Pascal MXM GPU Upgrades

I would get two into P870DM. Instantly.

 

Maybe... only if it overclocked as good or better than the NVIDIA counterpart and CrossFried worked correctly. I have zero interest in single GPU upgrades. I'm not taking any more chances wasting my hard earned money on garbage. And, I don't give a rat's butt if they are equal stock if the AMD GPU sucks at overclocking. That would be 100% absolutely worthless to me.

I'd want to see some hard proof in the hands of a few talented overclockers as well. No theoretical mumbo-jumbo based on some muppet's review on the internet. I'm talking multiple extreme OC benchmark tests in 3DMark 11, Fire Strike, Sky Diver, Time Spy and Vantage. None of the stupid bar graphs showing FPS in game comparison garbage. Don't get me wrong... those are nice to have as supporting information, but they don't mean jack without the extreme OC synthetic benchmarks. Those are what matter most to me. That is what separates the men from boys.

Assuming all the requirements are met, absolutely... I'd be an idiot to refuse such a blessing. But, I'm not lowering my standards and accepting less from AMD as a means of punishing NVIDIA. NVIDIA deserves to be punished, but not at our expense. Best way to punish NVIDIA is not buy their overpriced trash and keep your old hardware. If that old hardware runs as good or better than what AMD brings to the table it would be stupid to spend any money on new parts.

 

I kinda don't agree with measuring synthethic bench, because that doesn't give the consumer full detail on overall quality and power from the gpu. Driver support provides a big chunk of user experience in games too.


Also, Something more important that's often overlooked is the fps 0.1% low (see gamernexus video on that) to indicate sli/cf smooth implementation on top of multi gpu scaling.

 

Well, I really don't care about how much power it draws as long as I can give it as much as it needs. 200W, 500W, 1000W... whatever. As long as I have enough, everything is peachy.

I do care about quality, especially where it relates to durability, but we cannot extrapolate that based on test results. Only time will tell on that aspect.

For me gaming is secondary. Nice, but not essential. I can disable CF or SLI for playing games if the title is a coding abortion, and my 200W GTX 980 is already off the hook awesome for gaming. It would be idiotic for me to even think about spending more money on a GPU upgrade to play games, LOL. Single 980M is enough for that, and 980M SLI is beyond being more than enough for that. I have zero interest in VR, so there's that aspect also. Single 200W 980 is also plenty of horsepower for VR, but you'd never guess that now that NVGREEDIA has a shiny new object for them to make money from selling. They would like for everyone to believe they need a Pascal GPU to have a good VR experience, and that's marketing baloney. Unless, of course, they do some kind of extra proprietary witchcraft to implement an artificial blockage for VR and I definitely would not put that kind of shenanigan from the Green Goblin as a far fetched possibility. We already know they are as crooked as a dog's hind leg... no question they will stoop low and slither wherever they need to as often as they can identify a new stream of cash to extract more money from fanboys.

All I really care about for the most part is overclocked benching. Unless something changes dramatically, AMD does not offer the right tools for the job. If gaming were job #1 there would be no reason for me to care about Pascal... just expensive fluff and a big to do about nothing unless a person is limping along with something very old and inadequate.

 

If you wonder why I hate GameSux so much - it's a BlackBox (like all the proprietary software is, but this one's owned by nGREEDIA). No one really knows what's inside and how it works. The devs use the APIs provided by nGREEDIA, but the underlying implementation is unknown. If nGREEDIA's guys did their abstraction right, they can change the whole thing's inner-workings and no one would know a thing. Why the new GPUs just happened to perform better? You tell me. How long your old hardware would be better than whatever AMD? Again you tell me. Do you remember how much faster 780m was against 7970m/8970m? Do you know where that gap is now? You are free to chase the numbers, there's nothing wrong in that. You can buy every new hardware and beat whatever the hell you set before, there's also nothing wrong with that. BUT I see nothing wrong in getting a GPU and expect it to last me till I save-up for a new and NO ONE is forcing me to replace it sooner than that.

The current state of the "enthusiast" (in quotes, since it's almost gone, let's what the poll and the thread would result into) laptop market is sadder than a comedian at the Oscars.

 

I agree with just about everything you said here. 680M, 780M and 7970M are pretty close stock with a stock NVIDIA vBIOS. With a modded vBIOS and overclocking, the 680M gives it a spanking and 780M totally trashes 7970M. The gap between AMD and NVIDIA usually widens greatly once the GeForce demons are exorcised from the firmware. NVIDIA is their own worst enemy when it comes to firmware cancer. GeForce Experience is a joke. I use the @j95 approach to drivers and all but five folders are deleted from the driver package, leaving only the useful elements. All the worthless bloatware and spyware back-end trash gets dumped prior to installing the drivers.

You're being very smart about things, but that is not the norm. If you're not an overclocker/bencher there is even less of a reason to feel pressure to buy new GPUs. You are the proverbial nightmare customer for NVIDIA because you are not moved or impressed by hype and hoopla, and that's one of the things that makes you awesome. If more gamer-boys would just knock it off with the emo crap and stop freaking out about GPU upgrades for playing games when upgrades are not necessary NVIDIA would be more inclined to watch their P's and Q's. They are exploiting the emotional element that is associated with gaming and getting filthy rich in the process. This inconsistency is something I find humorous. They like to say synthetic benchmarks don't matter, but they like to run them stock or with a mild overclock, and then make lames excuses for the physics performance being a reflection of a wimpy CPU. They say it doesn't matter for gaming, but to a fairly large extent the same can be said of their shiny new GPU.

I'm angry about the manipulation and proprietary filth designed for the sole purpose of defrauding customers. That's something that is actually worth getting emotional about because it is a corrupt approach to doing business. I do want more horsepower for overclock benching. The only reason a gamer would actually need more horsepower than a 980M with a cancer-free vBIOS can deliver would be for 4K gaming or for NVIDIA CrapWorks bloatware that creates an artificially high demand using jacked up graphics settings.

 

What's next? Everything will be soldered? Ram, hdd, ssd, keyboard, keyboard keys, power switch, physically psu is soldered, LCD fully soldered. Fans riveted to laptop.

Sent from my ASUS_Z00AD using Tapatalk

 

O wait silly me. I nearly described an apple laptop.





Sent from my ASUS_Z00AD using Tapatalk

 

Yeah, everything is all goofed up everywhere we look. Apple wants Macs to be more like PC/Micro$haft and Micro$haft wants to make PCs more like a Mac. And, too many of the customers in both camps are not thrilled about it. Oil and water don't mix and trying to do that just creates a screwed up mess that nobody likes.

 

Don't forget your wallet being soldered to the company you buy that laptop from

 

I think universally what people who want mac laptops want is to have is an option for higher performance hardware. People whose sole goal isn't battery life or thickness.

I would be tempted greatly if I could get a MBP with a 1070 or 1080.




Sent from my iPhone using Tapatalk

 

Jony Ive not gonna let that happen

 

you'd be arrested for having a thermonuclear device

 

It would also be very useful as a space heater. Just load up GPU-Z rendering test and you've got a very fancy hand and lap warmer for the winter. Using Furmark, you could even roast marshmallows or grille hotdogs.

 

Mr. Fox is inviting to the annual Furmark-BBQ...I like that!

 

Its funny you say that. My friend had a 2011 MBP with the notoriously hot Radeon 6770M, managed to get a 1st degree burn from putting it on his lap during a gaming session. it gets NASTY hot with any sort of dGPU usage.

 

Charred silicon has a nice crispy texture to it and tastes like chicken. I prefer a spicy dry rub over sticky sauce.

 

you mean with the stupid T heatsink?



edit: hey @Papusan

this stupid T heatsink actually covers the PCH

 

Somehow in Apple's infinite engineering wisdom, they tought a double pipe heatsink shared by the CPU and GPU was a bright idea....but at least they incorporated a flat level portion on the heatsink to mate with the die. Razer apparently things that too much money to spend for cooling, so they decide to mate the copper piping directly over the GPU/CPU die without any sort of single copper plate shim to make a decent mating surface. Whoever makes these decisions should be fired on the spot.

 

That probably uses quantum heat displacement fields.

 

That post gave me cancer. How can anybody deny the ridiculous flaws in the Blade cooling system? It's beyond obscene for a computer priced that high to be so poorly designed internally

 

My cancer got cancer.

But in all seriousness. Its outrageous how he out rightly just denies that there is an issue, and they use "proprietary" way of applying thermal paste. Lol wut?

 

OK, just stop and think about it. This should come as no surprise. Look at the machine they are talking about. You cannot take their discussion or the product they are discussing seriously. One look at how it's made and it's obvious the engineers were not playing with a full deck. Granted, they didn't have much to work with trying to cram 2 pounds of sand into a half pound package.

Very similar conceptually to the MBP. The burned plenty of calories on aesthetic pleasantries. Fancy schmancy... all show and no go... turdbook city. How much you wanna bet the expert that claims they use a special process that others cannot replicate or improve upon thinks his own poop doesn't stink?

 

The only way I'd believe them is if they soldered the heatsink to the chips (which if you're going for mbp style, and they are), then I'd call that proprietary. But they don't and that would be one more component you couldn't replace...

Sent from my SM-G900P using Tapatalk

 

off topic

while a lot of laptops nowadays have permanent optimus (it's like being born fully-disabled, poor baby...), people say it's fine to do movie, etc on iGPU while the 3D is offloaded to dGPU

no, no it isn't



this is customized MPC-HC+ madVR running a 1440x1080 video file with uber-upsampling and various other quality enhancements ON A GTX 1080 DESKTOP CARD

quality video playback takes power you know....

 

Off topic but damn...they call this engineering???



Thats how they mate the surfaces. Shared double pipe heatsink without a proper copper "shim" to remove air bubbles. there are visible gaps in the pipes, they expect TIM to handle that gap? GTFO Razer, back to the drawing board with you.

 

Razer makes total crap. I have the misfortune of owning their orbweaver and have the widespread problem of the rubber palm rest glue failing. All they would need to do is use a better glue but no, for $140 they refuse to do that. They denied warranty service and I told them I would never purchase another Razer product again and I won't.

 

the first gen deathadder is really good though, if you don't mind 1800dpi

 

PCH has at least cooling. Worse with other brand's models

 

Is it just me or the middle GPU heat-pipe is actually CONCAVED!?!!! There are PLENTY direct contact desktop heat-sinks (there's a good reason behind it), but ALL of them are perfectly flat and there are no gaps!

 

Guys !! no MXM3.0b but why BGA Turdcity's (Tq Mr. Fox) Razer / crApple MBP ?? They are just worthless junk, not even to waste a pixel of space in this thread for that garbage...Lets bash the nGreedia / AMD / Failienware So that people could learn and give a single finger as an answer & vote, Micro$haft is getting flak too as of lately hope they get shafted soon..

Btw Any idea on the ACER Predator powered by Pascal, looks sick tho, Hope It's MXM 3.0b

 

I have new movie idea...

"BGA Turdcity" by Luc Besson and Quentin Tarantino lol

It would be the most SAD movie ever...

 

Guys the new GP100 Die shot is here sitting at 610mm^2 while the 980M has the bigger die size @398 than the new Pascal chips from 1080-70-60@310+ avg to closing @200 with 1060. I was semi-wrong regarding the whole heatsink mess coming up with HBM2 but the mechanism of heatsink would be much different (die should be fine) since there won't be anymore vRAM modules and the electronics will fully change, The Grid on the DM3 has all positions for contact with the vRAM chips and the power phases etc.., Seems like as technology is advancing at radical levels Maxwell > Pascal and maybe with Volta full architectural changes again, Add the PCIe 4.0 now we can imagine the similar position in future again within the NB realm.

Guess the modularity is coming to an end slowly like lead poisoning, design of the PCBs as a substitute here, I don't have a clue what's gonna happen next, really wish at-least 3-4 OEMs agree on a design standard like the HEDTs and go with full 18" DTR machines that can have space within the MXM slots to accommodate the changes at-least if they can't keep up the standard.

Only ASUS GX800VH/VI is the 18" machine but that's a huge misfire, ended up as a pompous colostomy bag with some dull tricks and hailing from the BGA Turdcity...
Edit : MSI GT83V is also there, dang ! Why didn't Clevo go that route given the panels already exist, could have much bigger space for everything from OCing to modularity, clearly one can notice that the 870DM & DM-2/3 are short of space.

Waiting for the NBR made CNC Milled DTR, even the concept of it's existence makes me happy at this saddest period..

GP100 Beast Die shot - Anandtech

 

I reckon an mxm rx 480 might pleasantly surprise us in terms of power consumption.

The desktop card was released as an unfinished product. I think can easily tweak voltages not only on core but also on the vram plus bin it well.

The result may be a card at 120w (780M) power consumption but performing at 980 levels.

Then a rx470 mobile card with under 100w would be a good 980M contender.

At a good price many us would be tempted.

Sent from my SM-A500FU using Tapatalk

 

There's a reason why the industry stopped calling them "laptops" and started calling them "notebooks".

 

The desktop REFERENCE card, with power draw limits disabled, pulls 192W in Witcher 3 at 1440p. There is ZERO way you're shoving that power in a 110W or so envelope. Clockspeed needs a sacrifice, and even the desktop card doesn't beat a 980 in most scenarios. So no, it won't be doing this. You don't make 82W vanish so easily.

 

I'm having trouble finding information on what exactly this "proprietary new Pascal crap" change entails - all I see are people talking about how angry they are with it. Can someone please explain, or at least link me to a reference on what exactly is happening here? All I can gather is either they're using a new MXM format for pascal only or the GPUs are soldered on with no exchangeable component.

I have to wonder if they'll take similar measures with their Quadro line for workstations. I bought my Dell 7510 because it has an MXM 3.0a Type A slot, for any 960m/m2000m pascal successor. If nvidia is ending MXM support completely, that would be a major bummer.

I've only spent about 30 minutes - an hour searching for a thorough explanation on the situation, so apologies if I missed something obvious. Thank you!

Edit: I think I may see the problem here.

Exhibit A:


Exhibit B: (1070)


So basically, they're constantly changing the board form factor, making upgrades impossible. Taking a page out of Intel's book, I see.

That sucks. I really hope this doesn't translate over into the workstation 15" form factor GPU line as well, but I won't be shocked if it does.

 

jony ive would disagree with you




but is true tho, you can't cheat thermals

 

well they kinda seem to go a separate way with quadros, at least thus far: the 980 quadro equivalent for notebooks, the M5500, is still in standard mxm 3.0b form factor with a whopping TDP of 150W and a juicy price tag at that... so lets see what nvidia is gonna do with pascal based mobile quadros...

Sent from my Huawei Mate 8 NXT-AL10

 

I meant in power draw actually. How do you keep same performance and magically keep a card drawing 82W less.

 

nvidia magic throttle

amd can make something up I guess?

 

I've checked the 7710 and it seems that there's little to no room for odd shapes, and I think that it is even better for the 15". DELL has just changed the chassis, so I doubt that they'll want to redesign it again so soon, but one should not underestimate nGREEDIA.

Depends on which side of the story you want to hear. That is the ABSOLUTE worse and it was on old drivers. There are also things like binning, better efficiency with lower clocks, and it was proven quite a few times that undervolt actually gains performance and the power draw is reduced by up to 30W. Some people intentionally miss these.

 

undervolt is doing the same thing with less power

overvolt is doing more things with more power...

@Mr. Fox

 

Not if the stock voltage is sky high. It was discussed in the Polaris topic. Also as you mentioned in the post right above my previous one - AMD already has such technology and can keep power consumption at given range.

 

Maybe AMD will continue the 100W MXM 3.0B form factor, but NVidia has stated it will not for consumer cards. AMD on the other hand came out with a card last year even, and never did the 180W/200W versions yet.

 

@Mr. Fox Phase cooling can you build yourself
A cheap one

 

I feel bad to see Tbonesan, Fox and other 870DM/P7 owners sell their Rigs and More will follow this path, Damn I guess the fun days of NBR / Notebook elitism are over...waiting to see update from MSI on their false promise of 2Gen upgrades...

 

I might be coming back for 1 more round (of punishment) . I still refuse to get soldered disposable tech on anything of substantial value. If I'm spending a lot of money on something I want to be able to maintain it at the very least. It will be sad if the crew disband but between here and TI I don't think that will happen

Clevo don't deserve a cent to be honest. I think MSI will come good but that they might pull a quick one and offer 1060's as upgrades so they can legally cover themselves. They know no one with a 980 will 'upgrade' to a 1060..."*cha-ching*$"... I could see that happening.

Clevo screwed their customers violently. If I didn't need a portable machine I'd likely be building a desktop or just give up all together.

 

You know, if you're bashing Clevo for it, MSI deserves rather similar bashing. And Prema said the reason they can't use reference MXM-B is because of nVidia, which leads me to believe it's a legal issue.

It's very unfortunate and you all have every right to be mad about it, but apparently if we didn't get those wonky out-of-spec designs, we'd ALL have soldered-only cards right now.