[fudzilla] Jan. 9th I'd paper launch only for hd7950

[tviceman]

http://www.fudzilla.com/graphics/item/25380-amd-hd-7950-wont-show-up-on-january-9th

Pretty crappy move by AMD to have back to back paper launches if this is true. There is no excuse or reason to do this within the context of the entirw situation.

 

[Panopticon]

If this is true I think it would be a very poor choice for AMD. I just built a new rig last night and was waiting to snag one of these on the 9th. If AMD does not release this card I would have to choose between the 7970 which is a bit overkill for 1080p or an older card... I also think Nvidia may drop some of their prices to counter what AMD is releasing. Imagine if they drop the price of the 1.5GB 580 to around $449-$399... I think AMD really needs to come out guns blazing and they should also be aware that alot of people are going to be building PC's with all this xmas money and giftcards floating around... Fud seems to have some decent sources but I feel like they may be wrong here... its not like the 7950 is a different card its just a slightly gimped 7970. I'm buying a card Jan 9 and I hope AMD will deliver the goods.

 

[MrTeal]

I wonder what this says about the yields at TSMC? On a new process I would have thought they would have a good supply of chips with one defective compute unit arrays to sell as 7950s.

 

[Lonyo]

You would rather not know how the cards will perform, and just have people talking crap?
If they aren't able to have the cards out, surely it's better to have information to enable people to make a purchasing decision when they know the cards are coming?

I never got the problem with paper launches, as long as they aren't too far in advance.

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row
Here's a Sandy Bridge preview from August 2010.
The chips arrived in January. That's a long time. Did you have a problem with it?

For most people on these forums, they are debating the performance of the new AMD chips, how they will perform relative to the old ones etc.
Many people are saying that X (new card) will be slower than Y (current card), so Y is a good purchase since we know the price that X will be.
Well we don't know how it will perform, but if we do know, people can either choose to buy the alternative now, or buy the new card when it does come out.

IMO there's nothing wrong with paper launches if they are very clear about when cards will arrive, and if cards do arrive on time.

I'd personally LOVE if NV paper launched their next gen cards right after AMD launched their 7000 series, I'm aiming to buy a new graphics card when Ivybridge comes out, but I could hold out on my HD6850 for a while if the future holds something that's better value, but if I don't know how future products will perform, should I wait and hope, or just bite the bullet?

 

[Arkadrel]

@MrTeal,

maybe they dont want 7950's disturbing lower tier cards (6xxx series)?
maybe they want to push people to go with a 7970 (instead of a 7950)?
maybe the amount of chips that fail to become 7970's (and end up as 7950's) is low?
(and their stockpileing some for abit, due to expected sales, and wanting to keep supply fed after relase)

who knows.... there s probably a reason for this though.

 

[rgallant]

You would rather not know how the cards will perform, and just have people talking crap?
If they aren't able to have the cards out, surely it's better to have information to enable people to make a purchasing decision when they know the cards are coming?

I never got the problem with paper launches, as long as they aren't too far in advance.

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row
Here's a Sandy Bridge preview from August 2010.
The chips arrived in January. That's a long time. Did you have a problem with it?

For most people on these forums, they are debating the performance of the new AMD chips, how they will perform relative to the old ones etc.
Many people are saying that X (new card) will be slower than Y (current card), so Y is a good purchase since we know the price that X will be.
Well we don't know how it will perform, but if we do know, people can either choose to buy the alternative now, or buy the new card when it does come out.

IMO there's nothing wrong with paper launches if they are very clear about when cards will arrive, and if cards do arrive on time.

I'd personally LOVE if NV paper launched their next gen cards right after AMD launched their 7000 series, I'm aiming to buy a new graphics card when Ivybridge comes out, but I could hold out on my HD6850 for a while if the future holds something that's better value, but if I don't know how future products will perform, should I wait and hope, or just bite the bullet?

+1
you got to love the post on the inter-vent

 

[Panopticon]

I agree that AMD might feel like the 7950 will make it harder to deplete stock of 6790s even though it costs more then a 6970. By the way have they even officially EOL the 6k series or are they still being manufactured?

 

[Hauk]

I wasn't holding my breath for 7950 on Jan 9th. It's not perfect science, but with no leaked photos circulating, it's not in the channel.

 

[Idontcare]

You would rather not know how the cards will perform, and just have people talking crap?
If they aren't able to have the cards out, surely it's better to have information to enable people to make a purchasing decision when they know the cards are coming?

I never got the problem with paper launches, as long as they aren't too far in advance.

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row
Here's a Sandy Bridge preview from August 2010.
The chips arrived in January. That's a long time. Did you have a problem with it?

For most people on these forums, they are debating the performance of the new AMD chips, how they will perform relative to the old ones etc.
Many people are saying that X (new card) will be slower than Y (current card), so Y is a good purchase since we know the price that X will be.
Well we don't know how it will perform, but if we do know, people can either choose to buy the alternative now, or buy the new card when it does come out.

IMO there's nothing wrong with paper launches if they are very clear about when cards will arrive, and if cards do arrive on time.

I'd personally LOVE if NV paper launched their next gen cards right after AMD launched their 7000 series, I'm aiming to buy a new graphics card when Ivybridge comes out, but I could hold out on my HD6850 for a while if the future holds something that's better value, but if I don't know how future products will perform, should I wait and hope, or just bite the bullet?

+1, totally agree

I wonder what this says about the yields at TSMC? On a new process I would have thought they would have a good supply of chips with one defective compute unit arrays to sell as 7950s.

Early ramp is always an issue of volume, poor yields compound the problem but even if yields were 100% there are still only so many wafers available for 28nm production capacity at this time.

I think you are seeing AMD doing what smart business would do with a limited supply of anything, they reserve it for the price-premium products and charge accordingly to whatever the market will bear such that supply/demand is balanced while supply (TSMC 28nm capacity limited) is maximized.

I agree that AMD might feel like the 7950 will make it harder to deplete stock of 6790s even though it costs more then a 6970. By the way have they even officially EOL the 6k series or are they still being manufactured?

I don't think this has anything to do with it. They managed the inventory transition from 5850 to 6950/6970 rather well.

I think this has more to do with not wanting to revisit the supply debacle that was the original 5850/5870 release when 40nm capacity was really tight.

They had a chance to do it all over again here with Tahiti, and I think we are seeing them apply lessons learned from 40nm to their approach in rolling out and transitioning their product lines to 28nm.

 

[MrTeal]

Early ramp is always an issue of volume, poor yields compound the problem but even if yields were 100% there are still only so many wafers available for 28nm production capacity at this time.

I think you are seeing AMD doing what smart business would do with a limited supply of anything, they reserve it for the price-premium products and charge accordingly to whatever the market will bear such that supply/demand is balanced while supply (TSMC 28nm capacity limited) is maximized.

Well, I suppose that is why I'm a little surprised by this. I would expect early in the new process that there would be plenty of dies with issues that would be recycled into 7950s, 7890s, etc. If you have a high defect density as we normally see at the start of a node you would expect to have a big supply of chips with issues vs chips that function perfectly.

I guess there's a few possibilities. One is that the 28nm process is already yielding quite good and there simply aren't that many dies that need to be written off. Another is that they just don't want to launch the two simultaneously, but if that was the case why schedule the launch dates at the same time?
Or, maybe those Sapphire leaks are correct about a 2304SP part, and that's not a new die but the full Tahiti, and the Tahiti XT is already a harvested part. That doesn't seem correct since my understanding is that is not the case and the 7970 is a complete die, but that scan in the other thread was interesting.

 

[Nintendesert]

I actually kinda like paper launches. The only people that really have an issue are those playing the red vs green game. Paper launches to me give me non-fud time with real numbers to decide whether I'm buying or not. Then when the cards actually hit I know whether I'm fighting to get a card or skipping. Due to my location I was able to weigh my choices and get a 5870 and a 570. Now I get to pass on the 7970 and judge all the reviews of the 7950 without rushing to claim inventory.

 

[Idontcare]

Well, I suppose that is why I'm a little surprised by this. I would expect early in the new process that there would be plenty of dies with issues that would be recycled into 7950s, 7890s, etc. If you have a high defect density as we normally see at the start of a node you would expect to have a big supply of chips with issues vs chips that function perfectly.

I guess there's a few possibilities. One is that the 28nm process is already yielding quite good and there simply aren't that many dies that need to be written off. Another is that they just don't want to launch the two simultaneously, but if that was the case why schedule the launch dates at the same time?
Or, maybe those Sapphire leaks are correct about a 2304SP part, and that's not a new die but the full Tahiti, and the Tahiti XT is already a harvested part. That doesn't seem correct since my understanding is that is not the case and the 7970 is a complete die, but that scan in the other thread was interesting.

Well, if the current Tahiti is a "limited run" for reasons relating to Apoppin's unpopular speculation article then it would make sense for them to not bother also releasing SKU's based on harvest die.

To much project overhead in that case. Just get your halo part out the door, meet your "we'll launch in 2011" on record obligations to the shareholders and the SEC, but hold back the volumes until its actually coming off the lines in the volumes that your AIBs expect and demand.

 

[tviceman]

You would rather not know how the cards will perform, and just have people talking crap?
If they aren't able to have the cards out, surely it's better to have information to enable people to make a purchasing decision when they know the cards are coming?

I never got the problem with paper launches, as long as they aren't too far in advance.

That is a slippery slope to walk. What is too far in advance and a particular point before the product is released be more relevant than any other point? What is the difference between Dec. 22nd and Jan. 9th for the hd7950 nda if it won't be available on either date anyways?

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row
Here's a Sandy Bridge preview from August 2010.
The chips arrived in January. That's a long time. Did you have a problem with it?

I had no problems with that article since it was a PREVIEW of the product and it was also noted that performance could be changed when the actual product launches.

 

[GaiaHunter]

This paper launch vs no info at all is very debatable and quite prone to different opinions.

For example, when there was no information about BD everyone wanted info and wished AMD released info about BD.

I think these "paper launches" are ok if and only if:
- Final product available to public is equal (or better) to the one being previewed;
- Release date of the product is given and then it is actually met;
- At release date product availability needs to be good and then product needs to be decently available for the near term future;
- Information about MSRP should be given if final release is close (couple of weeks or lower seems to be close) and at release should be no higher than MSRP given.

If these criteria are met these releases are fine by me.

 

[Stuka87]

Paper launches in advance of actual product availability is nothing new. Intel typically paper launches a few months in advance, and even AMD did when they released Bulldozer and the 5K series cards.

Personally I like it, it lets me know what to buy. And since I want a new card, this lets me better choose what i want.

 

[Fallen Kell]

Intel and AMD paper launch CPU's months ahead of time because there is an entire software stack that needs to make their updates to take advantage of any new features in the CPU. Specifically, compilers need to all be re-written to include any new instruction sets, with many of the most important compilers being open source (the GCC set and all its derivatives). Then you need to have actual software re-compiled to include detection of the new CPU and use code paths that were rebuilt for use with that CPU. In other words, something that takes months, if not years, to happen. That is why CPUs get a paper launch 6+ months in advance.

Video cards, on the other hand, simply execute existing frameworks (i.e. DX9, DX10, DX11, OpenGL, etc). The video card's drivers are in a sense like the compiler in my above CPU reference, the difference here is that it is more like a scripting language in that you don't re-compile the code since it is done on the fly. And since Nvidia/AMD are the ones in charge of releasing and updating the driver, they already had the 6+ months time it takes to implement those changes since they designed the hardware and knew of all the added features months if not years ahead of time.

 

[ddarko]

The problem is not paper launches. The problem is paper availability for weeks extending to months after the product goes on sale. No one would care about a paper launch a month before its retail availability date if you could buy the thing on that date. But that hasn't been the case with graphics cards for years now, even for products that supposedly had "hard" launches but sold out in five minutes. I wish the companies would build up enough stock before launching. Having adequate stock would also dampen or eliminate the other scourge of these launches, the retailer markups. Unfortunately, I expect the 7970 retail availability will follow the familiar pattern of selling out instantly, retail prices shooting up to $600-650 even when the cards do briefly come back into stock before finally falling to the $550 launch price two months after it became "available."

 

[Leyawiin]

The problem is not paper launches. The problem is poor availability for weeks extending to months after the product allegedly goes on sale. No one would care about a paper launch a month before its retail availability date if you could buy the thing on that date or reasonably soon after. But that hasn't been the case with graphics cards for years now, even for products that supposedly had "hard" launches but sold out within 15 minutes. I wish the companies would build up enough stock before launching. Having adequate stock would also dampen or eliminate the other scourge of these launches, the retailer markups.

Exactly. These GPU paper launches serve nothing more than to stymie the competition's high end sales by getting people to wait...and wait...and wait to purchase a card until they build up enough inventory to actually ship to vendors.

 

[Lonyo]

OK, lets put it this way:

MOST of the advice given on these forums recently has been along the lines of the following:
"New cards are coming soon. If you must buy now, then buy X (whatever is best). BUT, if you don't need to buy now, then you should wait if you can, and see how the new cards perform, because they might be a better deal or push prices down".

And you guys are saying that by giving us performance indication and pricing, AMD is doing something harmful?

Like ddarko says, paper launches aren't the problem IF they are given a date for a VERY hard launch. A paper launch with a promised release date which never really materialises, either due to more lateness, or insufficient stocks, is a bad paper launch and a bad thing, but you can have EXACTLY the same with a "hard" launch, as we have had in the past, so it's not a paper launch vs hard launch issue.


For all the people on the forums who are or were thinking of upgrading, but wanted to wait to see what happened, having this information is USEFUL.
For some of the people who weren't going to wait, this information could persuade them to wait and they might get a better deal.
For some of the people who weren't going to wait, it doesn't affect them.
For all the people who were going to buy no matter what, they get given performance information ahead of time, they just have to wait like they would do anyway, or they might change their mind and pick a different product up, e.g. a second card for Crossfire or SLI, if they don't think the gains are worth it.

Who loses out? Which group doesn't benefit or have no change?

 

[wahdangun]

http://www.fudzilla.com/graphics/item/25380-amd-hd-7950-wont-show-up-on-january-9th

Pretty crappy move by AMD to have back to back paper launches if this is true. There is no excuse or reason to do this within the context of the entirw situation.

the only crappy move is nvdia didn't even have the card to compete with AMD

 

[notty22]

AMD Radeon HD 7950 delayed to February Week 1

AMD Radeon HD 7950 delayed to February Week 1
By Hilbert Hagedoorn, January 1, 2012 - 1:27 AM N/A



We just got notified by AMD that thje launch of the Radeon HD 7980 has been delayed by at least three weeks. During the recent launch of the AMD Radeon HD 7970, a number of media outlets expressed varying levels of concern regarding the separation of the launch and market availability.
To avoid similar concerns relative to the AMD Radeon HD 7950, AMD have decided to move the launch of the AMD Radeon HD 7950 to early February. This will ensure the launch aligns with availability of AMD Radeon HD 7950 cards in the channel.
Availability of the AMD Radeon HD 7970 has not changed. It will be available on January 9th.

 

[VulgarDisplay]

I personally would have preferred the paper launch simply to give me time to digest some benchmarks before I decide on the 7970 or 7950.

 

[MrTeal]

Will the NDA on HD7950 reviews be moved to February as well, or can we hope to see reviews and benchmarks in a week at the original release date?

 

[Janooo]

Paper launch a few weeks before real availability is a good thing.
This helps Anand and other sites to do their job properly.

When the cards are siting at partner's warehouses that's the time when leaks happen and unofficial reviews pop up on the internet. That does not make any sense.

I have not seen any 7950 review yet only 7970.
If we see reviews of 7950 around Jan 9th then I would expect 3-4 weeks availability after that.

Hard launch is a hard thing to do without screwing up review sites.

 

[Panopticon]

Was thinking about this release strategy earlier and something crossed my mind that hadn't been mentioned yet so I thought I would chime in. After looking at the initial benchmarks of the 7970 and then seeing all the overclock benchmarks start seeping through I couldn't help but wonder if AMD is concerned about the performance gap and is holding the 7950 back until non reference 7970's come out.

The reason I think this is a possibility is because if an AIB puts out a 7950 with much better cooling and maybe even a factory oc then it might seriously detract from 7970 sales if its cheaper and the performance is close to that of a ref 7970. Don't forget the earlier rumors that non reference 7950's would be available day 1.

The tahiti core seems to take well to overclocking and I was just on a forum where someone has an early 7970 and is running 12XX mhz with the stock cooler. Can't wait to see if non reference oc 7950=ref 7970