AMD VEGA confirmed for 2017 H1

akarypid · 9 Jan 2017 at 09:22

Doom112 said:
Some AMD ** Please do not start using that term - EVH ** are still in a denial that even on Vulkan VEGA is only on par with GTX 1080. It is really a matter of concern for AMD when VEGA performs in 95% of games ,which are mostly based on DX11.

This is exactly what happens when you are short of R&D, customers and logic. You are launching a product nearly after 1 year and still it is on par against competitions second best card.

DOOOOOOOOOOOOOOOOOOOOOOOOOOOMED

Apache14 · 9 Jan 2017 at 09:46

Ohh a picture of the demo card, as always take with a massive amount of salt from this site.

http://wccftech.com/amd-confirms-vega-10-doom-4k60fps-vulkan-demo-outperform-gtx-1080-opengl/

Doom112 said:
Some AMD ** Please do not start using that term - EVH ** are still in a denial that even on Vulkan VEGA is only on par with GTX 1080. It is really a matter of concern for AMD when VEGA performs in 95% of games ,which are mostly based on DX11.

This is exactly what happens when you are short of R&D, customers and logic. You are launching a product nearly after 1 year and still it is on par against competitions second best card.

Where are you getting this information from. . . So far we have see a hand full of demos that show vega running 4k ultra gaming either locked to 60fps (battlefront) or unlocked for the doom demo. This is beyond that of a stock 1080, it's more like a overcooked 1080

AMD do a lot of R&D not just on gfx cards, look at adaptive sync and GPUOpen. That being said they are the first to make use of HBM.

africanos23 · 9 Jan 2017 at 09:49

hyperseven · 9 Jan 2017 at 09:59

africanos23 said:

Aren't they stills from Linus's video?

AMD that strapped for cash they need to use duct tape to hold their hardware together?

Apache14 · 9 Jan 2017 at 10:02

hyperseven said:
Aren't they stills from Linus's video?

AMD that strapped for cash they need to use duct tape to hold their hardware together?

Hah i think it's more to hide how many PCI power connections are there

if they had three the Internet would take that as fact..... Regardless of if it was just a engineering card

Also good thing to note, it's looking like it's running on an air cooler unlike the fury x thank God :-D

fs123 · 9 Jan 2017 at 10:05

hyperseven said:
Aren't they stills from Linus's video?

AMD that strapped for cash they need to use duct tape to hold their hardware together?

I think it's to hide the power connectors. The amount of power can indicate the gpu power, etc since they are claiming 4x performance per watt.
If, for example, it only has 1x6pin connector then it maybe small Vega and is an amazing achievement or if it's 2x8pin then it's probably the big Vega and underwhelming.

EDIT:

Apparently it's 1x8pin according to someone on reddit. The PSU is a Thermaltake SMART M1200W.

http://i.imgur.com/i6jE5dj.jpg

Here's the post on Reddit:

https://www.reddit.com/r/Amd/comments/5mwspq/just_came_here_to_say_that_the_vega_sample_ces_is/

Erik13 · 9 Jan 2017 at 10:09

So the unfinished card with Fury drivers matchs or beats 1080 and that is being doomed ? Lol seriously if u hate amd so much why bother even come here ? ( unless u r getting paid by Nvidia ) Also even if Vega fails which i don't see happening ( sorry ) they still have Zen so stop that Amd is dead bs, if that ever happened it should have been happened before not now. This year is prolly Amds most competitive year ( considering both gpu and cpu ) in a long time...

r7slayer · 9 Jan 2017 at 10:16

Exactly a unfinished engineering sample on drivers which was never intended for that card is doomed. I mean vega has the most architectual changes since amd intoduced gcn. So there is a lot that could change when proper drivers are made. We dont even know if the cards are boosting bit like zen. They could just have base clocks for now as the boost clocks are still being finialised which means more performance.

Erik13 · 9 Jan 2017 at 10:31

Exactly arch changed dramatically so just because previous amd cards with old gcn arch performed good with Vulcan/Dx12 and worse with dx11 doesn't mean Vega will do the same. It can be on par or perform even better in dx11 titles. To summarize everything u don't no **** sorry except of doom and battlefront performance maybe lol

fs123 · 9 Jan 2017 at 10:34

Erik13 said:
So the unfinished card with Fury drivers matchs or beats 1080 and that is being doomed ? Lol seriously if u hate amd so much why bother even come here ? ( unless u r getting paid by Nvidia ) Also even if Vega fails which i don't see happening ( sorry ) they still have Zen so stop that Amd is dead bs, if that ever happened it should have been happened before not now. This year is prolly Amds most competitive year ( considering both gpu and cpu ) in a long time...

The Doom12 guy appears to be one of those blind people who will do and say anything to try and put down AMD.
He must be too slow in the head to realise that no competition against Nvidia will only push prices up like we can see already with 1080/1070 cards. The mid tier 1080 is priced the same or higher than the top end 980ti of last gen but some fanboys are ok with that.

nashathedog · 9 Jan 2017 at 10:37

fs123 said:
I think it's to hide the power connectors. The amount of power can indicate the gpu power, etc since they are claiming 4x performance per watt.
If, for example, it only has 1x6pin connector then it maybe small Vega and is an amazing achievement or if it's 2x8pin then it's probably the big Vega and underwhelming.

EDIT:

Apparently it's 1x8pin according to someone on reddit.

https://www.reddit.com/r/Amd/comments/5mwspq/just_came_here_to_say_that_the_vega_sample_ces_is/

My psu has an 8 pin and a 6 pin coming out of one psu cable and while it may not be a good idea to run cards like that it's possible they have done it for this show to keep people guessing, I'm sure a top quality psu could cope for a day or two.

Or couldn't it?
I'm only guessing and I'm no expert.

fs123 · 9 Jan 2017 at 10:44

nashathedog said:
My psu has an 8 pin and a 6 pin coming out of one psu cable and while it may not be a good idea to run cards like that it's possible they have done it for this show to keep people guessing, I'm sure a top quality psu could cope for a day or two.

Or couldn't it?
I'm only guessing and I'm no expert.

You are reading too much into it. The system would need the sata connector for the OS drive loaded with the game.

The card is supplied with 1x8pin (150W) + 75W pcie slot for a total of 225W which is similar to a 1080/TitanXP. We don't know how much it actually draws though.

ScoTTyBEEE · 9 Jan 2017 at 10:47

nashathedog said:
My psu has an 8 pin and a 6 pin coming out of one psu cable and while it may not be a good idea to run cards like that it's possible they have done it for this show to keep people guessing, I'm sure a top quality psu could cope for a day or two.

Or couldn't it?
I'm only guessing and I'm no expert.

That one port can deliver max 225w. Doesn't matter how many splitters you use.

Apache14 · 9 Jan 2017 at 10:47

fs123 said:
I think it's to hide the power connectors. The amount of power can indicate the gpu power, etc since they are claiming 4x performance per watt.
If, for example, it only has 1x6pin connector then it maybe small Vega and is an amazing achievement or if it's 2x8pin then it's probably the big Vega and underwhelming.

EDIT:

Apparently it's 1x8pin according to someone on reddit.

https://www.reddit.com/r/Amd/comments/5mwspq/just_came_here_to_say_that_the_vega_sample_ces_is/

Can't be 100% there could be a PCIE power splitter or a MOLEX adapter on the other accessories lead.

That being said IF it is 1 pcie power plug, its some amazing performance based on that theoretical maximum power draw and the DOOM and battlefront demo FPS. At least gtx 1080 performance in those to examples on 1 pcie power connector.... :-D

Freddie1980 · 9 Jan 2017 at 10:53

1x 8pin at 150 watts plus 75 watts of PCI bus power certainly gives credence to the 225 watts TDP that's been touted which puts is drawing 23% more power then a vanilla 1080. We don't know what card was being used but in the Doom demo using Vega 4k Doom benchmarks puts it 17% faster then a 1080. If that was the same card that's not bad we just need to know what it's going cost. Of course if AMD were using the sampler Vega chip the results went from good to very impressive.

Boomstick777 · 9 Jan 2017 at 11:17

VEGA looking really good, but like it's months more than weeks away. Guess we won't see a GTX 1080 Ti anytime soon either ...

KillBoY_UK · 9 Jan 2017 at 11:44

fs123 said:
I think it's to hide the power connectors. The amount of power can indicate the gpu power, etc since they are claiming 4x performance per watt.
If, for example, it only has 1x6pin connector then it maybe small Vega and is an amazing achievement or if it's 2x8pin then it's probably the big Vega and underwhelming.

EDIT:

Apparently it's 1x8pin according to someone on reddit.

https://www.reddit.com/r/Amd/comments/5mwspq/just_came_here_to_say_that_the_vega_sample_ces_is/

nice spot!

fs123 · 9 Jan 2017 at 11:56

Freddie1980 said:
1x 8pin at 150 watts plus 75 watts of PCI bus power certainly gives credence to the 225 watts TDP that's been touted which puts is drawing 23% more power then a vanilla 1080. We don't know what card was being used but in the Doom demo using Vega 4k Doom benchmarks puts it 17% faster then a 1080. If that was the same card that's not bad we just need to know what it's going cost. Of course if AMD were using the sampler Vega chip the results went from good to very impressive.

The 225W is what is available but unlikely that it will be drawing the full amount. If it did then I'd expect AMD to put an extra 6-pin on the board too since it is not wise to draw 100% from the pci-e+8pin.

The 1080 draws about 206W if the temperature is below target temp otherwise the power target is enforced to drop it to around 180W.

http://www.tomshardware.co.uk/nvidia-geforce-gtx-1080-pascal,review-33557-10.html

Panos · 9 Jan 2017 at 13:06

Few things related to how many pins the power connectors of the vets has... (from above pick)

A) gtx1080 starts downclocking from 22C, having power limit to 1.093v.
Same power draw and overclocking abilities exist on the ones with 2 connectors (8+6) or those that came with a single 8pin.

B) R9 Nano, has a single 8pin. Consuming a lot of more power than the 1080.
But even theoretically power limited compared to the furyx still can do 1140 core, as much as any twin plug FuryX.

C) Given that polaris was true to the AMD power claim, with the Vega having 2.5x efficiency compared to Fiji core, a single 8pin will be sufficient to deliver at least a card twice as fast than the FuryX if not more...
Or running at much less power than the 1080 for same perf.

And AMD only has to deliver a card +45% more powerful than FuryX to beat the 1080.

But I dont care about power efficiency, not should anyone else for high end gpu in an overclockers forum imho.

That's mentality for downclockers or greenclockers

Rroff · 9 Jan 2017 at 13:19

Panos said:
A) gtx1080 starts downclocking from 22C

AFAIK first downclock is potentially at ~40C. There are a couple of graphs that appear to show 22C but its a data logging error on start up.

(Also while below 40C can allow for higher stable overclocks that isn't part of GPU Boost 3.0 behaviour).

EDIT: My 1070 boost to 1911MHz out the box all the way from 10C actual upto 79C before it drops a speed bin due to temperature - though depending on power load, etc. some cards will see 1-2 drops typically at around 51C and 69C ish.