Geforce GTX 1050 / 1050 Ti Thread

[happy medium]

to be fair im kinda disappointed by the gtx1050 ti performance..10k vs 17k of the gtx1060 is a huge gap.

the gtx950 once overclocked it went way ahead of the gtx960..
http://www.guru3d.com/articles_pages/asus_geforce_gtx_950_strix_review,30.html

but i guess price difference was closer on maxwell around $50 even less in most cases...now with pascal it should be around $100..

to be fair im kinda disappointed by the gtx1050 ti performance..10k vs 17k of the gtx1060 is a huge gap.

the gtx950 once overclocked it went way ahead of the gtx960..
http://www.guru3d.com/articles_pages/asus_geforce_gtx_950_strix_review,30.html

but i guess price difference was closer on maxwell around $50 even less in most cases...now with pascal it should be around $100..

here is the 768 core the gtx950 @ 1.550 ghz vs the gtx960.
If these are the same specs as the 1050ti,this should be close to the performance.

 

[happy medium]

If those benchmarks are legit, the 1050 TI of going to be 25% faster than the RX 460.

Yea but if the 1050ti is 150$ vs the 4gb 460 @ 115$. That's right where they should be. You can get a 2gb 460 at newegg for 85$ AR and promo today.
The 460 with 2gb is no different than a 4gb model performance wise.

The cheaper 2gb 1050 is the 460's competition.

 

[tviceman]

Yea but if the 1050ti is 150$ vs the 4gb 460 @ 115$. That's right where they should be. You can get a 2gb 460 at newegg for 85$ AR and promo today.
The 460 with 2gb is no different than a 4gb model performance wise.

The cheaper 2gb 1050 is the 460's competition.

I didn't realize 460's were going for so cheap already. I thought they were all $139-149.

 

[happy medium]

I didn't realize 460's were going for so cheap already. I thought they were all $139-149.

supply and demand, mabe nobodys buying them?or they know what the 1050 can do?
price has dropped like a rock fast.

 

[Glo.]

supply and demand, mabe nobodys buying them?or they know what the 1050 can do?
price has dropped like a rock fast.

Or they are preparing room for RX 465 with full 16 CU design?

There is a very good reason why GTX 1050 Ti is launching on precisely 24th October...

 

[Arachnotronic]

Or they are preparing room for RX 465 with full 16 CU design?

There is a very good reason why GTX 1050 Ti is launching on precisely 24th October...

I doubt that. If AMD had such a potential RX 465 up its sleeve, it would have launched it instead of waiting a couple of months to release it. The more performance AMD could have offered, the more attractive the RX 460 would have been, helping to boost its sales.

 

[dark zero]

I doubt that. If AMD had such a potential RX 465 up its sleeve, it would have launched it instead of waiting a couple of months to release it. The more performance AMD could have offered, the more attractive the RX 460 would have been, helping to boost its sales.

I feel that AMD didn't released that product launch details for nothing... I see that AMD is preparing their refresh with better process (aka Samsung) with the following products:
- RX 465 with full CU and better process.
- RX 475 with GDDR5X
- RX 485 with GDDR5X
- RX 490 being dual RX 480 with GDDR5X

Considering that Vega is about to hit by June of the next year, is recomended for AMD to refresh the line between November and December.

Also remember that it's time for the price slash for AMD to make room for the superior models.

 

[daxzy]

RX 460 2GB MSRP was supposed to be $99.
RX 460 4GB MSRP was supposedly $119.

We're actually seeing prices be more "normal". Too bad RX 470/480 prices are still way above MSRP, probably due to mining.

 

[Enigmoid]

I feel that AMD didn't released that product launch details for nothing... I see that AMD is preparing their refresh with better process (aka Samsung) with the following products:
- RX 465 with full CU and better process.
- RX 475 with GDDR5X
- RX 485 with GDDR5X
- RX 490 being dual RX 480 with GDDR5X

Considering that Vega is about to hit by June of the next year, is recomended for AMD to refresh the line between November and December.

Also remember that it's time for the price slash for AMD to make room for the superior models.

Is the 485 bandwidth limited to such an extent that it justifies much more expensive GDDR5X?

 

[Bacon1]

Is the 485 bandwidth limited to such an extent that it justifies much more expensive GDDR5X?

People that OC the ram found very good gains, more so than ocing the core.

 

[dark zero]

Is the 485 bandwidth limited to such an extent that it justifies much more expensive GDDR5X?

Indeed, people who OC it found that. Similar thing happened with the 1070.
GDDR5 is not enough for both cards.

Similar thing happens with the 470.

 

[crisium]

Well it's either the bandwidth or the ROPs, or probably both, that cause the 480 to compete with the 390X at 1080 but fall behind fast at 1440p+. It has a bottleneck and is clearly a very unbalanced design compared to the 1060. Too bad there's no "32ROPs5X" to refresh Polaris 10 with.

 

[dark zero]

Well it's either the bandwidth or the ROPs, or probably both, that cause the 480 to compete with the 390X at 1080 but fall behind fast at 1440p+. It has a bottleneck and is clearly a very unbalanced design compared to the 1060. Too bad there's no "32ROPs5X" to refresh Polaris 10 with.

Hardly with the ROPs. Some people found that OCing the memory got more gains than OCing the core. And that explains that even with better drivers the improvements are not to big. Also that explains why the RX 470 has their performance levels near to the performance levels of the RX 480.

 

[Enigmoid]

People that OC the ram found very good gains, more so than ocing the core.

Thanks. Is there any suggestion that the memory controller on Polaris can actually run at those speeds? AMD's memory controller generally has never been able to hit the same speeds that Nvidia can post Fermi. (AMD's last few APU revisions have used a generic out of house memory controller for instance).

 

[LTC8K6]

Will they/could they bring out the same cards again with a slightly different name and GDDR5X ram?

 

[happy medium]

Will they/could they bring out the same cards again with a slightly different name and GDDR5X ram?

I don't see this happening, it seem like another "wait and see" fanboy rumor.
Mabe next year when Nvidia refreshes their entire lineup, that I would believe.
I believe it was the chief AMD guy Raja or something like that, he said there will be 3 AMD cards released this year. The 460, 470 and 480.

So lets stomp out this bull before it starts.

AND this is all off topic, this is a 1050 thread not a 460, 470 and 480 rumor thread.

 

[LTC8K6]

I don't see this happening, it seem like another "wait and see" fanboy rumor.
Mabe next year when Nvidia refreshes their entire lineup, that I would believe.
I believe it was the chief AMD guy Raja or something like that, he said there will be 3 AMD cards released this year. The 460, 470 and 480.

So lets stomp out this bull before it starts.

AND this is all off topic, this is a 1050 thread not a 460, 470 and 480 rumor thread.

I didn't name a card mfg...

 

[Bacon1]

I don't see this happening, it seem like another "wait and see" fanboy rumor.
Mabe next year when Nvidia refreshes their entire lineup, that I would believe.
I believe it was the chief AMD guy Raja or something like that, he said there will be 3 AMD cards released this year. The 460, 470 and 480.

So lets stomp out this bull before it starts.

AND this is all off topic, this is a 1050 thread not a 460, 470 and 480 rumor thread.

It comes from this:

http://www.guru3d.com/index.php?ct=news&action=file&id=15570

 

[crisium]

Hardly with the ROPs. Some people found that OCing the memory got more gains than OCing the core. And that explains that even with better drivers the improvements are not to big. Also that explains why the RX 470 has their performance levels near to the performance levels of the RX 480.

It seems to me what you are describing also indicates a possible ROP bottleneck. If the ratio of Shaders-to-ROPs is unbalanced, as it likely is in Polaris, then wouldn't it result in what you are describing too? If I'm off base, can you explain what a ROP bottleneck would look like when OCing?

 

[LTC8K6]

It wouldn't surprise me to see either mfg release "new" versions of the mid-level gpus with GDDR5X, but I'd think NV is more likely to do it if it were going to happen.

OTOH, I have no idea if these GPUs can use GDDR5X.

 

[dark zero]

It comes from this:

http://www.guru3d.com/index.php?ct=news&action=file&id=15570

Also returning to nVIDIA, they already are hinting that they will do a Refresh of the 1060 and 1070 cards too with the GDDR5X which is now cheaper than before.

 

[Sweepr]

Exclusive: GP107 Die + 1050 Ti 4GB Board

Can anyone estimate the die size?

 

[Glo.]

6 pin connector?

 

[Sweepr]

Might be a custom board or an ES.

 

[Shivansps]

Custom 750TI boards used to have a 6pin connector as well.