True, but back then AMD had even less money, and more debt.
They simply don't have the resources to have poured 3 Billion into a 815mm2 behemoth; never mind 3 B into any GPU.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD VEGA confirmed for 2017 H1
- Thread starter Doom112
- Start date
- Status
They simply don't have the resources to have poured 3 Billion into a 815mm2 behemoth; never mind 3 B into any GPU.
Plenty companies have made better products with less money than their competitors. It is not the be all and end all.
Not the end all and be all, but it plays a massive role when they're billions in debt, and have more than one major product line for different segments.
Nor have competitors quite as powerful as Intel and NVIDIA; all those have died out, or moved to a different market segment or niché.
Hopefully AMD's talented engineers knocks Vega out of the park as well as Zen has been doing for CPUs.
Soldato
- Joined
- 28 Sep 2014
- Posts
- 3,574
- Location
- Scotland
Q1 2017 databook is old now, actually SK Hynix got databook updated to Q2 2017.
https://videocardz.com/69504/sk-hynixs-updated-memory-catalog-features-hbm2-and-gddr6
https://www.skhynix.com/static/filedata/fileDownload.do?seq=421
HBM2 part numbers in both Q1 and Q2 are different.
https://devblogs.nvidia.com/parallelforall/inside-volta/
That is as much as we ever found out about Pascal, Maxwell, Kepler and Fermi etc.
Nvidia are, they are fighting Intel and others for HPC market share, which id why they threw 3bn at Volta and decided that an 815mm^2 chip was still desirable, otherwise they would have waited for a a good 10nm process to come along. They are having similar battles in the autonomous driving space
Nothing has changed for availability for 2.0GBps 4GB modules, still MIA.
Worse is 1.6Gbps is now listed as Q2 2017, up from Q1 2017, and before that Q3 2016.
Not good news for general availability.
2GB/pin is not a technical issue, is just hynx diverting resources to flash memory with much higher demand, still doesn't mean they cannot accomodate AMD's need, at least i hope so, they co-developed the memory with them...
Last edited:
Aye, I really hope the reason we're not seeing he higher versions of HBM2 is because SK Hynix is currently providing it all for AMD. If not, it'll certainly sting us all.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
In general such a data book would be for prospective customers to know what's available. IF they are producing X amount and that capacity is already allocated and they won't have extra capacity till a later date, then that is the date you put in such a book. Same way OCUK might have 100 graphics cards coming in tomorrow, but show no stock till 2 months later because those cards are already sold.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
As expected we get WCCF pointing at someone else as the source and giving it the usual salty disclaimer.
I have a question that is going to sound super dumb, but I am going to ask it anyway since I am in the graphics card sub-forum.
How much extra heat does having the RAM on the GPU package cause, vs. using the traditional RAM chips on the PCB, or is it less over all heat, just more concentrated?
How much extra heat does having the RAM on the GPU package cause, vs. using the traditional RAM chips on the PCB, or is it less over all heat, just more concentrated?
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
Vega does need to compete,
They gave the high end a miss last year which was meant to be so they could put the time and effort into getting ahead of the game with Vega.
Basically the High-end AMD users went a year with nothing on offer in order to leapfrog over Pascal and get the next gen first.
I was secretly hoping that was the masterplan.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
I have a question that is going to sound super dumb, but I am going to ask it anyway since I am in the graphics card sub-forum.
How much extra heat does having the RAM on the GPU package cause, vs. using the traditional RAM chips on the PCB, or is it less over all heat, just more concentrated?
I don't think it adds a lot heat. My air cooled Fury is great for temps, a lot better than the MSI 290x gaming it replaced but I do think it's an issue for stability with overclocking which is why Nvidia aren't using HBM with gpu boost on their GTX gaming cards yet.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 31,179
If I can get Ti-level performance from Vega then I'll be very very happy. But as you can imagine, I'm not in that much of a rush!! It's just a pain turning on the ULMB mode every time I fancy a quick blast
(frame skips else at 75Hz)it's better to have them on the chip, because whatever cooling there is will be focused on the chip, while traditional GDDR module get stuck with passive, with temperatures exceeding 90° or even a 100°I have a question that is going to sound super dumb, but I am going to ask it anyway since I am in the graphics card sub-forum.
How much extra heat does having the RAM on the GPU package cause, vs. using the traditional RAM chips on the PCB, or is it less over all heat, just more concentrated?
Soldato
- Joined
- 18 May 2010
- Posts
- 23,010
- Location
- London
Salt. June the 5th according to this.
if that's the pricing then vega is a fail, not only they have no SKU close to 300$ ballpark, Nvdia already sold truckloads of 1070/1080, with low-mid end volta few months away.
all nvidia needs to do is to cut the price by 50$ for the 1070 and 1080, this will automatically renders vega useless, even if they decide to follow up on a price war and even if vega is slightly faster than it's direct competition.
this pricing if really stupid especialy for the low end, AMD could be gearing up to another break even architecture, with no market share benefit at all, maybe even pushing more towards their competition.
low vega cannot top 350$.
- Status