AMD Says Vega Delays Necessary to Increase Stock for Gamers

[Vayra86]

After GCN was finalized back in 2012 by their entirely Canadian design team, AMD has outsourced pretty much all of their graphics IP design to China. No wonder their graphics chips haven't improved in any noticeable way since GCN1.

Got a source? I couldn't find one. They started a joint venture and struck licensing deals for x86 in China, and have their own office in Shanghai.

And people are surprised why they have stagnated…
Saving money by outsourcing development rarely pays off in the long run.

...In the land of the blind... don't believe everything everyone says.

Yep, because random internet guy always knows more than the company that makes the product. I would agree with you if it was obvious. In this case, it is not, especially seeing as keeping product in stock has been an issue recently. Also, point to HBM as the problem when you have no proof, yeah. It could be the HBM or it could be that AMD has had extremely limited funds for some time now. If you haven't noticed, AMD has not redesigned it's GPU architecture since GCN and the 7970.

AMD has made attempts to redesign and improve GCN and has been mildly successful on some occasions. Polaris is a good example of a good architectural tweak towards efficiency. Tonga however, was not that much of a boost and more of a cost savings attempt / quite comparable to Nvidia's early Maxwell with the 750ti. Either way you're quite misinformed here. The GCN improvements on Vega are also really there, they just aren't showing their effectiveness yet, and look to be another Tonga. But the changes are there. On the green side of the fence, Nvidia has been riding on Kepler since about the same year, and has made its own tweaks to that architecture and now likes to call it Pascal. Somewhere along Kepler Refresh, Nvidia started splitting off the compute resources and pushed full on gaming efficiency throughout Maxwell > Pascal. But it is still essentially Kepler at its core, same SMX, GPC setup with additional resources per GPC.

http://www.nvidia.com/object/nvidia-kepler.html > have some fun here and compare the marketing slogans with those of GP100 and try to find 10 differences. Good challenge

About HBM: AMD has admitted themselves that HBM stock isn't the easiest thing to come by, prior to VEGA's launch. Common sense, try to apply some.

 

[bug]

Got a source? I couldn't find one. They started a joint venture and struck licensing deals for x86 in China, and have their own office in Shanghai.

I believe this is the source: https://www.hardocp.com/article/2016/05/27/from_ati_to_amd_back_journey_in_futility

 

[Vayra86]

I believe this is the source: https://www.hardocp.com/article/2016/05/27/from_ati_to_amd_back_journey_in_futility

All I can read there is that HardOCP is rather butthurt about getting placed on the shitlist. Look at the tone of the whole article, its also full of opinionated rumor that hasn't become fact.

Thanks anyway, its stuff like this that confirms the website belongs on my list of places to avoid, just like TweakTown.

 

[bug]

All I can read there is that HardOCP is rather butthurt about getting placed on the shitlist. Look at the tone of the whole article, its also full of opinionated rumor that hasn't become fact.

Thanks anyway, its stuff like this that confirms the website belongs on my list of places to avoid, just like TweakTown.

It depends on how you look at it. These movements cannot be officially acknowledged, if that's what you were expecting. And the RTG separation is a fact. The writing about Polaris is also pretty much spot on.
At the same time, the best lies are half truths, so...

 

[Vayra86]

It depends on how you look at it. These movements cannot be officially acknowledged, if that's what you were expecting. And the RTG separation is a fact. The writing about Polaris is also pretty much spot on.
At the same time, the best lies are half truths, so...

RTG separation yes, outsourcing to China no. Those are still different things.

 

[Captain_Tom]

All I can read there is that HardOCP is rather butthurt about getting placed on the shitlist. Look at the tone of the whole article, its also full of opinionated rumor that hasn't become fact.

Thanks anyway, its stuff like this that confirms the website belongs on my list of places to avoid, just like TweakTown.

Haha. HardOCP does some interesting articles from time to time, but at the same time I find most of their GPU reviews idiotic and inconsistent. The conclusions they come to (for both AMD and Nvidia) make some pretty odd mental leaps at times...

 

[Vayra86]

Haha. HardOCP does some interesting articles from time to time, but at the same time I find most of their GPU reviews idiotic and inconsistent. The conclusions they come to (for both AMD and Nvidia) make some pretty odd mental leaps at times...

Too much time on Youtube does that to any sane mind I guess.

 

[MBTP]

Rubbish: they could allow reviewers to fully benchmark the cards while waiting for availability but they don't.

Stock increase could be a reason but it's most definitely not the only reason.

Yep, there is something going on behind the scenes.

 

[Hood]

If Vega was any good at all, it would have been released a year ago, when it could have mattered. These delays are obviously because it's a huge disappointment in every way. This is the beginning of the end of AMD's time in the GPU market, no surprise they want to delay the inevitable.

 

[Vayra86]

If Vega was any good at all, it would have been released a year ago, when it could have mattered. These delays are obviously because it's a huge disappointment in every way. This is the beginning of the end of AMD's time in the GPU market, no surprise they want to delay the inevitable.

Well 'the end' for AMD has been called around alot, still hasn't happened...

 

[rtwjunkie]

If Vega was any good at all, it would have been released a year ago, when it could have mattered. These delays are obviously because it's a huge disappointment in every way. This is the beginning of the end of AMD's time in the GPU market, no surprise they want to delay the inevitable.

Hood my man, isn't that a little extreme? I'm pretty sure it's been public knowledge that HBM is in short supply. To me, it appears that they could do nothing but delay.

 

[Hood]

I didn't say "the end of AMD", I said "the beginning of the end of AMD's time in the GPU market". Not just because of Vega, it's only a symptom of bigger problems, with management mostly. HBM is just more smoke and mirrors, a bad decision they made against all advice, which came in handy when excuses were needed. I think they knew Vega was a failure almost 2 years ago, and everything since is damage control. Just my opinion, no real facts to back it up, but lots of red flags and things that don't add up. There's serious money to be made in the dGPU market, increasingly higher monitor resolution drives up demand for high-end cards, and AMD is losing money in their graphics division because of bad management for years now. Financial analysts have said that AMD needs a viable product in both markets (CPU and GPU), and "If AMD can't gain traction with their next series of product launches then the chapter after that may well be the realization of one of many acquisition rumors and at that point, all bets are off". Ryzen and Threadripper are viable, but HEDT is a limited market, and can't save AMD, not without a good showing in the graphics market. Vega's not good enough.

 

[efikkan]

If Vega was any good at all, it would have been released a year ago, when it could have mattered. These delays are obviously because it's a huge disappointment in every way. This is the beginning of the end of AMD's time in the GPU market, no surprise they want to delay the inevitable.

Just a little correction, Vega10 was taped out in June last year, so it couldn't have been released last year. The rumors about a Vega launch in October were bogus, even though I received a "shitstorm" when I pointed out the impossibility of an October release. Vega10 was intended to be released in May 2017.

 

[bug]

Hood my man, isn't that a little extreme? I'm pretty sure it's been public knowledge that HBM is in short supply. To me, it appears that they could do nothing but delay.

They could have done the sensible thing and go for GDDR5(X). Especially seeing that HBM does nothing for performance or power draw.

 

[RejZoR]

I didn't say "the end of AMD", I said "the beginning of the end of AMD's time in the GPU market". Not just because of Vega, it's only a symptom of bigger problems, with management mostly. HBM is just more smoke and mirrors, a bad decision they made against all advice, which came in handy when excuses were needed. I think they knew Vega was a failure almost 2 years ago, and everything since is damage control. Just my opinion, no real facts to back it up, but lots of red flags and things that don't add up. There's serious money to be made in the dGPU market, increasingly higher monitor resolution drives up demand for high-end cards, and AMD is losing money in their graphics division because of bad management for years now. Financial analysts have said that AMD needs a viable product in both markets (CPU and GPU), and "If AMD can't gain traction with their next series of product launches then the chapter after that may well be the realization of one of many acquisition rumors and at that point, all bets are off". Ryzen and Threadripper are viable, but HEDT is a limited market, and can't save AMD, not without a good showing in the graphics market. Vega's not good enough.

Without AMD investing into HBM, we'd be stuck with GDDR5 for the next 10 years. With HBM, others started investing in GDDR5X which bumps the bar quite a bit. And thanks to AMD, NVIDIA has HBM2 on Tesla thingies. Everyone always takes a piss at AMD, but they quickly forget they innovated bunch of really major things everyone uses now. There is no doubt HBM is the future. Every new radical feature had major problems in the beginning.

 

[bug]

Without AMD investing into HBM, we'd be stuck with GDDR5 for the next 10 years. With HBM, others started investing in GDDR5X which bumps the bar quite a bit. And thanks to AMD, NVIDIA has HBM2 on Tesla thingies. Everyone always takes a piss at AMD, but they quickly forget they innovated bunch of really major things everyone uses now. There is no doubt HBM is the future. Every new radical feature had major problems in the beginning.

Yes, I'm sure that if AMD hadn't jumped the gun on HBM, nobody would have thought of it
And this day, it's such a great solution, it doesn't beat GDDR in either performance or power usage. It may be the way forward, but if it doesn't offer anything in its second incarnation, it's pretty clear it's a product that was rushed to the market unfinished.

 

[RejZoR]

Because same couldn't be said for hundreds of other things that revolutionized everything we use these days?

 

[EarthDog]

Its still too early to market considering gddr5x...

...when is it going to matter? When 4k is mainstream (currently less than 1% use it according to steam stats).

 

[RejZoR]

Because 1080p people don't demand high performance somehow?

 

[EarthDog]

Because 1080p people don't demand high performance somehow?

They arent getting with gddr5? Gddr5x?? Seems like there are several high performing cards out that dont use it...

What is it offering at 1080p that gddr5x doesnt aleady fullfill? A couple watts power savings? Surely youve seen hbm and hbm2 does well where the bandwidth is actually needed...4k. Otherwise, there really isnt much use for it over gddr5x...

 

[bug]

Because same couldn't be said for hundreds of other things that revolutionized everything we use these days?

Here we go again. "Revolutionized" - so far other than inflating price and limiting stock, HBM has yielded little other benefits (which I have just told you in my previous post). And then you wonder why people call you out for kissing AMD's rear whenever you get a chance.

 

[RejZoR]

Here we go again with the AMD fanboy bullshit. Grow up. And get your head out of NVIDIA's rear.

 

[Th3pwn3r]

If Vega was any good at all, it would have been released a year ago, when it could have mattered. These delays are obviously because it's a huge disappointment in every way. This is the beginning of the end of AMD's time in the GPU market, no surprise they want to delay the inevitable.

Well 'the end' for AMD has been called around alot, still hasn't happened...

You can tell he doesn't know about consoles and how AMD is involved with them.

 

[efikkan]

Without AMD investing into HBM, we'd be stuck with GDDR5 for the next 10 years. With HBM, others started investing in GDDR5X which bumps the bar quite a bit. And thanks to AMD, NVIDIA has HBM2 on Tesla thingies. Everyone always takes a piss at AMD, but they quickly forget they innovated bunch of really major things everyone uses now. There is no doubt HBM is the future. Every new radical feature had major problems in the beginning.

Of course something will eventually replace GDDR for consumer graphics, but that doesn't excuse jumping on a shiny new technology just because it's new, while offering no benefit to the end user.

And BTW, AMD did not invent HBM.

 

[RejZoR]

Then what does? Pushing the limits is what drives the innovation. If everyone played it safe we'd still be stuck on SDR memory with everything. Or DDR1 at best. And yes, AMD did invent HBM. It was a joint development with Hynix. C'mon, can you even use Google?

Saying HBM doesn't matter because it doesn't have any significant benefits. How cramming 16GB of memory into a size of a SINGLE GDDR5 memory module isn't significant enough?

Seems like these days, TPU is filled with hardcore NVIDIOTS, who will refuse to accept facts and call anyone who presents them an "AMD fanboy"...