Drivers won't give you a 20% performance increase. In an ideal world it would, but in reality it won't.
-
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.
Poll: ** The AMD VEGA Thread **
- Thread starter TNA
- Start date
- Status
Drivers won't give you a 20% performance increase. In an ideal world it would, but in reality it won't.
Soldato
- Joined
- 26 Oct 2013
- Posts
- 4,236
- Location
- Scotland
That slide clearly is talking about different segments and the value of the TAM. If they are only in it for 6 months they will only be able to potentially gather half of it. To me the graph is clearly talking about money and growth potential.
Thanks for that.That slide has been interpreted in many different ways. Some say it's the segment they are attacking, others say the line merely indicates the end of the financial quarter they are referring to.
The slide is very obvious in its meaning, not at all ambiguous.
TAM axis on the left. Quarter on the bottom. In 2016 AMD were able to address all the mainstream segment with Polaris (their words). In 2017 Polaris addresses all the mainstream and Vega addresses some of the Premium/Pro market, but not all. Logically, this can only mean they don't have top-end performance, if you think about it.
The slide goes on to show that in 2018 they (believe they) will address all the Premium/Pro market. This likely means new Vega SKUs in 2018, possibly on a more mature process, and probably factoring in some improvement in drivers also.
But in 2017 they are not able to address the whole Premium/Pro market. Reason? Can only be lacking top-end performance relative to nV. No other conclusion makes sense. If they had top-end perf they could address the whole market.
Soldato
- Joined
- 26 Oct 2013
- Posts
- 4,236
- Location
- Scotland
How would they address the whole market for the first half of the year without Vega being launched yet even if it's twice as fast as a Titan XP?
The line moves the up the whole stack in 2018 because AMD believe they will have hardware to cover the high end market for the entire 12 months rather than 5-6.
The line moves the up the whole stack in 2018 because AMD believe they will have hardware to cover the high end market for the entire 12 months rather than 5-6.
And just how AMD could have prevented consumer sites from getting Vega FE?
By telling shops not to sell them to anyone not being in list of "AMD approved companies/persons" or by saying any site who reviews Vega FE stops getting any product samples/support from AMD?
Bet you would be picking up pitchforks and torches and screaming bloody murder then!
FE's results pretty much feel like more of same old GCN which doesn't make any sense:
With AMD having not had fresh higher end card in sale for some time why would they waste this long in making new big chip high end card if it didn't have any real architectural improvements?
"Die shrinking" old Fiji wouldn't take this long.
Also with Nvidia very succesfully using GDDR there would have been plenty of time to use it instead of waiting for uncertain HBM2.
Provably false.
The 2016 line shows the whole mainstream segment being covered by Polaris.
When was Polaris launched? Answer: June. 1/2 through 2016.
I'm expecting it to be same old ATI/AMD over again - packed full of forward looking technology 1-3 generations before it actually makes sense to put it into a released GPU and will never actually see utilisation throughout Vega's useful lifespan. That said there might be a small saving grace for AMD there in that they might be able to use it as a basis for future console hardware where some of those features might get used but only within the limits of that ecosystem.
Soldato
- Joined
- 26 Oct 2013
- Posts
- 4,236
- Location
- Scotland
They had multiple other cards all covering this segment before Polaris aswell though. Do you expect them to name all graphics architectures that were covering it?
LOL, clutching at straws mate.
Vega isn't going to be a 1080Ti competitor, it's going to top out somewhere near 1080 perf.
Feel free to quote me on release day... (not 9 months later after they fix the drivers...)
Yes
I normally use the Terrae_Expugnandae_5.1 mod as you get to play on a huge map as a single faction and there are 199 provinces you can conquer.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 31,192
For me if it performs as well as a Ti in Vulkan/DX12 but is priced closer to the 1080 then it's onto a winner imo. I'm hoping for strong forward-looking API performance myself
Doesn't matter if priced correctly.
You would save a lot of mony just buying a potato PC to play them on though
All that power you have is meant for pretty games.
I have always been a very big gamer but I am also very interested in technology.
For gaming I am happy to use anything that works so a single card @2160p will do fine as long as it runs.
My interest in technology is where all the GPUs come in as I like to find out stuff for myself rather than relying on tech sites who often get it wrong or are influenced by hardware manufacturers.
All the indications are they can't afford to undercut nV's prices, since it costs so much more to manufacture.
Unless AMD want to make a loss to gain market share... but can they afford to?
The funny thing is I, and many others I'm sure, took a look at the Vega launch details (what little there was of it) and the limited disclosure in the AMA and immediately knew it wasn't targeting gamer's as a gaming card. In the context of their presentations and the intended Vega H1 window, it was clearly a move to release Vega silicon so that they can claim they executed to schedule. From a corporate perspective it had to be released and so they worked it into a niche product. Where are the tech journalists and what do they do? Their competence has fallen dramatically such that most cannot produce or convey any worthwhile commentary or even trivial insight.
On the subject of Nvidia, their messaging is relatively straight forward as they are the performance leaders. Someone else said it earlier in the thread, since their engineers reliably churn out a newer faster products, their marketing department could sit back and twiddle their thumbs and the cards will release to consumer fanfare as they are the first to market at that the next performance level.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 31,192
No, but supply is going to be very very tight just like the FuryX was. This is just being released so they can say they actually released it, if that makes sense. They know they'll make nothing from it, it's just a stop-gap til they can get back into the higher-end game.
Caporegime
- Joined
- 8 Jul 2003
- Posts
- 30,063
- Location
- In a house
As i said earlier, AMD can't use GDDR, as they do a one for all.
Soldato
- Joined
- 28 May 2007
- Posts
- 10,222
I have always been a very big gamer but I am also very interested in technology.
For gaming I am happy to use anything that works so a single card @2160p will do fine as long as it runs.
My interest in technology is where all the GPUs come in as I like to find out stuff for myself rather than relying on tech sites who often get it wrong or are influenced by hardware manufacturers.
Yea what i said was only in jest. You wouldn't be spending the amounts you do if there was no real interest. just a shame the benching scene around here is not as prominent as it was. Probably due to the slowing down of tech and the lack of competition.
Soldato
- Joined
- 28 May 2007
- Posts
- 10,222
Amd themselves said that Vega can use gddr5.
https://www.pcgamesn.com/amd/amd-vega-gpu-specifications
Associate
- Joined
- 31 Oct 2012
- Posts
- 2,243
- Location
- Edinburgh
That's a weird article in that it's been edited a few times including recently, but is a very old one & doesn't separate new & old information or give sources. You may be right, I'd like to see it elsewhere tho too!
(Also strictly being able to use GDDR5 on GCN doesn't mean on Vega necessarily)
(Also strictly being able to use GDDR5 on GCN doesn't mean on Vega necessarily)
- Status