Geforce GTX 1050 / 1050 Ti Thread

[ultimatebob]

Nope.. is replacing 950. We need to wait if nVIDIA launchs a GT 1040 in order to see a competitor against the comparable opponent. And also the 1040 could replace the GTX 750Ti

How many people are really going to want something like a 1040 as an add-on board? It wouldn't be much faster than the integrated graphics on most recent processors.

 

[jpiniero]

Isn't the 1040 basically a rebrand of the 750 Ti? Even that seems like a tall task for any IGP.

 

[dark zero]

How many people are really going to want something like a 1040 as an add-on board? It wouldn't be much faster than the integrated graphics on most recent processors.

2 reasons:
1.- Memory available which is better than DDR4.
2.- Dedicated memory is better than integrated one.

 

[RussianSensation]

BTW this is how the leaked score would fit in Guru3D's chart:

You forgot 1 major detail -- 3dMark11 scores are 100% worthless. Using those scores, GTX780Ti outperforms RX 480 by 6%, beats R9 390X, beats R9 290X by 9% and is a whopping 21% ahead of the R9 290.

Meanwhile, in the real world, in modern games 780Ti is only 1% faster than R9 290, and slower than every other AMD card I mentioned, including losing to the R9 390X by 14%.

Now that we have objectively concluded that synthetic benchmarks like 3DMark11 are in no way related to real world gaming performance, we can look at real world gaming performance.

Reference RX 470 is 48% faster than GTX960
MSI Gaming RX 470 = $185
Asus ROG Strix RX 470 = $185
PowerColor Red Devil RX 470 = $185

Reference RX 480 is 79% faster than GTX960
MSI Gaming RX 480 = $205

GTX1060 3GB is 79% faster than GTX960
PNY GTX1060 3GB = $190

With those 3 cards on the market selling for $185-205, at least in the U.S. market, a GTX1050Ti will just continue the legendary overpricing x50/Ti SKU from NV. I am not trying to pick on NV either. It's just a fact that since 2012, no newly launched GPU below $159 has been worth buying. RX 460 can also be thrown into this category. In today's GPU market, for budget gaming, it makes sense to go towards the $185-200 price bracket and skip all the trash underneath; and while at it, also learn how to build or upgrade the PSU so that it has a single 6-8-pin power connector (yawn).

400-450W XFX/Corsair PSU often go on sale for $17-25. There is 0, I repeat, 0 excuse to overpay for under-powered GPUs in 2016 just because you "require" a dGPU without a 6-pin/8-pin power connector. These excuses are inexcusable on a technical forum imho. If a PC gamer cannot afford to spend $25-30 on a new PSU that will last 5+ years, they must be pirating AAA PC games or only play free to play games. A decent PSU lasts 5-10 years and as I have repeated many times, 750/750Ti users overpaid for those cards despite upgrading their PSU, and the circle of over-payment will continue until they learn how to build a system with a bare minimum acceptable PSU.

As a friendly reminder, R9 280 was on sale for $185 + $20 Newegg gift card + 3 free games during September 2014. There were many deals like that 2 years ago. I bet GTX1050Ti OC will be barely faster 15-20% than R9 280 OC (> R9 280X) despite 2 years apart, the 1050Ti being 2 full generations ahead of R9 280 (Maxwell + Pascal) and a new 16nm node.

For tech-savvy PC builders who have a modern PSU, it makes sense to go into the used market OR simply save up $30-40 more for a GTX1060 3GB/RX 470/RX 480 4GB.

On extreme budget, it's possible to hunt down a GTX960 2GB for $84, or mini 960 4GB for $94. I highly doubt that a $149 GTX1050Ti will be 59%-77% faster than those GTX960 cards.
http://galaxstore.net/9-Series_c_15.html

Now, you may object that it's unfair to compare a brand new card to discontinued last gen cards, but it's fair since consumers who are going to be cross-shopping GTX1050/1050Ti can still purchase new GTX960s.

In any case, there simply is no more great value left in the sub-$160 desktop dGPU market anymore. With 0dBA fanless modes on modern AIB cards from AMD/NV, low-power cards no longer even have the idle noise levels as an advantage.

 

[Sweepr]

You forgot 1 major detail -- 3dMark11 scores are 100% worthless.

And you forgot that I didn't say anything about 3DMark 11's relevence in NVIDIA vs AMD comparisons. Would save a hour+ texting. Whether you like it or not this will be a very successful card if it matches/beats a GTX 960 - at the right price and with the option of no power connector. Especially considering that its main competitor disappointed in the performance departament.

 

[maddie]

And you forgot that I didn't say anything about 3DMark 11's relevence in NVIDIA vs AMD comparisons. Would save a hour+ texting. Whether you like it or not this will be a very successful card if it matches/beats a GTX 960 - at the right price and with the option of no power connector. Especially considering that its main competitor disappointed in the performance departament.

Trying to understand.

You posted: "BTW this is how the leaked score would fit in Guru3D's chart"

If not trying to show comparative performance, why exactly did you say that?

 

[Sweepr]

If not trying to show comparative performance, why exactly did you say that?

To give a rough idea of how this card stack up to the existing Maxwell VGAs it replaces, assuming the leaked bench is real. And posting rumors doesn't mean you agree 100% with all their content.

 

[greatnoob]

And you forgot that I didn't say anything about 3DMark 11's relevence in NVIDIA vs AMD comparisons. Would save a hour+ texting. Whether you like it or not this will be a very successful card if it matches/beats a GTX 960 - at the right price and with the option of no power connector. Especially considering that its main competitor disappointed in the performance departament.

Yes you are correct as a couple of my friends are planning to get into PC gaming in the uni break and I'll most likely be looking at buying 1050s (Ti) for them. The only problem, as you alluded to, is whether or not the pricing will be 'right'... and knowing Nvidia's (and now also AMD's) pricing schemes I'd say these cards won't be priced accordingly. The 1050 and the 460 are disguised versions of the 1040 and 450 respectively, they are overpriced and barely move the needle for "minimum viable cards for a budget gaming PCs" since the GTX 650 Ti's release in 2012. Intel very well could have forced Nvidia and AMD to up their game in the low end/budget card market by beefing up their mainstream iGPUS (at the hand of cannibalising their Iris Pro lineup) and forcing both Nvidia and AMD to bring better products... but of course, Intel isn't really worried about competition. It seems like the entire CPU and budget graphics market is a cartel. AMD, Nvidia and Intel squeezing out as much money from consumers by offering very marginal 'improvements' at the same (and now, higher) prices.

 

[happy medium]

It seems there will be a large gap between the 1050ti and the 1060 3gb.

My guess is the 1050ti will overclock 30% or more like my gtx960.

 

[RussianSensation]

And you forgot that I didn't say anything about 3DMark 11's relevence in NVIDIA vs AMD comparisons. Would save a hour+ texting. Whether you like it or not this will be a very successful card if it matches/beats a GTX 960 - at the right price and with the option of no power connector. Especially considering that its main competitor disappointed in the performance departament.

What's the point of referencing an unreliable sunthetic benchmark if it has 0 correlation with real world games? If you think it has no correlation with real world games, then why are you spamming?/giving our forum a bad reputation? You posted an unreliable benchmark and got called out on it because any real world gaming comparisons can't be derived from those 3DMark11 scores. Then why are you posting the comparison?!

If the benchmark itself is unreliable, it's impossible to draw accurate conclusions on how much better 1050Ti will be against 950/950/970/R9 290. Yet, you erroneously tried to extrapolate real world gaming performance for 1050 cards based on a worthless synthetic benchmark. You try to come off impartial as if comparing Pascal to Maxwell excuses using a synthetic bench? It makes no difference if you compare NV to NV or NV to AMD if the benchmark itself is unreliable marketing trash -- GeekBench says hi BTW! Another garbage marketing bench.

Your comment that this card will be successful if it matches or barely beats a 960 just shows how ill informed the average PC consumer/NV loyalist is. If garbage sells well, the product is good? Lame. Forums such as ours should educate PC consumers to spend their $ wisely instead of spending it on marketing turds. I guess you are offended that I questioned the value of 1050/1050Ti tier overall that you had to bring up the overpriced turd RX 460 to make a point? If you had read my post, my advice was to skip all AMD and NV cards <$159 as they are a waste of $.

You know when someone has no objectivity about the dGPU market is when year after year they recommend bottom of the barrel underperforming and overpriced NV GPUs such as 8500/8600GT/S, GTS450, GFX550Ti, GTX650/Ti, 750/750Ti. Every single one of those cards was a horrible buy at its launch price. 1050/1050Ti will be no exception = overpriced + underperforming. This entire market segment is like flushing $ down the toilet.

You also didn't address how an $85-95 960 is not a better low budget card than a $150 1050Ti or how a $185-205 RX 470/480/1060 are all better as low-end gaming cards for 1080p and below.

Face it, you would always find some way to recommend ANY NV card, no matter the price, performance or price/performance. In your mind there is no such thing as a crap NV product because as long as an NV product is better than AMD's, it's worth recommending. I feel sorry for all the PC noobs/illiterates/fooled by marketing PC enthusiasts who purchased GTX750/750Ti/950/960 and now will "upgrade" to 1050/1050Ti instead of just buying a $250-330 R9 290/970 and enjoying that level of performance since September 2014. The fact of the matter is certain segments of dGPUs are not both worth buying from either AMD or NV brands but that train of thought is alien to you.

Here is another fun fact - even my 6700k downclocked to 2Ghz with memory tuned to DDR4-2133 easily handles 4K content. That means even for an HTPC box, 1050/1050Ti/460 are a waste of $ over a modern Skylake/Kaby Lake CPU. Pure marketing for 4K video acceleration that a $110 i3 6100 can do.

 

[cbn]

400-450W XFX/Corsair PSU often go on sale for $17-25. There is 0, I repeat, 0 excuse to overpay for under-powered GPUs in 2016 just because you "require" a dGPU without a 6-pin/8-pin power connector. These excuses are inexcusable on a technical forum imho. If a PC gamer cannot afford to spend $25-30 on a new PSU that will last 5+ years, they must be pirating AAA PC games or only play free to play games. A decent PSU lasts 5-10 years and as I have repeated many times, 750/750Ti users overpaid for those cards despite upgrading their PSU, and the circle of over-payment will continue until they learn how to build a system with a bare minimum acceptable PSU.

I definitely agree with your argument on value. However, one group that benefits from these cards without power connector would be the discount Pre-built market.

In this case despite these power connector less cards being a worse value, the discount price of the Pre-built can compensate.

With that mentioned, a person could always replace the Pre-built's PSU with one of higher wattage, but that would void the warranty. Another option, though not as well known, would be to use a 2 x molex to PCIe 6 pin adapter (but I believe each molex need to be on their own string as each molex supports 40W and perhaps not all OEM PSU have two separate molex strings).

 

[frozentundra123456]

I definitely agree with your argument on value. However, one group that benefits from these cards without power connector would be the discount Pre-built market.

In this case despite these power connector less cards being a worse value, the discount price of the Pre-built can compensate.

With that mentioned, a person could always replace the Pre-built's PSU with one of higher wattage, but that would void the warranty. Another option, though not as well known, would be to use a 2 x molex to PCIe 6 pin adapter (but I believe each molex need to be on their own string as each molex supports 40W and perhaps not all OEM PSU have two separate molex strings).

Well as far as that goes, opening the case and putting in a dgpu will most likely void the warranty as well. I do agree however, that there is a place for these cards that do not require a 6 pin connector. Even though it is generally better the replace the psu and get a more powerful card, one has to remember that not everyone is an enthusiast or technically adept person like the majority in these forums, and simply sticking a dgpu into a slot is a lot simpler and less intimidating than removing and replacing a power supply. And yes, there are OEM systems without 2 extra molex connectors, because I have a system like that. It is just too bad that the gap between the 460 and 470 is so large. A card slotted between those two could be a pretty competent 1080p card if the price was right.

But for a person on a limited budget with an OEM box, you could save a hundred dollars by using a 460 instead of a 470/1060 plus new psu. (460 for 100.00 vs a 170.00 card plus a 30.00 psu)
Personally, I would not consider a 460 because I dont consider it a sufficient upgrade from even the ancient HD7770 that I am using, but for someone coming from an OEM box with integrated graphics or a low end card like a GT730, it would be a huge upgrade.

 

[Sweepr]

What's the point of referencing an unreliable sunthetic benchmark if it has 0 correlation with real world games? If you think it has no correlation with real world games, then why are you spamming?/giving our forum a bad reputation? You posted an unreliable benchmark and got called out on it because any real world gaming comparisons can't be derived from those 3DMark11 scores. Then why are you posting the comparison?! ...Yet, you erroneously tried to extrapolate real world gaming performance for 1050 cards based on a worthless synthetic benchmark.

So you're saying we can't use this benchmark to compare something like GTX 1050 Ti to GTX 1060 (exact same architecture)? Based on TechPowerUp's 'real world' 1080p performance chart you just posted above GTX 1060 6GB delivers 72.1% the performance of GTX 1070 - meanwhile the former card is 72.5% as fast as GTX 1070 in Guru3D's 3DMark 11 Extreme chart. Similar with GTX 1070 vs GTX 1080 (83.9% @ TPU 1080p / 82.2% @ 3DMark 11 Extreme). I guess it's not as useless for same brand comparisons as your long rant indicates.

 

[nvgpu]

Always IGNORE the extremely delusional you know who that loves to write long irrelevant rants, the markets the GTX 1050 Ti is in it not even the same as the 290W room heaters from the competitor. Not the first time either, tried the same useless rants during GTX 960 threads also, some people just need to shill for the inferior red competitor all the time to derail threads.

Comparing cards that are totally different in capabilities and target markets is STUPID, period. Besides, if you wanted a gaming card, a GTX 1060 is FASTER, QUIETER & COOLER than any competitor 290W heater while supporting must have modern features like HEVC hardware decoding, VP9 hardware decoding, DisplayPort 1.4, HDMI 2.0b.

HEVC Main10/Main12 hardware decoding & VP9 hardware decoding which the competitor's 290W space heaters don't support is a big deal, as you can see from the Kaby Lake comparison.

https://forums.anandtech.com/threads/intel-skylake-kaby-lake-thread.2428363/page-300#post-38446251

GTX 1050 Ti is for less demanding games like MOBAs, HTPC usage, OEM PCs for easy upgrades without external PCIe power connector.

http://www.pcper.com/reviews/Graphics-Cards/Upgrade-Story-Can-GTX-750-Ti-Convert-OEMs-PCs-Gaming-PCs



Member insults are still not alllowed.


esquared
Anandtech Forum Director

 

[ShintaiDK]

Here is another fun fact - even my 6700k downclocked to 2Ghz with memory tuned to DDR4-2133 easily handles 4K content. That means even for an HTPC box, 1050/1050Ti/460 are a waste of $ over a modern Skylake/Kaby Lake CPU. Pure marketing for 4K video acceleration that a $110 i3 6100 can do.

Haha, no. It doesnt handle the required content. Maybe you can find some things it can do. But it certainly cant do all.

What's the point of referencing an unreliable sunthetic benchmark if it has 0 correlation with real world games? If you think it has no correlation with real world games, then why are you spamming?/giving our forum a bad reputation? You posted an unreliable benchmark and got called out on it because any real world gaming comparisons can't be derived from those 3DMark11 scores. Then why are you posting the comparison?!

If the benchmark itself is unreliable, it's impossible to draw accurate conclusions on how much better 1050Ti will be against 950/950/970/R9 290. Yet, you erroneously tried to extrapolate real world gaming performance for 1050 cards based on a worthless synthetic benchmark. You try to come off impartial as if comparing Pascal to Maxwell excuses using a synthetic bench? It makes no difference if you compare NV to NV or NV to AMD if the benchmark itself is unreliable marketing trash -- GeekBench says hi BTW! Another garbage marketing bench.

Your comment that this card will be successful if it matches or barely beats a 960 just shows how ill informed the average PC consumer/NV loyalist is. If garbage sells well, the product is good? Lame. Forums such as ours should educate PC consumers to spend their $ wisely instead of spending it on marketing turds. I guess you are offended that I questioned the value of 1050/1050Ti tier overall that you had to bring up the overpriced turd RX 460 to make a point? If you had read my post, my advice was to skip all AMD and NV cards <$159 as they are a waste of $.

You know when someone has no objectivity about the dGPU market is when year after year they recommend bottom of the barrel underperforming and overpriced NV GPUs such as 8500/8600GT/S, GTS450, GFX550Ti, GTX650/Ti, 750/750Ti. Every single one of those cards was a horrible buy at its launch price. 1050/1050Ti will be no exception = overpriced + underperforming. This entire market segment is like flushing $ down the toilet.

You also didn't address how an $85-95 960 is not a better low budget card than a $150 1050Ti or how a $185-205 RX 470/480/1060 are all better as low-end gaming cards for 1080p and below.

Face it, you would always find some way to recommend ANY NV card, no matter the price, performance or price/performance. In your mind there is no such thing as a crap NV product because as long as an NV product is better than AMD's, it's worth recommending. I feel sorry for all the PC noobs/illiterates/fooled by marketing PC enthusiasts who purchased GTX750/750Ti/950/960 and now will "upgrade" to 1050/1050Ti instead of just buying a $250-330 R9 290/970 and enjoying that level of performance since September 2014. The fact of the matter is certain segments of dGPUs are not both worth buying from either AMD or NV brands but that train of thought is alien to you.

And you say other people are spamming?




Personal attacks are not allowed.


esquared
Anandtech Forum Director

 

[Sweepr]

Geforce GTX 1050 Ti (Reference) - Ahead of GTX 960 in New Benchmarks

Let's start this Sunday with a new round of Geforce GTX 1050 Ti leaks:

- Reference card is ~21 cm and doesn't require an extra 6-pin power connector
- Custom cards will include 6-pin
- Specifications: GP107-400, 768 SPs, 64 TMUs, 32 ROPs, base frequency 1354 MHz, turbo frequency 1468 MHz, 128-bit 4 GB GDDR5
- Frequencies range from 1354 MHz all the way to 1797 MHz
- Launch between October 18-25th

Die

BIOS

GPU-Z

Geforce GTX 1050 Ti (Reference) - Back

Clocks

And the most interesting part, benchmarks.

- 3DMark Fire Strike Ultra (4K)

Scores from existing cards

- 3DMark Time Spy

Leaker didn't mention if Async is On or Off, but here's scores from existing cards with Async On

 

[jpiniero]

Much better if the actual clocks are 1364/1468 without the power.

 

[antihelten]

And the most interesting part, benchmarks.

- 3DMark Fire Strike Ultra (4K)

Scores from existing cards

You guys just had a long discussion about how comparing across brands with 3DMark is useless (something that you even agreed with), and then you go and post a 3DMark result for the 1050 Ti and compare it with a graph containing only AMD cards. Why?!

 

[Sweepr]

You guys just had a long discussion about how comparing across brands with 3DMark is useless (something that you even agreed with), and then you go and post a 3DMark result for the 1050 Ti and compare it with a graph containing only AMD cards. Why?!

You can draw your own conclusions. It most likely is faster than RX 460 at the same TDP if synthetic benchmarks show a massive
advantage.

 

[antihelten]

You can draw your own conclusions. It most likely is faster than RX 460 at the same TDP if synthetic benchmarks show a massive
advantage.

I have no doubt that it will be faster than the RX 460, but I don't think anyone was really doubting that at this point. So the question isn't whether or not it will be faster, but rather how much faster it will be, and as pointed out by RS, 3DMark is far too unreliable when doing cross-brand comparisons to say anything useful about that.

 

[maddie]

You guys just had a long discussion about how comparing across brands with 3DMark is useless (something that you even agreed with), and then you go and post a 3DMark result for the 1050 Ti and compare it with a graph containing only AMD cards. Why?!

You can draw your own conclusions. It most likely is faster than RX 460 at the same TDP if synthetic benchmarks show a massive
advantage.

I am mystified and have to ask a direct question.

Sweepr, have you been and/or are you running a marketing campaign on this forum?

You seem to be repeatedly contradicting yourself with successive posts. Looking for clarification.


Politely asking someone if they are a shill, is not allowed.


esquared
Anandtech Forum Director

 

[kraatus77]

Answer is pretty obvious, you shouldn't have asked that ^.

 

[Sweepr]

It's very simple maddie. 3DMark doesn't reflect overall gaming performance across different brands - but when you see one card beating the other by >40% while competing Pascal/Polaris products like GTX 1060 and RX 480 are so close in the same tests - they can't be ignored. Can't understand why some people are mad about the idea that GP107 might give Polaris 11 a run for its money in terms of performance and performance per watt. Worst case scenario you can buy your favourite card at lower prices.

 

[frozentundra123456]

What is the problem with synthetic benchmarks? That is all the data we have now. If someone has gaming benchmarks, they are certainly free to post them. And if one doesnt believe they are valid for some particular reason, they are certainly free to ignore them. After all, it is just preliminary data on a peice of hardware, hardly seems worth getting so worked up about and hurling personal accusations around.

 

[frozentundra123456]

Answer is pretty obvious, you shouldn't have asked that ^.

Yea, it is called a member call out and is against the rules of this forum. If one has a problem with a particular poster, they are certainly free to report them and let the mods sort it out.