Brace yourself: NEW MAXWELL CARDS INCOMING!

I had the same thing with my Note 2, but it really is different with phones. We now have fantastic screens, and all the power you really need so there's not as much to improve on Also the gaming benchmarks vs real world testing is really only a mobile comparison + who really games that much on their phone? For high end graphics cards, benchmarks and real fps are much more closely related, they don't differ from card to card and from the system they're put in like with smartphones. Anyways, my Note 2 has been great, more than powerful enough and I'm currently waiting for the Note 4 to be announced this September. Excited to see the new 2k screen, hopefully they keep it Amoled and improve the camera over the S5.

The updated Alienware 18 has a new IPS 1080p 60hz display. I would have liked a higher resolution for non-gaming things, but I'm glad they didn't force 120hz/3d display on us. The one thing that's disappointing, is the screen size in the Alienware 18. It could clearly be larger if they bothered to make it edge to edge.

 

Why would you not want a superior 120hz LCD? I find myself wondering how I ever lived without mine before and when i play games on my TV it's a very noticeable difference going down to 60hz.

1080p is the sweet spot for displays. I'd be lying if I said I didn't miss the 1200p display in my M17xR1 but for games, 1080p is the best balance between resolution and performance, at least for now. LCDs don't look right when you don't have them on their native resolution so a 2k display would be unacceptable personally

Sent from my HTC One_M8 using Tapatalk

 

We won't see upgraded top tier Mobile silicon until mid-end Q1 2015 at earliest, probably Q2..

 

What's that supposed to mean, R9-295X is coming out next month and 880MX will probably follow not too long after it, if you are talking about <= 20nm process that won't be happening until 2H2015.

 

Can anyone please suggest, will it be fine to go with current 870M with the fact that it is pretty cheap than 880M ? Or is it expected that the newer ones will be in same price range and it is worthy to wait for 870MX or similar one like that.
I am asking this question just to get an idea that if suppose 870MX arrives but it will be around price range of current 880M or more then I won't need to wait since anyways it will be out of my budget.

 

ok this has bought up some intresting stuff while I was asleep.. is the 295X confirmed for next months release? if it is as you say I cant see nvidia sitting on the fence (unless the 295X dose not out preform the nvidia cards)

It would be nice if we got some kind of idea when they may be bringing out the new card (if they even do...)

I Have used ATI in my old m17x r2 crosfire 4870 and i didnt really have any major problems but thats old tech obviously! If i get the choice i may just wait it out for the new ati cards if im offered that or come to some sort of agreement that i will keep the nvidia card on the promise that when the new maxwell cards come out im upgraded to them.

seems a bit unfair on resellers that all the problems tho 880m has caused may end up costing them money to keep customers happy!

 

It is unfair. If you aren't happy, return the machine and get a refund, don't have unreasonable expectations for the reseller to keep you happy. That's some seriously backwards thinking right there. Either get it fixed while you're in your 30 days or send it back and get a refund and wait it out but you can't expect the reseller to eat the cost of depreciation on a refurbished video card just because you want something else...

 

but me returning the machine is still gonna cost them because all the parts are then classed as second hand and cant be used in said laptops anymore?? i like the lapton alot.. i dont like the 880m because of all its issues at the moment.

 

If you read that post a bit more closely you would see that he/she said it seems a bit unfair on the resellers themselves. Not the consumer...

 

Battery life is the biggest issue. If this wasn't a problem it would be fine. But at least when it runs on integrated graphics you can get 3-4 hours out of it.

 

I agree.

With Maxwell, it will probably come to the point that you need something better than a vanilla 1080p display.
120Hz will be nice, that can output the extra FPS over 60 FPS.

I`d still prefer a higher resolution display though. Preferably an IPS one.
15" notebook owners are lucky this way, because there is a growing number of 3K/4K displays put inside them. I hope Alienware makes that leap with the upcoming Alienware 18. Can`t wait to rock 2 GTX 880MX on my next computer

Awefully quiet on that part right now. I`m always searching for new information to put in the M295X thread, but very little is posted.
August is the month Tonga will be released yes, so hopefully the mobile Tonga as well. I think that will be a good upgrade over the GTX 880M

If you scale the resolution down to the same ratio, it doesnt look bad. GT60 owners have found nothing to complain about here with their 3K display running lower res.

The new cards usually replace the old ones with the same price level. For example, a GTX 880MX will cost roughly the same as a GTX 880M, and a GTX 870MX will cost the same as a GTX 870M.

GTX 880M is selling for like $850. Can`t expect to price the 880MX much more than that really. Bloody expensive

 

How certain are people that the 880mx will be available for laptops, and if they even plan to release it? I'd rather move to Nvidia than go back to AMD.
Will the power consumption be too great for the gtx880mx for it to be made available for laptops - example the gtx680mx? Are the sources reliable?

 

I'll believe it when I see it in a shipping laptop.

 

Wasn't the 680mx only in Apple machines? Could be a similar situation here but you never know.

 

That's definitely not gonna happen , they have a bunch of N16E and gtx 680mx was released after the gtx 680m

 

680MX was a low clocked 780M, so perfectly suitable for laptops.

 

BUT was this offered in laptops? As far as I'm aware the high end GTX680MX was only offered in the imac. The lower end mobile gpus carried the MX badge for a few series' but not the top of the range. Maybe we might see a gtx870mx or a gtx860mx but I'm really not too sure about the GTX880mx. It would basically force me to wait nearly 3 months.

 

There will be a 880mx , well most likely , if their is no 880mx then their won't be any lower level cards either , why make a 870mx beat your current 880m into the ground

 

I present to you the GTX 675mx vs the GTX 680m

GeForce GTX 680M vs 675MX
NVIDIA GTX 675MX or GTX 680M? - Hardware Hangout - Gaming eXtreme
Let’s Get Confused: GeForce GTX 680M vs. 680MX vs. 675M vs. 675MX | Gaming Laptop Report

I don't even think the gtx675mx was offered by alienware in their laptops?

Edit- pardon my previous typo. I meant the GTX680mx was only offered in the iMac.

 

Both are Kepler and 680M is CLEARLY better than 670MX. While current 880M is Kepler too but 870MX (if done) gonna be Maxwell with high overclocking potential.

 

Okay, let's play the guessing game. Say Nvidia bring out the gtx880mx and Alienware magically make it available for the A18. What performance difference could we possibly see compared to the GTX880m?
We can't really use the increases of the lower end GTX kepler > maxwell as a guide as it's not the same. Nvidia just seem to be providing better value cards in the lower end.

 

I think that if the 880MX does come out, then I think it will be offered in Alienware laptops, I don't see them wanting to fall behind their competitors with GPU performance. As to when it will be included is a whole other story.

 

Read some of Cloudfire's earlier posts about 880MX performance, it's an educated guess, but looks like it's going to be good! (They may not call it 880MX, but we're referring to their next high end 28nm Maxwell card that will likely replace the 880M)

 

680m was a cut down version of 680mx but the 680M came first. Also you have to look at this by chip name not the name of card and the 680m and MX are both gk104. The reason the 675mx came was due to the 670m being based on Fermi not kepler. The only reason why the 680mx was created was because of apple wanted a full gk104. But they did release the 780M for laptops which was basically the 680mx.

They will always update their top of the line card to latest chip name so the 880mx will be in a laptop. The MX there is differentiate it from the Kepler version. It won't be an exclusive to apple. If the 880m was released as a maxwell chip then it would be believable that the 880mx might will only been apple only product if the 880m was a cut down version of the gm204.

Overall the creation of the 680mx (gk104) was due to the 680m(gk104) already being out there. Hence they will have the 880mx in a laptop.

EDIT: your focusing too much on the name of card but not the chip name, they add MX to show its different from the old one. The most important thing is that we go from gk104 to gm204, they can call it whatever the hell they want , even gtx donkey kong 64mx doesn't change what the chip name.

 

I was more specifically talking about Alienware giving the option. We will see if/when they release it. And to be perfectly honest, "GTX donkey kong 64 mx" sounds pretty darn fantastic, you might be onto something!

 

Alienware will most likely give the option , I understand not offering the gtx 675mx but if they skip the gtx 880mx then they will lose customers.

Lol it would be cool to have a card call gtx Megatron.

 

Speculation when the AMD card will be released along with the 880mx?

 

Man. I haven't kept up in a week or so on this thread. I just skimmed over 2 50 posts pages and it just seems like a bunch of guys stroking it over a blank piece of paper. Seriously. Lol. Knowing nvidia they're probably gonna start milking the new cow. We're gonna see 10-15% performance boosts from m-mx or whatever naming scheme they use but if maxwell is all it's cracked up to be, some serious oc'ing potential or for the frugal, some nice undervolts at stock frequencies.

 

Only on my mobile phone, so that's gonna be less detailed than intended, pita to work with quotes on a small touch screen.

Yes, I totally expect some more GK110 desktop parts, and possibly GK210, though as I said I don't think that's happening for the mobile world.

More importantly, where exactly has it been *confirmed* that "GM204 have been worked on for a long time now" ?

You're just wildly assuming things and present them if they were facts... It's not hard to make up random numbers, but making things up doesn't make them true and especially doesn't make them a valid argument for comparison with task facts.

That is *your* theory.
Also I know what I wrote, yes, per Watt, however pumping twice the power through a chip doesn't result in double the performance. How well a possible new GM204 would scale... No way to know. Well I'm sure you can help us out and come up with a number

I really don't see the clocks being that much higher, actually the 750 Ti has a lower default clock than the 650 iirc.
My concerns regarding the clock limit are mainly due to physics and the current node. The revised GK110 stepping showed a nice improvement, but I'm skeptical that GMxxx can push the clock limit a lot further.
Keep in mind that all mobile chips are severely underclocked, those are no reference for how well a certain chip can overclock at all.

I act strange, just because I throw in some thoughts you don't like to hear?
You clearly presented the first post as if it was ALL Maxwell... I can tell you, that won't be the case. My assumption with yet another Fermi part is really not farfetched at all and was mainly made to try and get some people back on the ground and not blindly hovering away until they can't see any facts anymore due to all those clouds (ha, get it? )

I'm much more under the impression that not "pretty much anything" is welcomed here.
In fact it seems to me that speculation that doesn't agree with the assumptions you made, collected and posted gets harshly criticized.


"The roof is high"... Yeah, haha, but that also results in a nice fall down to the ground. Why all the hate against conclusions and speculation that relativizes and also questions some of the stuff you present here?


I'm aware that this is a thread for speculating, yet I'm under the impression that many of the rumors and made up stuff that got chewed over here again and again is suddenly presented as if it was the truth, and then you ask me why I'm skeptical...



Go on, hype it up hype it up. I'm saying no 880m successor in sight until Q1 '15.

 

If they do it, they are probably going to release it as 870M and 880M and just switch from Kepler to Maxwell, much like they've done with the existing 860M that exists in both Architectures. That they MAY do soon.. but it won't be terribly different performance wise. Who knows, I just have a feeling we're not going to see anything that's extremely compelling until late Q1 or Q2 2015.

 

Probably slight performance bumps, and MX nomenclature. But hey, I'm just guessing too.

 

Sadly yea, no major performance improvements.
I would be extremely excited if they debuted the mobile GK114 or GM204 GPUs with stacked DRAM. That would be truly beast.

 

Since we're on the topic of "speculations" and "predictions", I'm going to throw this bold claim out here:

28nm Maxwell (880MX or whatever it's called) will be at most 35% faster than 780M, and will NOT be worth the upgrade from 780M.

 

How do you think Maxwell will do in SLI? is it possible they will release the 860m as MXM ? Maybe Maxwell is a SLI beast..?

 

OK, there are way too many arguments here from both the extreme ends, so lets be a bit more realistic.
I said it before and I will say it again, 880XM (or 880M maxwell edition or whatever else they want to call it) will most likely be based on the GM206 GPU and here are the many reasons why.

1st and the most important is that GM204 is denser and has a huge die compared to GK104, it is a new and unoptimized GPU and therefore will use a lot more power than the old yet optimized mobile variant of the GK104.

2nd is the fact that it may also run so hot (again due to density and the huge die) that it may be impractical to cool it by current notebook cooling systems, causing it to be severely underclocked to keep temperatures and power consumption under check, and maybe cut down as well.

3rd, if they are going to release a 20nm (or smaller) shrink of the GM204 GPU, nVidia has no reason whatsoever to release the mobile variant of the GPU now and cause all kind of issues. (880M probably gave them enough headache)

4th is the fact that there is possibly a pattern here if you look closely. Notice how desktop 880 is supposed to replace the 780, which performs similar to the slower clocked Titan (at least in non memory limited cases such as 3DMark11) and it has core count somewhere between the 780 and Titan, but clocked much higher (most likely if the benchmark isn't fake or gotten through an overclock on A GOLDEN SAMPLE) than the Titan resulting in a 30-35% performance increase while having a smaller die, and the fact that from the leaked roadmap the GM206 has similar core count to GK104, will be much more efficient thanks to the new architecture, uses even though the same but more refined manufacturing process, and has a smaller die, it therefore can clock higher than the 780M it is supposed to replace and result in the same 30-35% performance increase that the desktop variant will get all while not having the disastrous issues that 880M is having while staying within 100W TDP.

5th, if the performance assumption in point (4) is true, it will land somewhere similar to Tonga and that is what nVidia most likely targets for. Why would they bother spending so much money on huge dies just to underclock and cut them down while a smaller GM206 can clock higher and perform no worse than a horribly underclocked big core while being much more efficient, a lot cheaper (most likely cost of the die and complexity increases exponentially compared to die size rather than linearly) and much better yields again saving them a ton of money while they charge the same $800 - $900 for the top tier GPU and keeping the larger GM204 core until it is refreshed by a die shrink release it as a GTX 980M and it will have enough of a performance improvement to keep people buying their cards and milking even more money. This is nVidia after all, this strategy is nothing new to them.

6th, the last, and one of the most important points, is that if they did release GM204 as the 880MX then they would release a more cut down variant of it as 870MX and in return wasting a perfectly good GPU that the GM206 is in the process.

Don't expect this to be a miracle and get a 50%+ improvement on the same old 20nm process. Also to those who tell "DOOM" and "GLOOM" shouldn't expect a 5% improvement a year and a half later from the previous release, you know, the kind of sh*t that intel keeps pulling over and over again.

There it took me a while but I got my point across , even though the post ended up being much longer than I expected.

 

You're all wrong. It will be 500% faster than Kepler!

OK, let's close this thread now and never speak of Maxwell until it actually launches. Doesn't matter what you think, you simply do not know, unless you have a time machine, which I do. ..not.

 

That's not a time machine, it's a time masheen.

 

I think you're being a bit harsh svl7. I see cloudfire's predictions as being logical extrapolations from what is known currently about Maxwell 750ti, extrapolations based on performance comparisons between 750ti and close performing Kepler parts, extrapolations when comparing the TDP between those similarly performing Maxwell & Kepler & based on leaks about upcoming Maxwell product specifications. I think they're on the optimistic side, but I think it's based on logical extrapolations. I welcome that kind of speculation, I don't have a problem with it.

 

You said you have a working set of eyes, but I`ll help you anyway : https://www.zauba.com/import-gm204-hs-code.html

Going back as far to beginning of April. What you see there is GM204 parts and chips going back between Nvidia`s HQ in US to Nvidia`s GPU department in India. And from Taiwan (TSMC, that produce these chips you know..) to Nvidia`s GPU department.
You will also notice mobile GM204 being mentioned there if you look closely.

Its not random when they are based on previous Maxwell cards and what we know 1st generation Maxwell could do vs Kepler architecture. Its not making up things if you look at past cards and try to apply what we know.

Tonga (R9 M295x) is looking like its going to be a 350mm2 die. Much bigger than R9 M290X with 212mm2.
GTX 880MX (GM204) is looking like its going to be 430mm2 die, based on pixel counting the leaked GTX 880 die.
Both Nvidia and AMD have to increase the die to fit in more transistors since we are still on 28nm. But as long as the power consumption and TDP go down along with architecture gain, that is perfectly plausible. They are still a good deal smaller than GTX 480M which was based on the biggest Fermi core. GM204 is not the biggest Maxwell core. GM200 is. Also a confirmed product, if you use Zauba to search for it

I don`t think you understand the theory behind it at all.
You don`t put twice the power through the chip. You deliver the exact same power to the chip, but get twice the performance. Thats the whole basis behind double performance/watt.
Again, its a NEW architecture. They are more efficient. Its the sole reason why they can reduce TDP and power draw on the cards. And why we get more powerful cards each year...
Scaling on bigger cores is a valid concern, but what else can we make assumptions on other than existing Maxwell cards?

Probably because GTX 650 didnt have GPU Boost 2.0 while GTX 750 Ti does got that. Creating a more dynamic control of the GPU.

True, we dont know how the big chips will do when overclocked, but atleast its a good sign that the low end does run pretty high clocks.

No because the first post was sarcastic and a focus on the fermi rebrand rather than being hopeful for new Maxwell cards.

Its your view on the post, not mine. I didnt say all were gonna be Maxwell. I said because they move on to N16E there will be new Maxwell chips.

Well you are free to return to your little T|I forum. These Maxwell threads are the threads that are most active on this forum, the previous one almost reached 300 000 views, and this thread is already at page 29.
Wouldnt really be that popular if other people felt they couldnt post here now would it?
There is a big difference between attacking people and posting rude posts which I have never done, and making a debate to why I or other feel what the individual wrote could be wrong.

Its one of the reasons why these threads are popular svl.

Again, you are free to return to your own semi dead forum and govern your own set of rules there.
Let users here, included me, care about our own reputation, if we want to post wild speculations or not, even if they end up wrong. Nobody should feel bad if what they posted was wrong.
Nobody have a complete picture of the upcoming Maxwell chips, these threads are built up by using the rumors we come across, and getting more closer to the real deal as we go on. That is exactly what speculation is svl.

But go on, go enforce a boring policy on your own forum, where people can only post stuff about GPUs when they are first announced and officially launched. Atleast then you can sit comfortably in your own chair and knowing you didnt hurt your precious reputation.
But god what a boring forum it would be.

We will see. Like posted earlier in this thread, when the first codenames for Kepler cards was leaked on laptopvideo2go, it took 3 months until the products was launched by Nvidia

 

I`m not sure who this guy is, but he is hinting about GTX 980M coming to P150SM and P170SM
So the name might be GTX 980M for GM204 and not GTX 880MX

This is getting better each day

Source: Google Translate

 

He's just a random guy though isn't he? No inside sources? I suppose it doesn't matter too much what NVidia will call any possible flagship GM204, except if they do decide to call it a 980M it hints that it'll be sticking around for longer until the next flagship GPU - which isn't great news.

 

Random user would probably just spoil the details already since they have no NDA with Nvidia right? I don`t know.

Yeah I dont care what the GPU is called.
You could be right that since they go up a number to 900M series, it may signify that its here to stay for a while. But at the same time, it could also mean its quite powerful compared to 800M (880M)

 

From a marketing perspective a 980M sounds a lot more enticing than a 880M with an X plopped on the end. The closer we get to 2015, the more likely it'll be named as part of the 900M series.

 

or they can just do what they did with 680mx and 780m. same chips, slightly different performance, more cash$$ from rebrands

 

Maxwell isn't a re-brand, though? Even if there isn't a huge boost in performance, it still wouldn't be a re-brand. Maxwell is not Kepler.

 

no what I'm saying is that they could release the 880mx first with gm204 and make 980m a rebrand of 880mx with just higher clock speeds or no disabled SMM

 

I found a couple of mention about GTX 980M from chinese forums and a post from early July said the GTX 980M along with GT72 is coming in a couple of months.

Someone from Asus subforum said that G760 with GTX 980M is coming in the 10th month (October).

So I think September-October still is looking like the months we will see it on products. Which will be after when R9 M295X launches around August, which makes sense.

 

They may as well go straight to 900M with naming scheme. I don't really care. I just want performance!

So, what would Pascal be named, then? 1000M? Seems a bit ridiculous. They can't name have their names in the "thousands" like AMD. Noobs will be confused.

 

Well they were in the 1000s until they reached 9000 series and it became silly haha. So they will revert back to three digits and a new monkier after they exhaust their numbers! AMD did that too in 2004.

 

Well 28nm Maxwell being a 900 series means nVidia is going to keep it around for a while (aka milk it) before moving on, so I would tend to agree with Cakefish that it isn't good news.

Unless nVidia really does plan on skipping 20nm and going straight to 16nm Maxwell/Pascal. Otherwise 28nm Maxwell in Q4 2014 then 20nm Maxwell in Q2 2015 will just be another 880M style fiasco.