Wednesday, May 6th 2026

NVIDIA GeForce 10-series "Pascal" Generation Turns 10

NVIDIA GeForce GTX 10-series graphics cards turn 10 today. In May 2016, NVIDIA debuted the GeForce GTX 1080, powered by the "Pascal" graphics architecture on the TSMC 16 nm FinFET foundry node. This would go on to be the final generation of NVIDIA GeForce GPUs to truly comply with Moore's Law. To illustrate, the GeForce GTX 1080, a predecessor to the current RTX 5080, needed a single 150-Watt 8-pin power connector, and could play AAA games of its time at 4K Ultra HD with the game being rendered at native resolution (no upscaling). "Pascal" established phenomenal standards of performance-per-Watt over the GTX 9-series "Maxwell" generation.
Add your own comment

124 Comments on NVIDIA GeForce 10-series "Pascal" Generation Turns 10

#101
Dr. Dro
JustBenchingMy condolences. What happened? How did you notice?

EG1. I find it funny that you have to explain yourself for your 13900ks, haha. CPU is a beast.
Crashed while playing latest ZZZ update. Then it kept crashing after that. I decide to check and there it is. Connector melted, PSU side also singed (pins browned out), it must have happened during my last Metro Exodus EE playthrough and I didn't notice. A few heat and cold cycles after that, and there you have it. I think my card was working fine, but given there seems to be at least a little damage on the 12VHPWR connector, and that I have no means to test it on another machine, I decided to RMA it, if only for just servicing.

My PSU survived, I am running the P6000 off one of the terminals that got a browned pin from the overheating. The cable wasn't so fortunate, it is quite damaged (on both ends), see this thread from that post onwards:
Chomiq in Double 12VHPWR connectors in the future graphics cards may resolve overheating and melting cables?Sorry Doc, had to do it: 442624
Posted on Reply
#102
AusWolf
JustBenchingIt's weird cause we are talking about a card that is hitting the limits of what is possible. In car terms its the equivalent of saying "The SSC Tuatara is only 10 kmh faster than the previous fastest car" - as if that's not already impressive. Or the equivalent of a new pro athlete breaking the world record on 100m sprint by 0.05sec. That's impressive, cause it's at a segment that a tiny bit of more performance will require ungodly amounts of engineering , time investment etc etc.
People don't want the fastest car. They want a car that's a bit better than what they have. People are also not pro athletes. So who breaks what world record is irrelevant when talking about an entire product line.
JustBenchingBut the previous power was very high for your standards already. The 3060ti was drawing ~210-220 watts. The 4060ti was 12% faster at 145 watts. If people would prefer the 4060ti to be a 220 watts card but 50% faster then we are going to my point, power draw isn't important.
Ampere's power consumption was indeed very high. So while the 4060 Ti is more economical, it's still not a great upgrade over the 3060 Ti because it isn't much faster. It is a great upgrade over the 2060 and earlier, though. Don't forget that upgrades cost money (no shit, Sherlock), they don't just arrive at your doorstep automatically with every gen. Spending $3-400 on something that 's 10% faster is a huge waste.
JustBenchingI think so, isn't 40mm considered dual slot? Im not being ironic, I really don't know, but I think it's considered dual.

Just checked TPUs specs page and it says dual slot so I guess it is

www.techpowerup.com/gpu-specs/geforce-rtx-5090.c4216
It was already answered, but thanks anyway.
RulumiI would say current way and landscape is different now, hardware of consoles are just customized PCs now with more direct porting. And developers also need to make concessions to try to make the game as accesible as possible. Not that I necessarily thimk that's s bad thing, it's completely understandable with how pricey game development is nowdays and how risky a bad release can be for a company future now.
Are you implying that game devs are being lazy and just use whatever feature is available in current consoles for easy porting? If so, that's really not the PS5's fault.
RulumiWhile there can be good things for Zen 4, power efficiency is not really something I would note a Ryzen 9 7950X for, it's not that much lower in load than Raptor Lake-S 8P+12E and 8P+16E K models and idle and low load on it is noticeable higher. Zen 5 and Arrow Lake a good improvement in power efficiency to their predecessors in both cases, though Zem 5 keeps the high idle power consumption.

www.techpowerup.com/review/intel-core-ultra-7-270k-plus/23.html
Zen 5 efficiency is not much better than Zen 4. If you want a really efficient CPU, you should look for a 7800X3D or 7950X3D.
Posted on Reply
#103
Warigator
Dr. DroThe only reason I have this GPU installed right now is because my RTX 5090 took a smoke break earlier this week. Infamous 12VHPWR issue, not a problem until it happens to you yada yada
That really makes the 5090 look disappointing to me. It is enormously expensive, extremely power-hungry, and still seems to carry a real reliability concern with the 12VHPWR connector.

For such a high-end flagship card, that combination is hard to defend. I really hope the next generations will be much better, because a major jump in efficiency and value is badly needed. At this point, it feels not just desirable, but necessary, even with DLSS 6.
AusWolfAmpere's power consumption was indeed very high. So while the 4060 Ti is more economical, it's still not a great upgrade over the 3060 Ti because it isn't much faster. It is a great upgrade over the 2060 and earlier, though. Don't forget that upgrades cost money (no shit, Sherlock), they don't just arrive at your doorstep automatically with every gen. Spending $3-400 on something that 's 10% faster is a huge waste.
I agree that Nvidia Ampere power draw is rather high (it would be alright for 2019, not for 2021), but in some mobile products (not most of them, but a few) with low enough clocks and low enough voltage, like in the Switch 2, it's pretty low. Switch 2 is really weak compared to new desktop PCs with let's say 5070 Ti or RX 9070, but it's quite efficient. It uses only 8 watts in handheld mode typically. 8 watts for the whole device, from which half of that is for the Ampere GPU (4 watts!). The reason why Switch 2 only lasts 2.5 hours on a single charge is not high power draw, but battery technology which is still extremely bad in 2025.
Posted on Reply
#104
AusWolf
WarigatorI agree that Nvidia Ampere power draw is rather high (it would be alright for 2019, not for 2021), but in some mobile products (not most of them, but a few) with low enough clocks and low enough voltage, like in the Switch 2, it's pretty low. Switch 2 is really weak compared to new desktop PCs with let's say 5070 Ti or RX 9070, but it's quite efficient. It uses only 8 watts in handheld mode typically. 8 watts for the whole device, from which half of that is for the Ampere GPU (4 watts!). The reason why Switch 2 only lasts 2.5 hours on a single charge is not high power draw, but battery technology which is still extremely bad in 2025.
Well, everything can be efficient when properly tuned. RDNA 2 had some massive power spikes in the 6900 XT, but it's nice and calm in the Steam Deck. Or the 7800X3D and 7950X3D which are much more efficient than the 7700X and 7950X because of their more conservative voltage-frequency curves. A lot of our modern hardware is pushed to the limit out of the box, which I think is sad. In my opinion, it would be much nicer if hardware was more conservative by default, with room left for overclocking for the few enthusiasts who like doing it.
Posted on Reply
#105
Rulumi
AusWolfAre you implying that game devs are being lazy and just use whatever feature is available in current consoles for easy porting? If so, that's really not the PS5's fault.
Not necessarily, though it can be the case sometimes. More of game development getting increasingly more expensive and taking more time in all segments, so that developers kind of understandably so try to take less risks and be accesible in what works better to at least recover the development investment.

They already have precedents when 3D started to become the norm for producing in home systems, suddenly development costs went up and more risky to the company health if a title did poorly, a lot of companies disappeared back then in the late 90s and early 2000s, game development time and costs have only keep going higher and higher since then. Bit of sad effect is that variety and new ideas are more sparse because of so.
AusWolfZen 5 efficiency is not much better than Zen 4. If you want a really efficient CPU, you should look for a 7800X3D or 7950X3D.
It is more efficient at lower powers than Zen 4, then again Arrow Lake is also more power efficient below 100W and can scale below 50W properly unlike desktop Zen 4 and Zen 5, so all dependent in power scale target I imsgine. As you said though, the thing is indeed that AMD and Intel default settings for most models is higher by default now. But people often criticize when they go too conservative nowdays, so I don't know if it's the right or wrong choice, when they don't completely overdo it lf course.

I guess at least they still have the normal non K and non X models to put a safe option for people that want efficient default settings, AMD hasn't released much of that parts outside AM5 EPYC for Zen 5 for some reason though.
Posted on Reply
#106
AusWolf
Rulumigame development getting increasingly more expensive and taking more time
Then why do we have so many good indie games from small studios?

AAA games are getting more expensive because those devs focus on the wrong things. That's again not the PS5's fault.
RulumiIt is more efficient at lower powers than Zen 4, then again Arrow Lake is also more power efficient below 100W and can scale below 50W properly unlike desktop Zen 4 and Zen 5, so all dependent in power scale target I imsgine. As you said though, the thing is indeed that AMD and Intel default settings for most models is higher by default now. But people often criticize when they go too conservative nowdays, so I don't know if it's the right or wrong choice, when they don't completely overdo it lf course.

I guess at least they still have the normal non K and non X models to put a safe option for people that want efficient default settings, AMD hasn't released much of that parts outside AM5 EPYC for Zen 5 for some reason though.
The 9700X came with a default 65 W TDP (88 W PPT). My problem is not the power limit, but the voltage-frequency curve. You can limit even a 9950X to the same 65 W TDP, but it'll still overheat in low threaded work. This is why I praise the 7800X3D and 7950X3D, and think that throwing efficiency out of the window with 9000-series X3D was a bad choice. Power limiting is not a blanket answer to every problem.
Posted on Reply
#107
Rulumi
For temperatures really what AMD should have done is go with a different IHS design and not keep AM4 cooler compatibility. Current I/O die should be changed too, it's not very efficient and voltage and consumption goes up quickly if rising memory speed.

Zen 6 could improve the I/O die and IF if AMD wants, IHS wise I guess nothing can't be really changed substantially within AM5.
Posted on Reply
#108
Warigator
9850X3D efficiency is very disappointing

I expect +60% efficiency in the next architecture after Blackwell, as Blackwell increased efficiency only by 4%, so a large jump is overdue now. Similar stuff with RDNA 5 <--- it is necessary for PS6 handheld. They don't want it to be Series S with 2 hours of battery life after all. That would be too weak and too disappointing, unable to play PS6 games. They need something at least 1.6x faster than Series S, with much better RT and AI performance, 3+ hours of battery life somehow.
Posted on Reply
#109
AusWolf
RulumiFor temperatures really what AMD should have done is go with a different IHS design and not keep AM4 cooler compatibility.
That's not a problem with the 7800X3D and 7950X3D, but those are extremely efficient chips, so you do have a point.
RulumiCurrent I/O die should be changed too, it's not very efficient and voltage and consumption goes up quickly if rising memory speed.

Zen 6 could improve the I/O die and IF if AMD wants, IHS wise I guess nothing can't be really changed substantially within AM5.
That I agree with completely.
Posted on Reply
#110
Warigator
MLID said/claimed/rumored PlayStation handheld as follows: 4× Zen 6c cores + 2× Zen 6 low-power cores for OS/background tasks, RDNA 5 16-18 CUs at 1.2-1.8 GHz, 24 GB LPDDR6 around 200 GB/s, 512-1024 GB SSD (probably 2 variants available), $499 MSRP, Q4 2027.

He spoken that it will be 2/3 normal performance of PS5 + 3x RT and 6x AI + 50% more RAM, enough for 1080p gaming. May be optionally docked.
ghaziMan, the PS3 didn't even have unified shaders and they made Crysis 3 run on it! The main problem isn't hardware or developers but the practices and priorities of their employers.
Crysis 3 looked and run poorly on PlayStation 3!
Shadow of the Colossus was on PlayStation 2 in 2005. It looked very blurry and was running in 15 fps. They ought to had waited for PS3 and release that game on PS3 in 2008.
Posted on Reply
#111
Vayra86
WarigatorI don't applaud choosing i9-13900KS CPU because of high power draw. 7950 was a better choice, imo

anyway, with DLSS 4, you can basically triple resolution and triple framerate, so true performance is greater

in reality, 5070 Ti offers about 20x better efficiency when you take DLSS 4 into consideration

and RDNA 4 offers also about 20x better efficiency compared to GCN4 or Vega when you take FSR 4 into consideration
But you shouldn't - the technology is not available in every game nor will it continue to be.
Posted on Reply
#112
wolf
Better Than Native
WarigatorNvidia Ampere power draw is rather high (it would be alright for 2019, not for 2021), but in some mobile products (not most of them, but a few) with low enough clocks and low enough voltage, like in the Switch 2, it's pretty low.
And products like the A2000 fit that description too, just look at that efficiency at the time (not to mention clocks were still conservative, it oc's to +15% more performance at the same power). The desktop GeForce line up was just pushed really hard to meet its performance targets with what was quite a severe node disadvantage VS RDNA2. It made the jump to Ada huge in that regard but I can't help but wonder what Ampere would have been capable of on TSMC N7.
Posted on Reply
#113
chaoshusky
Dirt ChipStill going with the almighty 1080ti.
No reasonable upgrade to be seen anyway near..
I jumped from the 1080 Ti, to the RTX 3090 when that first released.. Took a month to get but more than worth it! The Ti was a great card, but i waited a generation to know the performance jump was more significant.. And its one of the few if only time i sold my old card to put money towards the 3090 thanks to shortages and price hikes, though i paid retail for it and didn't get scalped. I got a 1080 Ti cheap again though because i did love that card, wasn't even using it to its full potential. I do remember actually being annoyed when they did the RTX crap but 44fps (at 4K!) on Watch Dogs Legion tempted the upgrade more lol

Still working over 5 years on with a waterblock though so can't complain, just hope it lasts a bit longer as at today's prices i wouldn't want to buy a replacement card.. Nearly 3 grand at the moment for a 5090 when it should be the same price the 3090 was..somehow still scalping almost 6 years later. And AMD have been for a decade at this point, well longer than that really lol
Posted on Reply
#114
Warigator
wolfAnd products like the A2000 fit that description too, just look at that efficiency at the time (not to mention clocks were still conservative, it oc's to +15% more performance at the same power). The desktop GeForce line up was just pushed really hard to meet its performance targets with what was quite a severe node disadvantage VS RDNA2. It made the jump to Ada huge in that regard but I can't help but wonder what Ampere would have been capable of on TSMC N7.
It nicely shows that 6700 XT is twice as efficient as Vega 64 after 4 years (typical improvement over 4 years). And A2000 is exceptionally efficient, making it a good card for efficiency affictionados. 8 TFLOPS, 12 GB, 288 GB/s and 12B transistors for 70 watts (no cables!) and $549. :)

A2000 has usually about the same performance as 3050, but uses 60 watts less and emits 60 watts less heat. For years I was of opinion that 3050 uses too much energy for what it offers. 1050 Ti and 1650 have TDP and TGP of 75 watts, while 3050 has 130 (!!).
Posted on Reply
#115
AleXXX666
Dirt ChipStill going with the almighty 1080ti.
No reasonable upgrade to be seen anyway near..
as soon as it's properly cleaned up and serviced always at time. If such toast heater worked 24/7 full of dust for a 3+ yrs, it would be fried, so I wouldn't pay more than $50 in 2026 if I was like getting 1080 Ti.
Posted on Reply
#116
Dirt Chip
AleXXX666as soon as it's properly cleaned up and serviced always at time. If such toast heater worked 24/7 full of dust for a 3+ yrs, it would be fried, so I wouldn't pay more than $50 in 2026 if I was like getting 1080 Ti.
I usually use hardware to the point it's not worth anything by the time I'm considering upgrade.
My previous system, i5-2400 and 8GB ddr3, still working if I need it.
Posted on Reply
#117
Vayra86
WarigatorIt nicely shows that 6700 XT is twice as efficient as Vega 64 after 4 years (typical improvement over 4 years). And A2000 is exceptionally efficient, making it a good card for efficiency affictionados. 8 TFLOPS, 12 GB, 288 GB/s and 12B transistors for 70 watts (no cables!) and $549. :)

A2000 has usually about the same performance as 3050, but uses 60 watts less and emits 60 watts less heat. For years I was of opinion that 3050 uses too much energy for what it offers. 1050 Ti and 1650 have TDP and TGP of 75 watts, while 3050 has 130 (!!).
The entire Ampere line up is far too energy inefficient, the node simply is crap and you could see that from the moment the 3080 released. I also have always had the idea the failure rate was pretty high on this gen, even just going by what I saw pass us by on TPU. Ada, in comparison, is stellar in that sense, and uncoincidentally its also a gen with pretty conservative TDPs on a quality node and without a massive OC out of the box.
Posted on Reply
#118
lexluthermiester
Vayra86The entire Ampere line up is far too energy inefficient, the node simply is crap and you could see that from the moment the 3080 released.
Care to elaborate?
Vayra86I also have always had the idea the failure rate was pretty high on this gen, even just going by what I saw pass us by on TPU.
That is an inaccurate perspective, at least from my experience.
Vayra86Ada, in comparison, is stellar in that sense
Better power efficiency? Certainly. Stellar? Come back down to Earth my man.
Vayra86and uncoincidentally its also a gen with pretty conservative TDPs on a quality node and without a massive OC out of the box.
Maybe, but the performance uptick wasn't very compelling for anyone on Ampere unless you're talking about the 4090, but then the notions of power efficiency go out the window. Let's not get started on Blackwell.
Posted on Reply
#119
Warigator
Vayra86The entire Ampere line up is far too energy inefficient, the node simply is crap and you could see that from the moment the 3080 released. I also have always had the idea the failure rate was pretty high on this gen, even just going by what I saw pass us by on TPU. Ada, in comparison, is stellar in that sense, and uncoincidentally its also a gen with pretty conservative TDPs on a quality node and without a massive OC out of the box.
Switch 2 has very good energy efficiency.
It's just that current battery tech sucks!
Posted on Reply
#120
R-T-B
JustBenchingAnd nothing was different back then. The 1080ti was twice the transistor count of the 980ti, which is why it was faster.
But it had a die shrink which lowered power per transistor and increased efficiency... plus Pascals die was physically smaller. Thus, your statement is true, but pretty misleading.
Posted on Reply
#121
JustBenching
R-T-BBut it had a die shrink which lowered power per transistor and increased efficiency... plus Pascals die was physically smaller. Thus, your statement is true, but pretty misleading.
Die size is completely irrelevant though, it only determines the cost - not the performance. Performance is determined by the transistor count.
Posted on Reply
#122
R-T-B
JustBenchingDie size is completely irrelevant though, it only determines the cost - not the performance.
It also has a big correlation to overall heat output and power consumption, so no, it's not really totally irrelevant.
Posted on Reply
#123
lexluthermiester
JustBenchingDie size is completely irrelevant though, it only determines the cost - not the performance. Performance is determined by the transistor count.
Not true at all.
R-T-BIt also has a big correlation to overall heat output and power consumption, so no, it's not really totally irrelevant.
This is true but is only part of the equation.

Die size can also determine operating frequency thresholds as discrete power delivery can become a problem over longer distances on the die due to circuit pathway resistance. Power voltage can drop by as much as 5% from the entry point of power on the die to the central region of the chip, which can and does effect performance. And this is just one aspect of the engineering that has to be factored into performance.
Posted on Reply
#124
grammar_phreak
When is this series 10 years old? I thought the 1080 was released to the public on May 27th. All I remember is they tried this MSRP of $599usd and $699 FE pricing schtick, while AIBs charged the higher price for their cheaper models.

I do remember not being able to find a GTX 1080 for months. Once I saw the reviews for the Titan X Pascal, I decided to wait a bit longer for the 1080ti and that paid off handsomely.
Posted on Reply
Add your own comment
Nov 17th, 2026 17:48 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts