Gtx 485m?

Let's start making graphic cards guys.

 

If those numbers are true, then it's a hell of a card:
GTX 485M in Dirt 2:

Dirt 2 loves Nvidia cards, but still... the increase is bigger than anybody expected. 33% over the GTX 480M...

 

i hope its based on gf114 rather than gf110 according to wiki its 384 sps with 256 bit gddr5

 

There is not GF110 chip it could be based on. 480 shaders is just too much, even if you severely downclock them. 384 shaders for GTX 485M would be awesome, but even with a 100W TDP this is hard to pull off but seems like Nvidia did it. However, this would make the X7200 best laptop in the world without any competition from the R3, with 2x300W PSU and that kind of cooling you can run the GTX 485M at near desktop clocks...

Anyway, the 256 bit memory bus pretty much indicate that is has to be a 384 shader GF114. A GF110 chip with 256 bit memory bus would also have 384 shaders, so they would just use teh GF114.

Poor AMD, but if they go for 100W they should have an advantage as their design is still more efficient.

 

Beautiful !

It's pushing 12.000 GPU points in Vantage. That's amazing!

I want it! Too bad I can only upgrade next year .

 

generation G92b>>GF100 55nm>>40nm
285M>>470M increse about 35%

same generation same 40nm tec
470M>>485M increse about 45%
(vantage GPU 8561>>12166.5)

very doubtful
anyone has its gpuz spec?

 

All previous Nvidia GPUs just sucked because they did not use strong cards... the whole GTX 260/280/285M series was all about old G92 stuff. The GTX 480M was a bad joke, they ported one of the worst desktop cards ever made. GTX 460M and GTX 470M are still nothing compared to their desktop counterparts (192 shader / 288 shaders). I don't blame Nvidia because they did not have to bring anything that had power. They probably could have made a mobile card with 336 shaders already just as they could have made a Desktop GTX 460 with 384 shaders already, but there was no need to do so, because AMD was no bis threat, certainly not in the mobile market. Now Nvidia seems to be worried about Barts hitting the mobile market so they are finally releasing some 334/384 shader mobile card which (as we know from the desktop world) kicks.

It all makes sense and there isnt really any other chip that they could release. The new fermi surprised us all, look at the GTX 570 vs. GTX 480. I have no idea what Nvidia improved in their design, but they did a hell of a job, it's almost as if they intentionally screwed Fermi up to surprise everyone with the revised Fermi.

 

Why not just ask 許崑泰 direct?

 

he doesnt care notebook research, he cares about the bainaohui around china
ask 黃仁勳

 

I bet he has no clue about mobile cards ;-) They are developed by some crazy underground department.

 

oh! forgot 鄭春森
CLEVO/Field Application Engineer
James Cheng
E-Mail: no good
@@

 

er...... is it good to post personal email here?

 

Early numbers are showing AMD HD6900M winning in 3DMark06 so far.

 

We are still using 06 to evaluate new cards ?!

 

Most games are and will continue to be DirectX 9.0, so I guess that's why 3dmark06 still makes sense.

I personally prefer Vantage.

 

When it's the only measuring stick we currently have, then yes we are still using it.:yes:

 

disappointed at that die size ............480m n11e gtx remains the biggest .........ooooops gtx280 is the biggest
how can 1.95 billion transistors 485m fight 3.2 billion transistors 480m

 

That would be because the n11e is a FULL GF100 chip haha, it is a monster!

 

Only stick?

What about 3dM Vantage? or 11?

 

I think it's sort of a GTX 560 Desktop core by the looks of it.

 

If you've got those numbers post them.

 

I guess it's alright if you paraphrase it.

 

Sorry, was thinking abt the 485, we may well not have anything else for the 6900

 

We all know that AMD cards have an advantage in synthetic benchmarks, especially in the older ones like 3DMark06. If the 6900M is not winning by a fair margin, that it will mean nothing in games... It's no shame that the new fermi hit them by surprise, it will make them try harder on the 7000 series and that's a great thing. Besides, they will be withing 5% but at a much better price point and with Nvidias ridiculous pricing it still makes more sense to buy AMD cards, I'm wasnt trying to come off as a Nvidia fan boy.

 

You'll bury the HD6900M in a shallow grave because of a single game benchmark that you admit favors Nvidia but then defend a synthetic benchmark as favoring AMD to support Nvidia. If want to try and not sound like an Nvidia fanboy don't use the same kind of arguments that they would.

Also you might want to look at the specs of the desktop GTX 460 SE before you start believing a 256-bit bus confirms GTX 485M must be a full 384 core GF114 part.

 

Your right about the memory bus, but what else other than a 334 or 384 shader chip could they bring out?

I also don't want to bury the 6000M series, I'm just psyched about finally seeing some mobile Nvidia card worth buying and a bit disappointed with the desktop 6000 series. My post was just an expression of that. It should be clear to anyone that without proper benches you cannot make any educated statement about which card is better, but that wasn't my intention at this point. Nevertheless, the argument has some substance if the benchmark is true. I was talking about a fair margin. Of course Dirt 2 favors Nvidia cards, but 55% over the Mobility Radeon HD 5870 is a lot never the less even if you deduct 10% due to better support.

 

Like I mentioned in the other thread, Notebookcheck.com's scores aren't reliable because their testing methods and numbers are all over the place.


For example: Notebookcheck's chart show a 22% difference between the GTX 480M and the Mob. HD5870 while the Tom's Hardware review of the GTX 480M in the W880cu puts it less than 3% faster (at "Ultra" settings) than the Mob. HD5870 in the W860cu.

 

Yeap, when you read notebookcheck benches you always have to check what system they used, what CPU, etc. If you want to get a picture of what is going on. On the other hand, the do have the biggest database of notebook benchmarks on the internet.

 

It is a big database but it is mostly useless unfortunately because their test methodology is much less than ideal

They are presenting these scores and making comparisons as if the ONLY variable is the gpu itself which we obviously know is not the case

 

GTX 470M => GF104 / 288SP + 1,5GB /192Bit + 535/1070/1500MHz

GTX 480M => GF100 / 352SP + 2,0GB / 256Bit + 425/850/1200MHz

I think the 485M will be this:

GTX 485M => GF104 / 336SP + 2,0GB / 256Bit + ~550/1100/1300MHz

It will be a combination of 480M (~shader/memory) and 470M (~clocks) by using the efficiently GF104 (or GF114) core.

 

A full GF104 would be NICE! Then that would essentially be a downclocked full desktop GTX 460 1GB (with twice the vram)

As it is, the GTX 470m is a downclocked GTX 460 768MB. A $170 desktop card.....or a $600 laptop card! lol

 

SO they test the cards with different specs?.. lol finally I realized why everyone says they are not reliable.

After putting a computer together with a single 460 and seeing its performance, I only wonder when are notebook users going to have such price to performance ratio lol.. We are burning wallets to get our notebooks close to one desktop 460M..

 

That is the world of high end notebooks.....paying 3-4x the price to get comparable desktop performance

My desktop and laptop are within $100 of each other....here are the desktop specs though (I got very good deals on both the laptop and the desktop....)

Corsair 700D
Watercooled Rampage II Extreme
Watercooled Core i7 Extreme 965
6GB DDR3 1600 Dominator
2x Watercooled Radeon 5870 CrossFireX
GeForce 8800GTS 512MB for PhysX
2x SandForce SF-1222 64GB RAID 0
WD6400AAKS
Corsair AX1200

 

A downclocked GTX 460 wouldn't excite me one bit, unless the 485M goes for $450.

I'm not buying the 12,000 Vantage GPU score either. There's no way that Nvidia's current technology can find a 30% jump in performance over the 480M.

 

fixed for ya

 

It's possible. The GTX 485M could be based off the desktop GTX 560 die with 384 SPs and a slightly more efficient architecture.

 

Well the GTX 560 is 180W, and a 12K GPU score sits right between the GTX 285 and GTX 460. A full bore 6850 is "only" getting ~14K.

Don't get me wrong. I want it to be true, but it's hard to believe, for now. I really hope Nvidia silences my pessimism.

I wonder how much TDP falls off, if the GTX 560 core is downclocked from 820MHz -> 500MHz.

 

If GTX 485M were using the GF114 GPU it would have gotten a 500M series model number, just like the desktop GTX 560, GTX 570 and GTX 580.

Nvidia releasing the GPU as a 485M sounds a lot to me like it's still GF104.

 

You act like they are not notorious for rebranding old tech haha....(G92 GTX 285m?!)

 

thinking about GF104 or GF114

lets compare vantage GPU (all factory stock)
980X + GTX460 1GB 12482
2920XM + GTX485M 2GB 12166
very very close

but how about the tdp?
GTX460 1GB 6+6pin 160W
GTX485M?100W???can 100W hold 2GB and without downclock????
so it may be "featured" GF104 or GF114

480m N11E-GTX-A3
470m N11E-GTS-A1
485m N12E-GTX-A1

 

The manual for the PM15x0HM/P170HM indicates that only the P170HM which has a 220W PSU supports the GTX 470M. The GTX 485M is supported in both PM150HM (180W PSU) and P170HM. So it sounds like the GTX 485M has a lower TDP than the GTX 470M, but how would that be possible?

 

Maybe it's related to the CPU that can be used in those models.

 

But the i7-2920XM can be used in both models...

 

No, I use Nvidia's infamy for rebranding old tech as evidence to support my claim that the GTX 485M is still GF104. If Nvidia was so free and willing to slap a new 285M number on the already old G92 GPU why would they then hesitate to number a new respun GF114 GPU as a GTX 580M? Especially after the 500M series cards have already started showing up.

 

so that may be the strongest GF104
bcz 485m scored nearly the same with gtx460 1gb! which requires a dual 6pin and 160w tdp

 

You and kaltmond always seem to know more than you post.

Give us the goods.

 

Or maybe that 12K vantage score for the GTX 485M is junk....along with that Dirt2 benchmark that is now removed from Notebookcheck's GTX 485M page.

A 75W TDP card can hold 2GB modules since they're 40nm and use less power and no more chips than 50nm 1GB modules. Even downclocked the GTX 485M's 256-bit bus could still produce a higher memory bandwidth then the GTX 470M's 192-bit bus.

 

yes, very doubtful......and 3dmark06 removed too...........NDA!?
a gf104 mobile low tdp card with twice amount ddr5 can score like a desktop 160w gf104
that's..................= =

 

This hype train is looking to end in a derail. Slow down peeps.

Anyone expecting on par performance with a GTX 460 is flat out tripping.

 

GOOGLE:
ok....google knows everything.............