roosnam1980
Gold Member
Article: RTX 5070 Ti reaches the 107°C thermal limit while Windows reports 68°C.
Brazilian repair specialists......................
Last edited:
-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
Repair technicians access hidden RTX 50 hotspot data with NVIDIA MODS
- Thread starter roosnam1980
- Start date
Brazilian repair specialists......................
Black_Stride
do not tempt fate do not contrain Wonder Woman's thighs do not do not
Hiding the hotspot sensor in RTX50s but leaving it on in RTX40s/30s tells you Nvidia knew this was an issue.
Sardaukarama
Member
kiphalfton
Member
When's the class action settlement. Ready to reclaim some of that money I got ripped off.
Skifi28
Member
Hold on, let them cook.
bender
What time is it?
Magic Carpet
Gold Member
Just this past weekend I installed a new AIO for my AMD 7900X.
For funs I went into the BIOS and messed around with the TDP settings. If I lowered the settings below 105C Steam stopped showing the CPU temps entirely.
I've ended up keeping it at 105 and forced all the fans to stay at 20% until temps get to 85C. Then let the fans ramp up from there.
So far it's keeping the PC nice and quiet while playing games.
For funs I went into the BIOS and messed around with the TDP settings. If I lowered the settings below 105C Steam stopped showing the CPU temps entirely.
I've ended up keeping it at 105 and forced all the fans to stay at 20% until temps get to 85C. Then let the fans ramp up from there.
So far it's keeping the PC nice and quiet while playing games.
OLED_Gamer
Member
Damn. My 9070xt sits cool at 80ish degrees hotspot with my fans barely at 1000rpms.
CuteFaceJay
Member
up until it melts the power cableOn top of being a Good GPU it is a good cooking utility as well.
impressive
//DEVIL//
Member
What is the alternative ?
Shitty AMD card ? Or weak ass consoles ? Even the PS6 isn't as powerful as a 5080 and that system is 2 years away.
No thank you. I would rather play with the 5090 at 70% power limit. Lose 5 to 10% performance and enjoy the gaming at its peak. ( Assuming you are too lazy / too stupid to do a proper undervolting ( not speaking to you directly, in general ))
5090 is an overkill. Even the 4090 was, and still is, more than Enough. Not just enough. So yeah.. I think I will stick with Nvidia for now.
And when the 6090 is out, I will also probably still gonna buy it ( assuming it can be found for a decent price and not 5k US )
Last edited:
A.Romero
Member
While the reading is not easily read according to the article, the change in the clocks should be obvious, right?
The article describes a card that had poor thermal contact and when they solved that it came down 7 degrees. I wonder if we are getting short changed by this or it's just a defect on that particylar card...
Undervolting is pretty easy on this cards so if you haven't done it yet, go and do it. You'll get better performance at better temps.
The article describes a card that had poor thermal contact and when they solved that it came down 7 degrees. I wonder if we are getting short changed by this or it's just a defect on that particylar card...
Undervolting is pretty easy on this cards so if you haven't done it yet, go and do it. You'll get better performance at better temps.
Buggy Loop
Member
Crimson Desert on the 5070 Ti
The Gigabyte SSF edition, the smaller coolers of the lineup and in a meshilicious case which again, is typically low liters, not that much fan circulation compared to big towers
How the fuck videocardz got 107°C
The Gigabyte SSF edition, the smaller coolers of the lineup and in a meshilicious case which again, is typically low liters, not that much fan circulation compared to big towers
How the fuck videocardz got 107°C
//DEVIL//
Member
Yeah even the 5090 Astral while playing call of duty didnt reach 80 degrees. I was playing at locked 240 hz with dlss on.Crimson Desert on the 5070 Ti
The Gigabyte SSF edition, the smaller coolers of the lineup and in a meshilicious case which again, is typically low liters, not that much fan circulation compared to big towers
How the fuck videocardz got 107°C
I am going to assume i am not hitting the max power of the 5090, but still this is very impressive for a cooler.
BlackTron
Member
How are you supposed to know you are hitting the desired number if they obscured the sensor data? All he said is that you can't defend that they did that, and you can't.
SoloCamo
Member
Describe what is shitty about them? For the 20+ years of my pc gaming I usually had the top tier GPU, now I've settled with a RX9070 Powercolor Hellhound for $550 new and it's certainly not shitty.
I'm FAR from a power efficiency guy but the 5090 is kind of ridiculous.
Last edited:
nemiroff
Gold Member
I'm slightly confused by this "story". One card, one internal tool, one technician, No calibrated FLIR reading? That's pretty odd. It's almost like no one gives a shit about context and science
AFAIK, the failure rate of these cards isn't particularly high (last I checked it was around 1–3%?), stability has been normal, no particular image artefact issues. - unless I have completely missed some previous reports. The wattage = heat dissipation remain "the same" no matter. So what's the claim here. That Nvidia wants to cut down on support headaches because the readings could be misleading when the thermal paste application is fucked? Idk.
Of course this guy would immediately pop up here, stories like this is like the best christmas ever for him
AFAIK, the failure rate of these cards isn't particularly high (last I checked it was around 1–3%?), stability has been normal, no particular image artefact issues. - unless I have completely missed some previous reports. The wattage = heat dissipation remain "the same" no matter. So what's the claim here. That Nvidia wants to cut down on support headaches because the readings could be misleading when the thermal paste application is fucked? Idk.
Of course this guy would immediately pop up here, stories like this is like the best christmas ever for him
Last edited:
FalconPunch
Member
Of course the Nvidia defenders would show up. Having access to the hotspot temperature is important for diagnostics and preventative maintenance. If I can see that my hotspot is really high, I can immediately take corrective action like changing the thermal paste etc. If I buy a new gpu, I can test it right away to ensure that the hotspot temps are inline. If it's not, I can exchange the card for another one.
There's no good reason to not have that information available. The only reason Nvidia would remove it is if they had something to hide which they do.... First 12vhpwr, then hotspots temps, etc. Everything about the 50 series design is shoddy and I say that as someone who owns a 50 series card.
Puscifer
Gold Member
I'm not lol. This is pretty bullshit and I'm wondering if like the 7900 xtx we can repaste with some PTM.
I love his videos because my brother is a giant math nerd and the only difference is Daniel teaches it and my brother buys calculus workbooks lolI'm slightly confused by this "story". One card, one internal tool, one technician, No calibrated FLIR reading? That's pretty odd. It's almost like no one gives a shit about context and science
AFAIK, the failure rate of these cards isn't particularly high (last I checked it was around 1–3%?), stability has been normal, no particular image artefact issues. - unless I have completely missed some previous reports. The wattage = heat dissipation remain "the same" no matter. So what's the claim here. That Nvidia wants to cut down on support headaches because the readings could be misleading when the thermal paste application is fucked? Idk.
Of course this guy would immediately pop up here, stories like this is like the best christmas ever for him
Last edited:
Puscifer
Gold Member
Yeah but it's slots so cutely in SFF builds
Of course the Nvidia defenders would show up. Having access to the hotspot temperature is important for diagnostics and preventative maintenance. If I can see that my hotspot is really high, I can immediately take corrective action like changing the thermal paste etc. If I buy a new gpu, I can test it right away to ensure that the hotspot temps are inline. If it's not, I can exchange the card for another one.
There's no good reason to not have that information available. The only reason Nvidia would remove it is if they had something to hide which they do.... First 12vhpwr, then hotspots temps, etc. Everything about the 50 series design is shoddy and I say that as someone who owns a 50 series card.
Uguuu kawaiiii
nemiroff
Gold Member
Does this happen often to you? Do real world failure rates reflect this concern? Or is it just a feeling of concern not backed by data. Idk, I'm just asking because I'd like to know the facts because "I have Asperger's" and need to (try to) know EVERYTHING at ALL TIMES - and especially if there's a chance I'm wrong
I didn't understand that response. I took out the "shearing" part of my post because I wasn't sure if it was proper English (I was trying to point at ~metal fatigue from hotspotting). (not my native language - Ohbivousslyy). It seems it meant something different from what I thought? Is that it?
Also, as a critic of mega corporations like Nvidia (which I constantly give flak) I don't understand the labeling thing. "Nvidia defender", lol, that's a cute one.
Last edited:
Hawk The Slayer
Member
I undervolt my 5070ti and it's usually in the 50cs range, sometimes lower in the 40cs or higher in the 60cs even depending on the game shown in HWINFO64. How accurate is that? Usually stays around the 50cs range for most games. Hope these HWINFO64 info are accurate. Hopefully the hotspot is only a bit higher than these numbers like 10c to 20c max higher.
Last edited:
nemiroff
Gold Member
Nobody knows, because the wonderful scientists behind this story didn't even bother to measure the physical temperature - Or even test more than the one card.
But I'm supposedly a "Nvidia defender", so what the hell do I know
Last edited:
Dirk Benedict
Gold Member
This is straight up insidious shit, if absolutely true... in the back of my mind, I always have it that these tech companies are so good at designing these things, that planned obsolescence can be baked into the hardware design.
Buggy Loop
Member
And what's that they're hiding? What today changed is that peoples now see their GPU has been fine for over 2 years?
FalconPunch
Member
My apologies dude. I was a bit snarky in my reply. Well it has happened to me and more with old gpus. It's a sign of when I need to do maintenance to extend it's life span. Changing the thermal paste and pads.Does this happen often to you? Do real world failure rates reflect this concern? Or is it just a feeling of concern not backed by data. Idk, I'm just asking because I'd like to know the facts because "I have Asperger's" and need to (try to) know EVERYTHING at ALL TIMES - and especially if there's a chance I'm wrong
Yea I was just poking fun at the spelling mistake. Nothing serious.
//DEVIL//
Member
I am not defending anything. nor do I care or love Nvidia as a company? People being a fanboy of a company or a plastic box aren't really smart people ( no offense, but I stand by what I say ).
Its just that there is no alternative ( which was my post ). Intel is a flop. AMDs most powerful card isnt as good as an 80 Series from Nvidia, let alone 90 Series. us looking for the highest quality of graphics possible have no choice but to stick with Nvidia.
You do you. If that's what you like, then more power to you, bud. I look for high-end cards, and AMD left that market. They only focus on mid-range products. That alone, to me, is garbage and not worth looking at anything AMD does in terms of GPUs. I promise you not even the new AMD GPU will be as powerful as the 5090, and that next-gen card isn't coming till late 2027.
To me, AMD has been nothing but problems for years. Let's not even go all the way back. Let's just compare AMD's 6000 series to NVIDIA's 3000 series from six years ago until today.
AMD has almost always been behind when it comes to performance. They make questionable decisions, especially with memory. NVIDIA has consistently offered higher memory bandwidth in comparable cards, while AMD's choices have often left their GPUs choking at 4K because they simply don't have enough bandwidth.
They're still using GDDR6 ( Not even 6x that was on some on some nvidia cards from the year 2020 !!!!! ) their 9000 series because it's cheaper than GDDR7. It's just one bad decision after another.
Driver support has also been slower. To this day, your NVIDIA card will usually get a Game Ready driver on release day—or even a few days before a game's launch. With AMD, you're often waiting a week or two at least, unless it's an AMD-sponsored title.
Their drivers were buggy for years. It wasn't really until 2025 that they finally started getting their act together, and FSR was absolute shit compared to DLSS 2, 3, and 4. Only with the 9000 series and FSR 4 did they finally become competitive. Even now, we still don't have Multi Frame Generation, and NVIDIA has had it for well over a year. DLSS 5 will probably launch this fall, and AMD hasn't even hinted at a response.
They announced the FSR diamond. They already hinted that it's not coming to the 9000 series, while Nvidia is still supporting their cards from the 2000 series with the newest DLSS, giving the user the option to use it and lose a few more frames or stick to whatever DLSS they want within the Nvidia app itself. AMD? like a fart in the wind.
You might hate DLSS 5. Hell, I might hate it too. Who knows? But that's not the point. The point is that AMD is always behind. Always.
Not to mention the recent FSR 4 drama. The RX 6000 and 7000 series were clearly capable of running FSR 4, but AMD chose to screw over those customers by keeping it locked to the 9000 series. Then the DLL leaked—because of their own mistake—and people proved it worked. AMD got called out for it.
After that, they stayed completely silent on the issue for more than a year after the 9000 series launched. Why? Because they knew people who bought a 7900 XTX wouldn't have much reason to upgrade to a 9070 XT if FSR 4 officially worked on their existing card.
And then there's this...
How the fuck does your newest card end up not only being weaker, but also having less VRAM than the card it replaces? What kind of backwards company makes decisions like that?
All this so they can say, "Hey, look! Our card is $600 instead of NVIDIA's $750." Who gives a flying fuck?
These mother fuckers cant even get their drivers to work best on release day. The 9070XT was losing to the 5070 Ti until recently, with more driver updates over the year till it became the same performance. Some morons like to call it AMD fine wine, I call it engineering retardness. I am buying an AMD card with drivers that don't utilize the card to its potential, and I have to wait months or years to get that extra performance. GTFO with that garbage.
I'm not saying NVIDIA is perfect. Far from it. But AMD somehow finds a way to drop the ball every single generation in the dumbest ways possible.
Why would I buy their product when the competing product is simply better?
The PC gaming market has room for everything. Budgets range from $600 all the way to $6,000. Give me the option to buy a true high-end AMD GPU. Give me something exclusive that actually benefits gamers instead of constantly playing catch-up with NVIDIA.
Hell, even Intel's XeSS was better than FSR 1 through 3.5, and that came from a company that had barely entered the discrete GPU market before scaling it back.
The word pathetic doesn't even come close to describing AMD.
Last edited:
Buggy Loop
Member
Yea?
Founders edition 5090 ~600W beast
What's the issue? What's shoddy?
Anyway, good on this being now available and it'll spread to better monitor tool, imo hwinfo already has very good performance limiter details that are more critical than hotspot but the more the merrier.
I expected a lot worse to be honest in the delta from GPU temp to hotspot, a good ol 30-40° delta ala RDNA™. Overall this is a nothingburger. Pretty much in-line with Ada series.
Last edited:
ShadowkillXNA
Member
Yet again PC Master Race not giving Sony credit for something they did 20 years earlier SMDH.
Show Ken some respect dammit!
Last edited:
roosnam1980
Gold Member
it was crazy KEN , kaz is innocent...................Yet again PC Master Race not giving Sony credit for something they did 20 years earlier SMDH.
Show Kaz some respect dammit!
Gp1
Member
In Nvidia's defense, AMD does the same thing. The diference is you can access the the memory temperature (which is a good 20° C above the hotspot) directly from the adrenaline OCD and it does throttle at the right temperature (i believe it's 105 C°).
But the thing is that the hotspot isn't the hotspot of your card (just the hotspot of the GPU die)
I remember one of the infamous "first" videos on the internet of someone frying an egg over an Athlon XP
But the thing is that the hotspot isn't the hotspot of your card (just the hotspot of the GPU die)
I remember one of the infamous "first" videos on the internet of someone frying an egg over an Athlon XP
Last edited:
ShadowkillXNA
Member
it was crazy KEN , kaz is innocent...................
Shit you're right, I got them mixed up lol!
Sorry Kaz!
Justin9mm
Member
As a console gamer who does own an older gaming PC(3080) I agree 100%. There is no other option if wanting top PC Performance. The tech/gaming industry right now is in tatters. AI is slowly ruining our consumption of tech. I'm actually quite content with the Pro, but by no means does console gaming or the Pro come close to PC still if you are comparing high end GPUs.
YeulEmeralda
Linux User
Of it's designed to work at these temperatures who cares?
mclaren777
Member
I own two RTX 5070s and I rarely (if ever) run them above 80% load.
I keep most games capped in the 60-120 fps range depending on the genre and it greatly reduces temperatures.
I keep most games capped in the 60-120 fps range depending on the genre and it greatly reduces temperatures.
winjer
Gold Member
Of course this guy would immediately pop up here, stories like this is like the best christmas ever for him
Did you even see the test he did?
He basically showed how what software to use, said what to expect in temperature differences between hotspot and the average temp. Then showed his 5090 was normal.
It was just pure information. Nothing more.
CJ_75
Member
Yes, here I am. It's nice to finally have a topic that isn't about PlayStation discs or physical media.
I've undervolted my 5090 FE, so it performs more like a 4090 while running cooler. It is still pushing a lot of flames...sorry, I meant frames.
In winter, though, it doubles as a living room heater, truly a premium feature.
Seriously, I don't see what all the fuss is about. All this heated discussions is starting to make my blood boil.
Last edited:
OLED_Gamer
Member
Techpowerup themselves say the 9070xt is high end.
What you are talking about is enthusiast tier.