The “Order of 10”

[airfathaaaaa]

Based on what?

both are launching on june (launching aka market release not paper launch)
unless someone is delusional and thinks they gonna launch in 3 days and then they will go on computex empty handed and just showcase pretty much nothing

 

[maddie]

No, the whole context of that from question answering leads that statement to be confusing. It's definitely being taken out of context like no other when you read the full context it is vastly different

Also if Gp106 is coming later than gp104 and Polaris is 10 supposed to compete with Gp106 then it would be ahead of mainstream Nvidia offerings.

Just like Nvidia is ahead in high end offering...

That is what I thought he was trying to say. Did you think I meant something else?

 

[RaulF]

Good for you. Doesn't change the fact the card was hard to come by for quite a while due to yield issues on HBM1.

My point was toward the statement you made that it was no where to be found.

If you really wanted one, it could be purchased just had to keep an eye out for them. Newegg would get stock almost every other day.

 

[Jaydip]

Based on what?

Based on the launching campaign they have started and their ED event.I bet we we will see reviews pretty soon.

 

[maddie]

What's the possibility of getting prices this week? Might we only get a tech intro to Pascal?

 

[FatherMurphy]

I'm sure it's to the order of 10 in wood screws, deep learning or something trivial/inconsequential.

This type of marketing is quite infantile. It's like giving a ferret something shiny, if it's shiny the ferret likes it even if it is a polished turd.

That's my concern. When JHH used his CEO Math last year to state that Pascal was 10x faster than Maxwell, it was in the deep learning context, and the performance improvement stemmed from FP16 support, more bandwidth, and NVLink. Gaming and DX12 are not deep learning.

Much like the "Power of 3" mentioned above, perhaps the Order of 10 represents 10 "new" features of Pascal. It's anybody's guess.

 

[linkgoron]

That's my concern. When JHH used his CEO Math last year to state that Pascal was 10x faster than Maxwell, it was in the deep learning context, and the performance improvement stemmed from FP16 support, more bandwidth, and NVLink. Gaming and DX12 are not deep learning.

Much like the "Power of 3" mentioned above, perhaps the Order of 10 represents 10 "new" features of Pascal. It's anybody's guess.

The 10 represents the 10xx lineup

 

[Ferzerp]

Why so much anger over a little teaser website that doesn't even make any claims?

 

[FatherMurphy]

The 10 represents the 10xx lineup

That's as good a guess as any but, unless you know something no one else seems to know, it's just a guess. It seems strange to me that they would emphasize the "10xx" series when Nvidia tends to focus on its architecture name (and has been focusing on the name "Pascal").

 

[Timmah!]

So much fuss about revealing bunch of mid-range GPUs. Better be worth it now.

 

[FatherMurphy]

@Timmah... if your upgrade path is mid-range to mid-range, then hopefully now is the time for celebration. If you are sitting on a 980ti or Fury X, you may have to wait longer.

 

[Magee_MC]

It's definitely an interesting campaign, and it is getting people talking. I really like it as a whole, however the "Humanity shall be enlightened" phrase just strikes me as quite a bit over the top. When I think of enlightenment I'm more likely to think of Buddah than a GPU.

 

[DooKey]

Sigh, GDDR5X is available as RISK PRODUCTION. It means low amounts will be available NOW. 1080 will probably be like Fury X and have low availability initially.

 

[Aristotelian]

The beginning of June will be a very expensive time for me with Pascal and Broadwell-E releasing at the same time.

I (kind of) get the 5960x to 6950x upgrade, but you have 4 Titan Xs and if you replace those with 4 1080 cards you'd get a performance decrease, no? If only you were right, and the Titan X successor would be released then

I posted earlier in this thread and sometimes I think people make a big deal about the name - all I was excited to see beyond the initial page was the timer. It's a countdown, plain and simple, and is more than what Nvidia's competitor is offering. So when that countdown hits 0 I'm expecting a major reveal, and no more 'teasing'.

 

[RussianSensation]

They might, if this was anything like that... which it isn't.

If you announce a product May 6th and it's not on sale until May 31st-June XX, that's a Paper Launch. HardOCP slammed AMD over and over for this. If they don't write same articles about Pascal paper launching this week, it's a double standard and would be more proof that they are not impartial.

The 10X is the performance increase NV quoted for machine/deep learning, etc. to hype up Pascal over Maxwell. But I guess with "5 Miracles", they can tell us what 10 next generation features Pascal has.

+1000
This is brilliant. Like it or not, they are miles ahead of anything AMD does. A totally different level. This is why mainstream buyers are so much Nvidia towards.
It also shows that Pascal's launch have been meticulously prepared. Hopefully we will have a new kickass tech demo. I love them :wub:

No doubt, you have to be miles ahead in marketing when you try to sell GTX560Ti lineage successor mid-range die 680 for $500, then raise it again to $550 with 980, and then keep going with mid-range Pascal (offering sub-400GB/sec midrange card at flagship prices while having designed a 700-1000GB/sec HBM2 as the real flagship). Then use this brilliant marketing strategy to get the same people to upgrade yet again to Big Pascal in the same generation.

---

Hopefully NV doesn't artificially neuter 1070 too much, resulting in a 20-25% performance delta with a 1080. If they do, it will be akin to taking a 1060Ti, renaming it 1070, and jacking up the price from 660Ti's $299. I am just putting that out there now given that NV didn't even bother giving 1070 legit next gen memory. Big red flag to me right now.

 

[Aristotelian]

No doubt, you have to be miles ahead in marketing when you try to sell GTX560Ti lineage successor mid-range die 680 for $500, then raise it again to $550 with 980, and then keep going with mid-range Pascal (offering sub-400GB/sec midrange card at flagship prices while having designed a 700-1000GB/sec HBM2 as the real flagship). Then use this brilliant marketing strategy to get the same people to upgrade yet again to Big Pascal in the same generation.

Are so many of Nvidia's sales from people upgrading from generation X to x+1? I bought my 580s in SLI on release day and paid a huge amount, and was too lazy to upgrade to 680s. By the time 780s came around I started thinking about building a new PC but still, I was able to use my 580s enough to play things, of course not with maxed eye candy.

I'll be buying 2 of whatever the fastest gpus are this year. From the steam user data doesn't it show that a rather large amount of gamers are NOT upgrading year to year?

Don't get me wrong RS. I wish the HBM2 flagships were out now because I'm locked into buying a PC this year and it seems they'll be coming out next year, but isn't AMD doing something similar by using their smaller parts as a refresh of the current lineup - selling it as perf/watt increases - and then coming out with real flagship cards (Vega, HBM 2, larger die sizes) next year, just as Nvidia is with the real Pascal flagships (with HBM2)? I'd say both companies try to milk their customer base as much as they can in order to maximise their revenue, and I think (overall) that it's just the terrible position PC gamers are in. Massive amounts of investment and some games are looking no better really on a flagship PC than they do on a console like the PS4 (e.g., The Division).

 

[krumme]

Are so many of Nvidia's sales from people upgrading from generation X to x+1? I bought my 580s in SLI on release day and paid a huge amount, and was too lazy to upgrade to 680s. By the time 780s came around I started thinking about building a new PC but still, I was able to use my 580s enough to play things, of course not with maxed eye candy.

I'll be buying 2 of whatever the fastest gpus are this year. From the steam user data doesn't it show that a rather large amount of gamers are NOT upgrading year to year?

Don't get me wrong RS. I wish the HBM2 flagships were out now because I'm locked into buying a PC this year and it seems they'll be coming out next year, but isn't AMD doing something similar by using their smaller parts as a refresh of the current lineup - selling it as perf/watt increases - and then coming out with real flagship cards (Vega, HBM 2, larger die sizes) next year, just as Nvidia is with the real Pascal flagships (with HBM2)? I'd say both companies try to milk their customer base as much as they can in order to maximise their revenue, and I think (overall) that it's just the terrible position PC gamers are in. Massive amounts of investment and some games are looking no better really on a flagship PC than they do on a console like the PS4 (e.g., The Division).

So half a year later the gpus is twice as fast?? Does not make sense imo not to wait for big gen this time if you want that kind of perf. Hell bend on 31 is for what - tax ? Lol

 

[CakeMonster]

Sigh, GDDR5X is available as RISK PRODUCTION. It means low amounts will be available NOW. 1080 will probably be like Fury X and have low availability initially.

Knowing NVidia, they'll probably want to also have a hugely expensive product available, something for every kind of buyer. So just because GDDR5X will have limited availability for a few months, this is a good opportunity to cash in on that.

The risk they run is 1070 being too close to 980Ti or even 980 (ie not impressive), and 1080 being too expensive.

 

[CakeMonster]

@Timmah... if your upgrade path is mid-range to mid-range, then hopefully now is the time for celebration. If you are sitting on a 980ti or Fury X, you may have to wait longer.

I have a 980 but I'm still skeptical as to how expensive this generation might possibly be, which might leave me with 1070 as the alternative and not 1080, and the performance difference could end up not being enough to justify a buy.

Make no mistake, I desperately want something new, I still have that upgrade itch after close to 20 years of PC gaming. But I think I am justified in having high expectations for a 28->16/14nm move. That, along with myself getting more and more picky with games I play, could still make me sit out this time. I bought my 980 for Dragon Age: Inquisiton, and 780 for Bioshock Infinite, and it was worth it both times. But I can't think of any games in the next 6 months that I'm excited about. And I am actively _looking_ for any excuse to get some sweet new GPU tech here....

 

[Game_dev]

Weren't we led to believe pascal has a 6 month lead ahead of Polaris? It seems everyone was wrong

Pascal is more than 6 months ahead of Vega. Which is what matters.

 

[monstercameron]

Pascal is more than 6 months ahead of Vega. Which is what matters.

Source? Or do you just opine?

 

[Game_dev]

Source? Or do you just opine?

Technically Pascal already launched with Tesla. AMD's road map showed Vega in 2017

 

[swilli89]

Why a new thread over a little teaser website that doesn't even make any claims?

Yeah i know.






You deliberately misquoted a member to post a thread crap.


esquared
Anandtech Forum Director

 

[Silverforce11]

Sigh, GDDR5X is available as RISK PRODUCTION. It means low amounts will be available NOW. 1080 will probably be like Fury X and have low availability initially.

If it's priced according to rumors, $599 to $649, it will be low volume anyway so sampling GDDR5X will suffice.

 

[Game_dev]

If it's priced according to rumors, $599 to $649, it will be low volume anyway so sampling GDDR5X will suffice.

Even $699 wouldn't surprise me.