New M17x R4, M14X R2 and M18X R2 are coming - Speculation thread.

[YodaGoneMad]

After seeing Cebit, with all the new laptops having "GPU specs to be released" cards on them, I feel pretty certain nothing exciting is coming to mobile anytime soon. Probably late this year for a 680m type card, 675m is just a renamed 580m, and Nvidia has been pretty clear that they expect the 675m to beat any of the new AMD mobile parts that are coming soon.

AMD is also said to have horridly low yields, and so they are planning to put most of their focus on desktop cards with the high end mobile cards coming late this year. That means the 580m/675m is likely going to be top dog for most of this year in the mobile space, and then AMD will release a 7990m maybe in the fall, and Nvidia will come after them with the 680m in late fall or early holiday season.

MXM 3.1 is kind of a lame standard. All it really does is add PCIE 3.0 speed support, but it is a backward and forward compatible standard, which means new 3.1 cards will run in 3.0b slots. This further means that there is no TDP increase or anything really worthwhile, it will just be a tiny bit more bandwidth that really isn't needed by mobile cards.

Finally, Ivy reviews show a 5-15% performance increase and of course the faster GPU. Overall, nothing to get excited about, really means this is going to be a boring year for mobile, next year should be pretty amazing though. With mature GPUs on these new architectures and a brand new Intel architecture. Next year will probably see 50%+ performance gains in all areas.

[TheBluePill]

After seeing Cebit, with all the new laptops having "GPU specs to be released" cards on them, I feel pretty certain nothing exciting is coming to mobile anytime soon. Probably late this year for a 680m type card, 675m is just a renamed 580m, and Nvidia has been pretty clear that they expect the 675m to beat any of the new AMD mobile parts that are coming soon.

AMD is also said to have horridly low yields, and so they are planning to put most of their focus on desktop cards with the high end mobile cards coming late this year. That means the 580m/675m is likely going to be top dog for most of this year in the mobile space, and then AMD will release a 7990m maybe in the fall, and Nvidia will come after them with the 680m in late fall or early holiday season.

MXM 3.1 is kind of a lame standard. All it really does is add PCIE 3.0 speed support, but it is a backward and forward compatible standard, which means new 3.1 cards will run in 3.0b slots. This further means that there is no TDP increase or anything really worthwhile, it will just be a tiny bit more bandwidth that really isn't needed by mobile cards.

Finally, Ivy reviews show a 5-15% performance increase and of course the faster GPU. Overall, nothing to get excited about, really means this is going to be a boring year for mobile, next year should be pretty amazing though. With mature GPUs on these new architectures and a brand new Intel architecture. Next year will probably see 50%+ performance gains in all areas.

You were right all-along. How did you know?

[YodaGoneMad]

You were right all-along. How did you know?

The information has been out there since December, obviously not the rebadge stuff, but just that yields are bad, Ivy is not very exciting, and GCN is underwhelming.

I ordered my new M17x in December because I knew nothing amazing was coming for almost a year, and I have thus been advising other people to do the same.

It really puts some people in a frustrating spot now, if you had been waiting a few months for the "next-gen" stuff you end up with a 580m/675m that is a few months closer to being obsolete and an inconsequential CPU upgrade.

[Aikimox]

After seeing Cebit, with all the new laptops having "GPU specs to be released" cards on them, I feel pretty certain nothing exciting is coming to mobile anytime soon. Probably late this year for a 680m type card, 675m is just a renamed 580m, and Nvidia has been pretty clear that they expect the 675m to beat any of the new AMD mobile parts that are coming soon.

AMD is also said to have horridly low yields, and so they are planning to put most of their focus on desktop cards with the high end mobile cards coming late this year. That means the 580m/675m is likely going to be top dog for most of this year in the mobile space, and then AMD will release a 7990m maybe in the fall, and Nvidia will come after them with the 680m in late fall or early holiday season.

MXM 3.1 is kind of a lame standard. All it really does is add PCIE 3.0 speed support, but it is a backward and forward compatible standard, which means new 3.1 cards will run in 3.0b slots. This further means that there is no TDP increase or anything really worthwhile, it will just be a tiny bit more bandwidth that really isn't needed by mobile cards.

Finally, Ivy reviews show a 5-15% performance increase and of course the faster GPU. Overall, nothing to get excited about, really means this is going to be a boring year for mobile, next year should be pretty amazing though. With mature GPUs on these new architectures and a brand new Intel architecture. Next year will probably see 50%+ performance gains in all areas.

While I agree with your comments regarding Ivy and PCIE 3.0, the GPU part is a mess, lol. First, 580M is on par with the 6990M. Only after tweaking the throttling, updating vBIOS to 0.92 and OC'ing does it start to shine. At stock clocks and voltages the cards are pretty much on par except for heavy tess'ed games. Second, the 7970M is almost here (SI's have the first samples already) and it will definitely be a major upgrade with bumped tess and overall performance. I doubt the 675M will be a major threat being a rebrand. And third, the last 2 quarters were rather successful for AMD in the GPU sector.

[Kevin]

GCN is underwhelming.

You must've missed the 7870/7850 tests. If those are the 7900Ms, we're looking at easy 50% jumps over the current mobile chips.

[YodaGoneMad]

We shall see. I think you guys are way too optimistic. The 7990m and 680m chips MIGHT be a 50% boost, my money is on it being less than that by quite a bit, but those are still a long way from wide spread consumer availability, especially with low yields and everyone focusing on desktop cards.

My prediction is the initial 7970m will be like 15% faster than the 580m, but it will vary a lot game to game, with more modern DX11 games seeing a much smaller boost. Then the 7990m and 680m will be 30-45% faster than the 580m, but are still a long way from consumer availability.

Also, the 675m doesn't have to be a threat, it just has to hold its own. Unless the 7970m is significantly faster the 675m will really hurt it. Nvidia can sell it much cheaper, which will force prices down, they won't have supply problems, and my guess is it will still be roughly the same performance wise. You say the 6990m is on par with the 580m, and it is true, but there is probably at least a 10% gap between them in modern games, and with OCing the 580m will widen that gap even more. The 7970m and 675m will be the same deal, just switched, the 7970m will be faster, but most people will see them as similar because the 7970m won't be THAT much faster.

[Kevin]

Explain to me, technologically, how the 7970M could only manage to be 15% faster than the 580M, if it comes from either of the 7800 cards.

The 7850 at peak consumption is only 96W; the 7870 is 116W, and they're both faster than the GTX 570 and 6950, sometimes even the 6970 and 7950.

AMD would have to purposely hack the cards to bits for them to only manage the performance of a barely overclocked 6990M.

[bigtonyman]

Explain to me, technologically, how the 7970M could only manage to be 15% faster than the 580M, if it comes from either of the 7800 cards.

The 7850 at peak consumption is only 96W; the 7870 is 116W, and they're both faster than the GTX 570 and 6950, sometimes even the 6970 and 7950.

AMD would have to purposely hack the cards to bits for them to only manage the performance of a barely overclocked 6990M.

your comparing apples to oranges i think. The desktop cards are completely different than their laptop counterparts. an example is the 6990m. Its actually a down-clocked 6870, not dual 6970's like the desktop card although that would be pretty sweet. Just wanted to point out the differences between the two classes of cards. I may just be understanding what you trying to say here too, and if that's the case im sorry.

Edit: oops read through it again and you made sense, forget my rant

[goussama]

After seeing Cebit, with all the new laptops having "GPU specs to be released" cards on them, I feel pretty certain nothing exciting is coming to mobile anytime soon. Probably late this year for a 680m type card, 675m is just a renamed 580m, and Nvidia has been pretty clear that they expect the 675m to beat any of the new AMD mobile parts that are coming soon.

AMD is also said to have horridly low yields, and so they are planning to put most of their focus on desktop cards with the high end mobile cards coming late this year. That means the 580m/675m is likely going to be top dog for most of this year in the mobile space, and then AMD will release a 7990m maybe in the fall, and Nvidia will come after them with the 680m in late fall or early holiday season.

MXM 3.1 is kind of a lame standard. All it really does is add PCIE 3.0 speed support, but it is a backward and forward compatible standard, which means new 3.1 cards will run in 3.0b slots. This further means that there is no TDP increase or anything really worthwhile, it will just be a tiny bit more bandwidth that really isn't needed by mobile cards.

Finally, Ivy reviews show a 5-15% performance increase and of course the faster GPU. Overall, nothing to get excited about, really means this is going to be a boring year for mobile, next year should be pretty amazing though. With mature GPUs on these new architectures and a brand new Intel architecture. Next year will probably see 50%+ performance gains in all areas.

So just got a really quick question, I got the m18x in December, will it be compatible with the 680m?

[YodaGoneMad]

So just got a really quick question, I got the m18x in December, will it be compatible with the 680m?

Yes, the MXM 3.1 slot and cards are backward and forward compatible. A 3.1 card will work in your 3.0b slot.

Explain to me, technologically, how the 7970M could only manage to be 15% faster than the 580M, if it comes from either of the 7800 cards.

The 7850 at peak consumption is only 96W; the 7870 is 116W, and they're both faster than the GTX 570 and 6950, sometimes even the 6970 and 7950.

AMD would have to purposely hack the cards to bits for them to only manage the performance of a barely overclocked 6990M.

I have no idea where you get the idea that the 7850/7870 desktop cards are such amazing cards.

The 580m/675m are based on the GTX 560 TI with lower clock speeds. The 560 TI gets a 3dMark11 GPU score of 2224. The 7850 gets 2536, the 7870 gets 3152. This means the 7850 is 14% faster than the 560 TI, and the 7870 is 24% faster.

I predict that the mobile cards will hold the same ratios, the mobile AMD cards will be down clocked versions of the desktop cards. If the 7970m is based off the 7850 then I expect it to be roughly 15% faster than the 580m, and the 7990m roughly 25% faster than the 580m. Maybe I will be wrong, but all of my past predictions based on similar data analysis have proven true. These numbers also fall in line with my prediction earlier that the 7990m and 680m would be around 30% faster than current cards.
http://www.hardwaresecrets.com/artic...Review/1502/12