AMD Polaris architecture – GCN 4.0

[CAT-THE-FIFTH]

I'm not desperate at all, the sooner Polaris is shipped the better. I'm just stating the facts about what back to school is, as I said, I live in the Us and I know when places like Best Buy have Back to school specials and have a load of dirt cheap laptops and desktops being shoved in parent's faces. American's don't do back to school shopping in July, period. However, Polaris will have to start shipping to the OEMs in June to have a hope of shopping out in July and appearing on shelves in August.


The headline form your link:

That is around September 9th. From the data 80% of purchases are done within3-4 weeks of start of school, so August 12th-19th or so is when sales pick up. Exactly as I said.

From Wikipedia:

https://en.wikipedia.org/wiki/Back_to_school_(marketing)

The back to school period of time usually lasts from mid-July through early/mid-September, before the school year starts in the United States, Europe, and Canada.

From the second link from the US marketing site:

Propelled by back-to-school buying, US retail sales will grow a relatively strong 4.6% during the core back-to-school shopping months of July and August 2015

Another link:

http://adage.com/article/news/july-i-back-school-ads/242979/

Staples said it will begin setting up back-to-school displays in early July. Marketing for Target and JC Penney won't get under way until later in the month. And a message on Amazon's back-to-school landing page takes a humorous approach: "Shopping for back-to-school supplies already? We hate to tell you ... you're early." The page instructs shoppers to come back later in July.

Another one:

http://www.marketingcharts.com/traditional/back-to-school-stats-2015-57232/

Retail e-commerce sales during the core back-to-school season of July and August

http://www.pfsweb.com/blog/back-to-school-ecommerce-a-new-u-s-digital-holiday/

In 2014 and moving forward, it appears as if back-to-school (July and August) shopping will be even bigger, with sales projected to rise exponentially.

48% of U.S. internet users made back-to-school purchases in August, only 35% made purchases in July.

That is over a third of purchases made in July alone with something like 65% of people doing research during that time.


The first day of summer in the US this year is the 20th of June:

http://www.apples4theteacher.com/holidays/summer/first-day-of-summer.html

No AMD show during July will probably mean less interest in Polaris as they will probably only see Maxwell based systems in laptops and desktops.

The target is mid-2016, particularly in time for the major back-to-school season in the united states. Polaris should be available not just in the form of desktop solutions, but also notebook graphics chips.

Hence,if what the AMD CEO is saying is correct,for Polaris to be ready for the back to school season and be available in the middle of the year that would be either late June or sometime during July.

August/September is not the middle of the year.

If it is any later it would technically be not the middle of the year and would not in time for the back to school season as July is still consider an important month for sales too. It would be technically delayed by what the CEO said.

Whether it is a limited launch is another question but both the HD7970 and Fury X pretty much launched first and it took weeks for stock to actually being available,since it was all on pre-order IIRC.

I expect reviews will drop sometime during late June or sometime during July and then the cards will be on pre-order or have limited quantities in computer hardware retailers.

That is most likely what we will see IMHO.

 

[james.miller]

I'm happy to hold on to my 970 for the rest of the year, its still eating everything i throw at it and i doubt that going to change any time soon.

I'm going to wait a while until i see a deal from either side that i can't resist, i'm not rushing into anything.

Really? I'm struggling with The Division. Sure is pretty though.

 

[D.P.]

From Wikipedia:
Wall of junk

You seriously have issues if you went to that length.

 

[Orangey]

Care to clarify? as AFAIK Samsung only ramped upto volume production for high performance ASICs a few weeks back while TSMC was at the very end of last year - so basically nothing in it as far as that goes depending on when they start actually making stuff for the respective companies, while there might be some design considerations with Samsung having a head start with LPE its unlikely that aspect alone is going to dictate the schedule for GPUs.

Not to mention Polaris is being made at GF, not Samsung. And GF lagged behind massively.

source: AMD

 

[Dave2150]

From Wikipedia:

https://en.wikipedia.org/wiki/Back_to_school_(marketing)



From the second link from the US marketing site:



Another link:

http://adage.com/article/news/july-i-back-school-ads/242979/

Another one:

http://www.marketingcharts.com/traditional/back-to-school-stats-2015-57232/



http://www.pfsweb.com/blog/back-to-school-ecommerce-a-new-u-s-digital-holiday/





That is over a third of purchases made in July alone with something like 65% of people doing research during that time.


The first day of summer in the US this year is the 20th of June:

http://www.apples4theteacher.com/holidays/summer/first-day-of-summer.html

No AMD show during July will probably mean less interest in Polaris as they will probably only see Maxwell based systems in laptops and desktops.



Hence,if what the AMD CEO is saying is correct,for Polaris to be ready for the back to school season and be available in the middle of the year that would be either late June or sometime during July.

August/September is not the middle of the year.

If it is any later it would technically be not the middle of the year and would not in time for the back to school season as July is still consider an important month for sales too. It would be technically delayed by what the CEO said.

Whether it is a limited launch is another question but both the HD7970 and Fury X pretty much launched first and it took weeks for stock to actually being available,since it was all on pre-order IIRC.

I expect reviews will drop sometime during late June or sometime during July and then the cards will be on pre-order or have limited quantities in computer hardware retailers.

That is most likely what we will see IMHO.

Nice one, July is indeed within the US back to school period.

 

[SimBy]

Not to mention Polaris is being made at GF, not Samsung. And GF lagged behind massively.

source: AMD

It's made at GF but on Samsungs process. So tehnically no difference.

 

[D.P.]

Nice one, July is indeed within the US back to school period.

As is August and September, with the end of August and early September being the peak. So what does that mean, AMD want Polaris laptops on the best buy shelves by mid-august at the latest, which means shipping n July, and if they want to maximize sales get things out a month earlier than that.

 

[Mauller]

Not to mention Polaris is being made at GF, not Samsung. And GF lagged behind massively.

source: AMD

It would not take them long to ramp up. Any improvements on Samsung side would get sent to GF.

 

[Hamfist]

I think a lot of people thought AMD were going to release first, because they were ahead of the game with HBM2. Most rumours are now saying HBM1/GDDR5X this year, so Nvidia could be first to release at 14/16nm.

 

[drunkenmaster]

I think a lot of people thought AMD were going to release first, because they were ahead of the game with HBM2. Most rumours are now saying HBM1/GDDR5X this year, so Nvidia could be first to release at 14/16nm.

HBM2 was always going to be q3 earliest this year, no one expected HBM2 on the low end and the midrange most people thought also wouldn't be HBM though I still suspect it might be HBM1. The reason most people thought AMD would be first was not anything to do with HBM2. Likewise there is zero indication of GDDR5X becoming available for Q3 at the very earliest, late Q3 at that, probably more like late Q4 and we could easily not see any cards from it till 2017.

 

[drunkenmaster]

Not to mention Polaris is being made at GF, not Samsung. And GF lagged behind massively.

source: AMD

One of the very reasons Samsung agreed to license the process to GloFo was getting themselves a second source, which offers the chance for increased capacity and for back up production should something fail. If a production line fails at Samsung's fab then Apple can get some production going at GloFo to make up for the short fall. AMD may intend to manufacture at GloFo but it doesn't mean they can't take that exact design and produce it at Samsung should the need arise.

 

[Orangey]

It's being made at GF purely to cover the WSA. They're doing other designs (Vega?) at TSMC, which I believe hints at them having the superior process.

----------

Also I mentioned laptops in a previous post:

http://www.kitguru.net/apple/matthew-wilson/apple-to-refresh-13-and-15-inch-macbook-pros-in-july/

 

[drunkenmaster]

It's being made at GF purely to cover the WSA. They're doing other designs (Vega?) at TSMC, which I believe hints at them having the superior process.

----------

Also I mentioned laptops in a previous post:

http://www.kitguru.net/apple/matthew-wilson/apple-to-refresh-13-and-15-inch-macbook-pros-in-july/

If they can't produce chips at GloFo... then they lose nothing producing them elsewhere. Also from what I recall the WSA was renegotiated a couple of years ago. It was heavily punishing in the years right after GloFo was formed, a financial recession hit and Bulldozer wasn't successful but again the newer terms I believe were much less punishing and easier to fill. However again if they physically can't produce chips at GloFo, they stand to gain in increased sales of newer 14nm chips drastically more than penalties from failing to meet the WSA terms. Sales for AMD GPUs would increase dramatically as well as being able to sign big deals with the likes of Apple and others.

The WSA would be an incredibly poor reason to not make the chips at Samsung if Samsung had any capacity and Glofo didn't. As for AMD making other things at TSMC, I'm not sure I've seen anything that suggests they will, just a couple of years ago AMD not committing to either process because depending on plans and pricing either might be the better option. From what I've seen in the past year or so AMD amongst many others have moved away from TSMC and decided to make chips with Samsung.

 

[SimBy]

It's being made at GF purely to cover the WSA. They're doing other designs (Vega?) at TSMC, which I believe hints at them having the superior process.

----------

Also I mentioned laptops in a previous post:

http://www.kitguru.net/apple/matthew-wilson/apple-to-refresh-13-and-15-inch-macbook-pros-in-july/

Who said they are using TSMC for Vega?

 

[AllBodies]

Who said they are using TSMC for Vega?

No one.

And since they're claiming 2.5x performance per watt for Polaris on 14nm LPP, and without the power reductions of HBM, I doubt TSMC's 16FF+ process is actually better.

As far as I gather, when people were claiming 16FF+ was better, that was comparing it to Samsung 14LPE (E being for 'early'). And Samsung say LPP is 15% lower power AND 15% higher performance than LPE. Which makes a 35% increase in performance per watt over LPE.

 

[Orangey]

There's a lot of assumptions there. They didn't say "without HBM" and nobody was comparing LPE to 16FF+. You're also massively overstating the difference between LPE and LPP.

I don't know why people feel the need to pump 14nm up into something it's not.

 

[drunkenmaster]

No one.

And since they're claiming 2.5x performance per watt for Polaris on 14nm LPP, and without the power reductions of HBM, I doubt TSMC's 16FF+ process is actually better.

As far as I gather, when people were claiming 16FF+ was better, that was comparing it to Samsung 14LPE (E being for 'early'). And Samsung say LPP is 15% lower power AND 15% higher performance than LPE. Which makes a 35% increase in performance per watt over LPE.

It doesn't, it's up to 15% performance and 15% lower power, meaning, in general 15% performance = same power, 15% lower power = same performance, best combination probably something like 10% more performance with same power.

 

[mahius]

I think a lot of people thought AMD were going to release first, because they were ahead of the game with HBM2. Most rumours are now saying HBM1/GDDR5X this year, so Nvidia could be first to release at 14/16nm.

Nah, I think AMD will launch first since they already mentioned the 'back to school' thing and all, whereas Nvidia have kept mum about Pascal's release.

At that, AMD have actually shown demos of 2 different Polaris GPUs, whereas Nvidia has shown and said nothing at all about Pascal. Either they don't have anything yet, AMDs demos caused them to re-evaluate if Pascal isn't as good or Nvidia are going to surprise us when it does launch. Maybe even all 3.

A shame that, would be nice to be able to choose from both, but I guess AMD will get the early adopters now.

 

[Kaapstad]

Nah, I think AMD will launch first since they already mentioned the 'back to school' thing and all, whereas Nvidia have kept mum about Pascal's release.

At that, AMD have actually shown demos of 2 different Polaris GPUs, whereas Nvidia has shown and said nothing at all about Pascal. Either they don't have anything yet, AMDs demos caused them to re-evaluate if Pascal isn't as good or Nvidia are going to surprise us when it does launch. Maybe even all 3.

A shame that, would be nice to be able to choose from both, but I guess AMD will get the early adopters now.

AMD will try and get to the market first come what may as they have no other option in the high end graphics card market. Fiji cards are nowhere near as popular as the GTX 980 Ti. NVidia on the other hand are under no pressure to get Pascal on the market as their old Maxwell cards are still doing ok.

I would be interested to know what yields NVidia are getting on old GM200 chips compared to upcoming Pascal chips. I suspect there is quite a difference which makes going to a new node less attractive if their existing cards are still selling well.

 

[Orangey]

If NV do somehow get out first it will be a compromise, using aggressively tuned early drivers (which get cranked back later) to grab some popcorn headlines in Ubisoft DX11 games or some such, maybe an overhyped feature or two like VXGI in a PR blitz to get everyone thinking NV are taking the world by storm when really it's more of a wet fart.

However it's more likely they'll lay in the cut and confirm what AMD put out first, then position against it.