AMD VEGA confirmed for 2017 H1

[CAT-THE-FIFTH]

I do not like blind, teenage mind, and fanboys running the company. You will get the same result as AMD if you hire these kind of people ,which never learn from their failure and still blame their competitors for mismanagement.

He worked at Nvidia for years:

• Vice President of Telco Relations
  NVIDIA
  March 2008 – November 2010 (2 years 9 months)
  Established new division to develop relationships and explore technology development opportunities with the world’s leading telephone operators including China Mobile, Orange, Vodafone, Verizon, AT&T.
  - Established brand new group, hired world class key executives in Europe, Asia and the America’s
  - Created technology exchange programs and established value propositions
  - Responsible for ecosystem between ODM, OEM, the telco’s and the content world.
  - Design wins for at all targeted carriers
  
  
• 
  Vice President Content Relations. CTO for the PCGA (PC Gaming Alliance).
  NVIDIA
  September 2006 – March 2008 (1 year 7 months)
  Responsible for worldwide team managing third party partnerships with the worlds leading 3D content developers and publishers. Notable successes include TWIMTBP (The Way Its Meant To Be Played) program and contracts with EA, Ubisoft and Activision.
  
  In March of 2008 I officially joined the PCGA as NVIDIA's representative on the board and assumed the role of CTO.
  
  
• 
  Vice President GPU Sales
  NVIDIA
  June 2005 – September 2006 (1 year 4 months)
  International role responsible for worldwide sales and marketing of GPU (Graphics Processor Units) to ODM, OEM and retail channel partners. Responsible for $1bn sales.
  
  
• 
  Vice President of EMEA
  NVIDIA
  February 2000 – June 2005 (5 years 5 months)London, United Kingdom
  Regional role. NVIDIA’s first employee in Europe. Set up and established sales and marketing and later operations and finance in Europe. Responsible for opening offices in London, Munich, Paris, Moscow and Mumbai. Grew sales from zero to in excess of $300M.

I make that nearly 11 years. He was a key driving force behind TWIMTBP.

 

[N19h7m4r3]

Not a chance.

Check the bench thread, those GPUs struggle at 1080p and at higher resolutions you need a lot more than 4gb of VRAM per GPU.

Not a chance eh? Not what Adored TV found with 2x 480s, or what Hardware Unboxed found testing the 295X2 against it, with the AMD dual card over 20% faster in DX12 Tomb Raider.

It was found when people were looking into Drivers and Ryzen.

https://youtu.be/nLRCK7RfbUg?t=7m18s

 

[shankly1985]

but still developers prefer gameworks. Every year there are 7-8 AAA gameworks title.

And usually poorly optimised pieces of ****...

Yep You only after look at titles that AMD work on VS what Nvidia work on and see what gets rated the best for performance.
Nvidia titles are known for some of the "WORST" PC releases BATMAN for example! While everything AMD has worked on is known for its excellent PC performance for all hardware!! "Correct me with a AMD Title with poor performance"! The Nvidia list is quite endless
And its not just from AMD GPU users either CrapWorks is known for being poor even for Nvidia users and like above adds nothing to the game worth the performance hit!!

Maybe, just maybe now that Nvidia has opened up the Gameworks remains to be seen this will start to get better. but right now its doing nothing to push PC games forward, instead it just creates a ground for backlash!! Past titles back me up here and the term CrapWorks / GameDontWorks doesn't come from no where...

 

[TNA]

You should care about the total lack of competition allowing Nvidia to charge what they like. AMD need to get competing cards out, be it on value or performance.

Well yes, this makes more sense than to care about Roy. But how is bad mouthing them with every single post helping this?

I do not like blind, teenage mind, and fanboys running the company. You will get the same result as AMD if you hire these kind of people ,which never learn from their failure and still blame their competitors for mismanagement.

You did not answer my question.

Plus you make it seem like what they are doing is easy. Relative to Intel and Nvidia they have a very small budget, it is surprising they manage what they have. But you are too busy talking about minuscule things they slip up on. lol

 

[Kaapstad]

You should care about the total lack of competition allowing Nvidia to charge what they like. AMD need to get competing cards out, be it on value or performance.

This^^^

NVidia designed the Titan Xp to use GDDR5, it is a faster card than anything Vega is likely to bring to the table and it is an older design too.

AMD made a serious mistake going for HBM2 when they did not have to for Vega.

Worse still the Titan Xp using GDDR5X is capable of memory bandwidth far higher than anything that has been claimed for 2 stack HBM2 on a Vega card.

 

[CAT-THE-FIFTH]

NVM.

 

[Kaapstad]

Not a chance eh? Not what Adored TV found with 2x 480s, or what Hardware Unboxed found testing the 295X2 against it, with the AMD dual card over 20% faster in DX12 Tomb Raider.

It was found when people were looking into Drivers and Ryzen.

https://youtu.be/nLRCK7RfbUg?t=7m18s

I think I will put my trust in the guys on this forum when it comes to performance.

1080p
1 GPU

1. Score 106.35, Min 21.88, GPU 1080 Ti @2106/3150, CPU 6700k @4.7, Besty Link DX12 378.78 Drivers
2. Score 96.16, Min 30.45, GPU TitanP @2076/2777, CPU 5820k @4.7, Greebo Link DX12 373.06 Drivers
3. Score 94.41, Min 49.45, GPU TitanP @2088/2754, CPU 6900k @4.2, whyscotty Link DX12 369.05 Drivers
4. Score 92.87, Min 53.81, GPU TitanP @2088/2754, CPU 6900k @4.2, whyscotty Link DX11 369.05 Drivers
5. Score 74.74, Min 38.02, GPU 1080 @2063/2727, CPU 5650 @4.4, easyrider Link DX11 368.39 Drivers
6. Score 71.93, Min 8.06, GPU 980 Ti @1600/2100, CPU 6700k @4.7, Besty Link DX12 368.69 Drivers
7. Score 67.59, Min 30.56, GPU 1080 @2025/2610, CPU 6700 @3.4, regulus Link DX12 368.81 Drivers
8. Score 64.84, Min 31.99, GPU 1080 @1860/2502, CPU 6700k @4.0, hegemon Link DX12 368.81 Drivers
9. Score 64.76, Min 38.66, GPU 980 Ti @1537/1977, CPU 7700k @5.2, zia Link DX11 378.92 Drivers
10. Score 64.16, Min 39.31, GPU 980 Ti @1570/2102, CPU 4930k @4.0, Kaapstad Link DX11 359.91 Drivers
11. Score 63.25, Min 30.43, GPU 980 Ti @1570/2127, CPU 4930k @4.5, Kaapstad Link DX12 364.51 Drivers
12. Score 63.11, Min 30.87, GPU 980 Ti @1518/1928, CPU 4790k @4.8, khemist Link DX11 364.51 Drivers
13. Score 62.94, Min 40.59, GPU 980 Ti @1567/2000, CPU 3770k @4.5, David230 Link DX12 364.51 Drivers
14. Score 62.49, Min 41.82, GPU 980 Ti @1522/2001, CPU 6700k @4.4, Doom112 Link DX12 368.39 Drivers
15. Score 61.66, Min 17.36, GPU 980 Ti @1537/1977, CPU 7700k @5.3, zia Link DX12 378.92 Drivers
16. Score 60.46, Min 34.75, GPU 980 Ti @1518/1928, CPU 5930k @4.4, String Link DX11 364.51 Drivers
17. Score 56.93, Min 31.83, GPU TitanM @1392/2005, CPU 3930k @4.4, Gregster Link DX12 364.51 Drivers
18. Score 54.29, Min 37.98, GPU TitanM @1316/1752, CPU 5960X @4.0, Kaapstad Link DX12 362.00 Drivers
19. Score 48.67, Min 26.38, GPU 980 Ti @1203/1753, CPU 4790k @4.0, easyrider Link DX12 364.51 Drivers
20. Score 45.03, Min 25.03, GPU 980 Ti @1203/1753, CPU 4790k @4.0, easyrider Link DX11 364.51 Drivers
21. Score 44.31, Min 14.07, GPU 980 @1581/1852, CPU 4790k @4.5, SouthEastBlue Link DX11 364.51 Drivers
22. Score 43.72, Min 7.82, GPU 980 @1515/2102, CPU 3930k @4.0, Kaapstad Link DX11 364.51 Drivers
23. Score 37.66, Min 8.92, GPU 290X @1220/1650, CPU 4790k @4.7, thebennyboy Link DX12 16.7.2 Drivers
24. Score 36.82, Min 8.86, GPU Fury P @1040/500, CPU 930 @4.0, sothur Link DX12 16.7.1 Drivers
25. Score 35.99, Min 11.22, GPU 970 @1535/1801, CPU 3770k @4.5, tommybhoy Link DX11 364.51 Drivers
26. Score 33.87, Min 8.15, GPU 290X @1240/1600, CPU 4790k @4.7, thebennyboy Link DX11 16.3 Drivers
27. Score 32.16, Min 7.20, GPU Fury P @1040/500, CPU 930 @4.0, sothur Link DX11 16.7.1 Drivers
28. Score 30.16, Min 13.19, GPU 970 @1380/1753, CPU 3770k @4.5, tommybhoy Link DX12 364.51 Drivers
29. Score 27.21, Min 10.19, GPU 780 @1176/1552, CPU 4790k @4.6, Aretak Link DX11 362.00 Drivers
30. Score 25.51, Min 11.57, GPU 780 @1176/1552, CPU 4790k @4.6, Aretak Link DX12 362.00 Drivers

2 GPU

1. Score 94.61, Min 21.51, GPU 980 Ti @1455/1950, CPU 5930k @4.4, String Link DX11 365.10 Drivers

3 GPU

4 GPU


[*]Score , Min , GPU 00 @/, CPU @, 00 Link 00 378.78 Drivers

there is no point in posting the higher resolutions as there are no AMD results higher than 1080p.

 

[N19h7m4r3]

Aye, AMD putting all their eggs in the HBM2 basket was a massive mistake. Hynix let them down several times, considering their mass production was scheduled for Q3 2016 originally.

I think I will put my trust in the guys on this forum when it comes to performance.

That's the canned benchmark, not in-game results found by one of the most reputable reviewers and tech journalist there is.

The top result on your board is 100FPS average, 295X2 was matching that in the video I linked you as well; and that's a highly overclocked 1080Ti.

We all know the ingame performance varies depending on areas. Many people replicated it; as it seemed to have stemmed from people looking into the talk of NVIDIA drivers being odd on Ryzen. So many tested with AMD cards against NVIDIA, and some added in some dual cards to see what's going on.

In the video he also found the Titan XP was barely faster than the GTX 1070 during his tests.

It's fine if you don't want to believe it, but it's happened, and been recorded, and logged by several reviewers; even Adored TV; as much as people dislike him; but TechSpot/Hardware Unboxed is far more reliable and replicated it all.

 

[shankly1985]

I think I will put my trust in the guys on this forum when it comes to performance.




there is no point in posting the higher resolutions as there are no AMD results higher than 1080p.

TBH Kap while I think its excellent what you do on this forum you do the best Benchmark threads.. I have spoken quite a lot of my hate towards just still image or numbers.. For me Raw gameplay will always tell the biggest picture. Its 2017 benchmarking of games has moved on for me!! We now need a new way of benchmarking games!

Here is the 295x vs TitanXp running on DX12 and Ryzen CPU, tells a bigger picture dont it Plus its also known that the Tomb Raider bench dont added up to the actual gameplay of the game. So can we ever trust build in benchmarks?

 

[Gregster]

You should care about the total lack of competition allowing Nvidia to charge what they like. AMD need to get competing cards out, be it on value or performance.

100% this. Even the die hard NVidia fans should be wanting AMD to do really well, as regardless of brand loyalty, a monopoly (which it is as it stands at the top end right now) isn't good for anyone's wallet. Competition brings price competition and innovation. With only one competitor, you will only see Intel style drip feed 5%-10% gains each gen.

 

[CAT-THE-FIFTH]

AMD said they pushed most of their R and D spend to Zen in the last two years,so this is probably why they had to either back HBM2 or GDDR5X. Regarding my own experience of doing some run-throughs of the Geothermal Valley,I found at least the DX12 performance driver set released with the GTX1080TI to be buggy. In some run throughs with my GTX1080 there were decent DX12 gains but in others performance would tank and it was because it seems the driver or the game was not doing proper VRAM allocation,ie,when I hit issues,simply turning the texture settings down one notch solved it - I was running an older generation IB Core i7 though.

 

[shankly1985]

100% this. Even the die hard NVidia fans should be wanting AMD to do really well, as regardless of brand loyalty, a monopoly (which it is as it stands at the top end right now) isn't good for anyone's wallet. Competition brings price competition and innovation. With only one competitor, you will only see Intel style drip feed 5%-10% gains each gen.

100% agreed
Lets see Doom agree

 

[CAT-THE-FIFTH]

100% this. Even the die hard NVidia fans should be wanting AMD to do really well, as regardless of brand loyalty, a monopoly (which it is as it stands at the top end right now) isn't good for anyone's wallet. Competition brings price competition and innovation. With only one competitor, you will only see Intel style drip feed 5%-10% gains each gen.

I do feel a bit sorry for them that it is second time unlucky,but yeah the cards are probably 3 to 6 months delayed at this point(going from the slide which came out around the time Polaris was launched) - but apparently they might be dual sourcing from Samsung as AMD for the first time mentioned their HBM2 products.

 

[Gregster]

I get HBM2 and why they are using it but first time it was the interposers that made stock rarer than a Man U fan living in Manchester! And now it is looking like HBM2 is messing them about (is that confirmed btw or just rumour?). I just hope they have a beast and have it out soon.

 

[LoadsaMoney]

100% this. Even the die hard NVidia fans should be wanting AMD to do really well, as regardless of brand loyalty, a monopoly (which it is as it stands at the top end right now) isn't good for anyone's wallet. Competition brings price competition and innovation. With only one competitor, you will only see Intel style drip feed 5%-10% gains each gen.

Yup

 

[CAT-THE-FIFTH]

I get HBM2 and why they are using it but first time it was the interposers that made stock rarer than a Man U fan living in Manchester! And now it is looking like HBM2 is messing them about (is that confirmed btw or just rumour?). I just hope they have a beast and have it out soon.

On the flip side at least hopefully it gives them time to tweak the other aspects of the card before launch,ie,the cooler,drivers,etc.

 

[Armageus]

100% this. Even the die hard NVidia fans should be wanting AMD to do really well, as regardless of brand loyalty, a monopoly (which it is as it stands at the top end right now) isn't good for anyone's wallet. Competition brings price competition and innovation. With only one competitor, you will only see Intel style drip feed 5%-10% gains each gen.

Absolutely - whilst I rarely buy anything high end anymore, without competition at the top, there are no improvements to trickle down - only have to look at Intel I3's to see how that worked out (same dual cores for years, but with ever increasing prices)

Lets see Doom agree

Might be waiting a while

 

[TNA]

100% this. Even the die hard NVidia fans should be wanting AMD to do really well, as regardless of brand loyalty, a monopoly (which it is as it stands at the top end right now) isn't good for anyone's wallet. Competition brings price competition and innovation. With only one competitor, you will only see Intel style drip feed 5%-10% gains each gen.

Well said Gregster. This is how I see it also. I want AMD to do well for this reason, not because of some kind of blind loyalty or something.

100% agreed
Lets see Doom agree

He won't, as that would not fit with is mission parameters

He went all silent and won't answer my question.

 

[D.P.]

I get HBM2 and why they are using it but first time it was the interposers that made stock rarer than a Man U fan living in Manchester! And now it is looking like HBM2 is messing them about (is that confirmed btw or just rumour?). I just hope they have a beast and have it out soon.

There is no confirmation but what Hynix says in their product catalogs is pretty telling. It would be one thing if Hynix never publicly mentioned HBM2 modules and speed, then one could think that maybe they have some secret deal to supply AMD blah b;ah blah, but the fact is they publicly released a catalogue saying HBM2 2.0Gbp would be available in Q3 2016, and then ever quarter since then they have delayed and delay HBM2 and dropped the 2.0Gbps down to 1.6. If AMD really had some contract to use ever single HM2 chip produced then hynix would have never publicly advertised the chips for sale. So the fact that there is not currently volume HBM2 at 2.0Gbps from Hynix is pretty much fact.

Maybe AMD will try and use Samsung although Samsung is committed to giving Nvidia a lot of volume they might have some spare capacity.



Alternatively, if we assume HBM2 is plentiful then why is Vega delayed so much? Its not the fab process, AMD having been using GF's 14nmFF+ for a long time now, it is very mature so doesn't have the teething problems of a brand new node shrink. There is absolutely zero business reason to delay, and its not like vega replaces Polaris so they have to clear the supply chain (they did the opiste and released the RX580). So if its no HBM2 supply issues the oly possible explanations is a serious deign flaw requiring a new revision and testing of engineer samples. Burt then AMD have announced availability of their workstation version so there can't be a serious design flaw. The workstation model can be sold at significant profits and will be low volume so makes sense if HBM2 is rare and expensive.

Really, there is no other explanation. HBM2 is late.

 

[Gregster]

Yup

She is a stunner!