Pascal: What do we know? Discussion, Latest News & Updates: 1000M Series GPU's

I am sort of in the waiting game too. I'm not really interested in Pascal or Polaris performance, I was just following things from the architectural standpoint and Pascal didn't impress at all (yeah besides clock speeds) so I am still hoping that Polaris will. I bought a Sapphire Nitro R9 380X back in December for $230 and until 4K HDR monitors start showing up I have no reason whatsoever to upgrade (my current monitor is 1080p and sucks). And also the value of this card is just amazing, heck the price hasn't even went down since launch. What's even more amazing is it started beating a desktop 970 in some newer games (went from about 780 level to 780Ti). So I'm hoping Vega will come out in Q4 and new monitors should become available by that point also so I can upgrade then. Vega 10 with 4096SP and 16GB HBM2 should at least be 60% faster than Fury X and along with 4K 60Hz HDR monitor should make some beastly system.

 

Yeah... a few AMD Oompa Loompa console gamer boys is not something we need to lose any sleep over, but an army of Oompa Loompa console gamers is something we should all be worried about. Having the best weapons money can buy is not a guarantee of victory when you are outnumbered like 100:1, LOL. Even an army of handicapped peasants armed with nothing more than sticks and rocks can win the battle if there are enough of them.

 

You forgot the pitchforks and torches.

 

I don't get what the laugh is about.
I am about 99% sure that Polaris 10 will be weaker than GP104. It probably won't even pass Fiji. If AMD wanted to challenge their lineup, they'll need to run Fiji on 14nm, bump the performance and add overclockability and reduce power consumption (the 14nm node should automagically do this), add HBM2.0 and sell it at like $400. If AMD however gets CLOSE to GP104 performance with some of their cards, rebranded or not, and reduces their prices very competitively, they will make a big shakeup. Why pay $150-200 USD more for a card that's 5% stronger? It's the same 980Ti v Titan X arguement, and the 980Ti won.
Until nVidia's driver crap starting last year, and their inane attempts to control everything about a GPU and wrest control from the end-users as much as possible without making someone feel like they bought a Mac, I preferred them too. In fact, I STILL prefer them for multi-GPU, and for nVidia Inspector's absolute usefulness... but that's the hardware/software potential I'm thinking about. They're going all out with tricks and lies to the public, snuffing out people that speak ill of them to the point where the internet either sucks their meowmix or simply shuts up, and locking off their good tech behind proprietary gates or simple paywalls. They need to get knocked all the way back down to about 30% market share, and actually have to work their way back up.

But of course... they will never have that happen. Know what my twitter timeline was after the 1080 that they lied about performance and efficiency was revealed at the nice overpriced $600 USD mark? "OMG THEY MADE IT AFFORDABLE!" "OMG IT'S SO CHEAP!" "I CAN GET ONE!" "NEW CARD IN 3 WEEKS!" "I LOVE YOU @Nvidia" "GOD THEY'RE AMAZING" with everybody retweeting everything and nobody caring. AMD isn't going to beat that with Polaris and they sure as hell are not going to do it with rebranding Fiji. Vega needs to be something like... Tri-SLI Titan X performance at $600 or less to have them win anything.

That doesn't help marketshare and their ability to stay afloat, though. I've seen the master plan, and while there's a lot of good facts in there, a small portion, where AMD praise is happening, is cherry-picked. As it stands, it doesn't matter if they have console contracts. It just means they're getting a steady inflow of money in general, which might be good, but if it doesn't supersede the costs of their output, it doesn't matter. The fact that consoles haven't been dying rampantly after 1-2 years means that the 7970M cut down variant they added in the PS4 and the R7 260 they put in the Xbox 1 (which is straight cut from 7970Ms too, as far as I remember, though it matches another direct architecture codename they had that wasn't Pitcairn) are very resilient, unlike many users' mobile GPUs on this forum and on the web in general, that tended to die after about a year. Which means they're trying really hard, unlike what they felt was safe to sell us.

 

I laughed at Ethrem's description of Pascal (pig with lipstick)

Anyhow, I've been disgusted with the whole GPU scene and even more so, with the so-called "next-gen" consoles.
I have been gaming since the days of Atari and Dos and I can't remember a time where we, as consumers, have been lied to so openly.
I honestly can't believe how Ponies can actually be excited at the prospect of a "new" PS4 so early after the launch of that console. Did they forget that these "next-gen" consoles were supposed to last at least 5 years?

I doubt AMD will match Nvidia's performance because I think they don't care about this market segment. Dominating the consoles market is more than enough for them.
The lack of competition in PC is the worst thing to happen to us. Why would Intel or Nvidia care what we think of them? Do we have any other choice than buying their product? Same goes for Apple with the pathetic iPhone with a death timer triggered by each new iOS update. Or Microsoft with their increasingly crappy OS.
Monopoly frickin sucks and it's a shame that we are forced to deal with it.

 

For those paying close enough attention, there was nothing positive in my previous statements about AMD. In fact, it was exactly the opposite. Reality can be brutal... as you have pointed out. It's also the primary reason BGA turdbooks have become so popular. A product does not actually need to be any good so long as enough people can be led to believe that it is good.

 

Although, off-topic, Meerkats aren't that stupid, have great teamwork & social lives and successfully eat snakes as part of their diet, they can also survive snake venom bites even when they're that small - just wanted to defend the Meerkats is all!

 

I prefer the phrase "never underestimate the power of a single stupid person".

 

Does anyone have a source to the confirmation of AMD securing contracts for Polaris architecture with regards to the newer consoles? I've only heard speculation and hypothetical musings from the likes of DigitalFoundy and others. Is this actually confirmed?

 

I'm not aware of 7970ms dying within an year (two and more, yes, and then again not all of them), but I'm well aware of 980ms dying within couple of months, and there are no new loud and flashy topics about that. There are also dying 680ms and 780ms, but there's not a single topic to be found about it either. I found it "a bit" strange.

It's funny how precisely this matches the Pascal announcement

 

I had four 7970M in the span of just a few months that malfunctioned severely, although they did not completely die. They were early production models. I was extremely excited about the 7970M CrossFried upgrade and equally disappointed that all four GPUs were garbage. That was a very long time ago. They haven't done anything important or remarkable for notebook owners since then, and that's terribly sad.

I agree on Pascal. I'm not ready to be excited. Excitement would be premature at this point. Skepticism is in order based on their track record. Heck, we don't even know 100% for sure it will be available on MXM... or compatible for upgrades with existing systems. I've heard rumors that the Jolly Green Giant is pressing OEMs to make it incompatible with previous generation notebooks and only available in BGA. We will have to wait and see whether they obey or disregard those wishes. And, this talk about overclocking better... *burp* ...remember, their position is they "do not support overclocking on notebooks" so we shall see. Just because desktops do, doesn't mean mobile will. A GPU that doesn't overclock like crazy is not worthy of being purchased as far as I am concerned. Using it stock and being happy about it is for school girls.

And, if it is available on MXM and the price for the mobile variant follows the path that Maxwell did, it will be more expensive than the severely overpriced GTX 980M. I have had four 980M and one 970M that are crappy samples that don't overclock as well as I expect them to and all 5 have black screen problems if pushed as hard as I have seen others push theirs. In addition to those, two more 980M that died (as in no longer bootable). They are building them with crappy (weak) MOSFETs and QC is at a historical low for NVIDIA.

It will also be really interesting to see if we are in for another round of digital genocide, should the NVIDIOTS decide to cripple Maxwell with cancer drivers the way they did Kepler when Maxwell was released, and Fermi when Kepler was released.

It's a pretty sorry state we live in now. No expectation for anything resembling excellence to surface from the red side and and an expectation for another round of abuse and treachery from the criminally insane on the green side. Wouldn't it be nice if they were both good companies that produced consistently good products and treated their customers right? NVIDIA demonstrates about as much respect for customers as I do for someone else's chewing gum stuck on the bottom of a restaurant dining table.

The best approach, and the one I plan on taking, is to let a bunch of other people blow their money on the new red and green garbage. I will watch and see what happens and then decide if there is a any good reason to spend money on something new. Early adoption isn't a wise approach any more.

 

I wouldn't call it a pig. Maxwell was an enormous success and loved by what it delivered in performance and heat. So if Pascal is a natural evolution of Kepler->Maxwell->Maxwell (2nd edition)->Maxwell (3rd edition or "Pascal") we can hope for a reliable card, good driver support and stability.

I rather take this one than a new Kepler II which has to struggle yet with all the child diseases.

 

All I know is from desktop side, I'm waiting until a card is at least twice the performance of 980 Ti and preferably with HBM2 before I swap out my 980 Ti. From mobile standpoint, I'm currently running only my trusty old P650SE with GTX 970m 3GB. And honestly both are super capable machines. It's funny how after this presser that we still are in the dark about exactly what we're getting and when.

 

The 1080Ti and the next Titan should have HBM2. Probably late this year, though.

I told my friend with a 780 to just wait for the 1080Ti because he's still happy with the 780, but it should be more powerful than 3x 780's.

 

I've noticed the gpu's touted a vapor chamber cooling design.

How does that compare to what we have now and would we be able to see that in mobile?

Sent from my SGH-M919V using Tapatalk

 

Is there any news on mobile versions of the GTX10** yet, I've googled and come up with really old articles?

 

Nothing , wait till computex or sometime in June

 

After a long day at work where I managed to squeeze in a few minutes of "brainstorming" on the situation at hand, I have decided to stick to my trusty AW17 and the 980m (provided it doesn't fail me like the 1st one I got which died after 2 months) for a while.
Anyway, I have invested quite a bit in quality holidays for the family so I don't think it's a bad thing to postpone this, especially when I am part of those disgusted with Nvidia's practices and not in awe after Pascal's unveiling. I swear, some websites are posting about this as if it was the 2nd coming of Christ!

This will only be the 2nd time I postpone building my "glorious" HTPC... But at least I won't be lying to myself and buying a product I am not 100% convinced by. After all this is STILL a $600 purchase.

I had also planned getting a PS4 this month as Uncharted 4 is the game I was waiting for, but with the rumors of PS4K, I have also decided against buying that console as well. Will wait for E3 to get more details on that and decide if it's worth the wait (AND the to-be expected premium price of course)

Oh man, how things are complicated now... Kids these days don't have it easy. Remember the days of NES/SNES? Or DOS games? The only worry I had was a damaged floppy disk...

 

LinusTechTips was invited to a special presentation from NVIDIA and hinted to us that Computex is where NVIDIA plans to officially release Pascal.

About 3 weeks from now, folks!

 

http://www.dsogaming.com/news/report-nvidia-gtx1080-is-up-to-50-faster-than-the-gtx980ti/

As expected and discussed here but more simpler way to understand the nGreedia's overly hyped show.

GO AMD GO !

 

That article is ridiculous, lol. No offense to you... It's aimed at them.

It says, "As we can see below, the GTX1080 is 1.7X and 1.8X faster than the GTX980 in The Witcher 3" when NVIDIA said it was 2x faster. COME ON, REALLY?! Making such a big deal about that? It's still a good performance jump. Wait until driver support is good and we may actually see it reach above 2x... And it definitely is 2x faster when overclocked to 2000Mhz.

These attention whores will say anything.

 

Yeah that poster is full of fact ignorance. Even one mongoose/meerkat can take a cobra.

Sent from a 128th Legion Stormtrooper 6P

 

Um...that cobra is about to be lunch for a hungry gang of mongooses

 

As a confirmation, I had TWO 7970ms in my laptop, as the first one DIED within a year of owning it, so yes, the rumors and talk of them being poorly-made and badly designed are indeed true and real.
The 2nd one died last year.
I would suspect that there were a great many people that bought that card only to see it die in less than a year, as I have seen dozens of comments from users who reported their experiences testifying to that.
Let's not also forget that the first batch that came out in the Spring of 2012 were bad, and they had to remake them due to faulty design and construction issues.
That second generation batch was also bad, as quite a few people here and elsewhere that owned them like myself, reported that their GPUs had died within a year of use.
Lucky for me that I was able to get mine replaced under warranty only 3 weeks before it expired.
It was a poorly-made GPU, and is the reason why I and others have such a strong dislike for AMD products in general.
That's not even touching on the AMD driver issues, delays, and of course, Enduro problems that have plagued that GPU for almost 2 years.

 

Dont expect too much performance on 2000mhz OC. The stock boost clock will likely be around 1.8ghz already. Hardly a massive perf boost.

 

The base clock is 1600~. Boost is 1700~. They overclocked to over 2000Mhz and was stable with a max temperature of 67C.

 

Nvidia stated boost clock is 1700, the actual boost clock would be around 1800ish. We have seen this on almost all nvidia cards feature boost clock. It boosts higher than whats stated.

I want to see some actual overclocking numbers on water/aftermarket air before making any decision on this one. Either way, the "founder edition" cards seems like a quick cash grab since the 599USD cards are launching later. I believe only the FE cards are launching in May 27rd, so a quick cash grab for those without patience. Whatever nvidia may say, just a 8 pin with a massive vaporchamber is not what I want for serious overclocking.

I am extremely skeptical of this launch right now. So far, there is a lot of marketing languages used. For example,"twice as fast as titan x for only 599!"(in VR only). Also a launch price of 699USD for a glorified mid end card is not pleasing at all.

GTX 1080 is nothing but a glorified mid end card priced like a high end GPU. Too bad AMD's competition is lacking. What happened to the days where a x60 card beat last gen's x80 card and are actually PRICED like a x60 card?

I can always pick up a used 980 TI and just clock it to GTX 1080 performance for way less.

 

I didn't say that there was no problem at all. The point was that unlike the separate topics about pretty much every 7970m failure*, I can't see A SINGLE SEPARATE topic about 680/780/980 failure. And trust me there are quite a few. Drivers... I guess you were under a rock the last year alone...

*Every single topic was more or less the same, so it's not like they can't Google and find their answer and possible resolution, but NO, they have to post that very topic in disgust. Again I cannot see a SINGLE EFFFIN topic about nGreedia.

 

Btw that Vulkan Implementation was good at-least some positive results in someway or other compared to using that shackled POS Win10 DX12 B$.

 

I was looking for one to illustrate the folly of stupidity en masse and that was the best I could find. Just pretend they are mice and it will make perfect sense.

 

_________________________________

LOL. Under a rock? No, it was too small a place for two people, so I let you be there all alone by yourself....

FYI, I was pointing out that from personal experience in context of what has been talked about here and elsewhere for 4 years now about the 7970m and it's driver issues, so I don't know where you have been, since we have already discussed this topic ad nauseum already since 2012.
Additionally, no one has kept Nvidia faultless here either, which also has already been discussed here at great length. I owned an Nvidia GPU as well, which died in 3 years of usage, so I can speak to both issues, again from personal experience and knowledge.
I can personally attest to both companies taking their customer bases too lightly, and not caring enough about QC and CS issues. which has precipitated literally hundreds of threads here and elsewhere on these two topics alone.

 

__________________________________
Yeah, but every now and then, it would help the Mongoose to have GI Joe give a helping hand to take out Cobra.
"Knowing is half the battle"....
LOL.

 

NVIDIA drivers have been garbage for more than a year for Kepler and Maxwell and for more than a couple of years for Fermi. I think one of the big differences between AMD and NVIDIA as far as durability is the idiots on the green side of the house commit digital genocide and kill or otherwise cripple performance of their own products with cancer drivers, and it certainly seems like they do it on purpose (to make the new junk seem more amazing) from where I am sitting. The stuff that goes wrong with AMD is due to QC issues. But, QC problems are setting new records for NVIDIA now, so I find it harder than ever to cast stones at AMD at this point. If there would be one major fault for AMD it is the lack of innovation, and lackluster performance, most especially in the mobile arena. They haven't really done anything except recycle the same old stuff since 7970M and their MXM GPUs are very crappy overclockers. If their MXM GPUs would overclock as well as their unlocked desktop CPUs we would have some pretty amazing stuff to work with.

@triturbo - what we lack in NVIDIA hate threads at NBR and Tech|Inferno is more than made up for in the endless supply of them at GeForce Community. Seems NVIDIA owners take their complaints directly to the source of their misery, but it falls on deaf ears and they often get trolled by fanboys that think the Jolly Green Giant can do no wrong.

 

Well, LinusTechTips experienced the card first-hand and witnessed its clock speeds, temperatures and performance. They said it was 1600~ and 1700~ respectively, overclocked to over 2000Mhz. Don't know what to tell ya... NVIDIA lies?

 

Lol Nvidia surely stirred up everything, it's all over the internet and the updates are very frequent.
Check this out, http://wccftech.com/nvidia-gtx-1080-asynchronous-compute/ wtf is this they didn't utter a word on the ASync tech 5B$ features and now the press release ??

This is the golden chance for AMD to strike & turn everything. Even if they manage to get a good Polaris mobile that would be a direct blow !

 

Is that really so surprising to you? Ofcourse nvidia lies. Need a list?

And get this: http://www.gamersnexus.net/news-pc/...080-founders-edition-and-reference#!/ccomment

We are paying more for a ****ty lesser reference card just because they called it "founder edition". Yeah no.
Aftermarket card for me and that cooler looks like crap anyways.

 

Stock clock is 1607, boost is 1733. Basically it sounds like these cards can't even hold their boost if he was seeing them in the 1600s...

 

Black one might not be too bad. Very Batman.

Sent from my Nexus 6P using Tapatalk

 

I had to go look it up and I don't remember what page it was on in this thread where it was asked if AMD making the new console chips was confirmed but this Ars article talks about AMD expecting revenue from three new SoCs between 2H 2016 and through 2017.

http://arstechnica.com/gaming/2016/...aming-processors-coming-soon-is-one-for-xbox/

 

If I had more money to burn, I would grab the Radeon Pro Duo, 16 teraflops on a single PCB....this card impresses

 

It's severely limited by its 4GB VRAM. Not worth it.

 

He was saying what they were and reporting seeing over 2000Mhz, stable, at 67C.

 

Yeah 2144MHz if I remember correctly. Still though, if the clocks were ever in the 1600s then nVidia has again given too low of a power limit to sustain boost which is unacceptable. Bump the TDP up so that people actually get what they pay for. It's even worse that the behavior was seen at an nVidia event because those are always highly binned chips. I can only imagine what we are looking at in retail products if that's the case.

 

Anyway, 25-30 fps is not worth 1200 dollars. Waiting to see what the TI offers or simply going with AMD next round.

Edit: gsync monitor be damned. Actually considering buying a second freesync monitor and switching between the primary depending on which company is best that year.

Sent from my Nexus 6P using Tapatalk

 

4K benches show it beating cards with more vram like the Titan X . Wonder if the bandwidth makes up for it

 

could be, havent seen u in a bit man. good to see you around.

 

I need to see it in a situation that expressly uses more than 4GB. Bandwidth has little to do with straining the capacity.

It will choke when, for example, the situation calls for 6GB of VRAM.

 

Could be dependent on the game engine. Storing the textures vs streaming them. Admittedly I know little about this topic.

Sent from my Nexus 6P using Tapatalk

 

Apparently DX12 doesn't mirror VRAM for SLI/crossfire. So you could actually work with 8GB on the pro duo(or 12GB for 980ti SLI )

Sent from my Micromax AQ4501 using Tapatalk

 

Not if the photographer had to super glue each of those mongooses into place to get the picture. IOW you only see what the photographer shows you. Something to think about with marketing.