I'm going to predict the mobile version will see the same performance jump from 480M to 485M.
-
I'm just going by the numbers in that benchmark, assuming it's a legit 880 run.
If this was 20nm Maxwell I can a see a 580M to 680M style jump, but for 28nm Maxwell I would say 50% is the absolute upper limit. If I had to put down a number I'd say 35% over 780M. -
Killerinstinct Notebook Evangelist
But also factor in that maxwell was built ground up for laptops , and the phenomenal performance of the gtx 860M and 850M maxwell
I say 50% isn't that far fetch but we will see , all depends on power consumption and tdp
-
580 to 680 was almost a 100% increase. I wish we could see a doubling in performance but I'm seriously doubtful. I think we'll get a 15-30% increase with some lower temperatures that hopefully beg to be overclocked to show a worthwhile difference. I hope I'm wrong by the way and Cloudfires 780ti power level predictions some true, but I just don't see it happening.
I'm pretty confident the decent OCing part won't happen without some love from Svl7 and Johnksss. -
I know it seems a little desperate, but it's all we've got right now.
Also on a side note the price of the G750JZ dropped on amazon last night, granted it was only a few percent more, but the price has been rock solid for months until now and it was the official Asus reseller. -
We here have a GM204 beating the previous GK110 by 30%. GM204 which will be targeted around GK104 TDP.
We had GTX 680 (GK104) which was about 40% faster than GTX 680M (GK104).
If GTX 880MX is anything like that, we are looking at a 3DMark11 score around 13000-13500 (GPU score). In comparison, the GTX 880M score 8500.
Should be a pretty good upgrade over a very hot 880M while possibly having lower TDP -
Huh? Captain obvious has just made a suicide LOL.
-
Well if nvidia is gonna release a maxwell GTX 880m for laptops seems Scan will be replacing my keplar 880m cards because they just dont preform!
-
According to the latest posts here it doesnt look it was very obvious.
We are looking at a 50-60% increase in performance if everything goes as planned and the bench isnt fake.
Lots of If's but its certainly possible
-
thing is are they gonna bring out a maxwell 880m gpu? and how long are we gonna have to wait for it? i need one! lol
-
@Cloudfire, true. The shocker gonna be if that "Ultimate is coming" is not about Maxwell at all
Like in the past when we were expecting Maxwell 880M 3 months ago. Oh I believe some of us will make suicide in this case for sure... in Diablo or whatever game people are playing now
Cloudfire likes this. -
i can see my hardcore barb commiting suicide. and then i will reset my bf4 stats and delete my WoW chars.
-
Robbo99999 Notebook Prophet
Brace yourself - it was the shield tablet!:
SHIELD Tablet | NVIDIA
[snore!]
(this was the countdown timer for the 'Ultimate is coming thing')Cloudfire likes this. -
Killerinstinct Notebook Evangelist
Lol man I was hoping for more ... Well gamescon should make us happy , also the tablet isn't that bad , console mode might be interesting
-
Nvidia embarrasses themselves.
Or like they like to say: ULTIMATE EMBARRASSMENT. like it's meant to beMr Najsman likes this. -
And they did... I put too much faith in NVIDIA.
-
meh.. thanks nvidia :/
-
New driver support for upcoming Quadro (desktop) cards have been added to a Nvidia driver.
Quadro K4200/K5200 might be Maxwell GM204. Good chances about that.
LaptopVideo2Go: NVIDIA & Laptop News
They might be presented the first time during SIGGRAPH in mid August.
Nvidia have announced Quadro cards there in previous years, like GK110 last year
http://alienbabeltech.com/main/nvidias-first-fully-enabled-gk110-arrives-quadro-6000-packing-12gb-of-vram/
---------------------------------------------------------------------------------------
That was pretty obvious but I agree that they made way too much out of it "Ultimate coming". The background on that countdown was the Shield`s tablet speakers.
That other countdown toward mid August on the other hand, that one is looking very interesting
Killerinstinct likes this. -
Quadro K2200M which was mentioned on the first post of this thread:
INF v5.008 - News - LaptopVideo2Go Forums
Is a Mobile Quadro Card based on GM107 Maxwell (Same chip as GTX 860M and GTX 850M use)
Revealed by latest AIDA64 update
-
so were probably looking at october release if its actually coming out? or even later?
-
Who knows.
-
Killerinstinct Notebook Evangelist
Rumors point to September and august release
Cloudfire likes this. -
Nobody knows. I say sometime during August - October.
-
-
i just dont want to be left with a chip that is basically only good for keeping the room warm. stupid keplar 880!!!
reborn2003 and Cloudfire like this. -
thegreatsquare Notebook Deity
Of all the speculation, hints and leaks, I think the the AMD/Nvidia roadmap...
http://forum.notebookreview.com/att...-grafikchip-portolio-roadmap-10-juli-2014.png
...offers the most realistic expectations.
Upcoming Tonga/Maxwell will likely offer performance above HD 7950 [R9 280] and at best, may snag the bottom range of the GTX 780. That's the range I'm expecting. -
Yeah the GTX 880 does run pretty hot. Its squeezed to the upper limit on what you can shove in to a notebook.
Here is hoping GTX 880MX will have lower TDP which I think it will have
kevin_172 and Killerinstinct like this. -
Its been said a million times that there is something else that makes the chip hot, NOT the core clock as 780s running the same clock speed (and in most cases slightly higher voltage to achieve them) are running COOLER than 880M does! NVidia botched something with these cards that makes them hotter than their predecessors.
Sent from my HTC One_M8 using TapatalkTBoneSan likes this. -
I`m not saying you are wrong, but I guess you are comparing your Clevo machine against Alienware M18x/Alienware 18 again? Not exactly accurate.
But who knows, GTX 880M was just a milking cow anyway to recieve money while working on the GM204. Wouldnt surprise me that they paid less attention to binning this round, or did something on vbios level that does somewhat get the 880M to run at the sub 1GHz clocks, but is still unstable. Like I said, GTX 880M is pushing Kepler to the max of what you can put inside a notebook. Its a massive leap over GTX 680M which had 1344 cores @ 720 Mhz. You now have 1536 cores at almost 1GHz. Better silicon is one thing, but that extra strain on the chip gotta go somewhere.
This is why we need the new GM204 chips ASAP. The way I see it, GTX 680M was the only card from Kepler worth buying. GTX 780M was a little better in terms of high voltage memory and 1SMX more, but in my head barely worth it. GTX 880M was just pathetic.
So my plan for the upcoming Maxwell cards is get the GTX 880MX and sticking with it until Pascal is out. Unless they somehow push out 16nm FinFET on Maxwell which I highly doubt -
Actually the findings were by johnksss in his Alienware. He compared the two cards side by side in the same machine and the 880M was significantly hotter and more unstable than the 780M. Thats what started the uproar was finding that it's both hotter and less stable than the 780M at the same clock speeds and the 780M actually went higher in clock speed than the 880M.
The stability has gotten better but the heat has not. And the vbios throttle is just plain bad. I'm sorry but 888MHz on a card that is supposed to be 954MHz with a 993MHz boost is just plain broken.
Sent from my HTC One_M8 using TapatalkMr Najsman likes this. -
Robbo99999 Notebook Prophet
I followed this whole 880M situation since they launched, and read the reviews. I don't remember it being said or concluded that 880M runs hotter than 780M when at the same clocks & voltage - but yes there are definitely throttling problems with the 880M that aren't present in the 780M. (The 880M at stock is obviously hotter than the 780M at stock due to higher clocks & voltage.)
EDIT: actually, having said that, the 880M has more VRAM so it probably is gonna run just a little bit hotter actually, but don't remember any proof of this being shown. -
Johnksss said that it's very hot but it's not the core that is the problem, I believe he said that in the TI forums. I'm on my phone so I can't dig it up right now but I remember what he said because it made no sense and he never explained what he meant he just said that something makes the 880M hot and it isn't the core.
Sent from my HTC One_M8 using Tapatalk -
I seriously doubt that something like that will happen, most likely it will be based on the GM206 core and in the best case run cooler at high clock speeds, that puts it at around the same area as Tonga, I just don't see GM204 in a notebook at 28nm, they will have to cut it down badly and clock it really low, at which point it costs a lot more and doesn't perform much better than GM206.
-
Actually... Johnksss (as well as I) got better scores with the 880m vs 780m, as well as higher max clocks. Just as Johnksss I tested the 880m in the same system as I tested the 780m.
It benched really well in my experience, at least with the proper vbios, but the 780m is crap as well with the stock vbios.
As for games... no idea, but I know that Nvidia is releasing one crappy driver after the next, and that's been going on for ages already.
If your 880m runs at less than 954MHz at 3d clocks then it runs below specs and it's either a heat issue or a board power limit on your mobo. Or really crappy drivers of course.
Where exactly do we have a GM204 that beats a GK110? I don't see a GM204 anywhere at all.
Dream on. You say that the supposed 880mx is going to be more or less on the level of a 780Ti?
Seriously... as long as Nvidia is still using the same manufacturing process I don't see how this could be possible at all. GK110 is almost twice the size of a GK104, ever since the 480m fiasco Nvidia hasn't put a die monster in a mobile system, so I'm not gonna see that happen.
Also I don't see 28nm Maxwell to be that much superior to Kepler... Nvidia boasts with twice the performance per watt, now that doesn't mean that Maxwell is twice as fast as Kepler. I don't know how well the supposed GM204 can handle high clocks, but the limits of the silicon are still there. Of course the biggest limit will still be that green stuff agreement of Nvidia, aiming for lower power consumption instead of more raw processing power.
Either way, I sure hope we'll see a performance increase, it feels like ages since we first saw Kepler running in some systems. I just think that some estimations here are way blown up.
Also, that speculation stuff has been going on for ages now, and it's not like there is any more concrete information. We've seen people (including you
) getting some crazy numbers and dates out of nowhere... yet all that seems to be forgotten as soon as someone comes up with "new data". The first quote of you in this post is exactly what I mean.
What about GK210? I think the next desktop stuff we see is GK210. -
You are forgeting that you are talking about Maxwell, which is extremely efficient, and we don´t know the shader count of a GM 204 chip.
-
Killerinstinct Notebook Evangelist
Here is a bench of gtx 880 a couple of pages back , that's what cloud is basing his info on , but we don't know if its fake. Only time will tell
-
Well, since I'm blessed with two working eyes I've seen that post. Even if it wasn't fake and is in fact showing a 3dMark run of a GTX 880 it still doesn't mean that it's a GM204 chip - at all.
It's just so ridiculous jumping on every tiny and most far-fetched piece of information possible and then adding some "facts" if one feels like it... That's exactly what's happening here most of the time. -
Killerinstinct Notebook Evangelist
Lol oh good started to get worried here
, thought your eye sight started to fade
, but i just brought it up , thought u might have missed , which I found highly unlikely but you are right could be a different chip. we just assumed that it is maxwell since most rumors point to Q4/2014 launch of the 28nm maxwell. But you could be right , might be gk210 but why did someone post that in a maxwell thread. But again we will know at gamescon apparently or whenever they launch gtx 880.
EDIT: its fun to jump on the most far-fetched piece of info , its just like a cat with a laser pointer,Robbo99999 likes this. -
Guys don't forget that all we have is a benchmark score with a slight hint of 880 in there. We don't know if it's stock or highly overclocked, also there is the possibility of this being a GTX 880MX SLI, (for reference a 280X Crossfire scores 6300 in Exteme Total score), and of course we can't even be sure that it isn't fake.
-
What other chip than GM204 would it be? Another GK110? GK210? Hell no!
Its already confirmed that GM204 have been worked on for a long time now. Its in the cards (pun intended)
+/-10%, thats exactly what I`m saying. The GTX 680M was more or less on a GTX 580 level. The flagship from the Fermi series. It doesnt look like Maxwell is a less efficient architecture than Kepler, looks like its the other way around actually.
If the 3DMark bench posted a few pages back is legit, the GTX 880MX can be like 40% slower and it will end up inside that +/-10% window.
Then explain to me why:
GTX 660 (28nm - Kepler) beat a GTX 560 Ti (40nm - Fermi) by 30% and have the TDP reduced from 170W to 120W (-30%)
vs
GTX 750 Ti (28nm - Maxwell) beat a GTX 650 Ti (28nm - Kepler) by 25% and have the TDP reduced from 110W to 60W (-45%)
Nvidia got better results on the same process than they did with a new process. I think you are underestimating new architecture and what they can do with a process they have learned inside out for 3 years.
For one, the GM204 is looking like its roughly 100mm2 less in size than GTX 480M ("the fiasco").
Secondly, how do you know it was a failure because of the die size and not the architecture at that stage?
PER WATT, svl. That is the key here.
They can make a GTX 880MX twice as fast as GTX 680M and keep TDP on a reasonable size, or they can make another GTX 880M with very high temperatures. That is the theory.
They can go down in TDP, but use the +35% advantage per core over Kepler.
Sure, but they changed the architecture to a more efficient one. Currently if you look at the stock clocks on the GM107, it is already much higher than stock clocks on the GK107 when it launched. Not sure why you are so sceptical.
You may be right, you may be wrong. I still want to see why you feel why it shouldnt. Havent seen why yet just semi ranting. Like the first post on this thread where you focused on ultra low end Fermi that they would pop up in N16E instead of talking about the new cards in that series. Not that I dont appriciate your input but sometimes you act a bit strange
.
Gotta hype it up, hype it up.
What a boring thread it would be if I just posted "Here are new entries about N16E cards that are coming. That is all folks. Dont post anything more because we dont know anything until they are released..."
No this is a speculation thread. Speculations based on tiny fragments of pretty much anything regarding Maxwell is welcomed here. The roof is high, and nobody should feel bad about throwing out their opinions or expectations even if they are wrong. Discussions are always fun, especially about graphic cards that have a great impact on our computersMr Najsman, Robbo99999, heibk201 and 1 other person like this. -
I just really hope the people that are posting on this forum who own gaming machines do...well...gaming from time to time
Technology moves forward and then there are just plain limitations that occur. Everything is speculation until we get working pieces into the actual machines. Again, I have said this before but we need higher resolution screens in order to really justify all this extra speculated horsepower. I love technology but power is only as good as it's needed outside of paper benching.
I have the HTC One M8 and it will very soon be 'obsolete' once the new SoC gets released that doubles the performance. Unfortunately there are absolutely no games or apps that really take advantage of all this horsepower. The LG G3 is a great example of loading up the best specs to only be the 'biggest loser' in real world testing.
Now back to notebook gaming....until we get real tests, let's 'frag on' and see what actually materializes. If the 880mx or whatever it's called is released in a new AW 18 and the AW 18 is built EXACTLY how it is now...I would never upgrade. The machine is so locked down and gimped I would actually pay a ton more money to have a machine with a real screen, appropriate resolution and less restrictions coupled WITH software (games) that need it. I don't see that happening along with these cards just yet. When the notebook manufacturers actually upgrade other components and when games are released that necessitate an upgrade only then will I be looking into my piggy bank. Of course that would then beg the question of "But X is coming out in X period of time." Thus the speculation cycle continues.
Now my hypocritical Donkey is going back to my paused Crysis Blackmod experience
Cheers
-
This new SHIELD tablet seems a bit ridiculous to me.
You also have to have it connected to a power source to enjoy 1080p gameplay. So, you spend hundreds of dollars on a tablet to go wireless, yet you can't play wirelessly? That definitely puts "ultimate" in "The Ultimate is Coming." I think NVIDIA should focus on Maxwell (and other future gen technology) and stop screwing around with this crap.
NVIDIA making tablets is like a car company making R/C cars. It's just not right.daveh98, Cloudfire, TBoneSan and 1 other person like this. -
This was a pain to dig up on mobile...
The post where johnksss says that the heat from the 880M is not from the die.
http://forum.notebookreview.com/showthread.php?p=9707702
Overclock tests I didn't look terribly closely at but my understanding was that clock for clock the 880M was still performing slower than the 780M at the same speed. That may have been early driver related, who knows. Mine are fine with the modded vbios they're just hot as hell and the 2D clocks don't kick in which keeps my fans on all the time so I can't use the mod.
As for my card, I have no idea what is up with it. Modded vbios is 993MHz until it hits the thermal ceiling but stock vbios the second card drops to 888MHz after awhile with a max temp of 87C on auto fans and an average of 81C after the fan profile kicks in to around 70%
Sent from my HTC One_M8 using Tapatalk -
Gotta sell those Tegra chips somehow you know...
They are trying to spread out income on more things than just graphic cards. Not sure if you noticed but they also have began delivering chips to the car manufacturers to use in luxurious cars from Audi etc -
That's interesting. Didn't know that.
-
Rumor is that nVidia is inking a contract with Google for the next tablet processor... Would be an insanely big win for them if it happens.
Sent from my HTC One_M8 using Tapatalk -
Pretty much anything linked with Google is a "big thing" these days. Bunch of smart people working there.
I'm much more interested in seeing what Mark Zuckerberg does with Oculus technology. Forget tablet gaming, let's go virtual.
And now this has gone way off-topic. My bad.
-
Robbo99999 Notebook Prophet
Yeah, but gaming at 120fps is pretty good if you get the 3D capable screen - that needs some serious horsepower, even at 1080p.Cloudfire likes this.
Brace yourself: NEW MAXWELL CARDS INCOMING!
Discussion in 'Gaming (Software and Graphics Cards)' started by Cloudfire, Jul 14, 2014.