AMD VEGA confirmed for 2017 H1

[NeoWave]

I suspect the lack of info is b/c Vega is 'delayed' from AMD's initial planned release date due to Ryzen & issues with HBM2 supply but yeah; endless speculation can help promote it for free. It has to launch in June for AMD not to have major egg on their faces & if not they'll have to say why it isn't launching; so official news is coming and likely this or next week in the form of invites to journos.

 

[Sargatanas2511]

I suspect the lack of info is b/c Vega is 'delayed' from AMD's initial planned release date due to Ryzen & issues with HBM2 supply but yeah; endless speculation can help promote it for free. It has to launch in June for AMD not to have major egg on their faces & if not they'll have to say why it isn't launching; so official news is coming and likely this or next week in the form of invites to journos.

There was little info on Ryzen, similar to what we have now for Vega and that wasn't delayed at all. We have little information on Naples but that isn't delayed to the best of anyone's knowledge. Just because they're sticking to a plan of secrecy doesn't mean anything is wrong.

They're still perfectly on track for the release schedule that they mentioned about a year ago.

 

[Wrinkly]

if that's the pricing then vega is a fail, not only they have no SKU close to 300$ ballpark, Nvdia already sold truckloads of 1070/1080, with low-mid end volta few months away.
all nvidia needs to do is to cut the price by 50$ for the 1070 and 1080, this will automatically renders vega useless, even if they decide to follow up on a price war and even if vega is slightly faster than it's direct competition.
this pricing if really stupid especialy for the low end, AMD could be gearing up to another break even architecture, with no market share benefit at all, maybe even pushing more towards their competition.
low vega cannot top 350$.

You still have the cost of G-Sync versus Freesync. Nvidia still do not have the hardware to support Adaptive Sync.

 

[Journey]

I suspect the lack of info is b/c Vega is 'delayed' from AMD's initial planned release date due to Ryzen & issues with HBM2 supply but yeah; endless speculation can help promote it for free. It has to launch in June for AMD not to have major egg on their faces & if not they'll have to say why it isn't launching; so official news is coming and likely this or next week in the form of invites to journos.

How is it delayed if since last year they stated H1 '17, if they meant Q1 '17 they would have said that. It's like when some one says it'll take 2-4 weeks for something, and then saying it's been 2 weeks, now it's late. Also all of the invites have technically gone out, since anyone who is anyone, in the hardware world will be at Computex in Taiwan, which is where it will be unveiled.

 

[beany_bot]

How is it delayed if since last year they stated H1 '17, if they meant Q1 '17 they would have said that. It's like when some one says it'll take 2-4 weeks for something, and then saying it's been 2 weeks, now it's late. Also all of the invites have technically gone out, since anyone who is anyone, in the hardware world will be at Computex in Taiwan, which is where it will be unveiled.

Well... technically you could say it is delayed. Or equally not delayed.
Depends how you understand their weird vague graph. As they released Polaris in the middle of last year then graph 1 would be accurate. And yes, Vega is late. (From this graph, not subsequent announcements)

...or Polaris was late. And Vega is on time. Either way, what does it matter. I'm just filling time. Computex will reveal all.

OR

 

[djbully]

Will there be pre-orders Gibbo? I want to get on in the top dog ASAP!

Cheers

 

[Geck0]

Comedy gold, soon we shall see. It will be interesting no doubt.

 

[Simon Belmont]

What will Nvidia respond with if Vega is significantly faster then 1080Ti or TXp?

 

[Bazzards]

Price cuts I would imagine.

 

[TNA]

What will Nvidia respond with if Vega is significantly faster then 1080Ti and TXp?

Nothing probably, just price cuts. They will just try and get Volta out asap. Not like Nvidia have always had the best graphics card on the market. Even if they are 3-6 months behind by a little then crush Vega with Volta, it won't make a huge difference to them.

 

[Apache14]

I don't know how people around the web can be so sure Vega will not be competing with the 1080ti / Titan Xp.


In my mind there are two ways to look at Vega (with the information from the latest leaks e.g a 64CU / 4096 Core card clocking in at ~1500mhz - 1600mhz) without taking into account any gains that have been made on how efficient the architecture is.

1) Basing on a Fury X at 1500mhz
2) Basing on a 480 with 4096 cores

I understand both of the above are not really scientific as the Fury X had issues with allocating tasks to the cores it did have and the 480's Polaris arch will not scale 100% with more cores. But then again we know that AMD has shown of some upgrades to Vega on paper.

By using the results from a 1080ti results here http://www.guru3d.com/articles_pages/msi_gtx_1080_ti_gaming_x_review,29.html we see that the MSI Gaming X 1080ti gets around 23,000 in firestrike on defaults.

Using the Fury X results (14374) and its default clock of 1050mhz we can guesstimate that a 1500mhz Fury X would hit around 20500 (right between the 1080 and the 1080ti) (calculated on scaling the freq (result ÷ base_freq) * target_freq -- (14374 ÷ 1050)×1500)

Now using the RX 480 (2304 core)@ 1266mhz boost results of (11298) and guesstimate a 4096 core 480 @ 1266mhz would hit around 20100 again between the 1080 and 1080ti (calculated on scaling the cores (result ÷ base_cores) * target_cores -- (11298 ÷ 2304)×4096)


Now this is very non scientific but its about as good as what anyone else has argued against the Vega performing near the 1080ti or even the 1080. But from simple scaling we can guess that even if all of AMD's efforts to upgrade efficiency of the new cards fail (tiled rendering, better task allocation ect,) and we only get the performance per clock or core as the Fury X or RX 480 it should still bet somewhere between the 1080 and the 1080ti..... Or at least at the movement there is as much of an argument for that than Vega being slower than the 1080.

 

[JediFragger]

1) Basing on a Fury X at 1500mhz

Don't forget that Vega has a longer pipeline than Fury (in order to reach higher clockspeeds) so this may affect performance somewhat (I'm guessing they've done it to improve efficiency due to said clocks).

 

[TNA]

I don't know how people around the web can be so sure Vega will not be competing with the 1080ti / Titan Xp.


In my mind there are two ways to look at Vega (with the information from the latest leaks e.g a 64CU / 4096 Core card clocking in at ~1500mhz - 1600mhz) without taking into account any gains that have been made on how efficient the architecture is.

1) Basing on a Fury X at 1500mhz
2) Basing on a 480 with 4096 cores

I understand both of the above are not really scientific as the Fury X had issues with allocating tasks to the cores it did have and the 480's Polaris arch will not scale 100% with more cores. But then again we know that AMD has shown of some upgrades to Vega on paper.

By using the results from a 1080ti results here http://www.guru3d.com/articles_pages/msi_gtx_1080_ti_gaming_x_review,29.html we see that the MSI Gaming X 1080ti gets around 23,000 in firestrike on defaults.

Using the Fury X results (14374) and its default clock of 1050mhz we can guesstimate that a 1500mhz Fury X would hit around 20500 (right between the 1080 and the 1080ti) (calculated on scaling the freq (result ÷ base_freq) * target_freq -- (14374 ÷ 1050)×1500)

Now using the RX 480 (2304 core)@ 1266mhz boost results of (11298) and guesstimate a 4096 core 480 @ 1266mhz would hit around 20100 again between the 1080 and 1080ti (calculated on scaling the cores (result ÷ base_cores) * target_cores -- (11298 ÷ 2304)×4096)


Now this is very non scientific but its about as good as what anyone else has argued against the Vega performing near the 1080ti or even the 1080. But from simple scaling we can guess that even if all of AMD's efforts to upgrade efficiency of the new cards fail (tiled rendering, better task allocation ect,) and we only get the performance per clock or core as the Fury X or RX 480 it should still bet somewhere between the 1080 and the 1080ti..... Or at least at the movement there is as much of an argument for that than Vega being slower than the 1080.

Yeah. This is why I have always expected 1080ti performance all the way back from when this thread was created. But people were like it will be 1070 performance, when asked why, they sight R&D and because you know, AMD... lol

 

[Apache14]

Don't forget that Vega has a longer pipeline than Fury (in order to reach higher clock speeds) so this may affect performance somewhat (I'm guessing they've done it to improve efficiency due to said clocks).

Yep but as we don't know the affects of any of the architecture changes in Vega all we can do is scale current chips we know the results for. I'm in no way saying that i think BIG Vega will be between the 1080 and 1080ti, but just that using the only metrics we have no one can rule out 1080ti or better performance

Also note my 480 with 4098 cores was at 1266mhz, we scale that up again to the 1500+mhx thats been rumored for Vega we would get something thats just a bit ~1000 points faster than the 1080ti in firestrike

 

[TonyTurbo78]

In 2 weeks we find out they have changed Vegas name to "CoreRipper", will urinate all over Nvidias flagships and will be priced at 50% less than a 1080ti.
..
..
..
..
Now wouldn't that be something Hey, we gotta dream

 

[Apache14]

Also what some people forget is that GPU revisions normally go in pairs, node change (28nm -> 14nm) and then refinement on the node

So for the last generations of nvidia, Maxwell was refinement (increases performance per clock / core) then Pascal was a node change with very little per clock gains (the main gain was from the higher clocking of the new process) and Volta again will be a refinement where efficiency is gained / refined

Now remember this isn't AMD's node change (that was polaris, and they did gain a little when comparing to the 290/390 series) Vega is AMD's 2nd generation on the new process so would be strange to see no gains in terms of performance.

 

[benjii]

The AMD Financial Analyst meeting is today at 9pm. It'll be interesting to see their projected profits over the course of this financial year. We may well get an updated road map too.

 

[RBImGuy]

New King of Gpu is coming, a star is born called Vega

 

[Rroff]

Well... technically you could say it is delayed. Or equally not delayed.
Depends how you understand their weird vague graph. As they released Polaris in the middle of last year then graph 1 would be accurate. And yes, Vega is late. (From this graph, not subsequent announcements)

...or Polaris was late. And Vega is on time. Either way, what does it matter. I'm just filling time. Computex will reveal all.

Can't really use roadmap graphics like that - they rarely can be broken down to correlate exactly with dates.

 

[TNA]

New King of Gpu is coming, a star is born called Vega

Flopper, is that you?