That would have been the best thing for AMD. And I understand why Jen-Hsun would have wanted it. Since then, AMD has been so mis-managed all the way from the top.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
AMD's Computing and Graphics division down 54.2%
- Thread starter PRIME1
- Start date
That would have been the best thing for AMD. And I understand why Jen-Hsun would have wanted it. Since then, AMD has been so mis-managed all the way from the top.
To me, that's the story of AMD. Good technology, poor management. The CPU side has been hard to defend for a while now. I think that if they spun off ATI now, any number of companies would consider buying up ATI (but not for anywhere near what AMD paid for them).
Given how heavily AMD has invested in the 'APU' I do wonder what an ATI-less AMD would look like.
Given how heavily AMD has invested in the 'APU' I do wonder what an ATI-less AMD would look like.
If you think about it this has kind of been a long time coming where we've all for the past decade just kept our eyes closed to whats going on. Everyone was selling off their fabs a decade ago, but a few major players. Everyone was leaving the x86 world, which arguably is literally owned by Intel alone both in licensing and market share. Everyone realized back in the early 2000's that it wasn't worth the investment to stay in the GPU business except the top two major players.
The market has pushing toward a oligarchy of Fab's and a one vender x86 and GPU solution market for years. We just all wished for competitive pricing sake that it wouldn't happen. Truth is it costs a lot of money and the well established players aren't going anywhere and nobody is going to come in and do it themselves from the ground up. Only hope is a bigger fish eats the smaller fish. It's how every industry goes when they become big enough that the cost to enter is colossal. Even if that happened I could honestly not see things being the same if Samsung or even Intel bought out AMD's GPU division.
The market has pushing toward a oligarchy of Fab's and a one vender x86 and GPU solution market for years. We just all wished for competitive pricing sake that it wouldn't happen. Truth is it costs a lot of money and the well established players aren't going anywhere and nobody is going to come in and do it themselves from the ground up. Only hope is a bigger fish eats the smaller fish. It's how every industry goes when they become big enough that the cost to enter is colossal. Even if that happened I could honestly not see things being the same if Samsung or even Intel bought out AMD's GPU division.
Fury would not have saved AMD, the things you people say is ridiculous. You cant count on one product to save an entire company. Even if Fury was better than a 980ti, AMD wouldnt have the supply to satisfy the demand. Prices would once again skyrocket at etailers which wouldnt benefit AMD in any way since they dont get any of that extra cash.
AMD should restructure their divisions, or simply go private like Dell. The constant battle of pleasing investors is a never ending one so going private would be ideals. If they do that they need to secure funds on their own then, which would be a whole different challenge.
AMD should restructure their divisions, or simply go private like Dell. The constant battle of pleasing investors is a never ending one so going private would be ideals. If they do that they need to secure funds on their own then, which would be a whole different challenge.
Thats not the main issue. The main issue is OVERpaying for ATI and pretty much spending all reserves AMD had. Its like a teenager financing an expensive car with all the money they have and working at a job which doesn't cover the monthly cost of gas, payments, insurance and everyday spending he/she does monthly.
It was a stupid move to throw all their cash away without any reserves. After the buyout it was pretty much constant lending from institutions and investors, which will eventually come to a stop as there is nothing to show for it.
so you believe that there will be new GPUs at 28 or 20 or 22nm before 14nm?
I cant predict the future, so who knows. But there is this piece http://wccftech.com/amd-greenland-gpus-feature-hbm2-14nm-coming-2016/
What I do know is that a future iteration of AMD's cards will use HBM2 which will allow for more than 4GB maximum VRAM capacity.
Nvidia skipped HBM1 because their Pascal cards will utilize HBM2, which doesnt limit the VRAM size to 4GB maximum.
This is where another issue lies - AMD had to use HBM1 in order to compete with 980ti and Titan cards in some shape or form which drained even more resources for production, marketing, etc. HBM1 could have been skipped by AMD, but then they would be even more behind with only 390 series cards at their disposal. So in a way Fury is a loss/win for AMD.
Nvidia had the resources to make something decent out of GDDR5 memory which is why they stuck with it and offered Maxwell. With Pascal they are jumping straight into HBM2 which will most likely be an 8GB standard for VRAM in their top tier cards, although 32GB is the maximum so the sources say in terms of tech specs.
SamiiRoss
Gawd
- Joined
- Jan 8, 2012
- Messages
- 720
Yes but, pascal is going to be delayed. AMD is going to use these hbm2 chips first because they have priority. And there won't be too much chips for the first months. So AMD still has a chance.
You must be confusing Pascal with Volta. A few years ago Pascal wasnt even supposed to be released or be on the roadmap at all, see my previous 2 posts with official Nvidia roadmaps. With Volta's development constantly being pushed back Nvidia had to make a stop gap solution which is what Pascal is.
AMD does have priority, but who knows whether the development of the next card isnt going to be delayed due to technical reasons or financial? I highly doubt key suppliers will wait on a potential "loser" because they cant release a product on time, so that HBM priority could fly out of the window.
I am rooting for AMD because we need competition and innovation, but the news keep getting worse and worse on management and income side of business.
Did you hear AMD's q2 2015 conference call? I think you should before you think HBM 2 is going to be the cause of that
.AMD's priority is on going from HBM, so they have one year since they started using HBM. It has nothing to do with priority of HBM 2.
buttons
2[H]4U
- Joined
- Oct 12, 2011
- Messages
- 2,174
The sad part is, there is also nothing new worth buying either. I stopped building AMD when Ivy bridge came out and started using intel i7-920 d0 that i could pick up for $40-50... The used market is so much better value in my opinion. especially xeon chips from data centers upgrading. i was able to get a six core hyper threaded intel 60 watt chip for $70
I don't even think the Volta delay is that big of a deal based on the roadmaps we know of though. From the looks of it they just stole all the features we knew of Volta and renamed it Pascal. Thus the *New* Volta is still relatively unknown as far as what it'll bring to the table. Not sure why they did the codename switcharoo in the first place.
Unless Pascal brings no further Maxwell architecture improvements except shader core counts and a smaller node, while Volta was supposed to originally have major architectural advancements, then I can see the delay making sense. So far Nvidia hasn't alluded to that at all though.
What I've read is that Volta is delayed by a fair bit. Especially if the rumor is true that they want to run Volta on 10nm, well damn thats a huge jump from what we have right now. Some rumors pointed to Pascal running on 16nm and some said 20nm. 20nm architecture is pretty backed up in terms of orders, it seems Apple got first dibs on it so others will have to be in line.
Motley fool, which is primarily an investment news source states "NVIDIA's Pascal might be a 16-nanometer chip. Its successor, Volta, will likely be built on a 10-nanometer manufacturing node"
http://www.fool.com/investing/gener...acturing-technology-will-nvidia-corporat.aspx
IMO Nvidia thought everything would go smooth with TSMC fabs and it would be a smooth sailing trip for Volta when it was announced 2 years ago, but turned out to be the exact opposite. That's why they had to plug something in between Maxwell and Volta.
SamiiRoss
Gawd
- Joined
- Jan 8, 2012
- Messages
- 720
Did you hear AMD's q2 2015 conference call? I think you should before you think HBM 2 is going to be the cause of that
AMD's priority is on going from HBM, so they have one year since they started using HBM. It has nothing to do with priority of HBM 2.
Rumor: AMD gets priority advantage to Hynix HBM2
http://www.guru3d.com/news-story/rumor-amd-gets-priority-advantage-to-hynix-hbm2.html
5150Joker
Supreme [H]ardness
- Joined
- Aug 1, 2005
- Messages
- 4,568
LOL! Yeah I'm sure Hynix would give priority to a sinking ship like AMD with probably less than 20% marketshare by now over the dominant player in the GPU industry. Yep, makes perfect business sense.
.."The rumor comes from wccftech".
Yep, no credibility.As Joker stated its from the same source, and go listen to the conference call
If AMD plans on releasing their new cards in lets say Q1 2016 yes they will still have priority because priority for HBM doesn't run out till end of q2 2016.
SamiiRoss
Gawd
- Joined
- Jan 8, 2012
- Messages
- 720
LOL! Yeah I'm sure Hynix would give priority to a sinking ship like AMD with probably less than 20% marketshare by now over the dominant player in the GPU industry. Yep, makes perfect business sense.
.."The rumor comes from wccftech".
i bet you amd will have priority with HBM2.
They won't unless they release prior to q2 of 2016, simple, actually its not even priority its exclusivity.
Richthofen
Limp Gawd
- Joined
- Feb 20, 2005
- Messages
- 134
I don't think anyone will give a damn about discrete GPUs as a market segment. It will be either overall GPUs or nothing and overall Nvidia has plenty of competitors (Intel, the ARM based SoC army etc.). So from a legal perspective there is nothing against an acquisition of AMD by Nvidia and i would highly recommend Huang to plan such a thing. It absolutely makes sense. However the CPU part should be sold or closed quickly afterwards.
- Joined
- Apr 10, 2003
- Messages
- 23,513
Of course the inventor of the technology of the future is going to have priority on chips. Just makes sense. Best Nvidia can hope for is that AMD's next generation card in 2016 has low sales so that they don't need the full HBM2 capacity. Going by this year's HBM 1 chip sales, all Fury cards sold out, they won't have an issue running through the stockpile of HBM 2 chips.
Nvidia fans shouldn't worry. Nvidia will have the pick of the litter when it comes to DDR5 next year. They have proven that they can make great cards with the older technology so Nvidia being possibly locked out of HBM 2 shouldn't matter. Pascal being delayed until a later date is fine as Nvidia fans have proven that they will buy the same card over again if Nvidia will unlock one more shader.
- Joined
- Feb 11, 2013
- Messages
- 6,478
SKHynix only gave AMD 1 year advantage with HBM by being a developer of the technology (indistinct of HBM1 or HBM2) of the market. that's all... that's the pure reason Nvidia didn't used HBM with maxwell and it's the same reason nvidia will use HBM2 straight with Pascal.
I dont know about the whole AMD having dibs on HBM2 considering the fact that rumors are circling around Pascal being released Q2 of 2016.
It doesnt make sense to have AMDs rights expire at Q2 2016 when Nvidia will be releasing their cards during the same time. That would mean they have a limited priority over Nvidia, but not 100%, not even close. If AMD had such rights then Nvidia wouldnt be able to get anything out in 2016, which is doubtful. For Nvidia to build Pascal/Volta they would need to have access to HBM2 for a very long time, otherwise it would be impossible to design the cards.
It doesnt make sense to have AMDs rights expire at Q2 2016 when Nvidia will be releasing their cards during the same time. That would mean they have a limited priority over Nvidia, but not 100%, not even close. If AMD had such rights then Nvidia wouldnt be able to get anything out in 2016, which is doubtful. For Nvidia to build Pascal/Volta they would need to have access to HBM2 for a very long time, otherwise it would be impossible to design the cards.
rumartinez89
Limp Gawd
- Joined
- Apr 16, 2012
- Messages
- 392
HBM2 might only be available on GP100 if the AMD rumor is true. That could actually end up being the next titan's feature for justifying a $1000 pricetag
That's a lot of wishful thinking. In the end, money talks. And NVIDIA has much more of it to talk with. No plant is going to wave off a high volume NVIDIA order because AMD wants to get a low volume one in - one which they won't even be ready with for another year.
And Fury cards are "sold out" because they barely made any since yields are terrible. But you already knew that.
Last edited:
- Joined
- Jul 28, 2009
- Messages
- 8,421
Im not all that worried about it. AMD has been presumed dead now for several years but they keep on plugging along. I dont know that its as bad as it seems. [H] posts every bad piece of press AMD gets while ignoring the exact same stories about Intel so the perception is much worse than it really is.
I do think it is worse right now that Fury was a disappointment and the 300's were just a rebadge but I dont think its hopeless. Zen looks to be pretty promising and this is the first time since they acquired ATI that theyve come up with a lackluster GPU line. Theyve always been competitive with GPU's and I feel pretty good that they will again. Yeah this is definitely a low point for AMD with no new CPU's worth a shit in 2 years now and a flat launch of their new GPU line but Im not ready to pull the plug yet.
I do think it is worse right now that Fury was a disappointment and the 300's were just a rebadge but I dont think its hopeless. Zen looks to be pretty promising and this is the first time since they acquired ATI that theyve come up with a lackluster GPU line. Theyve always been competitive with GPU's and I feel pretty good that they will again. Yeah this is definitely a low point for AMD with no new CPU's worth a shit in 2 years now and a flat launch of their new GPU line but Im not ready to pull the plug yet.
Hornet
Supreme [H]ardness
- Joined
- Oct 4, 2005
- Messages
- 6,624
I'm not sure why some people are assuming AMD's next architecture will be release at the same time as nVidia's. Judging by the fact that Fury was late to the party, my guess is Pascal will arrive first in less than 12 months time, while AMD's next architecture is at least 12 months away. Maybe even more, if Fury's delay is any indication.
So, do people really believe that Hynix will just hold back HBM 2 and waste everyone's time until AMD is ready to use them? Priority make sense if 2 companies need the same thing at the same time. But if they don't, then it's a simple matter of first come first served, no?
I agree that AMD may still be competitive in the GPU department but I don't think that's the main concern. The important question is can they at the very least break even, and stop losing money. We have to keep in mind that even before Fury, AMD have been selling graphics card too. Their previous hardware wasn't bad at all, and people were buying them. Yet somehow, they continue to lose money. This to me, shows that their problem is bigger than just GPU sales.
So, do people really believe that Hynix will just hold back HBM 2 and waste everyone's time until AMD is ready to use them? Priority make sense if 2 companies need the same thing at the same time. But if they don't, then it's a simple matter of first come first served, no?
Not sure what you are referring to, but the only "bad press" [H] posted about AMD in recent time is their financial report, just as they do with Intel and nVidia. And that's the bottom line, isn't it? Every companies have a bad product every now and then, that's not why people are less confident in AMD's future. Rather, it's the fact that they are still losing millions of dollars. That is the real concern regarding AMD. It isn't about how Fury performed that is just one of AMD's product, but the concern is about how they still show a negative figure each time these financial report is release.
I agree that AMD may still be competitive in the GPU department but I don't think that's the main concern. The important question is can they at the very least break even, and stop losing money. We have to keep in mind that even before Fury, AMD have been selling graphics card too. Their previous hardware wasn't bad at all, and people were buying them. Yet somehow, they continue to lose money. This to me, shows that their problem is bigger than just GPU sales.
Last edited:
Remember those rumors of how AMDs OEMs were stuck sitting on huge piles of stock and one of the reasons AMD was taking so long to release new cards was due to them having to clear channels?
Seems if AMD channel partners were having to clear a large back log of product before AMD could release the next generation then AMD would probably have a few months where revenue tanked but without a corresponding drop in market share, because OEMs were still selling parts that AMD had already been paid for.
Huh...
Seems if AMD channel partners were having to clear a large back log of product before AMD could release the next generation then AMD would probably have a few months where revenue tanked but without a corresponding drop in market share, because OEMs were still selling parts that AMD had already been paid for.
Huh...
That's the real question, because rumors are pointing towards Pascal being 16nm in Q2 2016. That seems crazy to me with all the BS surrounding TSMC fabs.
And there seemed to be some veiled discussion in recent months that Pascal could be 28nm.
I haven't even heard of 14/16 SoC being ready yet at TSMC. iPhones are typically released in September?
But Samsung is SoC only, right?
defaultluser
[H]F Junkie
- Joined
- Jan 14, 2006
- Messages
- 14,398
Yeah, with 256-bit GDDDR5 powering their high-end video card this generation, they have the option to move to 384-bit next generation, and save HBM for just the top-end "enthusiast" card.
Since they are actually getting a die shrink this time it would not be as much of a cost increase as AMD incurred with Tahiti->Hawaii (they occupied the same price brackets on release), so 384-bit GTX 1080 sounds feasible.
And they have a history of selling 192/256-bit parts down as low as $200, so the GTX 1060 could bump up to 192/256-bit.
The GTX 1050 Ti - the 750 Ti still has plenty of gas in that 128-biit bus to burn, so that would cost them nothing to attach a faster chip tot he same bus.
All in all I expect Pascal to just be a Maxwell die shrink, with a few tweaks, with the architecture extended on the Enthusiast parts (ala GTX 280). I would be impressed to see HBM across the board, but they don't need it (like HD 4850 being GDDR3).
defaultluser
[H]F Junkie
- Joined
- Jan 14, 2006
- Messages
- 14,398
No, you're right, it will have enhancements, I was just thinking along the lines of Kepler, where it was more emphasis on tweaks like Boost and less FP64, and more efficient use of SMs. Not a major new architectural overhaul like Maxwell or Fermi.
Well, at some point the PC just took over as a gamers choice for me...
i was all happy with my amiga 500 untill... tadaa.. ultima underworld..
after that, the Amiga just got used less and less..
i found something better.