Discuss like an adult fella. He was quite correct in his statement, as was you if you are both going on rumours. AMD have priority with one supplier and Nvidia have priority with another. The telling thing will be on what level of GPUs come with HBM/2.
-
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 Polaris architecture – GCN 4.0
- Thread starter Boomstick777
- Start date
Discuss like an adult fella. He was quite correct in his statement, as was you if you are both going on rumours. AMD have priority with one supplier and Nvidia have priority with another. The telling thing will be on what level of GPUs come with HBM/2.
HBM2 reserved for entusiast cards that needs bandwidth for 4k/VR.
I expect gddr5 for the rest.
Small HBM2 HDR Polaris card with 1.3 displayport is the big thing for the consumer. New screens that offers a unprecedented image visual quality using HDR will be displayed with AMD hardware the best way on 4K.
While I love my 1440p 144hz screen a new HDR one will be a great addition both for desktop and gaming usage. You want bandwidth for the entusiast end as the consumer who buys that are going for 4k.
Polaris the brighter future from AMD
The GPU is only made in the fab, nothing else - and they lack the facilities for assembly. Assembly plants are located elsewhere. The plants AMD use for assembly are in China and Malaysia. All stock PCB cards are assembled at them. So it doesn't matter if they're GDDR5 or not.
It may be that they manage to wriggle out of their contract in future with said assembly plants ... but they literally just sold them (October), and were contracted at the time.
I don't think there are any suitable facilities in the US that would be sufficiently cheap, unless Samsung are building a large assembly facility in Texas to go with their 14nm FF fab that either is or soon will be online there ... there's been no word of it though, and doubt there will be. Assembly plants are almost always in Thailand / Malaysia / China / Vietnam due to cheap labour.
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,493
Lower power usage trough finfet tech.
AMD power tech to control transistors and set fps rate for power target.
New memory tech further advance the compression algorithm.
You then reach a unprecedented power goal.
Thats why you see 88w for AMD Polaris and 150w for Nvidia 950.
Its not just finfet but the tech behind the tech.
60w difference on a low level gpu is a ton of difference.
AMD power tech to control transistors and set fps rate for power target.
New memory tech further advance the compression algorithm.
You then reach a unprecedented power goal.
Thats why you see 88w for AMD Polaris and 150w for Nvidia 950.
Its not just finfet but the tech behind the tech.
60w difference on a low level gpu is a ton of difference.
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,493
Except TSMC quote a 70% reduction in power usage going from 28nm to 16ff+
Hexus said this,
A couple of months ago we heard rumours that AMD had secured priority access to SK hynix's HBM2 chip output. The news sources explained that AMD had worked closely with SK hynix on HBM(2) development and the pair of companies has a long established relationship. Furthermore AMD was expected to strive to capture "as much of the initial production capacity as possible," which could have derailed Nvidia's Pascal GPU launch proper."
If the source is accurate that sounds bad for Nvidia really bad.
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,493
Hexus also said that hynix are not the only supplier, duh.
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,493
yeah, again, TSMC themselves quote 70% power usage reduction from their process... that IS 3x P/W
Yeah but let's not get too excited until we see the hardware, and that Amd can make big chips on the process.
It could all be an illusion with their frtc and low voltage and freesync.My old 7950 in a game under 100% load used to use 1.18v for a 925 MHz boost clock, in stock mode with a 0% power limit the card would throttle between 800 and 925 MHz, using 130w-160w. If I manually undervolted to 0.8v for 800mhz the 7950 would use barely 85w, or around 100w at 0.9v 900mhz.
No body cares about whose foundry they use ^^^^
It's carefully augistrated of course, with the new Transistor Switching tech it's obvious why they had V-Sync on.
But at the end of the day that is not really so important, if its running flatout it will no doubt use more power but also perform better. if its so frugal in this test just how much overhead performance is left in it before it gets anything like the power consumption of the 950?
It's carefully augistrated of course, with the new Transistor Switching tech it's obvious why they had V-Sync on.
But at the end of the day that is not really so important, if its running flatout it will no doubt use more power but also perform better. if its so frugal in this test just how much overhead performance is left in it before it gets anything like the power consumption of the 950?
Last edited:
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,462
- Location
- Planet Earth
The 32NM and 28NM plants GF use are located in Dresden and New York:
https://en.m.wikipedia.org/wiki/GlobalFoundries
This covers not only AMD CPUs but also the SOCs in the PS4 and XBoxOne too. All are assembled in Asia.
AMD produced most of its CPUs at the fab in Dresden and they have WSA considerations too and Samsung had a head start over TSMC and has capacity which is not contested by a billion other companies too. WSA is really an important consideration here. It's one of the main reasons AMD has lost so much money through penalty clauses and inventory writedowns(since they need to produce a fixed number of chips based on how much capacity GF expects them to buy).
You also need to consider that Samsung fab is probably producing chips for iPhones which are being assembled in China.
Also using a low power process for mainstream chips will be advantageous due to the target market and also since it will be using GDDR5,so they cannot rely on power saving measures with HBM2.
You need to consider shipping costs are a tiny percentage of overall costs. FFS, a while ago I saw some Young's Frozen Cod which was caught in the Atlantic, packed in China and then shipped back here!
But we are probably going to disagree on this so best keep it at that.
Last edited:
CAT.
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,462
- Location
- Planet Earth
It's interesting about the tessellation bit though - it appears(if I am reading it right though),that they will only actually render what can be seen on screen and Nvidia have been doing something similar for a while.
IF true, no wonder Nvidia have better performance in extreme tessellation scenarios. It is not that their Tesselator is more powerful, just that they are performing Hardware based screen-space culling on tessellated vertices.
Yup.
Nvidia also have more aggressive driver level Geometry LOD.
Nvidia are very good at optimising their hardware through software, better than AMD, or at least right now.
The down side is there is a minor IQ difference, with driver level Geometry LOD for example distant objects can sometimes appear less complete and / or less detailed than on AMD.
Personally i just think its dumbing down IQ to gain that extra 1 FPS here...and there.... but it all adds up and TBH more often than not you'd have to know what you are looking at to see it and even then its hardly obvious.
I'd rather it just be the game developers cut, rather than vendors hacking and picking away at the IQ to look good in benchmarks.
Last edited: