Geforce GTX 1050 / 1050 Ti Launch Thread ($109 / $139 - October 25th)

[Phynaz]

An unlocked BIOS on the Ti is probably going to make some screaming overclocks achievable.

 

[hojnikb]

Anyone knows if its made on tsmc 16nm or glofo 14nm node ?

 

[nathanddrews]

I can't wait to see some solid reviews of the 1050. I'm guessing that you could get away with a fanless mod with an undervolt.

Anyone knows if its made on tsmc 16nm or glofo 14nm node ?

At least according to NVIDIA's PR, all things Pascal have been 16nm, so there's little reason to think this is anything but sloughed off GP107 chips from the same process.

 

[antihelten]

Seems odd...the 950 2gb card is beating the Nitro 460 4gb card in both of those at your links...so the 1050 should at least be very close, and the Ti a little out front.

It really depends upon how high the various 1050 models boost. 950 cards tend to boost to somewhere around 1350-1450 MHz (I'm not sure exactly which model CB used), which puts them 10-20% ahead of a 1050 at 1455 MHz FLOPS wise.

It could also be that Nvidia has been a bit conservative with their numbers, since you don't want people complaining when they can't hit exactly the FPS values advertised, so you leave a bit of leeway.

The 1050 Ti should of course be a good bit faster than both the 1050 and the RX 460, especially if rumors of models boosting to 1700 MHz are true.

 

[Det0x]

Hmm

This is taken from Videocardz newest post...

NVIDIA announces GeForce GTX 1050 Ti and GeForce GTX 1050 @ http://videocardz.com/63832/nvidia-announces-geforce-gtx-1050-ti-and-geforce-gtx-1050

 

[Sweepr]

Beware - lots of pics ahead:

Spoiler

Spoiler

 

[Sweepr]

Samsung 14nm FinFET confirmed:

These two new GPUs will use Nvidia's GP107 GPU, which has been created using Samsung's 14nm FinFET manufacturing process, though Nvidia has not confirmed exactly why they have decided to use Samsung instead of TSMC.

Nvidia has given the below statement, stating that they are using Samsung as a secondary source for their Pascal GPUs.

''TSMC is the best foundry and our primary supplier. We have always used a second source for some of our supply and have worked with Samsung since 28nm.''

http://www.overclock3d.net/news/gpu...nnounced_their_gtx_1050_and_gtx_1050ti_gpus/1


Here's some of the cards announced so far:

http://inno3d.com/products_detail.php?refid=266
http://inno3d.com/products_detail.php?refid=267
https://www.zotac.com/cn/product/graphics_card/GeForce-GTX-1050/all
https://www.zotac.com/cn/product/graphics_card/GeForce-GTX-1050Ti/all
http://www.manli.com/en/products/NVIDIA_Graphics_Cards/10_Series/products/95
http://www.manli.com/en/products/NVIDIA_Graphics_Cards/10_Series/products/96
http://www.galax.com/en/graphics-card/10-series/galax-geforce-gtx-1050-oc.html
http://www.galax.com/en/graphics-card/10-series/galax-geforce-gtx-1050-exoc.html
http://www.gainward.cn/Home/ProductDetail?id=80&isStop=0

 

[Det0x]

Samsung 14nm FinFET confirmed:



http://www.overclock3d.net/news/gpu...nnounced_their_gtx_1050_and_gtx_1050ti_gpus/1

Looking forward to the (real) overclock/boost clocks then, could be interesting

 

[krumme]

Confirms gf/gf is a good deal lower clock than tsmc. But to be expected. Now diesize is interesting. Tdp seems high imo so perhaps 14nm ff is simply not that well behaved.
Samsung is ramping 10nm and is in mass production now so they probably is moved on.
Remember this is nearly half a year after first pascal. P11 embedded is supposed to boost to 1400 too as for Ryans writeup and its about same time as this gpu.

 

[Sweepr]

If this leak is true, boost can hit nearly 1.8 GHz for the Ti:

http://i1024.photobucket.com/albums/y303/martmail55/ab5cfb0ed27f47febac712871686179b_zpsxxntwq49.jpg

Die size is 135 mm², 3.3 billion transistors.

 

[krumme]

If this leak is true, boost can hit nearly 1.8 GHz for the Ti:

http://i1024.photobucket.com/albums/y303/martmail55/ab5cfb0ed27f47febac712871686179b_zpsxxntwq49.jpg

That would be nice and lets hope so but ref clock is there for a reason.
To me it seems there is a good deal of smart binning here. 1050 takes a voltage hike and stays at same tdp but that probably saves a lot of dies. Nice move. Then segmentation is tied to production.

 

[tviceman]

Samsung 14nm FinFET confirmed

This makes for a really close comparison to P11 since GloFo is a mirror of Samsung. I know the RX 460 isn't a full functioning die, but the fact that it's cut down and will still end up slower and less efficient is proof that Nvidia's higher R&D and more focused development is paying off.

 

[Det0x]

That would be nice and lets hope so but ref clock is there for a reason.
To me it seems there is a good deal of smart binning here. 1050 takes a voltage hike and stays at same tdp but that probably saves a lot of dies. Nice move. Then segmentation is tied to production.

Anandtech said:

"I’ll address what’s likely the elephant in the room first, which is the manufacturing process. To date all Pascal GPUs have been fabbed over at TSMC on their 16nm FinFET process. GP107 is not one of those GPUs. Instead, it’s fabbed on a 14nm process – NVIDIA’s specification sheet doesn’t technically state whose process – but by simple elimination it’s a very safe bet that they’re making the chip over at Samsung. Feature size is a red herring here, and instead the significance of this deal is that NVIDIA has not used a fab other than TSMC for GPUs for a long time. In fact we’d have to go back to 2003 to find an NVIDIA GPU fabbed somewhere else, when NVIDIA tapped IBM to help fab the ill-fated NV3x series (GeForce FX).

Suffice it to say, tapping another fab is a very big deal. There’s no second-sourcing here – GP107 is only being made on Samsung’s 14nm process and GP106+ only on TSMC’s 16nm process – but splitting orders like this may just as well be new territory for NVIDIA. As this is just a product announcement NVIDIA hasn’t said anything about the change in fabs, so let your imagination go wild here, but it definitely has some ramifications. I really need to get the GTX 1050 cards in house and on the testbed to figure out the full ramifications of this, but I think the most important change here is that a new process from a new vendor means that the voltage/frequency curve we’ve come to know with TSMC 16nm and Pascal has essentially been thrown out the window.

This in turn may explain the clockspeeds of the GTX 1050 cards. All of the other desktop GeForce 10-series cards have an official boost clock of 1600MHz+, with all but one of those cards being 1700Mhz+. The massive jump in clockspeed relative to Maxwell 2 is one of the signature elements of the Pascal architecture, and a major factor driving the significant performance gains of this generation compared to the last. The GTX 1050 series, by comparison, is only rated to boost up to 1455MHz for the GTX 1050, and lower still for the GTX 1050 Ti at 1392MHz.

Given that these are power-constrained cards, the final specifications of the cards are bound by a larger number of variables than usual – power curves, attainable frequency range, and now total power consumption – so I’m not even going to try to insinuate that the lower clockspeeds are solely a function of the change in fabs. However it’s very important to keep in mind that these lower clockspeeds come with a sometimes sizable increase in TDP relative to the GTX 750 series; instead of 55W/60W cards, we have 75W cards. So to use the fully enabled GTX 1050 Ti as an anchor point, power consumption has gone up 15W (25%) for a 28% increase in the boost clock, 1 more SM (20%), and somewhat decoupled from this, the doubled ROP count.

It’s telling then that NVIDIA has informed the press that the higher TDP cards with an external power connector are going to have much higher boost clocks. Whatever is going on under the hood, power plays a big part, and at a TDP limit of 75W, GP107 isn’t getting all the room it needs to stretch."

http://www.anandtech.com/show/10768/nvidia-announces-geforce-gtx-1050-ti-gtx-1050

And then we have this:

If this leak is true, boost can hit nearly 1.8 GHz for the Ti:

http://i1024.photobucket.com/albums/y303/martmail55/ab5cfb0ed27f47febac712871686179b_zpsxxntwq49.jpg

Die size is 135 mm², 3.3 billion transistors.

 

[DamZe]

OMG the hype machine is in full effect, 3x the performance of a GTX 650? Are they serious?

 

[mohit9206]

Pricing is good by Nvidia. So currently we have
RX460/1050 - $150
RX460 4GB - $180
1050Ti -$200
RX470 4GB - $250
1060 3GB-$270
1060 6GB - $320
RX480 8GB-$350
1070 - $600
1080 - $900
Prices are fine in the low and mid range but completely out of control in the hi end.

 

[LTC8K6]

Pricing is good by Nvidia. So currently we have
RX460/1050 - $150
RX460 4GB - $180
1050Ti -$200
RX470 4GB - $250
1060 3GB-$270
1060 6GB - $320
RX480 8GB-$350
1070 - $600
1080 - $900
Prices are fine in the low and mid range but completely out of control in the hi end.

Why do you have the 1050ti at $200 instead of $139?

 

[LTC8K6]

Presumably the 1050ti cards with power connectors and high boost clocks will be a little more than $139, though.

 

[antihelten]

Beware - lots of pics ahead:

Spoiler

A quick estimate based on those slides, puts the 1050 Ti at roughly 40% faster than the 950, and thus 20% faster than the 960, and 20% slower than the RX 470.

With the 1050 Ti at $140, the RX 460 and RX 470 (both 4GB) would have to be roughly $100 and $170, to maintain similar perf/$. It will be interesting to see exactly where this card lands performance wise, and what impact it will have on the pricing of RX 460 and RX 470.

 

[Blitzvogel]

I was tempted to purchase an RX 460 for my wife's computer, and the 1050Ti looks just as tempting, especially if a price war ensues.

 

[DidelisDiskas]

Hopefully the 1050ti will put the final nail in the anemic rx 460 coffin. The 4 gb models go for up to 170 euro around where i live, which is totally insane. I hope this forces amd to release the uncut polaris 11.

 

[Bacon1]

RX460/1050 - $150
RX460 4GB - $180
1050Ti -$200
RX470 4GB - $250

Where are you getting pricing from?

460 are $95-$110 not $150

470 are $185-$200

480 are $200 - 240

 

[mohit9206]

Why do you have the 1050ti at $200 instead of $139?

Where are you getting pricing from?

460 are $95-$110 not $150

470 are $185-$200

480 are $200 - 240

Outside USA where graphics cards are about 40% more expensive due to being classified as luxury products.

 

[RussianSensation]

Outside USA where graphics cards are about 40% more expensive due to being classified as luxury products.

Why can't you purchase GPUs on eBay or Amazon or Computeruniverse? Last time I checked Computeruniverse ships almost anywhere in the world, including India, China, Kazakhstan, Vietnam, Russia, etc.

Canada also has inflated GPU prices so I stopped buying new GPUs in Canada. Pretty simple since as consumers we do have real choices to purchase outside of our home country.

A quick estimate based on those slides, puts the 1050 Ti at roughly 40% faster than the 950, and thus 20% faster than the 960, and 20% slower than the RX 470.

Absolutely not. Cherry-picked settings in random titles NV picked don't matter to show the average performance delta. TimeSpy score shows that 1050Ti is barely faster than the GTX960. EVGA confirms it's not going to be even 25% faster than a GTX960:

"The EVGA GeForce GTX 1050 Ti and 1050 may look small, but they offer big performance. In fact the EVGA GeForce GTX 1050 Ti is about 26% faster than a GTX 950, and the GeForce GTX 1050 is about 16% faster." ~ Source

It also seems you never read reviews on RX 470 closely as you would known just how much faster than GTX960 it really is, especially since RX 470 has ~20% overclocking scaling headroom.

With the 1050 Ti at $140, the RX 460 and RX 470 (both 4GB) would have to be roughly $100 and $170, to maintain similar perf/$.

Nope, not happening.

First, RX 470 has massive overclocking headroom that starts off a much stronger base performance than 1050Ti's will.

An overclocked RX 470 beats a stock RX 480 and GTX980.

Second, 1050Ti's performance will be below 60 fps in modern titles in many cases where RX 470 won't be. In that context, the price/performance metric has to be taken into context. Just $30-40 more will get one to move into an RX 470/1060 3GB card that will actually provide solid 60 fps performance in AAA titles with minor settings turned down. It's amazing to reach on this forum how people are trying to align price/performance curves with the 1050Ti and RX 470 but at the same time advocate spending hundreds of dollars more to buy a GTX1070 instead of the Fury and don't even blink at the thought of recommending $220 more expensive GTX1080 over the GTX1070 for ~ 22-25% more performance. Why should a budget gamer try to save $30-40 and lose 20-30% more performance? That's a horrible trade-off.

Some of you clearly don't understand that in a blind-test, the average PC gamer who is the target market for sub-$200 dGPUs won't be able to tell the difference in playability between an RX 470/480/1060 3GB but they will be able to tell the difference between RX 470/1060 and the 1050Ti. In modern games, the RX 470 is very close to the RX 480/1060 cards.

Literally in recent AAA titles, RX 470 is up there with GTX1060 in performance. So how can it be that for just $30-40 savings the 1050Ti will be a good deal?

It seems some of you never bothered reading reviews on RX 470 when you are comparing GTX1050Ti to it. RX 470 is much closer in performance to the GTX1060 3GB/GTX970/980/R9 390.

20-25% faster than GTX960 won't even make a dent to come close to the RX 470.

Thirdly, the market has even better deals than $170 RX 470 such as the $180 MSI Gaming RX 480 4GB. There is no way a GTX1050Ti is worth $140 for gaming in the USA when AIB RX 480 4GB is dropping to $180. We are talking $180 card that has similar performance to the $250-$300 GTX1060s. Now that's value! GTX1050 and 1050Ti need to be $79 and $109 cards to make sense. In the context of overall performance, it simply isn't logical to save $30-40 over 2-2.5 years of GPU ownership to lose 20-40% GPU performance. That's like 1/2 price of a single AAA game. RX 470/480 4GB and GTX1060 3GB are all bare minimum GPUs that gamers should be purchasing. Everything below should be skipped, unless going into the used dGPU market where R9 390/290/290X are dirt cheap.

As I mentioned already, even if AMD had cards 30-40% faster than GTX1050/1050Ti, NV would still outsell them. So any logical argument why GTX1050/1050Ti won't be worth buying will fall on deaf ears just like 8600GT/8600GTS/GTS450/550Ti/GT610->650Ti/GT710->750Ti all sold well and they were all horrible gaming GPUs.

 

[AtenRa]

I really want to see RX 460 4GB vs GTX 1050 2GB in 2016 games both DX-11 and DX-12/Vulkan.

Also, hope AMD will release full P11 to get against GTX 1050Ti. Good times, competition is good for consumers

 

[Sweepr]

TimeSpy score shows that 1050Ti is barely faster than the GTX960. EVGA confirms it's not going to be even 25% faster than a GTX960:

"The EVGA GeForce GTX 1050 Ti and 1050 may look small, but they offer big performance. In fact the EVGA GeForce GTX 1050 Ti is about 26% faster than a GTX 950, and the GeForce GTX 1050 is about 16% faster." ~ Source

From your link:

So now synthetic benchmarks are valid for gaming performance comparisons?

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?

...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.