Laptop Gamer, should I wait for Pascal?

[tverma]

Yeah but that's the problem. If AMD leaves the discrete graphics or graphics business altogether, what will stop Nvidia from pulling an Intel on us?

 

[renz496]

actually nvidia already did that since kepler. back in 2010/2011 GF104 were fit into 200-250 range product. but now? 680 and 670 is based on GK104 were priced 500 and 400 respectively. 980 and 970 which based on GM204 was priced 550 and 330 at launch. usually product with code name ending with number 4 has always been nvidia mid ranger and priced in sub 250 market. now those 'mid range chip' has been move up to high end segment of sub 500 market. the flagship chip is even more expensive now in sub 700. but because of nvidia doing so it give AMD some room to breath. imagine if nvidia did sell 680 as 660 at 250 back in 2012. that will be disaster towards AMD 7970 which priced at 550 back then.

 

[tverma]

actually nvidia already did that since kepler. back in 2010/2011 GF104 were fit into 200-250 range product. but now? 680 and 670 is based on GK104 were priced 500 and 400 respectively. 980 and 970 which based on GM204 was priced 550 and 330 at launch. usually product with code name ending with number 4 has always been nvidia mid ranger and priced in sub 250 market. now those 'mid range chip' has been move up to high end segment of sub 500 market. the flagship chip is even more expensive now in sub 700. but because of nvidia doing so it give AMD some room to breath. imagine if nvidia did sell 680 as 660 at 250 back in 2012. that will be disaster towards AMD 7970 which priced at 550 back then.

Yeah you are right. Hopefully this competition between them is bigger than we expect and they really battle for the cheapest prices

 

[renz496]

If you're looking at the trend AMD actually want gpu price to increase. In fact they probably want to change people perspective that AMD gpu is cheaper alternative to nvidia. Hence Fury X was priced the same as 980ti and not much lower. Then look at nano. They believe the card worth the asking price of 650 because it was better binned for lower power consumption and it's smaller size can fit into mini itx cases. And initially the rumor talking about Fury line up will compete directly in both and price with nvidia Titan X. Means initially Fury X could cost between 800-1000. But nvidia jump the gun with 980ti at 650. Perform about the same as Titan X but with significantly cheaper price. And when you consider 980ti OC potential.....

In any case pretty much everyone aware the financial situation AMD was in. So they push AMD slowly in this angle. Just look the damage done by 970. Once AMD can easily sell 290/290X above 400. But when nvidia drop 970 amd have to lower the price of 290s below 300. And at the same time nvidia leisurely selling 980 at 550. If anything price war is probably something that AMD want to avoid with nvidia. It is something that they can't sustain in the long run against nvidia.

 

[tverma]

Yeah. Check this out: http/www.extremetech.com/gaming/218224-nvidia-unveils-pascal-specifics-up-to-16gb-of-vram-1tb-of-bandwidth
Does increasing the transistor count improve performance?

 

[renz496]

Within same architecture? Then yes. GM204 only have 5billion transistor but in gaming it is faster than GK110 with it's 7 billion transistor.

 

[tverma]

Within same architecture? Then yes. GM204 only have 5billion transistor but in gaming it is faster than GK110 with it's 7 billion transistor.

For example, lets say the GTX 980 has 8 billion transistors (I don't know if that is the case), Pascal is supposed to have around 17 billion transistors at the highest end. So, will this result in a direct performance improvement?

 

[renz496]

There will be improvement for sure but it won't translate in linear way. Means if for example pascal having 16billion transistor it might not double the performance. Maxwell gaining massive performance increase in games vs Kepler despite built on the same node because nvidia throw away compute stuff from maxwell so they can contribute those transistor count towards gaming performance. AMD did the same with Fiji or else Fiji will not be possible with current 28nm node. With Pascal nvidia was supposed to bring in those compute stuff back. So it is possible nvidia was hoping to gain more performance from node shrink itself. It will be interesting to see if nvidia can overcome this problem or not.

 

[tverma]

There will be improvement for sure but it won't translate in linear way. Means if for example pascal having 16billion transistor it might not double the performance. Maxwell gaining massive performance increase in games vs Kepler despite built on the same node because nvidia throw away compute stuff from maxwell so they can contribute those transistor count towards gaming performance. AMD did the same with Fiji or else Fiji will not be possible with current 28nm node. With Pascal nvidia was supposed to bring in those compute stuff back. So it is possible nvidia was hoping to gain more performance from node shrink itself. It will be interesting to see if nvidia can overcome this problem or not.

So how much of a gaming performance increase should we expect? 10%,20%, 30% if we are lucky?

 

[renz496]

Hard to say. I don't have crystal ball haha

But if Nvidia want to play it to maximize their profit they will release card that is 10% faster than 980ti first. Even if going from GM200 to GP100 can give you at least 60% performance increase.

 

[tverma]

Hard to say. I don't have crystal ball haha

But if Nvidia want to play it to maximize their profit they will release card that is 10% faster than 980ti first. Even if going from GM200 to GP100 can give you at least 60% performance increase.

So this brings me back to my original is it wroth it to wait?

 

[renz496]

For deskstop pc or laptop?

 

[velo3100]

My best suggestion with all tech things is, if you can wait then wait. Tech will always improve in terms of performance and efficiency. I am still waiting for the perfect laptop for me as my gaming needs are very specific. Also for $1000, laptops are not the best for gaming due to heat. This will limit the time it lasts. But I don't know how well the cooling will improve by then.

 

[tverma]

Yeah...well considering I'm running Intel HD 3000, I was unable to play any good game so I found this great deal and just decided to run with it. You wouldn't find a gaming laptop for that price. For $999:Aspire V15 Nitro Black Edition VN7-592G-70EN 15.6" Gaming Laptop Computer; Intel Core i7-6700HQ Processor 2.6GHz; Microsoft Windows 10; 16GB DDR4 RAM; 1TB 5,400RPM Hard Drive; 128GB Solid State Drive; NVIDIA GeForce GTX 960M 4GB GDDR5; Secure Digital Card Reader; 10/100/1000 Network; 802.11ac Wireless; Bluetooth 4.0; 15.6" Full HD Widescreen LED-backlit IPS Display. I'd imagine this is a good deal right? I did find a laptop with 970m for 999 but its processor was 5700, it was only 8Gb ddr3, no solid state, only one fan cooling, etc. This laptop that I bought is also meant for school so I feel like I got a great deal but opinions?

 

[bugnguts]

I would suggest waiting, but it could be Q3 2016. If in a move to maximize profit Nvidia only improves the performance by 10-15% than the rest of the benefits from the double node shrink will come in efficiency. 28nm to 16FF allows for around 70% less power use. In a desktop this matters less to gamers but in a laptop it means running a lot cooler. This will allow for quieter, thinner, lighter and cooler laptops. Nvidia is pushing more power in laptops; they recently announced support for full desktop 980gtx in laptops. If they push mobile performance and match the 980gtx desktop with a 16nmFF Pascal which draws much less power then you will have some great graphics power that runs cooler than today's mobile cards. Pascal will either give quite a bit more graphics power or be much more efficient. Win/Win if you don't mind waiting.

 

[tverma]

I would suggest waiting, but it could be Q3 2016. If in a move to maximize profit Nvidia only improves the performance by 10-15% than the rest of the benefits from the double node shrink will come in efficiency. 28nm to 16FF allows for around 70% less power use. In a desktop this matters less to gamers but in a laptop it means running a lot cooler. This will allow for quieter, thinner, lighter and cooler laptops. Nvidia is pushing more power in laptops; they recently announced support for full desktop 980gtx in laptops. If they push mobile performance and match the 980gtx desktop with a 16nmFF Pascal which draws much less power then you will have some great graphics power that runs cooler than today's mobile cards. Pascal will either give quite a bit more graphics power or be much more efficient. Win/Win if you don't mind waiting.

Yeah. I decided to buy a GTX 960m 4gbvram, i7-6700hq, 16gb ddr4, 128 ssd, 1TB acer aspire pro for $999. 'Twas a good deal so yeah plus I have TB3 on that laptop so external graphics are in my future

 

[wizao3]

I would suggest waiting, but it could be Q3 2016. If in a move to maximize profit Nvidia only improves the performance by 10-15% than the rest of the benefits from the double node shrink will come in efficiency. 28nm to 16FF allows for around 70% less power use. In a desktop this matters less to gamers but in a laptop it means running a lot cooler. This will allow for quieter, thinner, lighter and cooler laptops. Nvidia is pushing more power in laptops; they recently announced support for full desktop 980gtx in laptops. If they push mobile performance and match the 980gtx desktop with a 16nmFF Pascal which draws much less power then you will have some great graphics power that runs cooler than today's mobile cards. Pascal will either give quite a bit more graphics power or be much more efficient. Win/Win if you don't mind waiting.

Yeah. I decided to buy a GTX 960m 4gbvram, i7-6700hq, 16gb ddr4, 128 ssd, 1TB acer aspire pro for $999. 'Twas a good deal so yeah plus I have TB3 on that laptop so external graphics are in my future

Isn't 970m worth a lot more than 960m ? I mean performance wise 960m might die much sooner than the 970m, with 970m you might last one more year of decent gaming, at the same price ram and cpu architecture isn't as important for a gamer as a 50% performance increase in my opinion.

 

[tverma]

Isn't 970m worth a lot more than 960m ? I mean performance wise 960m might die much sooner than the 970m, with 970m you might last one more year of decent gaming, at the same price ram and cpu architecture isn't as important for a gamer as a 50% performance increase in my opinion.[/quotemsg]

970m is definitely better than the 960m. However what you have to take into account is the amount of money youll end up paying for it. Getting to a 970m, you have to shell out an additional 400-500 dollars. That to me seems like a pointless purchase. For example on everything ultra totally maxed out, I get ~30fps on fallout 4. That to me shows that this gpu has plenty of life as I don't mind turning it down to medium as the years pass.

 

[Garbis93]

nvidea just talked about the gtx 980 MX and 970 MX do you think pascal will be delayed for the laptop?

 

[tverma]

nvidea just talked about the gtx 980 MX and 970 MX do you think pascal will be delayed for the laptop?

This is Nvidia milking the market. Pascal will be delayed definitely