Will the RX 480 really be available on the 29th?

[R0H1T]

So AMD is going to get boned by miners again? Hard to develop brand loyalty with gamers when miners drive the price of your card through the roof. And then in a few months the used market gets flooded with RX 480 once this mining is unprofitable? Ouch.

Yup very likely, the bigger question is whether retailers also wet their beaks, like they did last time D:

 

[Stormflux]

I seriously hope we don't run into another Crypto Miner 290X situation. Was there any good reputation earned for AMD during that time? All I remember is that I was wanting the 290X but it was no where to be found near MSRP for months and I ended up with an nVidia Kepler card, that I still am using. If I were able to get a 290X at the time I'd still be sitting pretty... I wouldn't have had to deal with nVidias planned obsolescence.

Anyway... my budget for a card is $300 US, so here's hoping to wide availability to a $299 RX 480X. Or if Miners sweep all the cards... maybe I can grab a 390X for super cheap.

 

[RussianSensation]

damn it rs, leave some for gamers :\

Haha don't worry I will. July 1st is Canada Day Weekend (Canada's birthday) and also my cousin's birthday. There will be too much drinking happening that week for me to pay attention to hitting F5.

What's interesting is that if Polaris 11 is 1280 shaders 128-bit and less than 120mm2, how can Polaris 10 be only 2304 shaders 256-bit with a ~ 232mm2 die size? Usually it doesn't require double the die size to double everything, but let's say there is a linear relationship, it would have made more sense to design a 20 CU (1280 shader) Polaris 11 128-bit, then straight up double that and release a 40 CU (2560 shader) Polaris 10. I have a feeling AMD is using salvaged RX 480X/485 die which is how they are able to price RX 480 so aggressively. Most likely notebooks and Apple are getting the full C10 version. It's also possible that's the missing link which could reconcile the missing "up to $299" Polaris 10 SKU.

 

[RussianSensation]

So AMD is going to get boned by miners again? Hard to develop brand loyalty with gamers when miners drive the price of your card through the roof. And then in a few months the used market gets flooded with RX 480 once this mining is unprofitable? Ouch.

Let's say you didn't know miners were interested in the cards, what difference would it make? Once the cards launch, it's free for all. A friend of mine is buying one as an HTPC; nothing to do with mining or gaming. Some other people buy it for distributed computing, OpenCL on a budget, etc. point is a GCN graphics card is a general purpose computing device. I don't think it's fair to criticize one or another user base for desiring RX 480 for their specific use case. Keep in mind that MrTeal assumed 480 is going to be as valuable to miners as an R9 390/390X but it may not be true. With Warhammer game bundle and R9 390's excellent mining horsepower, fire sale 390/390X cards may be more valuable. Not all miners will even want to buy 100s of cards at launch since they won't know how well 480 mines. Either way, miners aside, if AMD truly did mean their goal is to target 84-85% of all PC gamers, we are talking 6-7M sales a month then in a 40-50M annual market for dGPUs, then they better have anticipated that they'd need millions of cards ready. As I mentioned before, I am not happy with the way either AMD or NV launch cards nowadays. To outdo each other, both companies rush their launches, in turn not allowing sufficient time to build up inventory in a way Sony/Nintendo/MS do with console launches. Also, AIBs tend to launch cards months later which is very unfortunate since many gamers don't prefer blower reference cards.

As a side note, May-late October is my favourite time of the year. I am already starting to hit the gym, playing sports, going out, I have almost no desire to stay indoors when it's sunny and +28-30C outside. Once gloomy and cold winter sets in, I cannot get the summer back. Maybe I am getting older but during the summer I want to be outdoors. That's why it always puzzled me why AMD/NV love these April-June GPU launches. I mean consoles aren't generally launched during that time either. The 970/980 launch made perfect sense to me as NV spent months building up inventory and then we are closer to winter where there is a higher propensity to game indoors during the cold winter.

Anyway, I am still unsure how risky AMD's strategy is. As soon as NV launches 1050/1060/1060Ti, they will have top-to-bottom covered for 2016. They are already launching 1070/1080 notebooks by Q3 2016. OTOH, AMD is forefeiting the entire $350+ space for 2016? I mean even RX480 will have no chance competing in laptops against 1070/1080 since it appears Pascal still has a lead in perf/watt. It seems the gamble with foregoing GDDR5X for HBM2 was a risky move.

Don't you have a 970? Looking at benchmark scores, I would say 1070 is the minimum worthwhile upgrade anyway.

 

[coercitiv]

I mean even RX480 will have no chance competing in laptops against 1070/1080 since it appears Pascal still has a lead in perf/watt.

That's my problem with their product as well: if they can't come close in perf/watt on desktop, there's little chance they can do that in mobile. Sure, the new 14nm might have let them down at higher frequencies and may actually be better than 16nm at lower speed, but I kinda find that hard to believe. (we know a bit about that from the iPhone launch, but that was a while ago and things might have already changed considerably)

Either they get somewhat close to the claimed 2.8x perf/watt improvement in desktop as well, with that "150W Board Power" being just a placeholder to indicate max card power design and TDP being significantly lower than that, or they likely lose in the mobile market as well.

The Kepler chip inside my notebook is grinning at me :twisted:

 

[Elixer]

I forgot all about the miners.
I don't think that AMD will care who buys it, but, I am betting the markup on these things on fleabay will be like the 290 fiasco.

Forgot where I read it, but, it was mentioned (one of the semi trade mags) that they can do around 100 wafers per hour, and each wafer has X amount of chips, and out of those X, we have Y good ones. I wonder how long validation testing takes as well?
Would be very interesting to know how many they can crank out per day, followed by how long the AIBs validate the cards.
We do know that fab 8 (NY) can do 60k 300 mm wafers/month, and if someone wants to do the math of the polaris chip die size, we can get a ballpark figure of how many possible chips there are, assuming AMD put up the cash for 100% run on the 480.

We also don't know if AMD is going to dual source these chips, Samsung has 3 fabs that can do 14nm, and there is only 1 fab that GF has that can do 14nm (NY one). Would make sense, since it is supposed to be Samsung's tech being used in the NY plant.

 

[Qwertilot]

Well basically anything other than relying entirely on a process being transported to GF from Samsung (dual sourcing, TSMC etc) would have been a good idea!

Sadly not really one AMD can afford with their requirements to order lots of stuff from GF every year and their CPU shipments having fallen so far

The money thing is also why they shipped chips these sizes - they're the biggest things that fit into consoles, semi custom chips etc, which is the market they seem to be relying on to survive long term. Also notebooks, if they can get any presence back there.
(With fair reasons to expect medium term success.).

 

[EndlessWargasm]

I don't think that AMD will care who buys it

I suspect they very much do. The sky high retail prices were partly responsible for the 290/290x (IMO underserved) terrible reputation.

Not only did it negatively impact AMD in the eyes of its most profitable customers (gamers), it didn't even get to reap the monetary reward, with retailers doing the actual gouging.

If AMD could have anticipated the mining craze it could have made fat stacks during this time, positioning the Hawaii GPU as a 'professional' mining chip initially, then re releasing it down the way as a gaming part.

Sadly AMD is not a company that appears to be particularly agile or able to pivot, as exemplified by it trudging down the bulldozer road for 5+ years.

I worry that a presumably supply constrained 14nm part (RX 480) is going to face similar pressures from Ethereum mining.

 

[maddie]

I'm not sure who I should have pity on, but as I said, I'm just a random person with a garage. There's a lot of people with a powered garage.

Regardless, I think you're being extremely optimistic. This isn't the PS4, and I would be shocked if they had several hundred thousand units available on launch day. In an environment where every 480 is possibly earning $30/week, miners are incentived to purchase and deploy early. I expect the initial shipments to sell out pretty much immediately, and supply to be short for awhile. You might poo poo mining, but it significantly affected the launch of Hawaii on a mature 28 nm process and the market cap and production rate of Litecoin (the primary driver at the launch of Hawaii) was less than what Ethereum is doing today. A 480 is going to be extremely challenging to find on the 29th of June, if that's when they do launch.

I don't know if I'm phrasing my words or thoughts badly, but how did you get that impression? Seriously, can you explain.

Also, I remembered 28nm as being newer than this 14 nm appears to be. Might be wrong.

 

[flopper]

the 480 looks like a slam dunk check mate card.
90% of users can upgrade/update into a VR enabled world to come.

Hard launch 29 June seems like its happening.
Great

 

[maddie]

I forgot all about the miners.
I don't think that AMD will care who buys it, but, I am betting the markup on these things on fleabay will be like the 290 fiasco.

Forgot where I read it, but, it was mentioned (one of the semi trade mags) that they can do around 100 wafers per hour, and each wafer has X amount of chips, and out of those X, we have Y good ones. I wonder how long validation testing takes as well?
Would be very interesting to know how many they can crank out per day, followed by how long the AIBs validate the cards.
We do know that fab 8 (NY) can do 60k 300 mm wafers/month, and if someone wants to do the math of the polaris chip die size, we can get a ballpark figure of how many possible chips there are, assuming AMD put up the cash for 100% run on the 480.

We also don't know if AMD is going to dual source these chips, Samsung has 3 fabs that can do 14nm, and there is only 1 fab that GF has that can do 14nm (NY one). Would make sense, since it is supposed to be Samsung's tech being used in the NY plant.

Try for yourself.
http://isine.com/DieYieldCalculator.html

Samsung was claiming around 0.2 defects/cm^2. Accurate? Play with higher.
You only get the Full working die yield not partial ones.
Rx480 specs strongly suggests a RX480X 40 CU model

In any case, not that many wafers are needed and we can safely assume that GloFlo is not swamped by orders.

edit:
I get 152 full die out of 238 possible using a 0.2/cm^2 defect density. Harvested will be a lot of the remaining 86.

Ballpark numbers, yes, but still somewhere to start.

 

[MrTeal]

I don't know if I'm phrasing my words or thoughts badly, but how did you get that impression? Seriously, can you explain.

Also, I remembered 28nm as being newer than this 14 nm appears to be. Might be wrong.

For 28nm, it was pretty mature by the time Hawaii launched in late October 2013. By that point they had been in volume production of Tahiti for almost two years, so the jump from a ~350mm² die to a 440mm² on that process wasn't a big step. 14nm LPP is much newer than that.

Sorry if I misinterpreted your intention though. It appeared that you were suggesting that mining card sales wouldn't make an impact on launch availability for the 480, as AMD is planning on selling many hundreds of thousands per month. I'd disagree with that assessment, though we'll have to wait to see how well it pans out.

 

[maddie]

For 28nm, it was pretty mature by the time Hawaii launched in late October 2013. By that point they had been in volume production of Tahiti for almost two years, so the jump from a ~350mm² die to a 440mm² on that process wasn't a big step. 14nm LPP is much newer than that.

Sorry if I misinterpreted your intention though. It appeared that you were suggesting that mining card sales wouldn't make an impact on launch availability for the 480, as AMD is planning on selling many hundreds of thousands per month. I'd disagree with that assessment, though we'll have to wait to see how well it pans out.

I now realize that you wrote Hawaii. I swear I saw Tahiti when reading the post, so yeah, 28nm was very mature.

I have wondered if Rory Read was a victim of that fiasco?

I will be doing a cost analysis of what would be the crossover price for 290 class and P10 models to justify the mining choice. My electricity cost is around $0.05 KWhr.

 

[thilanliyan]

So AMD is going to get boned by miners again? Hard to develop brand loyalty with gamers when miners drive the price of your card through the roof. And then in a few months the used market gets flooded with RX 480 once this mining is unprofitable? Ouch.

I don't think speed-wise it will match a 290 for example, so unless it really sips power, for someone like myself there is no reason to upgrade. My 290 for example can do 31MHs at ~250w. IF a 480 can get anywhere near that speed at much lower power usage, I'd bite. If it's only like 25-50w less I wouldn't bother. I'm kinda hoping it doesn't reach those speeds lol...disassembling my watercooling loop is gonna be a pain in the butt.

 

[Dave2150]

unless it really sips power, for someone like myself there is no reason to upgrade. My 290 for example can do 31MHs at ~250w. IF a 480 can get anywhere near that speed at much lower power usage, I'd bite. If it's only like 25-50w less I wouldn't bother. I'm kinda hoping it doesn't reach those speeds lol...disassembling my watercooling loop is gonna be a pain in the butt.

Didn't you read/see that it's a sub 150W part with only 1x6pin PCI-E power connector?

 

[thilanliyan]

Didn't you read/see that it's a sub 150W part with only 1x6pin PCI-E power connector?

Yeah I did, but I don't see anything about mining speeds, so how can I know whether MHs/w is good or not?

 

[Kenmitch]

We should have more info on availability in the next couple of weeks I'd imagine.

 

[poofyhairguy]

damn it rs, leave some for gamers :\

It will be like buying hot concert tickets but on Newegg, for sure. I plan to buy one or two even if there aren't any reviews yet just for mining. The gamers waiting to see if its actually any good will be screwed, by then Newegg will jack up whatever second shipments they get just to milk miners.

 

[SteveGrabowski]

Let's say you didn't know miners were interested in the cards, what difference would it make? Once the cards launch, it's free for all. A friend of mine is buying one as an HTPC; nothing to do with mining or gaming. Some other people buy it for distributed computing, OpenCL on a budget, etc. point is a GCN graphics card is a general purpose computing device. I don't think it's fair to criticize one or another user base for desiring RX 480 for their specific use case.

I'm not criticizing mining with it, I mean it's free money if this card ends up being a 150W 390x. I just meant it's a real kick in the nuts to AMD. Gamers who want these cards at MSRP are going to be pissed, and then when they buy these cards used with high fail rates after 24/7 mining use they'll blame AMD for making crap.

 

[maddie]

It will be like buying hot concert tickets but on Newegg, for sure. I plan to buy one or two even if there aren't any reviews yet just for mining. The gamers waiting to see if its actually any good will be screwed, by then Newegg will jack up whatever second shipments they get just to milk miners.

If AMD was smart, they would also sell some direct. Nvidia does it.

 

[poofyhairguy]

If AMD was smart, they would also sell some direct. Nvidia does it.

If we are all smart we will buy as many as we can when we can and scalp them to miners on Ebay.

I might buy two and sell one to help pay off the cost on the first one if I can get two. Come on Frys, don't let me down!!!

 

[maddie]

If we are all smart we will buy as many as we can when we can and scalp them to miners on Ebay.

I might buy two and sell one to help pay off the cost on the first one if I can get two. Come on Frys, don't let me down!!!

The GPU arbitrage market.

 

[boozzer]

If we are all smart we will buy as many as we can when we can and scalp them to miners on Ebay.

I might buy two and sell one to help pay off the cost on the first one if I can get two. Come on Frys, don't let me down!!!

if AMD were smart they would be prepare for the miners. don't get caught off guard like they did with the 200 series. have no stock due to miners for the first 6 months, when they do the market was then flooded with used mining cards.

the first initial 3-6 months is very important. especially if they want market share.

 

[poofyhairguy]

if AMD were smart they would be prepare for the miners..

What can AMD do? Have a mandatory survey before someone buys a GPU to make sure they aren't a miner? Force companies like Newegg to only sell one per customer? I just don't see any practical way AMD can "prepare" for the miners. They will make every card they can, and sell every card they make.


Well.....that isn't completely true. If AMD really really wants to keep miners away they need to sabotage mining in their driver and make it so Polaris only gets like 11MH. But then that GPU better damn well beat the GTX 1060 by a good margin day one or I don't know who is gonna buy them.

 

[boozzer]

What can AMD do? Have a mandatory survey before someone buys a GPU to make sure they aren't a miner? Force companies like Newegg to only sell one per customer? I just don't see any practical way AMD can "prepare" for the miners. They will make every card they can, and sell every card they make.


Well.....that isn't completely true. If AMD really really wants to keep miners away they need to sabotage mining in their driver and make it so Polaris only gets like 11MH. But then that GPU better damn well beat the GTX 1060 by a good margin day one or I don't know who is gonna buy them.

prepare as in having enough product to sell during the first 3-6 months without prices going up 100%-200% miners are customers too amd needs to cash in! both gamers and miners.