dangerous electronic design. Maybe there's nothing clearly defined in some safety standard that insists things like GPUs do their own power balancing/checking. I know it's typical for that to happen but maybe it's not hard-defined in regulation.
 
It’s not just gen on gen which is bad though, go back 4.5 years and you’re looking at +66% performance for 43% more money, so just a 23% increase in performance per dollar since 2020.

2016 - 2020 would have got you 126% performance for 16% more money or a 110% increase in performance per dollar which is over 4x higher than today.

I don’t dispute that but I think the most relevant question for people in the market for a new card today is… which high end card gives the best bang for buck?

It is the 5080… at MSRP pricing, anyway. It would still be the best bang for buck even if you could get a 4090 at MSRP. Well, unless you’re one of the people that thinks a card is terrible unless you can play a modded high res pack games that make the most of the vram (tiny % of buyers).

That stands unless AMD offer something compelling but I’m not holding my breath. Here’s hoping they can be competitive.
 
Last edited:
I really don't understand why you're not getting this. Many people skip generations. No one could possibly see into the future that the 50 series is just basically the 40 series again and without a time machine your only options are to buy it or just skip another generation and maybe not play the games you really wanted to play. TLDR; hindsight is a wonderful thing.

This!

I can remember posting here more than once that I was perfectly happy with the performance of my 3080 Ti, and was not going to upgrade until the 5 series.

However, (and fortunately so, looking at the 5 series debacle) I had a small change of circumstance which meant I had to spend some dosh and I chose to buy a new 4090 FE which luckily became available at the same time.

Had the above not happened, I would now be in the same boat, attempting to upgrade from 3 to 5 series.

And in light of the despicable descison by Nvidia to drop PhysX support, I'd more than likely be trying to buy a s/h 4090.
 
I reckon this is why Nvidia might get away with all this. They don't supply the end-to-end chain. Most electrical devices come with a power cable that plugs into the wall and it's clear you are at fault for dangerous electronic design. Maybe there's nothing clearly defined in some safety standard that insists things like GPUs do their own power balancing/checking. I know it's typical for that to happen but maybe it's not hard-defined in regulation.
Historically all standards like this would have been created by IEEE steering committees and it would have been massively over engineered but most critically, safe.

This time around Nvidia got their feet under the table to steer the creation of the new 12vHPWR. This probably left the standard full of holes which they've now tried their luck with.

Any second year EE undergrad could tell you the current implementation stinks though.
 
So how's everyone enjoying the launch? :D
In honest truth, I think I speak for most of us that the launch is the worst launch of an Nvidia GPU series, perhaps ever. Never has there been so many issues with such high-end products, all the while the customers are getting price gouged.

Sure, we could be here feeling sad and miserable that Nvidia's greed and incompetence, has killed off any hopes of upgrades for most folks.
But all we can do in this situation is just laugh about it, since even complaining about it or voting with our wallets seems to have gotten us all nowhere. So that's what I've been doing, having given up any hopes of 5000 being a good GPU series, make memes and have laugh. Same thing on the AMD side, fully expecting AMD to mess up their new GPU launch like they always do, might as well have ajoke about it, instead of crying over electronics that are too expensive and unavailable to buy.

9l4kk6.jpg


So yeah, this launch is just so bad it's hilarious for me at this point, no hopes of expecting anything good out of it.

@mahius Whats so funny about my Gaming Trio having the ROPS issue, its not a problem i wil just RMA but laughing at other misfortune is a d*** move

Apologies mate, you are correct, my mistake. It's not my intention to make fun of your situation personally or anyone else's for that matter :( The fact is that this shouldn't even be happening at all, it's atrocious that such expensive products have such problems. I was just on autopilot laughing at the fact that a trillion dollar company can mess up so badly, not at you mate (as mentioned above). I had intended to unclick given the likelyhood of misinterpretation, but forgot to do so. If it helps, feel free to laugh at my expense for buying a GTX 970 10 years ago with the 3.5GB VRAM. And that's something that can't be fixed by RMA. I hope you manage to get your 5090 sorted through RMA.

I feel bad that overall most people hate the 5000 series and are dunking on it, but if you enjoy it mate, at least someone out there got something positive from this launch.

The rest of us... we got nothing.
9l4l6o.jpg
 
I really don't understand why you're not getting this. Many people skip generations. No one could possibly see into the future that the 50 series is just basically the 40 series again and without a time machine your only options are to buy it or just skip another generation and maybe not play the games you really wanted to play. TLDR; hindsight is a wonderful thing.

So waiting to see what AMD bring with the 9070/XT would not be an option? Just get on with a cheaper stand in card until things normalise slightly is not an option. These are GPUs here, not life or death.

The idea that these people literally have zero option but to pay and feed the scalper nonsense is laughable.
 
I don’t dispute that but I think the most relevant question for people in the market for a new card today is… which high end card gives the best bang for buck?

It is the 5080… at MSRP pricing, anyway. It would still be the best bang for buck even if you could get a 4090 at MSRP. Well, unless you’re one of the people that thinks a card is terrible unless you can play a modded high res pack games that make the most of the vram (tiny % of buyers).

That stands unless AMD offer something compelling but I’m not holding my breath. Here’s hoping they can be competitive.
The 5070ti actually looks far better bang for buck if we are just sticking with MSRP pricing.

TPU has the 5080 just 11% faster than a 5070ti but costs 33% more.
 
So waiting to see what AMD bring with the 9070/XT would not be an option? Just get on with a cheaper stand in card until things normalise slightly is not an option. These are GPUs here, not life or death.

The idea that these people literally have zero option but to pay and feed the scalper nonsense is laughable.
I never said to feed scalpers. The chance of AMD making something better than the 5080 is approaching nil. What cheaper stand in cards?
 
Thanks for people like @mahius for laughing at my situation but i just hit jackpot
Card was purchased from probably teh worlds largest online retailer. They could not offer me a replacement as they have none so they have told me to keep the card and given me a patial refund of £625 (original price £2499) meaning RTX 5090 Caming Trio OC cost me £1874. I also have it in writing that my warranty with MSI is unaffected and am still claim a replacement card from them directly
 
Last edited:
Back
Top Bottom