No New GPUs from AMD for the Bulk of 2013 [TechPowerUp]

[BallaTheFeared]

Yikes, what an awful future you've painted 🙁

 

[ocre]

Look at the last 2 quarters of AMD's GPU division's earnings. Even if AMD didn't sell a single GPU in all of 2013, it wouldn't kill the company. And even if HD8000 was 10x faster than GTX700, it wouldn't save the company by any measures. The key markets where AMD is losing $ (CPUs, APUs, servers, etc.) are actually what can kill the company because the losses there are major and they seem never-ending. What AMD needs to execute well on are Richland, Kabini, Piledriver, etc., not HD8000 GPU that might make them $25 million in profits every 3 months. Even if AMD could launch a GPU as fast as the Titan for $1K, it wouldn't even make a dent in their financials.

wow.

why spend so much time preaching amd graphics then. Overall it means nothing. I am kinda blown away with your post really.

That second sentence would make the all time top quotes of all quotes.

But i completely disagree here. I think every million AMD makes is very important. Perhaps their management thinks more like what you say but i sure hope not.

 

[tviceman]

Look at the last 2 quarters of AMD's GPU division's earnings. Even if AMD didn't sell a single GPU in all of 2013, it wouldn't kill the company. And even if HD8000 was 10x faster than GTX700, it wouldn't save the company by any measures. The key markets where AMD is losing $ (CPUs, APUs, servers, etc.) are actually what can kill the company because the losses there are major and they seem never-ending. What AMD needs to execute well on are Richland, Kabini, Piledriver, etc., not HD8000 GPU that might make them $25 million in profits every 3 months. Even if AMD could launch a GPU as fast as the Titan for $1K, it wouldn't even make a dent in their financials.

Russian I'm kinda shocked you're defending this whole situation. And especially in this manner.

AMD needs all the help it can get right now. They can't afford to relax on a single segment within the company. The company needs to fire on all cylinders. Good companies do not postpone a release of a product line because "it won't kill them" nor do good managers breed that kind of complacency. The company is bleeding money in most of it's product divisions, the last thing they can possibly deal with is one of their only profitable divisions slipping in sales and profits due to stale products and / or increased competition. Fresh products garner interest, interest garners sales. Right now they are buying keys from publishers to just keep their products selling (in lesser quantities than a year ago). In other words, they are bringing in less revenue off their graphics cards than they were a year ago because they are having to market significantly more for a 15% decline year over year sales and a 5% decline vs. the previous quarter.

Complacency and bad management is what got AMD into their financial disaster to begin with. Delaying a product lineup by 2 quarters is exactly what they don't need to do right now.

 

[blastingcap]

Russian I'm kinda shocked you're defending this whole situation. And especially in this manner.

AMD needs all the help it can get right now. They can't afford to relax on a single segment within the company. The company needs to fire on all cylinders. Good companies do not postpone a release of a product line because "it won't kill them" nor do good managers breed that kind of complacency. The company is bleeding money in most of it's product divisions, the last thing they can possibly deal with is one of their only profitable divisions slipping in sales and profits due to stale products and / or increased competition. Fresh products garner interest, interest garners sales. Right now they are buying keys from publishers to just keep their products selling (in lesser quantities than a year ago). In other words, they are bringing in less revenue off their graphics cards than they were a year ago because they are having to market significantly more for a 15% decline year over year sales and a 5% decline vs. the previous quarter.

Complacency and bad management is what got AMD into their financial disaster to begin with. Delaying a product lineup by 2 quarters is exactly what they don't need to do right now.

Someone said that TSMC 28nm wafers cost 1/3 as much now as they did at launch, with presumably higher yields too, so in terms of profitability AMD is probably making more money on their GPUs than you might think. And it may actually make sense to stay on 28nm longer, whether with 7xxx or 8xxx. In fact if NVDA and AMD wanted to collude, they could both refuse to launch anything new and milk 28nm for as long as possible.

I've read reports that AMD says it definitely will be launching another series by the end of the year, but that doesn't contradict the OP which was talking about the "bulk of 2013," not "all of 2013." You could have a Christmastime launch of HD8xxx and have both statements be true.

 

[RussianSensation]

why spend so much time preaching amd graphics then. Overall it means nothing.

It's the only truly great product/business segment they have right now. What other products are they going to talk about? The crappy ones?

I am kinda blown away with your post really.

I separate finance from marketing/products. On the financial side, AMD's desktop GPU division is nearly irrelevant for the firm's long-term survival (i.e., to prevent them from going bankrupt) since it accounts for less than 10% of the firm's stock price. Even if AMD doubled the revenues of their GPU business and doubled the profits there, they'd lose more in all their other products lines, which would result in a total net cash flow loss for the firm. It would not save the company in any shape or form. It's pure finance, not emotions.

The only way for them to turn the company around is to either improve their CPU/APU business, focus on new products in different product segments or invent something new/revolutionary and/or license existing technologies.

AMD doesn't even have the resources to compete with NV in the Tesla / HPC markets and they surely won't double their GPU revenues any time soon either. Therefore, the answers for AMD's long-term sustainable financial health are not discrete desktop GPUs. If you've always followed what NV said, GeForce is a cost center for them. If NV's professional GPU segment didn't exist, the entire GeForce brand would be barely profitable. And the only way products like Tesla are possible is because of the R&D costs that are subsidizes by the GeForce brand. JHH has mentioned this on various occasions to investors.

Russian I'm kinda shocked you're defending this whole situation. And especially in this manner. Good companies do not postpone a release of a product line because "it won't kill them" nor do good managers breed that kind of complacency. Complacency and bad management is what got AMD into their financial disaster to begin with. Delaying a product lineup by 2 quarters is exactly what they don't need to do right now.

-- AMD never delayed HD8000 series by 2 quarters since it was never officially announced anywhere that they would be launching desktop HD8000 in before Q3 2013. You have a source for that?

Aren't you assuming desktop HD8000 parts were somehow slates to launch in the first half of 2013? Based on what? rumors? When is Maxwell a launching in 2014? We have no idea. All AMD has stated is that sometime in 2013 they will launch Sea Islands.

-- I am not in any way defending the delay of HD8000 series. What I am explaining and what I said for months last year, AMD is a CPU company, not a GPU company from a financial point of view. What do I mean by that? It means that on the whole the GPU division cannot sustain the firm. It's not financially possible. Unless AMD turns its critical business around (CPUs, servers, APUs), hit a home run in the smartphone/tablet/micro-servers markets, they are toast. Discrete GPU sales won't do anything to change the situation since AMD has $2 billion in debt. Again, this is pure finance, not wishful thinking.

-- You yourself assumed GTX700 is launching by June. Based on what? Rumours, assumptions? No evidence.
http://www.tomshardware.com/news/Radeon-GeForce-Delay-GPU-Next-Generation,20838.html

 

[tviceman]

Someone said that TSMC 28nm wafers cost 1/3 as much now as they did at launch, with presumably higher yields too, so in terms of profitability AMD is probably making more money on their GPUs than you might think. And it may actually make sense to stay on 28nm longer, whether with 7xxx or 8xxx. In fact if NVDA and AMD wanted to collude, they could both refuse to launch anything new and milk 28nm for as long as possible.

I've read reports that AMD says it definitely will be launching another series by the end of the year, but that doesn't contradict the OP which was talking about the "bulk of 2013," not "all of 2013." You could have a Christmastime launch of HD8xxx and have both statements be true.

20nm isn't coming until 2014. Their southern islands (Sea Islands) refresh will be on 28nm, I am fully aware of both of these facts. I also realize that 20nm cards might not come until Q2 2014, or even later. But AMD's own roadmaps from this time last year showed Sea Islands launching in Q1 of this year. Just like GF100 and the rest of Fermi's lineup was delayed - a delay is a delay is a delay. If they are delaying for technical reasons or because inventories are currently too high, then that is one thing. But if they are delaying for the sake of delaying, then stupid stupid stupid stupid stupid.

 

[sontin]

AMD doesn't even have the resources to compete with NV in the Tesla / HPC markets and they surely won't double their GPU revenues any time soon either. Therefore, the answers for AMD's long-term sustainable financial health are not discrete desktop GPUs. If you've always followed what NV said, GeForce is a cost center for them. If NV's professional GPU segment didn't exist, the entire GeForce brand would be barely profitable. And the only way products like Tesla are possible is because of the R&D costs that are subsidizes by the GeForce brand. JHH has mentioned this on various occasions to investors.

Nonsense.
The Geforce business is their bread and butter. It makes more than half of the revenue and 66% of the operative income.
The only reason nVidia had money to went into the Tesla and Tegra business was because of the consumer business. The whole R&D of the architecture comes from the Geforce business. That's the reason why their professional business has gross margins between 50-70%.

 

[ShintaiDK]

Someone said that TSMC 28nm wafers cost 1/3 as much now as they did at launch, with presumably higher yields too, so in terms of profitability AMD is probably making more money on their GPUs than you might think.

We already know the balance sheets, so its no secret. And they aint exactly earning much on GPUs. 22mio profit in Q4. And a -15% YoY on revenue. To compare they earned 34mio in Q1.

Losing revenue as chip design costs goes up aint helping either in terms of the future.

 

[RussianSensation]

I've read reports that AMD says it definitely will be launching another series by the end of the year, but that doesn't contradict the OP which was talking about the "bulk of 2013," not "all of 2013." You could have a Christmastime launch of HD8xxx and have both statements be true.

Bingo. AMD never once officially stated anywhere that HD8000 desktop parts are launching before end of summer 2013. Everything we've read on GTX700 / HD8000 launching by Q1-Q2 2013 were just rumours. It's incredible that people now write articles online and create headlines for hits, rather than factual information. A delay is this:

First official roadmap

Second official roadmap

Nonsense.
The Geforce business is their bread and butter. It makes more than half of the revenue and 66% of the operative income.
The only reason nVidia had money to went into the Tesla and Tegra business was because of the consumer business. The whole R&D of the architecture comes from the Geforce business. That's the reason why their professional business has gross margins between 50-70%.

This is why I hate getting into financial discussions here. You guys don't even get the basics of R&D allocation between different business segments. It's simply an accounting treatment. If you didn't have Tesla, the GeForce business on its own would be barely making NV any $. Go read up on NV's history and their profits during historical years and which business segments generated the most profits. It wasn't desktop NV GPUs, not by a mile. NV's GeForce business is doing a lot better recently because NV raised prices through the roof per mm2 of die sold with Kepler. Things weren't looking so hot previous generations.

Many of you also seem to keep missing where AMD has been losing the most market share in GPUs. AMD is refreshing the most important segment of GPUs - the mobile one, because this is where they lost the most market share. This is the more critical segment to get design wins, not desktop cards. On the desktop side HD7000 series did alright vs. GTX600. What they need are strong design wins with HD8000 mobile parts. The profits of desktop GPUs on their own for AMD are so small, they are meaningless in the context of the entire AMD business. In Q3 2012 alone, AMD lost more $ than the entire GPU business makes in 1 year. That's including profits in mobile and desktop dGPUs. In Q4 2012, AMD lost more $ in 1 quarter than they make in 3-4 years in GPUs.
http://www.anandtech.com/show/6690/...ing-out-a-rough-year-looking-towards-the-next

Look at AMD's Revenues and Net Income in 2011 and 2010.
http://www.anandtech.com/show/5465/...ort-169b-revenue-for-q4-657b-revenue-for-2011

What's been happening to the firm since then? What business segments suffered the most? It's not GPUs, not even close. AMD is dying as a CPU/server business. Those are the facts. The GPU business hasn't been strong in years. GPUs are not what's killing AMD. Your business doesn't drop from $6-7 billion in Revenue to barely over $1B because your GPU sales went down. It's all there, go read up financials. AMD's GPU graphics is starting to look more and more important (hyped) because everything else is losing so much $. It's the only bright spot left. The company is a former shelf of itself. AMD was a CPU/server business, the GPU business was just there to make sure the transition to APUs with ATI's graphics tech would allow AMD to have class leading CPUs. That strategy failed. What is left is a CPU business that's dying and a GPU business that hasn't made a lot of $ since ATI's acquisition to keep the firm afloat.

AMD's desktop GPU graphics would never sustain AMD. It's not financially possible given their balance sheet and other trends of their product lines and their cash flows. Unless AMD does any of the things I mentioned, or someone buys them out, their desktop GPUs will do nothing to keep them alive.

 

[RussianSensation]

Yikes, what an awful future you've painted 🙁

To the people who are shocked by my comments, how many AMD GPUs have you purchased in the last 5 years? If you are so afraid of AMD going down now but you did nothing to help them all this time, then it shouldn't worry you if they fail 🙂. Now you are nervous they might go bankrupt?

If they are delaying for technical reasons or because inventories are currently too high, then that is one thing. But if they are delaying for the sake of delaying, then stupid stupid stupid stupid stupid.

Since AMD launched HD7000 series with gimped GPU clocks and broken launch drivers, I am sure if HD8000 was ready and tapped out, it would have been launched. With AMD laying off engineers left right and center, how many resources do you think they have to work on next gen console GPUs for Sony/MS, rewrite the entire memory management of GCN, redesign HD7000 to make it worth it for more than a 15% performance boost given that HD7970GE reference card used up 238W at TPU?

AMD paid $4 million to include FC3 bundle. That's a lot cheaper to increase sales of old tech like HD7000 series than to do a redesign of a new chip on a 28nm node where your last refresh (HD7970Ghz) was pushing 240W TDP with a jet engine reference cooler. NV otoh has the resources and GK110 already made and manufactured. Now all they have to do is collected fail 14 SMX K20X chips, bundle them as 13 SMX cards, overclock to 1000-1050mhz, bam the Titan is ready (this is just my example, not Titan's specs btw). The work has already been done last year. AMD has no $, lacks engineering resources and GCN is less efficient in performance/watt which means they have to redesign Tahiti XT from the ground-up to get more out of it (perhaps start with Pitcairn XT) or wait for 28nm node to mature more. Since we are seeing AMD go all out with the NSR bundle, it seems HD8000 has either ran into 28nm node issues, AMD thinks it's better to include bundles vs. redesigning the chip since GTX700 hasn't launched yet, or they might try to squeeze 10% more performance from GCN drivers before launching HD8000 series later in the year. That would escape the PR and driver issues mess that plagued GCN 7000 parts for most of last year.

And again, all of this assumes the "delay" is even real since HD8000 desktop parts were never announced to launch in the first half of 2013 on any AMD roadmap.

 

[tviceman]

To the people who are shocked by my comments, how many AMD GPUs have you purchased in the last 5 years? You are so afraid of AMD going down now but you did nothing to help them all this time. Now you are nervous they might go bankrupt?

The fear mongering is silly. I sometimes emotionally root for the underdog, but I never, ever, ever make a purchase decision because "if I don't they might go out of business." If that was the case, I'd have bought lots of crap through the years from now defunct companies that I can't get warranty work on or product replacements for.

 

[Eureka]

Going bankrupt or not, why are we obligated to help them? If their products aren't up to par, we should put our money into mediocre solutions?

 

[3DVagabond]

Going bankrupt or not, why are we obligated to help them? If their products aren't up to par, we should put our money into mediocre solutions?

Well, if it was as simple as that, I'd agree. Even when their product is better, they fail at marketing it and nobody buys it. Sometimes it's because of underhanded anti-competitive schemes, like Intel. Sometimes it's AMD's own fail, like SI.

 

[RussianSensation]

But AMD's own roadmaps from this time last year showed Sea Islands launching in Q1 of this year.

No such thing was ever stated. This is media making up facts. I posted AMD's roadmaps above. There are no indications of desktop HD8000 launching in Q1.

The fear mongering is silly. I sometimes emotionally root for the underdog, but I never, ever, ever make a purchase decision because "if I don't they might go out of business." If that was the case, I'd have bought lots of crap through the years from now defunct companies that I can't get warranty work on or product replacements for.

You stated earlier something along the lines of "AMD needs all the help they can get." If you are so concerned you'd be buying their GPUs especially since this generation they were faster, cheaper and made $$. You didn't. Sounds to me imo, what you are concerned about is if AMD goes under, NV raises prices or delays progress for everyone of us.

Going bankrupt or not, why are we obligated to help them? If their products aren't up to par, we should put our money into mediocre solutions?

No, you aren't obligated. But then jump on the bandwagon how HD8000 launching later is a big deal. You did nothing to support the company when they made a superior card than a GTX670 and then you are upset they are in financial trouble and may need more time to get their next series out, which in turn might mean NV stalls too, or raises prices, etc.? Again, there isn't a single credible source of AMD ever publicly stating HD8000 desktop cards were to launch Q1 2013.

HD7970Ge is the fastest single GPU. Nearly everything else on the desktop they have has very competitive price/performance, overclocking and game bundles. Why exactly is AMD in a rush to replace that series? Your comment that their GPUs are sub-par / mediocre does not compute. HD7970GE spanks your 670 so hard, they are not even in the same performance class. Since July of last year HD7970 cards cost about the same as 670s and the majority of good after-market versions hit 1150-1200mhz (a lot of those hit 1150mhz on stock voltage too). Same story on the price/performance front. HardOCP's overview late summer pitted HD7950 OC vs. GTX670 OC and 670 couldn't beat it despite at the time costing $70-80 more.

Also, now people here are upset AMD's HD8000 might not be out until Q4 2013. Yet, again ignored all the links I provided that same is rumoured for GTX700 series. Could it be there are some 28nm issues / wafer price reasons that make it more profitable for both AMD and NV to launch refreshes later in the year than Q1 2013? And again, what are some of you worried so much that HD8000 series won't be out until Q4 2013? As if you planned on upgrading to it? What difference does it make since you already just said that AMD's products are sub-par. Doubt HD8000 series would change your mind. Again, it sounds like you are more upset this would stall NV and NV would raise prices, rather than anything related to AMD's products.

 

[tviceman]

No such thing was ever stated. This is media making up facts. I posted AMD's roadmaps above. There are no indications of desktop HD8000 launching in Q1.

You are right, I stand corrected. So either the floating specs and release dates of the various sea islands chips that have floated around are entirely fabricated, or (as I believe) there is still some kind of delay taking place between the retail 8000 series and 7000 series. The time between Cypress and Cayman was 14(ish) months. If the time between Tahiti and ]Sea Islands highest end chip] is looking to be a minimum of 18 months if it comes Q3 2013, 21 months if it comes Q4.

That looks like a delay to me.

 

[tviceman]

You stated earlier something along the lines of "AMD needs all the help they can get." If you are so concerned you'd be buying their GPUs. You didn't. What you are concerned about is if AMD goes under, NV raises prices or delays progress for everyone of us.

I'm sorry allow me to clarify to what you and I both know I meant: AMD needs to help themselves as much as possible and quit shooting their own feet with terrible mismanagement and product decisions.

 

[Eureka]

No, you aren't obligated. But then jump on the bandwagon how HD8000 launching later is a big deal. You did nothing to support the company when they made a superior card than a GTX670 and then you are upset they are in financial trouble and may need more time to get their next series out, which in turn might mean NV stalls too, or raises prices, etc.? Again, there isn't a single credible source of AMD ever publicly stating HD8000 desktop cards were to launch Q1 2013.

HD7970Ge is the fastest single GPU. Nearly everything else has class leading price/performance, overclocking and game bundles. Why exactly is AMD in a rush to replace that series? Your comment that their GPUs are sub-par / mediocre does not compute. HD7970GE spanks your 670 so hard, they are not even in the same performance class. Same story on the price/performance front. HardOCP's overview late summer pitted HD7950 OC vs. GTX670 OC and 670 couldn't beat it despite at the time costing $70-80 more.

Except, you forget I did buy a 7970 GE, which I always agreed was the top of the line solution at the moment. I switched to a 670 because I got a good price for the 7970 GE (thanks to everyone's hype) and I got the 670 for what ended up being $130 less than what I paid for the 7970 GE.

That's not the point. The point is, you seem pretty intent on rallying everyone to support AMD by buying their GPUs. There's a bit of bias in that alone. What I said, is that if they do produce mediocre components, then it's their fault they're in the hole. Which is why they are in the position they are now... they've had mediocre solutions for a number of years and only now have they come back with the 7000 series. Why they need to push on a 8000 series launch is that they don't lose their momentum, I'm happy to see AMD being relevant again.

I supported AMD whenever they provided a good solution (in the past, with their processors, with their GPUs, and now with a 7970). But if they consistently produce mediocre solutions with a few good ones, then I'm not obligated to continue supporting them. In my price range ($250), only the 7950 is price comparable, and it's not a huge performance difference.

Sorry, but you seem to think AMD is the win-all solution for everyone, when it's not. Look in my signature, my purchase history ranges over AMD, Intel and nvidia. In fact, I've bought more AMD-based products alone over everything else. But I don't believe in brand support, I don't harp a single solution. Why do you?

 

[RussianSensation]

You are right, I stand corrected. So either the floating specs and release dates of the various sea islands chips that have floated around are entirely fabricated, or (as I believe) there is still some kind of delay taking place between the retail 8000 series and 7000 series. The time between Cypress and Cayman was 14(ish) months. If the time between Tahiti and ]Sea Islands highest end chip] is looking to be a minimum of 18 months if it comes Q3 2013, 21 months if it comes Q4.

That looks like a delay to me.

Even floating specs for HD8970 hinted at June launch date (Q2 2013), not Q1.
http://www.kitguru.net/components/g...adeon-hd-8970-and-8950-launch-plans-revealed/

From where I am standing:

1) There are no next generation PC games with next generation graphics launching in all of 2013 by the looks of it (maybe Crysis 3 delivers, maybe Star Wars 1313/Watch Dogs);

2) There isn't a hint of GTX700 being more than a 20-25% faster GTX680 which isn't anything worth upgrading to for people rocking GTX670 OC/HD7970GE/GTX680, etc. level of cards.

3) The Titan is rumoured to bring a very solid and impressive performance increase but at a very high price. Chances are unless NV transitions some of those GK110 chips to lower price levels, we may need to wait to 20nm parts to get that 60-75% performance increase from HD7970GE/GTX680 without having to pay through the nose.

4) PS4 could be years ahead of PCs which would obsolete GK114/HD8970 GPUs very quickly (this happened when PS360 launched with GeForce 7900GTX/X1950XTX/8800GTS/2900XT/3870). I remember I just got an 8800GTS a bit after PS3 launched and that card was useless in 1.5 years. As far as I see it, GTX680/7970 are living on borrowed time. When next gen DX11 games launch like Witcher 3 and beyond, these GPUs are toast. 20-25% faster than GTX680/7970GE won't even make a good showing in those 2015 titles and beyond, which means we'll want Maxwell / Volcanic Islands as real upgrades, not refreshed Keplers and GCN parts. In the first 12 months after PS4/720 launch, most games probably won't be intensive enough which should allow GTX680/7970GE to pass by. Then once graphics picks up, 20-25% performance increase over those GPUs won't be enough to deal with next gen games, which kinda makes 28nm refreshes a waste of time to be honest. I think the real breakthrough are 20nm parts and they'll have 75%+ faster performance we'd need to play next gen PS4/720 ports.

Timothy Lottes:

“If PS4 has a real-time OS, with a libGCM style low level access to the GPU, then the PS4 1st party games will be years ahead of the PC simply because it opens up what is possible on the GPU. Note this won’t happen right away on launch, but once developers tool up for the platform, this will be the case. As a PC guy who knows hardware to the metal, I spend most of my days in frustration knowing damn well what I could do with the hardware, but what I cannot do because Microsoft and IHVs wont provide low-level GPU access in PC APIs. One simple example, drawcalls on PC have easily 10x to 100x the overhead of a console with a libGCM style API.”

Source

You guys think GK114 and HD8970 will be enough for next gen DX11 titles? Why even upgrade to those cards when 2013 is full of console ports and when next gen DX11 titles come out, those cards will be worthless for next gen games that will require GPUs 2-3x faster? History repeats itself (PS360 did it) 😛

"Though the architectures of the next-gen Xbox and PlayStation both resemble that of PCs, several development sources have told us that Sony’s solution is preferable when it comes to leveraging power. Studios working with the next-gen Xbox are currently being forced to work with only approved development libraries, while Sony is encouraging coders to get closer to the metal of its box. Furthermore, the operating system overhead of Microsoft’s next console is more oppressive than Sony’s equivalent, giving the PlayStation-badged unit another advantage."
http://www.edge-online.com/news/the...hand-games-50gb-blu-ray-discs-and-new-kinect/

 

[tviceman]

“If PS4 has a real-time OS, with a libGCM style low level access to the GPU, then the PS4 1st party games will be years ahead of the PC simply because it opens up what is possible on the GPU. Note this won’t happen right away on launch, but once developers tool up for the platform, this will be the case. As a PC guy who knows hardware to the metal, I spend most of my days in frustration knowing damn well what I could do with the hardware, but what I cannot do because Microsoft and IHVs wont provide low-level GPU access in PC APIs. One simple example, drawcalls on PC have easily 10x to 100x the overhead of a console with a libGCM style API.”

Which is why I argue in favor of Valve coming out with a steam-PC. The hardware might be mid-range compared to today's best, but having a unified hardware set without directx, and with which developers can squeeze performance out of by getting closer and closer to the hardware over time like what happens on consoles, it would be a great thing for PC gaming. Mods + controller input freedom + better graphics + cheaper games = better.

 

[RussianSensation]

Except, you forget I did buy a 7970 GE, which I always agreed was the top of the line solution at the moment. I switched to a 670 because I got a good price for the 7970 GE (thanks to everyone's hype) and I got the 670 for what ended up being $130 less than what I paid for the 7970 GE.

There were plenty of times when GTX670 was the best GPU to buy last year. I recommended it for nearly 2 months straight from launch to late June 2012. My point is you made a sweeping generalization regarding AMD's mediocre products. If you said Bulldozer, A10-5800K or overpriced $550 HD7970, I wouldn't have even pointed that out. Since this is a GPU thread, I am guessing you are discussing existing HD7000 cards. Problem is that generalization hasn't generalization applied in at least 6 months since HD7000 has delivered very competitive, if not leading, price/performance+overclocking since mid-summer. It certainly does not apply to desktop HD7000 products today, which is why continuing selling them for another 2 quarters isn't detrimental to losing any momentum.

The point is, you seem pretty intent on rallying everyone to support AMD by buying their GPUs. There's a bit of bias in that alone.

I recommend based on price/performance and overclocking. People here have short memory. When NV had products that delivered on those in spades (8800GT, GTX460/470, GTX670 in early 2012), I wholeheartedly recommended them. I don't see anything in NV's existing desktop line besides GTX690 & occasional sales on a GTX650Ti under $120 or GTX670 at $299 that stands out. What should I recommend, more expensive, worse overclocking, slower cards with inferior game bundles? You call that biased?

What I said, is that if they do produce mediocre components, then it's their fault they're in the hole. Which is why they are in the position they are now... they've had mediocre solutions for a number of years and only now have they come back with the 7000 series. Why they need to push on a 8000 series launch is that they don't lose their momentum, I'm happy to see AMD being relevant again.

HD4870 that beat GTX260 for $100 less was mediocre? It prompted NV to rush GTX260 216 just to match it. $260 HD4890 that matched $500 ($649 launch price) GTX280 for less 9 months later was mediocre? HD5850/5870 that beat Fermi by 6 months and cost just $259/369 were mediocre? HD6950 that unlocked into a 6970 that easily matched a GTX570 for less $ was mediocre? 2 HD6950 2GB unlocked that cost the same as a GTX580 3GB was mediocre? Whatever you say.

Further, you are saying AMD shouldn't lose momentum and then in the same post claim their products are sub-par to begin with. How can you lose momentum if your products are sub-par? That means you never had any positive momentum to begin with. And if AMD's products are not sub-par, they aren't losing any momentum since no one else is beating their product line convincingly for them to lose that momentum. Your comment is akin to stating that NV is losing momentum because their GTX700 series got "delayed" to Q4 2013, which according to the same sources it did. It seems to me both GTX700 and HD8000 desktop cards may not come out until Q3-Q4 2013 and Titan is the next big product for us to get excited about. All the rumours of GTX700 series also being pushed back are being ignored though.

I supported AMD whenever they provided a good solution (in the past, with their processors, with their GPUs, and now with a 7970). But if they consistently produce mediocre solutions with a few good ones, then I'm not obligated to continue supporting them.

What's your GPU history from HD4000 to 6000 series? With GTX200 vs. HD4000 series, you either got rapped by GTX280's early adopter price premium, or paid one hundred more for a barely faster GTX285 over 4890, or bought the slower GTX275. I can understand why someone bought GTX470/480 but HD6900 was a walk over most GTX500 line because for the price of a single GTX580, one could purchase nearly 2 unlocked 6950s. GTX570 was overpriced vs. 6950 unlocked too and couldn't beat it in games. In today's titles GTX570 has serious issues with just 1.28GB of VRAM which means in hindsight it was a far worse GPU than HD6950 2GB unlocked. GTX570 not only cost more, it had a more fragile VRM design which resulted in many of those cards simply failing from overclocking. Sounds peachy.

Again, I stayed consistent in my recommendations on price/performance + overclocking which is why I recommended HD4870 and 4890, then 5850 when it launched, then 460/470s when they launched, then HD6950s when they launched, and then GTX670 when it launched. No emotional attachment to those cards, just mathematics.

Sorry, but you seem to think AMD is the win-all solution for everyone, when it's not. Look in my signature, my purchase history ranges over AMD, Intel and nvidia. In fact, I've bought more AMD-based products alone over everything else. But I don't believe in brand support, I don't harp a single solution. Why do you?

No, I don't believe in that. I recommend based on price/performance and overclocking. It's that simple really. Whichever side is winning in those metrics I recommend. I also consider specific features someone might want if they explicitly state them (i.e., running 3 2560x1600 monitors with 2 over DL-DVI, SLI vs. CF, PhysX is a big deal to them, etc.). The reason I bring up bitcoin mining is because it ties into price/performance since it effectively pays for the cards. PhysX, 3D Vision, Eyefinity, they don't affect those metrics which is why I assume the person has already done their research on these specific features. Price/performance + overclocking can be measured. For these other features, each buyer should assess their importance for themselves because they are qualitative in nature, not quantitative. By me focusing on the quants, the buyer can focus on qualitative features. The quants is what I focus on.

 

[BallaTheFeared]

Hey hey, maybe this will make Windows 9 not suck!

One can dream...

 

[paul878]

I think AMD is stuck with a lot of HD7xxx inventory. The are giving $100 worth of free games to try to move it.

 

[tviceman]

I think AMD is stuck with a lot of HD7xxx inventory. The are giving $100 worth of free games to try to move it.

This is what I think as well.

 

[RussianSensation]

Which is why I argue in favor of Valve coming out with a steam-PC. The hardware might be mid-range compared to today's best, but having a unified hardware set without directx, and with which developers can squeeze performance out of by getting closer and closer to the hardware over time like what happens on consoles, it would be a great thing for PC gaming. Mods + controller input freedom + better graphics + cheaper games = better.

I agree that it would be great for PC gaming/Steam if more users join and start buying more PC games. Since games won't necessarily be coded around Steam box as it won't be a fixed hardware over 6-7 years, I am not sure how much it would benefit from hardware level access though. What I got about Steam box is that it's supposed to be upgradable over time and there would be various SKUs of it. What do you think?

Valve is probably more worried that Apple would get to the living room before PCs do. If Apple ships TVs with gaming consoles/capable hardware inside and a full blown music/TV/gaming distribution system in place like iTunes, that would be a huge threat for Valve since they are going to try and attract some of those same casual gamers.

I think AMD is stuck with a lot of HD7xxx inventory. The are giving $100 worth of free games to try to move it.

9 months to move oversupplied inventory? Talking about a delay from Q1 to Q4 2013 to move millions of HD7000 cards that are just sitting there from being overproduced in the summer? How do you explain new versions of 7970 cards popping up? For example, XFX even replaced their 7970 with a all new redesign, new cooler, PCB, fans, etc.

Old XFX HD7970 from 2012. New XFX 7970 2013 model, just launched last week.

Since Never Settle bundle started last year, you are actually then talking about oversupply of HD7000 cards spanning Q4 2012, Q1 2013, Q2 2013 and Q3 2013, or 12 months of oversupply? Does not compute. You guys are reading way too much into this. AMD got NV's leading sales guy on their team and they are doing the NS bundles as the new thing moving forward. This appears to be a new initiative for the entire brand, not an HD7000 initiative only.

"PCWorld reached out to AMD in the wake of all the hub-bub to get a feel for the situation, and while there are still plenty of questions up in the air, I was told by a chortling AMD representative, "We will certainly have new [GPU] products in 2013." The company plans to clarify its 2013 plans for the Radeon brand later this week."
http://www.pcworld.com/article/2027...on-and-quickly-clarify-2013-radeon-plans.html

 

[tviceman]

I agree that it would be great for PC gaming/Steam if more users join and start buying more PC games. Since games won't necessarily be coded around Steam box as it won't be a fixed hardware over 6-7 years, I am not sure how much it would benefit from hardware level access though. What I got about Steam box is that it's supposed to be upgradable over time and there would be various SKUs of it. What do you think?

Well Valve has said other manufacturers might make their own "steam-enabled" PC, but with Valve's hardware specifically it won't be upgradeable. Gabe has said that much. (and insofar as the rest of what I am about to say, it will refer entirely to Valve's hardware) I envision it having a product cadence not unlike the iPad (although not as often). A new steam pc comes out every 24 months at the same price point but with improved capabilities and backwards compatibility. Although some games today attempt to detect hardware (like Skyrim or Oblivion) and set graphics settings that way, future games will simply detect which Valve-based steam PC the user has and auto-configure the graphics entirely transparent to the user. So a game that comes out in 2016 can take advantage of the Valve's newest steam pc, but will automatically dial down settings to maintain smoothness on the older hardware that came out in 2013/2014.

Insofar as Linux not having the support today - I think it's a non issue if Valve's hardware is even mildly successful. Valve will likely offer incentives to devs to port current / recently released projects over to Linux, and future titles developers are working on would have concurrent development. Developers will go where the money is, and if this thing sells in the millions in the first year (a few million sounds modest to me) then it will roll in it's own success much like consoles do. Also, it will be more lucrative for hardware makers because I'm pretty sure Gabe has said they'd sort of be using off-the shelf parts (but reshaped to custom fit the form factor). Buying the actual hardware from vendors like Intel, Nvidia, or AMD pays more than licensing the tech behind it (like Microsoft and Sony do for their consoles).

/end tangent