Another 6 months before VEGA?

ecmaster76 said:
They announced that a year ago. Can you buy a product with that part in it? If not then, there a few reasons that could be the case
Click to expand...
If that is your case, you should provide the parts catalogue of Samsung for their HBM2 and what they spec, until then we just have the official Samsung manufacturer production news.
Thanks
 
CSI_PC said:
Because the P100 NVLink card is at maximum power, it is impossible to increase the clocks of the memory without exceeding 300W, some articles have picked up on this in the past as well.
Sorry we will have to disagree with regards to meaning when one says "supports a data transfer rate of 3,200" and with HBM2 where they say "Samsung’s new DRAM package features 256GBps of bandwidth, which is double that of a HBM1 DRAM package".
Cheers
Click to expand...

Its clear what they compare it to. You should just have linked the whole sentence.

The new DRAM supports a data transfer rate of 3,200 megabits per second (Mbps), which is more than 30 percent faster than the 2,400Mbps rate of 20nm DDR4 DRAM
Click to expand...

In the end of the day the result is the same.
No HBM2 2Ghz. No DDR4 3200mhz.

Nvidia isn't the only one using HBM2 either.
 
Shintai said:
Its clear what they compare it to. You should just have linked the whole sentence.



In the end of the day the result is the same.
No HBM2 2Ghz. No DDR4 3200mhz.
Click to expand...
Context to me is that one supports speeds to 3,200Mbps (does not mention native which is what your argument is about them being slightly misleading), and the HBM2 product features 256GB/s.
For the 1st one to be native they would not use the word supports speeds.
But this is seriously digressing.

It is wrong to generally say no DDR4 3200, because the chips support speeds even over 4000 now.
If the support speed did not exist, you would not have any of these fast DDR4 modules.
Comes down to context IMO.
 
Last edited:
Spoonie_G said:
Not looking good. According to this article we're looking at the last months of Q3 2017. Didn't I say that it wont be until September before we see these parts? AMD leaves me no choice but to overpay for a GTX1080, and hope I don't have screen tearing issues. Who's going to be willing to wait until September? AMD is Doomed (no pun) if they think that releasing a competing product, which will be 26 months after the fury x, makes them a company that's worth sticking with. The longer they wait? The longer it's going to take for them to make any kind of impact on the high end. AMD need to stick to the mid-range. They are incapable of releasing high-end cards that fit any type of product life-cycle. If what is below is true i'm done with AMD for now. I'll sell my monitor, get a freesync monitor, overpay for both because NVidia can do as they please because AMD is incapable of competing at the high-end. Leaving no competition to help drive down prices. On top of that, by the time September comes (8 months now) NVDIA will probably have a card faster than the VEGA waiting in the wings to be released in October. WTF is wrong with AMD? Retaining costumers must not be in their project plan.

Here's the quote for the lazy:

"
AMD Vega 10 Release Date

Vega 10 GPUs, the first ones to utilize the new architecture of AMD, is expected to hit the store shelves in the last months of Q3 2017, according to PC Perspective. The GPU is said to come with a higher price compared to its predecessor because of its high-performing specs, but not as expensive as NVIDIA GTX 1080.

http://www.universityherald.com/art...a-10-vs-nvidia-gtx-1080-amd-vega-10-specs.htm
Click to expand...

You don't need to have tearing.

NVIDIA Maxwell and Pascal parts have FastSync that is a back end triple buffer. It drops unused frames over refresh rate, doesn't add a bunch of lag like Vsync. (works best with fps 2-3X refresh rate)
 
Okay this thread is to bitch about why Vega is not here and its done a very good job at that until now. lol
 
In AMDs Q4 results.

GPU products based on the Vega architecture are expected to ship in the second quarter of 2017.
Click to expand...

Note ship, not launch. May could become June/July.
 
Shintai said:
In AMDs Q4 results.



Note ship, not launch. May could become June/July.
Click to expand...

Could you link the Q4 results? Since [H] and every other tech website has yet to post it? And My GoogleFu seems to be failing me at the moment.
 
Thanks.

They still took a loss, even though they had the best consumer product catalog they've had in years, AND the new semi-custom work. It's a smaller loss, but I just don't see AMD surviving through 2019.

Maybe Ryzen will blow us away. And Vega will perform much better than 1080. But I'm not holding my breath. Intel and Nvidia certainly aren't taking them seriously.

I guess this all-but-confirms no license deal with Intel?
 
Their Outlook for Q1 is also telling. The Intel-AMD deal was nothing but fiction it seems.
 
Shintai said:
Their Outlook for Q1 is also telling. The Intel-AMD deal was nothing but fiction it seems.
Click to expand...

Yeah, that Q1 2017 17% year-on-year revenue outlook continues to build upon the 15% increase year-on-year they saw in Q4. So no new products available in quantity until Q2. Guess Ryzen won't be widely available until April?
 
whateverer said:
Thanks.

They still took a loss, even though they had the best consumer product catalog they've had in years, AND the new semi-custom work. It's a smaller loss, but I just don't see AMD surviving through 2019.

Maybe Ryzen will blow us away. And Vega will perform much better than 1080. But I'm not holding my breath. Intel and Nvidia certainly aren't taking them seriously.

I guess this all-but-confirms no license deal with Intel?
Click to expand...

Really you are thinking about dooms day scenario now. According to your analysis AMD should have failed 3 years ago. That is the smallest loss they took compared to how they have been bleeding money and with Vega coming up and Ryzen shipping soon, and They have a deal with google. If they can do what they did with fail of a polaris launch I doubt they would be worse shape by the way zen is looking and they will finally have a card to fill the gap they have had for a year. So I don't see AMD going out of business in 2 years lol.
 
NKD said:
Really you are thinking about dooms day scenario now. According to your analysis AMD should have failed 3 years ago. That is the smallest loss they took compared to how they have been bleeding money and with Vega coming up and Ryzen shipping soon, and They have a deal with google. If they can do what they did with fail of a polaris launch I doubt they would be worse shape by the way zen is looking and they will finally have a card to fill the gap they have had for a year. So I don't see AMD going out of business in 2 years lol.
Click to expand...

They still lost 50 million in their second-strongest quarter. They lost 500 million for the year.

One quarter of tiny profit does not mean the sinking ship has righted itself. When AMD cuts it's yearly losses IN HALF WHILE INCREASING REVENUES, then you can make such a claim.
 
whateverer said:
Yeah, that Q1 2017 17% year-on-year revenue outlook continues to build upon the 15% increase year-on-year they saw in Q4. So no new products available in quantity until Q2. Guess Ryzen won't be widely available until April?
Click to expand...

End of Feb - April is what is expected.
 
whateverer said:
They still lost 50 million in their second-strongest quarter. They lost 500 million for the year.

One quarter of tiny profit does not mean the sinking ship has righted itself. When AMD cuts it's losses IN HALF, then you can make such a claim.
Click to expand...


Yep Q4 is generally almost as good as Q3 "usually", but in this case, where did those updated PS4 sales go in semicustom ;)

See how they are screwing with numbers? Q3 they factored in charges for wafer procurement from other fabs, even with that gone they still ended up with a loss.

There is no amount of "believable" numbers AMD can give unless you go through the line items and determine for yourself what is going on and then compare with other quarters.......
 
whateverer said:
They still lost 50 million in their second-strongest quarter. They lost 500 million for the year.

One quarter of tiny profit does not mean the sinking ship has righted itself. When AMD cuts it's yearly losses IN HALF WHILE INCREASING REVENUES, then you can make such a claim.
Click to expand...

Agreed! But how are you predicting they are out of business in 2 years. Are you betting on it? They are improving, yea they lost money but do you see an improvement over previous quarter or do you see it getting worse? They improved despite the shit show of a launch "polaris", not having any high end shit. On top still no competition on the CPU side. Now they will have entire lineup of new products coming up. How to do come to conclusion they are out in 2 years?

I don't have issue with your explanation but I do with your bold prediction of them being out in 2019. I have heard that song for years now.
 
NKD said:
Agreed! But how are you predicting they are out of business in 2 years. Are you betting on it? They are improving, yea they lost money but do you see an improvement over previous quarter or do you see it getting worse? They improved despite the shit show of a launch "polaris", not having any high end shit. On top still no competition on the CPU side. Now they will have entire lineup of new products coming up. How to do come to conclusion they are out in 2 years?
Click to expand...


The only way AMD will survive is if Zen does it for them for more than 2 years, which is a risky proposition at this point, need to see what its capable of first then everything will be more clear, cause Zen is their foundation for the next 10 years ( Zen and Zen +) i'm assuming right now 10 years cause each gen of CPU's have been for 5 years.

Polaris has nothing to do with their bottom line. Lets not even go there, we saw full quarter of Polaris where they lost marketshare, just improved sales because of Q3 season strength, which had more benefits to nV. And that was without value card competition from nV. Margins stayed relatively flat up 1 % point, so no new products yet. Cause if Zen was shipping to OEM's, I would expect Margins to improve by 5%.

Why aren't they getting profits when they aren't paying for wafer deal anymore that was big bucks wasn't it? Like what 150, 250 million something like that?
 
razor1 said:
The only way AMD will survive is if Zen does it for them for more than 2 years, which is a risky proposition at this point, need to see what its capable of first then everything will be more clear, cause Zen is their foundation for the next 10 years ( Zen and Zen +) i'm assuming right now 10 years cause each gen of CPU's have been for 5 years.

Polaris has nothing to do with their bottom line. Lets not even go there, we saw full quarter of Polaris where they lost marketshare, just improved sales because of Q3 season strength, which had more benefits to nV. And that was without value card competition from nV. Margins stayed relatively flat up 1 % point, so no new products yet. Cause if Zen was shipping to OEM's, I would expect Margins to improve by 5%.

Why aren't they getting profits when they aren't paying for wafer deal anymore that was big bucks wasn't it? Like what 150, 250 million something like that?
Click to expand...

Like all i am asking is. Is he guaranteeing they are out of business in 2 years and if not is he ready to come and eat crow on his statement. I get the points he was making but to say yea they don't last past 2019 is kind of going over the top a little.
 
NKD said:
Agreed! But how are you predicting they are out of business in 2 years. Are you betting on it? They are improving, yea they lost money but do you see an improvement over previous quarter or do you see it getting worse? They improved despite the shit show of a launch "polaris", not having any high end shit. On top still no competition on the CPU side. Now they will have entire lineup of new products coming up. How to do come to conclusion they are out in 2 years?

I don't have issue with your explanation but I do with your bold prediction of them being out in 2019. I have heard that song for years now.
Click to expand...

Because AMD at this point is likely under the 2 billion mark in Total Current Assets. That's value that can be used to pay debts on the short-term. Yahoo hasn't run the 2016 numbers yet, but they were barely above 2 billion in 2015. Just click on "Balance Sheet."

https://finance.yahoo.com/quote/AMD/financials?p=AMD

They also have 2.2 Billion of debt that matures in 2019.

http://seekingalpha.com/article/3221666-amd-bonds-how-much-time-is-really-left-on-the-clock

So the likelihood of them getting another round of financing at a rate that won't eat the company alive is pretty much moot, given the continuing loses for each year.. Especially if they don't even have enough assets left to pay-off that 2.2 Billion. And with that big a debt, you either improve your balance sheet so you can get a more favorable rate in refinance, or you go bankrupt.
 
Last edited:
NKD said:
Like all i am asking is. Is he guaranteeing they are out of business in 2 years and if not is he ready to come and eat crow on his statement. I get the points he was making but to say yea they don't last past 2019 is kind of going over the top a little.
Click to expand...


Yeah their value as a company is less then their total debt which is not good at this point so if Zen doesn't bring them out of this funk, AMD is definitely gone in its current form.
 
razor1 said:
The only way AMD will survive is if Zen does it for them for more than 2 years, which is a risky proposition at this point, need to see what its capable of first then everything will be more clear, cause Zen is their foundation for the next 10 years ( Zen and Zen +) i'm assuming right now 10 years cause each gen of CPU's have been for 5 years.
Click to expand...

AMD is calling Ryzen a four year GPU, not 10. Apparently they are planning an architecture beyond Ryzen. I'm hoping that Ryzen keeps them in the game long enough to get there, too.
 
Algrim said:
AMD is calling Ryzen a four year GPU, not 10. Apparently they are planning an architecture beyond Ryzen. I'm hoping that Ryzen keeps them in the game long enough to get there, too.
Click to expand...


So 2 years for Zen and 2 years for Zen + ?
 
Clarification: Four year Ryzen Architecture. I would assume (though assumptions are dangerous in these parts) that Zen/Zen+ are considered the same architecture.
 
ah ok well that's yep assumptions are always dangerous lol.

Ok reading about Zen a bit, looks to me Zen + and then there will be another chip after that also built on Zen architecture so a full 3 generations of CPU's, 5-6 years for Zen, which is the same life span of AMD CPU architectures from the past.

http://www.techtimes.com/articles/1...at-powers-ryzen-chips-set-to-last-4-years.htm

That statement is a not-so-subtle dig at Intel's so-called "tick-tock" model, which implies a two-year lifespan for a chip architecture. Here, the "tick" represents the year Intel builds a CPU from a smaller process technology. The "tock" stands for the time a brandnew architecture is introduced.

It must also be underscored that the four-year lifespan does not mean that AMD's new microprocessor architecture will simply get replaced once that window is up. It appears that the company plans to improve on it further, expanding its life by as much as three generations.
Click to expand...
 
Last edited:
zen+ seems like probably mostly clock bumps may be little ipc gains. Yea I did hear them say, I think it was in a article after an interview that they want to make new and improved architecture every 4 years or so. We will see about that. I think success of zen will only set that up.
 
Spoonie_G said:
On a side note. I'm picking up my EVGA GeForce GTX 1070 FTW HYBRID today.
Click to expand...

Do you really need a watercooled 1070? Most nicely aircooled ones clock pretty good and run decent. I got the evga 1070 for 389 and rocking over close to 2ghz average boost clock. But I guess if thats what you want then thats what you want lol
 
NKD said:
Do you really need a watercooled 1070? Most nicely aircooled ones clock pretty good and run decent. I got the evga 1070 for 389 and rocking over close to 2ghz average boost clock. But I guess if thats what you want then thats what you want lol
Click to expand...


Yes... Air cooled cards throttle. Water cooling give you more overclocking headroom.. Nice try.
 
Spoonie_G said:
Yes... Air cooled cards throttle. Water cooling give you more overclocking headroom.. Nice try.
Click to expand...

LOL you do take little things to heart. I was just saying. lol at air cooled cards can get noisy. So your radiator is not air cooled on that card? Nice try! ;)
 
NKD said:
LOL you do take little things to heart. I was just saying. lol at air cooled cards can get noisy. So your radiator is not air cooled on that card? Nice try! ;)
Click to expand...


I'm not taking things to heart. It was you that responded with that snide remark.
 
NKD said:
Like all i am asking is. Is he guaranteeing they are out of business in 2 years and if not is he ready to come and eat crow on his statement. I get the points he was making but to say yea they don't last past 2019 is kind of going over the top a little.
Click to expand...

It is a very boring way of pretending to be interesting , yelling doom and gloom about AMD done to death already ....
 
CSI_PC said:
If that is your case, you should provide the parts catalogue of Samsung for their HBM2 and what they spec, until then we just have the official Samsung manufacturer production news.
Thanks
Click to expand...

A quick Google search doesn't pull anything up. But here at [H] anytime a product is announced and is isn't available we call it a paper launch. If it goes over a year then maybe vaporware.

Obviously there is no way for me to prove or disprove what Samsung can actually make. That said, either something is making the high speeds parts unavailable or unappealing to nVidia or they are strategically holding back. I cant prove that one way or another either.

BTW, This page is getting *way* off topic even by AMD Flavor standards
  • HBM memory tech, sure
  • AMDs financial health, I guess...
  • Zen roadmaps, umm guys?
  • GTX 1070 air vs water cooling, please take elsewhere
 
ecmaster76 said:
A quick Google search doesn't pull anything up. But here at [H] anytime a product is announced and is isn't available we call it a paper launch. If it goes over a year then maybe vaporware.

Obviously there is no way for me to prove or disprove what Samsung can actually make. That said, either something is making the high speeds parts unavailable or unappealing to nVidia or they are strategically holding back. I cant prove that one way or another either.

BTW, This page is getting *way* off topic even by AMD Flavor standards
  • HBM memory tech, sure
  • AMDs financial health, I guess...
  • Zen roadmaps, umm guys?
  • GTX 1070 air vs water cooling, please take elsewhere
Click to expand...
Unfortunately you wont find the parts catalogue, only one currently available is the 1H 2016 from Samsung and it has no mention of HBM2.
But it is not vaporware because Nvidia has been using it since summer 2016 in their two P100 models, and these really did ramp up in production late Q3/early Q4 2016.
Rumour (emphasis on that point) though last year was that Nvidia had tied up all of Samsung's HBM2, I think latest Samsung parts catalogue 1H 2017 comes out around May-June.
Cheers
 
Last edited:
Spoonie_G said:
I'm not taking things to heart. It was you that responded with that snide remark.
Click to expand...

LOL all I asked why you feel you need watercooling thats all. Cuz it does cost 60-70 more especially on pascal when it runs cool as hell. That's all. Now I did say what you want is what you want, no one can change that. You don't like being asked a question? Stop assuming everyone is out to get you, or you are mad over getting advice in a forum?

anyways, lets end it.

Cheers.
 
CSI_PC said:
Unfortunately you wont find the parts catalogue, only one currently available is the 1H 2016 from Samsung and it has no mention of HBM2.
But it is not vaporware because Nvidia has been using it since summer 2016 in their two P100 models, and these really did ramp up in production late Q3/early Q4 2016.
Rumour (emphasis on that point) though last year was that Nvidia had tied up all of Samsung's HBM2, I think latest Samsung parts catalogue 1H 2017 comes out around May-June.
Cheers
Click to expand...

That's the thing though, P100 only runs its HBM2 at 70% of what Samsung claimed they could produce - 180GB/s per stack on P100 vs 256/GBs per stack that Samsung announced (granted the marketeers may have simply omitted the usual "up to" qualifier for a long term goal).

Hynix is claiming they can get 205GB/s out of their stacks, almost 15% faster than each stack on P100

Ultimately my point is we are yet to see a commercially available product using full speed HBM2 which brings into question where the holdup is.
 
ecmaster76 said:
That's the thing though, P100 only runs its HBM2 at 70% of what Samsung claimed they could produce - 180GB/s per stack on P100 vs 256/GBs per stack that Samsung announced (granted the marketeers may have simply omitted the usual "up to" qualifier for a long term goal).

Hynix is claiming they can get 205GB/s out of their stacks, almost 15% faster than each stack on P100

Ultimately my point is we are yet to see a commercially available product using full speed HBM2 which brings into question where the holdup is.
Click to expand...
And you cannot run over 300W.
How many times have I mentioned the spec of the P100 NVLink is 300W, you try to run it at 256GB/s and you are over that.
The only evidence that exists is the official production news from Samsung.
 
CSI_PC said:
And you cannot run over 300W.
How many times have I mentioned the spec of the P100 NVLink is 300W, you try to run it at 256GB/s and you are over that.
The only evidence that exists is the official production news from Samsung.
Click to expand...

Do we know the actual power usage on P100 and how much of that is consumed by the HBM?

If the performance boost was worth it NV could have updated NVlink to support more power.
 
You must log in or register to reply here.
Back
Top