Finally, Nvidia announce GTX 700M series for mobile computers

I`m not sure if Sweclockers ( Source) are allowed to do post this now since they are the only sites that have posted the slides, plus the GPUs can not be found on geforce.com
Anyhow,

GTX 780M 31% faster than GTX 680M according to Nvidia. Seems pretty accurate according to what I have seen.
Official reviews from all the reputable reviewer sites coming when Haswell is announced (early next week) :thumbsup:

 

765m was confirmed to be 600MHz 192-bit with 700MHz GDDR5 not what is listed there. Unless there's two versions of the 765m... which wouldn't be unheard of from nVidia

Source: http://forum.notebookreview.com/sager-clevo/705795-w230st-13-weve-been-waiting-57.html#post9205820
GeForce GTX 765M | VideoCardz.com

If you compare:

GPU Core: 600MHz / 800MHz
Shaders: 768 / 768
Bus Width: 192-bit / 128-bit
vRAM Clock: 700MHz / 1000MHz

The performance would likely be about the same then, right?

And how many ROPs and TMUs?

 

According to Meaker, NVIDIA lies ( http://forum.notebookreview.com/ali...news-06-10-2013-save-date-22.html#post9211970) about the percentages of performance increase over the previous generation, so I don't know what to believe...

 

Who have confirmed 192bit in that thread? All I can see is 700Mhz for the GPU, but you have to consider that it is specifically clocked by Clevo for that tiny 13". As for all Nvidia mobile GPUs, OEMs are free to make their own versions with whatever clock they like. GTX 765M inside a 15" will be higher clocked, like the specifications

Pretty much every result we have seen so far have shown about 30% in average over the GTX 680M. So this time I actually believe them

 

That AMD/NVIDIA market share line graph made me laugh.

 

Read through the thread for a few pages, and note my comparison above. There could be two versions. Not to mention I've received a couple notes from people inside that have confirmed just that. But as always I'm skeptical until the product is actually released.

Plus this little nugget that started it all: http://forum.notebookreview.com/sager-clevo/705795-w230st-13-weve-been-waiting-28.html#post9163533

If it's no less than a 760m, then 128-bit 700MHz GDDR5 and core at 600MHz would be significantly slower than the 760m. 192-bit GDDR5 would boost that performance considerably over the 760m.

 

I dont think AMD is laughing...

-----------------------------------------------------------------

This is a real candidate for my next notebook with 780M. Looks sweet, plus it have option to use 120Hz screen and its built with aluminum. Combine that with the efficient and not noisy cooling system from Asus, I might have a winner here.

Look here HTWingNut. Notice that they write 2GB GTX 765M. That means it IS either 128bit or 256bit. We can rule 256bit out.


Even Clevo write this in their specifications for the Dreambook ( source)

So yeah, the specifications in the first post for GTX 765M seem to match. It is 128bit I`m afraid. Now there might be a 192bit version coming out as well, since there exist both 128bit and 192bit GK106. But that remains to be seen. I have seen nothing that confirms 192bit for GTX 765M (1.5/3GB listed)

 

Sure they are. It will change a lot this year.

 

GTX 660 is 192-bit with 2GB GDDR5... it can work.

Specifications | GeForce



I'm thinking Clevo is offering a 192-bit with the W230ST. 192-bit version could even come with 3GB GDDR5 and 128-bit with faster 2GB.

 

There's been AMD techs posting on industrial-oriented forums regarding the 8970m and 8990m cards.
8970m was not initially supposed to be a refresh, but a stronger chip in its own right, and 8990m was supposed to be a heck of a lot better than that, but with nothing extremely impressive from the nVidia side this year other than the 780m, they refreshed the 8970m and will probably be selling the initial 8970m as a 8990m.
Or they'll be skipping all this entirely and go Volcanic. If they do that enough months before Maxwell is out (like the 7970m release), coupled with the decent drivers (and the drivers have been getting much better), we might be seeing some goodies from the red team.

 

I've got some hope for AMD this year. Not only will they be breaking new records with their mobile GPU lineup, they're entering a new market entirely: gaming consoles. AMD is poviding Sony with processors for the new and improved PS4, hence their stock has doubled since Q1 this year. Too bad I'm not rich, 'cause I'd be much richer today.

 

i'm disappointed in 770m, its 192-bit? so 675mx will be faster as its 256-bit? :/

 

They also landed the Xbox One. Depending on margins for those chips it's a win. They had the Xbox last generation as well, but the real win is they scored the entire APU architecture.

 

I'm not so sure. While winning the ps4 and xflop one does give them a revenue stream (and a great presentation point for investors), the profit margins are razor thin. In fact, they'll likely be delivering APUs and GPUs at a net loss for the first year+ until manufacturing costs come down.

 

The GTX770M will beat GTX 675MX. Its 96GB/s vs 115GB/s, but on the other hand the 675MX is atleast 200+MHz clocked higher. And you might get memory that you can overclock quite a bit

 

My question exactly. Where is the replacement for my 675mx? Seeing this makes me glad I went the route I did with the 675. I didnt want to spend top $ for the 680, and it appears (correct me if i am wrong) that it will still be the 2nd most powerful card in the new 700 series lineup.

 

GTX 775M...
It may be 960 core with 256bit or it is 1344 core with 256bit. Its pretty much obvious that it will be GK104

 

its probably going to be 1344 CUDA cores unless ofcourse they are planning another winter release like they did with 675mx because the gap between 775m and 780m will be too large otherwise.

i have a small question, the driver updates we receive in the next year, will they also improve 600m series for the games or just 700m, or both but 700m having the major impact?

 

All the rumors can now come to an end. NOtebook check just released benchmarks and other game tests of 780, 770 and 765 in the upcoming CLEVO notebooks.

Im Test: Nvidia GeForce GTX 780M, GTX 770M & GTX 765M - Notebookcheck.com Tests

It's in german so use translate..

 

English Version of 7XXM Review!

 

Finally. Lets find out what the GTX 780M can do against the other 2 top GPUs, 7970M and GTX 680M
Im Test: Nvidia GeForce GTX 780M, GTX 770M & GTX 765M - Notebookcheck.com Tests

All games they tested. Benchmarks not included.

BioShock Infinite
GTX 780M: 46.6FPS
GTX 680M: 46.6FPS (0%, I call BS)
7970M: 43.7FPS (+6.6%)

SimCity
GTX 780M: 39.9FPS
GTX 680M: 32.1FPS (+24.3%)
7970M: 30.7FPS (+30%)

Tomb Raider
GTX 780M: 65.3FPS
GTX 680M: 32.2FPS (+102%)
7970M: 46FPS (+41.9%)

Crysis 3
GTX 780M: 29.3FPS
GTX 680M: 21.8FPS (+34.4%)
7970M: 20.3FPS (+44.3%)

Dead Space 3
GTX 780M: 144.2 FPS
GTX 680M: 111.1FPS (+29.8%)
7970M: 92.3FPS (+56.2%)

Far Cry 3:
GTX 780M: 30.4FPS
GTX 680M: 23.6FPS (+28.8%)
7970M: 25.4FPS (+19.7%)

Assassin Creed 3
GTX 780M: 42.6FPS
GTX 680M: 34.1FPS (+24.9%)
7970M: 24.7FPS (+72.5%)

Hitman
GTX 780M: 33.8FPS
GTX 680M: 22.6FPS (+49.6%)
7970M: 28.8FPS (+17.4%)

Black Ops 2
GTX 780M: 95.1FPS
GTX 680M: 75.3FPS (+26.3%)
7970M: 75FPS (+26.8FPS)

Need For Speed
GTX 780M: 59.9FPS
GTX 680M: 51.4FPS (+16.5%)
7970M: 48.4FPS (+23.8%)

Medal Of Honor:
GTX 780M: 54.8FPS
GTX 680M: 40.9FPS (+34%)
7970M: 46.2FPS (+18.6%)

Dishonored
GTX 780M: 125FPS
GTX 680M: 115.7FPS (+8%)
7970M: 99.9FPS (+25%)

Fifa 13
GTX 780M: 307.7FPS (lol)
GTX 680M: 256.9FPS (+19.8%)
7970M: 226.6FPS (+35.8%)

Borderlands 2
GTX 780M: 83.3FPS
GTX 680M: 71.1FPS (+17.1%)
7970M: 62.9FPS (+32.4%)

F1 2012
GTX 780M: 100FPS
GTX 680M: 87FPS (+15%)
7970M: Not included

Guild Wars 2
GTX 780M: 41.2FPS
GTX 680M: 31.9FPS (+29.1%)
7970M: 30.6FPS (+34.6%)

Counter Strike
GTX 780M: 194FPS
GTX 680M: 128FPS (+51.6%)
7970M: 153.5FPS (+26.5%)

Sleeping Dogs
GTX 780M: 30.8FPS
GTX 680M: 22.7FPS (+35.7%)
7970M: 28.5FPS (+8%)

Darksiders 2
GTX 780M: 136.9FPS
GTX 680M: 108.8FPS (+25.8%)
7970M: 104.1FPS (+31.5%)

Max Payne 3
GTX 780M: 54.5FPS
GTX 680M: 42.6FPS (+27.9%)
7970M: No data

Dirt Showdown
GTX 780M: 66.4FPS
GTX 680M: 52.9FPS (+25.5%)
7970M: 48.7FPS (+36.3%)

Diablo 3
GTX 780M: 188FPS
GTX 680M: 159.6FPS (+19.8%)
7970M: 100.8FPS (+86.5%)

Risen 2
GTX 780M: 53.8FPS
GTX 680M: 42.2FPS (+27.5%)
7970M: 43.1FPS (+24.8%)

MW3
GTX 780M: 117.3FPS
GTX 680M: 102.8FPS (+14.1%)
7970M: 90.7FPS (+29.3%)

Alan Wake:
GTX 780M: 54.1FPS
GTX 680M: 41.9FPS (+29.1%)
7970M: 44.7FPS (+21%)

TES: Skyrim
GTX 780M: 68.3FPS
GTX 680M: 54.2FPS (+26%)
7970M: 55.6FPS (+22.8%)

Anno 2070
GTX 780M: 72.6FPS
GTX 680M: 51FPS (+42.4%)
7970M: 55.4FPS (+31%)

Battlefield 3:
GTX 780M: 50FPS
GTX 680M: 34.7FPS (+44.1%)
7970M: 34.7FPS (+44.1%)

Total average difference:

GTX 780M is 29.6% faster than GTX 680M

GTX 780M is 32.6% faster than 7970M

Spoiler

GTX 680M: 829.1/28 = 29.6% 7970M: 847.4.4/26 = 32.6

 

Good info! 780m ~ 22% on average from 680m at stock according to notebookcheck. However 780m seems to be double the FPS in Tomb Raider. Can't explain that one, unless there's specific profiles they concentrated on for the 780m that may filter down to the other cards, but still double the performance? I think they botched the details in the game with that one.

765m isn't as great as I thought it would be, but then again, all those are run at 1080p ultra detail, so hard to judge.

 

Nope, GTX 780M stock is 29.6% faster than GTX 680M stock in games. Look at what I just calculated.
Notebookcheck included benchmarks in their average ratings.

 

Not impressed GTX 680M falls 10%-23% behind GTX 780M . The only two games where GTX 780M shines are Anno 2070 BF3, CS Go ,and hitman where there is increase 30% . We have remember that this is GTX 680M not GTX 680MX which will be likely trail 780M a little bit. So we can conclude that if you have 680M or 680MX you don't need to spend $$$ unless if you want to .

 

Look at bottom of review, they show -22% from 780m. Somebody's calculations are wrong then.

 

You`re welcome to go over the numbers I posted. I didn`t include the first game, Bioshock, since according to NBR GTX 780M get the same FPS as 680M

 

If it's identical then it won't affect the outcome. I was just saying, Notebookcheck showed a value you showed a different one. Something isn't right. I would also throw out the Tomb Raider because there is no way it's twice the performance, and throw out the ones that are 100FPS plus because it's really a moot point.

 

Might want to look over your numbers again... some of them are way off (lol f1 2012). I'd do the calculations, but I'm on a phone right now

 

I just updated the calculatons and included Bioshock too. So all games are included (You can`t just disregard games over 100FPS since its just as valid comparison).
My calculations is 100% correct.

Nope they are correct.

100FPS/87FPS = 1.15 = 15%
100FPS/46FPS = 2.17 = 217%

EDIT: Math fail for me ;(

 

Here's what I get, note equation at top:

 

...That's for the 670mx. The 7970m wasn't even in that benchmark.

And 100fps obviously isn't faster 217% faster than 46fps. Look at your equation right above it.

 

For the HD7970m, it is not 217%, it is 117%. A common mistake when handling %. If you are normalizing by HD7970m, getting twice the performance is a 100% increase. 200% is triple.

 

BAH, my brain hurts.

Ok, so 202% is 102%, 217% is 117%. Too late for me to figure this out but I`m relying that what you guys say is correct.

Calculation updated, again. Still different than what HT get, but thats my final answer lol

 

Excel wouldn't lie to me, I think...

 

We need someone unpartial to go over the numbers we posted. I still think mine is correct. If not I blame that its 2.30AM here

 

I think Excel is quite impartial, but it looks like R3d is willing.

 

Lol fine.. finding my ti89 right now...

edit: hold up, gotta clean off my desk first

 

I don`t trust computers. I rely on old calculators that runs on the solar energy

Excel probably right though, but Im sure I did it right this time

 

Solar Calculator
Excel
Ti-89

Hmm.

 

100FPS/46FPS = 2.17 = 117% but you knew that, right?

 

No I actually didn`t. Even google sided with me. I thought.

 

Just use Excel or GoogleDocs or whatever, that's what spreadsheets are designed to do.

And there's a difference though. It's 117% faster, but 217% times the performance.

 

I got what cloud got. (BTW cloud you made a minor mistake diablo 3, should be +17.8% when compared to the 680m)

I injured my hand so my handwriting is pretty bad right now.



HTWingNut, you're not supposed to average the fps, because that "dilutes" the advantage. You need to average the % differences to get the average % difference.

 

It depends on how you want to look at it. Average FPS % difference or average of the percentages difference. It doesn't really matter because you're diluting it all anyhow with an average. And I don't know how or why you guys insist on doing it by hand. Takes all of 60 seconds to calculate in a spreadsheet. This is a tech forum after all.

Take out Bioshock it goes over 30%, take out Bioshock and Tomb Raider it goes down to 28%, and those are the two in question in my mind. In any case 25-30%.

 

i dont understand the jump in tomb raider? and that 0% change in bioshock?
you can do all the calculations but in the end some games will perform faster then 30% and some will only get 10%..


by the way i had this question got lost in page 2 LOL

 

Captmario - you are correct. It's roughly a 25-30% improvement. All depends on the game and drivers.

Driver updates will likely affect both since they're the same architecture. Unless they do something unexpected with Boost 2.0.

 

Man, you are either wrong in the first example or in the second And pick yourself which one.
If you want to see the advantage one over another then you are looking the percentage level of increase aka 15% and 117% where your 2nd example is wrong.
But if you want to visualize one in the amount of another then you are looking for in how many times did it increased which is 1.15 (or 115%) and 2.17 (217%) times more and then your first example was wrong.
Pick yourself which is wrong

 

Haswell probably gives it 2-3% as well

 

I come away from this most impressed by the performance per dollar of the 7970M

 

Im Test: Nvidia GeForce GTX 780M, GTX 770M & GTX 765M - Notebookcheck.com Tests

I'm puzzled with this notebookcheck review about 700M series, they don't even mention the cpu, we just know it is a Haswell one...Nothing about power consumption...

That said the cpu does not seem to limit the 780M so it seems accurate.