Which is better the GTX 670 or the GTX 1050 Ti?

Mar 30, 2018
11
0
10
Hi, I'm new to the forum.

Well, my question is simple. I have a GTX 1050 Ti but I would like to know if the GTX 670 is more powerful than my GTX 1050 Ti. I have seen in the Userbenchmark page that the GTX 670 is 1% more powerful than my card, but in other similar sites like Passmark that my 1050 Ti is more powerful. Then in other forums, some say that the GTX 670 was much better than my 1050 Ti. I really do not understand, I've seen videos where my 1050 Ti is even better on average than a GTX 680. Even at TechpowerUp it's better https://www.techpowerup.com/gpudb/2885/geforce-gtx-1050- you

For what I do not understand, I am very confused with what some people say ... Someone who has both cards can answer it?

Thanks in advance
 
Solution
670 and 680 stock has about 5% difference. Example factory overclocked models from msi get higher performance than 680 stock. Just look it up theyre practically same when overclocked max stable clocks.

Too bad i can no longer test my 670 to see differences. And your right about userbenchmark giving crappy results but so does passmark. Superposition or valley are better or 3d mark tests.

Im sure your 1050 ti is bettr optimized and new drivers dont give you performance loss which can happen with older nvidia cards.

I hope you can have a nice overclock on it, have you tried yet?

Sorry if i sounded rude on first messages so busy at work and stressed. I am in no way trying to prove 670 is better in gaming than a 1050 ti, and not...
I'll let PassMark tel you what they think. (and I agree), the 1050ti is a better card because it consumes close to 100w less, has DX 12, yes is score a little less on single thread processes, but that is compensated by faster processor and higher multithread scores (gaming)

jfJ9PCF.png
 

fagetti

Notable
Mar 1, 2018
919
15
1,165
My GTX 670 windforce x3 had passmark G3D score of 6700 when overclocked without voltage raise. Somehow that stock score seems pretty low on passmark site. Ofcourse older card takes more power and can break any time but i woudlnt say its worse.
 
Mar 30, 2018
11
0
10


I know that the GTX 670 can do good oc as well as I know that they are not bad cards (they are still very good indeed). What I can not explain is why '' some '' people say that a 670 is better than a 1050 Ti when in all the videos I've seen it is compared to a 680 and then they put the Userbenchmark excuse (which has many inconsistencies even if they are from users, in fact I think passmark has more consistent scores in GPUS) to say (my 670 is better blah blah) I mean, if the 670 was better, why the 1050 Ti is compared to the 680 and not this one? why do I always compare the fps of my 1050 Ti with those of a video of a 680 in the same game my 1050 Ti gives the same or better results? Where is the 670 in that then? I mean if you overclock it reaches 680, but say how it '' overcomes '' the 1050 Ti, I do not know ... WTF

BTW the image of The Paladin shows is the average in G3D score, if you click on the respective card, the scores of each baseline of the respective card will be displayed. https://www.videocardbenchmark.net/gpu.php?gpu=GeForce+GTX+670&id=35 The highest score so far is on June 29, 670 is 6588 G3D (but I assume it has overclock) but I guess you can go a little bit longer since you got one of 6700.

 

fagetti

Notable
Mar 1, 2018
919
15
1,165
Here is comparison of passmark stock scores for 670, 680 and 1050 ti : https://www.videocardbenchmark.net/compare/GeForce-GTX-670-vs-GeForce-GTX-680-vs-GeForce-GTX-1050-Ti/35vs41vs3595

My 670 didnt just overclock to stock 680 (score 5678) , my score was 6700. Sadly i dont have this card anymore upgraded to 780 lightning.

Check this page for my superposition 1080p medium and 1080p high scores, somewhere after mid-page: http://www.tomshardware.co.uk/forum/id-3536560/tom-superposition-thread/page-7.html

These are overclocked scores so you can compare your 1050 ti oc with same pre settings (1080p high and 1080p medium)
 

fagetti

Notable
Mar 1, 2018
919
15
1,165
close all programs including browser if you benchmark with superposition, even chrome lowers score

EDIT: competely forgot 1050 ti has 4gb ram that means your graphics card beats 670 on every game and bencmark which gives benefit (like passmark) or requires more than 2GB ram. Nearly all modern games require more vram but fortnite is an exeption where 670 would dominate.
 
Mar 30, 2018
11
0
10


Yes, but as I said, the numbers there are not the real score, they are just an average score. In scores the 680 in stock should be close to 6700-800 G3D equal to that of the 1050 Ti (although this varies according to the processor) one day I saw that the 1050 Ti had a baseline with a score of 7500 points, but probably it had OC. A 680 that overclockes better, with oc should be a little higher (7600-700).

 
Mar 30, 2018
11
0
10


I do not think the 4GB of VRAM are the ones that give him the real advantage. Fornite, I do not know ... and a GTX 1050 of 2gb equal to the GTX 680 in modern games and maybe it does not get away much in the old games https://www.techspot.com/article/1557-fortnite- benchmarks / https://www.youtube.com/watch?v=RnOk5xhaUbM (they are the only ones I found where they compare both cards) In userbenchmark I do not know if it will have a fornite benchmark in the games section, but even if I would, I would not take it into account as it is a page constantly change and the numbers are not 100% successful to make a real test in real time. And in synthetic benchmarks, yes those GTX 670 with overclock are very good. But as I said I'm interested in performance in real time, I'm not a synthetic benchmark: S

I've also tested my 1050 Ti in old games like Bioshock Infinite, Sleepging Dogs, Crysis 2, Resident Evil 5 and 6 and the Stalkers (where Kleper should not be nerfed) and then compare the results with a 680 of some video from the same game. My 1050 Ti is winning by a few fps (3-4) so ​​it would still be in relative performance with the 680 unless it is overclocked that would have a 6-10 fps advantage. So if in the end my 1050 Ti gives the relative performance of a GTX 680, why are there people who say that a 670 is better?

 

fagetti

Notable
Mar 1, 2018
919
15
1,165
670 and 680 stock has about 5% difference. Example factory overclocked models from msi get higher performance than 680 stock. Just look it up theyre practically same when overclocked max stable clocks.

Too bad i can no longer test my 670 to see differences. And your right about userbenchmark giving crappy results but so does passmark. Superposition or valley are better or 3d mark tests.

Im sure your 1050 ti is bettr optimized and new drivers dont give you performance loss which can happen with older nvidia cards.

I hope you can have a nice overclock on it, have you tried yet?

Sorry if i sounded rude on first messages so busy at work and stressed. I am in no way trying to prove 670 is better in gaming than a 1050 ti, and not everyone try overclock theyre card that far like me.
 
Solution
Mar 30, 2018
11
0
10


No problem friend :) In fact, I'm glad that someone shares friendly discussions, so you share ideas and more. Is that obviously the GTX 670 if it is overclocked can reach or exceed a 680 in stock, after all a 670 is only a 680 only with fewer shaders and slightly lower frequencies. Is that I always wanted to have something similar to a GTX 680 (it was my biggest wish for 2012) and when the 1050 Ti came out I was happy, I was having similar performance but with only 75w of energy! and it only makes me feel a little bad that some people say that a 670 can "kill" as if nothing to my 1050 Ti ...

As for Passmark and Userbenchmark, I agree. Both are useless for real-time performance. Also looking at your superposition benchmarks and comparing them with those of the GTX 680, the 670 you had has practically the same score of a 680 in stock :) My 1050 Ti gives a slightly higher score, but a 680 with overclock exceeds my 1050 Ti for about 300 points (3fps more in average in superposition) those kleper overclocked like a beast!

Finally, the 1050 Ti as well as the pascal scale very badly with overclock, only about 4-6 fps more ... a waste of profit, but it is appreciated.

Greetings and thanks for taking the time to respond!

 

fagetti

Notable
Mar 1, 2018
919
15
1,165
This was a nice discussion in the end. If someone reading this thread is thinking about budget graphics card it depends on few issues. First is power supply, if you already have a high quality power supply with 650w or 750w then you might think about Gtx 670 or 680 over 1050 Ti, would be ridiculous to buy new psu for that old card. And ofcourse you drain 200-275w compared to 70w for 1050ti.

And leon is there a way to increase voltage for 1050ti? Im pretty sure little bumb wouldnt do any harm if you find a way like soft mod for it? You could get better overclock for sure.

 
Mar 30, 2018
11
0
10


unfortunately it is not possible ... In the 6-pin models the maximum you can upload is only a little power limit, but the Voltage is totally blocked :( one of the biggest reasons why the 1050 Ti and many pascal stink in the scope of the overclock, that's why even though a gtx 1070 is better in general than a 980 Ti for example, but it is annihilated if the 980 Ti is overclocked

 

fagetti

Notable
Mar 1, 2018
919
15
1,165
So you cant see voltage bar in msi afterburner or evga precision? Try both. There could be unlocked bios which gives you small voltage bumb to nvidia recommended maximum voltages, like my GTX 780 Lightning went from 1.15v-1.175 to 1.212v and i gained some performance. This card overclocks even better than my gtx 670. Look my scores compared to GTX 1060 stock and overclocked http://www.tomshardware.co.uk/forum/id-3536560/tom-superposition-thread.html
Even if i test in fortnite it shows improvement in fps compared to GTX 1060 stock.

If you open gpu-z sensors what voltage do you get under max load?
 

fagetti

Notable
Mar 1, 2018
919
15
1,165
I guess it depends on the model of 1050ti, what you have?

EDIT: open msi afterburner go options => general => press mark on "unlock voltage control" and "unlock voltage monitoring" also change option right of that "MSI standard" change that to "msi extended" after that click apply / ok on botton. You might need a restart before voltage control enabled.
 
Mar 30, 2018
11
0
10


No, none of that works. The voltage bar appears but can not be modified. I have a SC model that does not have a 6-pin connector, but apparently the 6-pin models do the same thing. It seems that the 6-pin models of the 1050 Ti is just marketing, the only good thing would be the dual fans but a 1050 Ti does not need as much refrigeration, even with OC. Also the 1050 Ti seems to not give more performance once the Core Clock makes the gpu boost 3.0 exceeds 1911mhz, so to raise more frequencies or even if hypothetically you could raise the voltage more, you will not gain more performance.

The good thing about the Kleper and the Maxwell in terms of voltage was that if a card had the voltage blocked, it could be unlocked with a modified bios as you say, but unfortunately for the pascal and even less for the 1050 Ti, not exist..

PD: In the post of superposition there is a guy (equinehero) that has the 1050 Ti, but says that in terms of performance is identical to a GTX 580 only because it could give a similar performance in a synthetic bench ?! WTF? Anyway, on the internet you find everything ... my 1050 Ti gives only 300 points below in medium preset in superposition than a 680 with OC (6300 points) but according to him, a 1050 Ti is only equal to a gtx 580, which means that the 580 must give in preset medium 6000 points? Honestly I do not think so, the 1050 Ti must be defective or simply does not know what it talks about, to the point where it underestimates the 1050 Ti ... in short ... things like this are what you find on the internet and forums.

Greetings friend.
 
Mar 30, 2018
11
0
10


Actually doing that can already change the voltage, the problem is that as I said in the previous comment, raise the voltage may not be useful now that I see for a 1050 Ti, since after 1911mhz in the core is not achieved nothing in performance and possibly when the voltage goes up, the card may be screwed ...

Anyway, thank you for your attention, friend, I really appreciate it :)
 

fagetti

Notable
Mar 1, 2018
919
15
1,165
So you capped it at 1911mhz? It might be hitting power limit also or frequency is somehow locked at 1911mhz and going further doesnt really change the frequency. Do a benchmark or gaming while gpu-z sensors on backround, what kind of TDP its hitting?

And check if frequency really changes (gpu-z readings) above 1911mhz if you change it in afterburner.