Poll: ** The AMD VEGA Thread **

[akarypid]

There is a danger here though that people equate market strategy to tech/R&D development progress. AMD managed to successfully catch Intel wrong footed in their product execution cycle but it doesn't mean - as some try to make out - that Intel has fallen behind and will be playing catch up on some of the tech that AMD have put into their CPUs, etc.

Oh I quite disagree on this. Intel won't have their new architecture ready before 2020 (the earliest). Even then, there's a risk their new architecture may not perform as well as they hope.

Right now, the fact that they have a process advantage saves them. The 10nm chips will give Intel a respite at the end of the year. But come 2019 they will be in trouble again with the next iteration of Zen. The 7nm GloFo process will be equal (expectations are that it will actually be marginally better) to Intel's 10nm, so no more process advantage. It is very likely the crown for highest clock will go to AMD and then there's still the power efficiency and the cost/scaling of the CCX approach. I expect Intel will bleed some more due to all this...

 

[old gamer]

Well I think we are going to end up with 1080ish performance and 1070 pricing, and tbh I will be happy with that, don't really want to spend any more on a graphics card

 

[akarypid]

Well I think we are going to end up with 1080ish performance and 1070 pricing, and tbh I will be happy with that, don't really want to spend any more on a graphics card

Let's just wait for proper drivers to see where it really lands, both performance-wise and consumption-wise.

I'm also interested to find out how the lower-core-count "salvaged" SKUs will perform. Something with 3072 shaders clocked at 1200MHz may be quite power-efficient (based on what we've seen so far from the 14nm process which seems to like low clocks a lot). Would be very interested to see how that stacks up against the 1070 and what it will cost.

 

[nashathedog]

I have a colleague going to SIGGRAPH and I am deeply envious. He probably won't even care about the GPUs

What a waste

Well I think we are going to end up with 1080ish performance and 1070 pricing, and tbh I will be happy with that, don't really want to spend any more on a graphics card

You get my vote for the guess I most want to see come true.

I'll be very happy if Vega can give us 1080 performance for the price the Fury cost.

 

[Devrij]

What a waste

I know, I'm going to ask him if he'll at least go check it out for me

 

[harney]

 

[Chuk_Chuk]

Companies like AMD/Intel/Nvidia have you sign NDA with your contract. And while you work for them if you leak something they basically drive you to be unemployed and homeless to the end of your days.
It's a lot of money there, and if AMD doesn't want you to know something, ofc they wont do so. Especially if they have the tech that slingshots them forward 4-5 years over the competition.

Just to add to this. The CPU/GPU market in terms of the number of companies is quite small and I think everyone knows everyone. So if word get round, you become very hard/impossible to employ with everyone knowing you broke an NDA.

Oh I quite disagree on this. Intel won't have their new architecture ready before 2020 (the earliest). Even then, there's a risk their new architecture may not perform as well as they hope.

Right now, the fact that they have a process advantage saves them. The 10nm chips will give Intel a respite at the end of the year. But come 2019 they will be in trouble again with the next iteration of Zen. The 7nm GloFo process will be equal (expectations are that it will actually be marginally better) to Intel's 10nm, so no more process advantage. It is very likely the crown for highest clock will go to AMD and then there's still the power efficiency and the cost/scaling of the CCX approach. I expect Intel will bleed some more due to all this...

I though 7nm was next year.

 

[silversurfer]

But is like trying to play games with a Quadro, without installing hacked Geforce drivers

how well does a Quadro play with those drivers ?

 

[Chuk_Chuk]

Potentially a base clock of 1630Mhz? With it being able to boost higher under the right conditions?

 

[five8five]

Potentially a base clock of 1630Mhz? With it being able to boost higher under the right conditions?

1630 will be the boost clock and possibly a base clock of 1412.

 

[melmac]

I know Adored TV gets a bad rap around here for been completely AMD biased, but that video seems pretty balanced and I agree with most of his conclusions. Vega will be somewhere between the 1080 and the 1080ti. It will probably get close to the 1080ti in games using Dx12 and Vulkan and close the gap in Dx11 over a period of time. I don't think it will take a year like Adored TV says, maybe a couple of months. AMD have been improving a lot lately in getting out performance drivers quicker.

Three things will decide if this is a success.

1. The price: Will they price it aggressively, based on it's performance, the fact that's it's coming late to the party and the current price of the 1080 cards, or will they just price it based on what it cost them to make with their margin on top? I don't know how much it costs to make a GPU but people keep saying HBM2 is expensive. Can they really sell the card at a price that will shake up the market? Of course the mining craze is going to muddy the waters as the 580 prices are really high at the moment.

2. Overclocking: It doesn't have to be an amazing overclocker, but, it would be nice if it saw even modest gains from overclocking.

3. Drivers: Have solid launch drivers.

I wonder will there be an AIO version as well as an air cooled version?

 

[Chuk_Chuk]

1630 will be the boost clock and possibly a base clock of 1412.

Videocardz stated that the benchmark doesn't properly detect overclocked unreleased cards. Isn't the boost clock, just an overclock of the base clock?

 

[nashathedog]

Three things will decide if this is a success.

1. The price: Will they price it aggressively, based on it's performance, the fact that's it's coming late to the party and the current price of the 1080 cards, or will they just price it based on what it cost them to make with their margin on top? I don't know how much it costs to make a GPU but people keep saying HBM2 is expensive. Can they really sell the card at a price that will shake up the market? Of course the mining craze is going to muddy the waters as the 580 prices are really high at the moment.

2. Overclocking: It doesn't have to be an amazing overclocker, but, it would be nice if it saw even modest gains from overclocking.

3. Drivers: Have solid launch drivers.

I wonder will there be an AIO version as well as an air cooled version?

1, The price is where it will win or lose,
2, I prefer to plug it in and play so I'll buy an OC model and the closer it is to the limit the better, I don't really care about overclocking it manually.
3, AMD's driver support was top notch for the year and a half I had with my Fury, Hopefully we'll get more of the same.

As for AIO's i missed out on a Fury X because they messed up on the quality control, I hope the same doesn't happen this time, If it doesn't I'll be having one.

 

[TNA]

I know Adored TV gets a bad rap around here for been completely AMD biased, but that video seems pretty balanced and I agree with most of his conclusions. Vega will be somewhere between the 1080 and the 1080ti. It will probably get close to the 1080ti in games using Dx12 and Vulkan and close the gap in Dx11 over a period of time. I don't think it will take a year like Adored TV says, maybe a couple of months. AMD have been improving a lot lately in getting out performance drivers quicker.

Three things will decide if this is a success.

1. The price: Will they price it aggressively, based on it's performance, the fact that's it's coming late to the party and the current price of the 1080 cards, or will they just price it based on what it cost them to make with their margin on top? I don't know how much it costs to make a GPU but people keep saying HBM2 is expensive. Can they really sell the card at a price that will shake up the market? Of course the mining craze is going to muddy the waters as the 580 prices are really high at the moment.

2. Overclocking: It doesn't have to be an amazing overclocker, but, it would be nice if it saw even modest gains from overclocking.

3. Drivers: Have solid launch drivers.

I wonder will there be an AIO version as well as an air cooled version?

+1

I would love to buy the AIO version as long as they do not price it silly.

 

[RavenXXX2]

 

[TNA]

So RX Vega will likely be very close to 1080Ti performance in this game. Maybe even beat it over time as drivers mature. But the real question is, how does it do on Nvidia friendly games.

 

[tekkas]

thanks for posting this i have subscribed what a great channel

seems like it will be 1080ti in dx12 vulkan they hope for however dx11 it may be behind the 1080ti which isnt bad tbf im looking forward to vega

 

[five8five]

Videocardz stated that the benchmark doesn't properly detect overclocked unreleased cards. Isn't the boost clock, just an overclock of the base clock?

You linked to the Adored vid where he mentions 1630 for the RX.

 

[five8five]

So RX Vega will likely be very close to 1080Ti performance in this game. Maybe even beat it over time as drivers mature. But the real question is, how does it do on Nvidia friendly games.

That's against a stock 1080 and Ti so I suppose it depends on how the RX overclocks.

 

[CAT-THE-FIFTH]

1630mhz means AMD have finally managed to have a stock GPU go past 1500mhz. That indicates there must some significant changes to the shaders as Polaris is based on the same technology in AMD Fiji.