[Videocardz rumour] Vega pushed forward to October

[MrTeal]

Take it with a grain of sale given the source, but if true that sure might make a Christmas build much more interesting.

http://videocardz.com/59808/amd-vega-gpu-allegedly-pushed-forward-to-october

 

[MrTeal]

It should be noted that the credibility of the original (German) 3Dcenter source is more than a little suspect, as they seem to place Vega 11 higher in the product stack than Vega 10.

 

[railven]

Quick, let's fit as many narrative pushing theories into the first few posts!

GO!

[Before the mods infract me I was kidding.]

Anyways, it does throw a wrench at NV's party. Question is, does the wrench hurt or it is it a "hey guys, we still exist" after everyone (journalist) started putting up the NV banners.

 

[Paul98]

Wouldn't surprise me, I was expecting around this time for it all along

 

[Mopetar]

Seems very optimistic given previous information. Assuming it were true it probably means a paper launch and an extra $100 for the privilege of purchasing a Founder^H^H^H^H^H^H^H reference card.

Something big would had to have happened since the Capsaicin event in March for the timeline to shift that much.

 

[n0x1ous]

an extra $100 for the privilege of purchasing a Founder^H^H^H^H^H^H^H reference card.

At least AMD liquid cools their reference cards adding SOME extra value!!!

 

[railven]

Seems very optimistic given previous information. Assuming it were true it probably means a paper launch and an extra $100 for the privilege of purchasing a Founder^H^H^H^H^H^H^H reference card.

Something big would had to have happened since the Capsaicin event in March for the timeline to shift that much.

AMD isn't that tasteless. They'd call it the XT PE version. DO IT AMD! Bring back suffixes that ran longer than the GPU name!!!

X800 XT PE BABY!!!!

EDIT:

At least AMD liquid cools their reference cards adding SOME extra value!!!

Somethine tells me this trend won't continue. Not if the node helps reign in their heat issue.

 

[piesquared]

After testing they must have found that first silicon came back in very good shape. Probably yielding very good also.

 

[lilltesaito]

Wouldn't surprise me, I was expecting around this time for it all along

I was thinking/hoping it would be sometime this year. Maybe they are just waiting on HBM2 production, I dont think that Vegas is much different over Polaris other than the memory.

 

[piesquared]

AMD isn't that tasteless. They'd call it the XT PE version. DO IT AMD! Bring back suffixes that ran longer than the GPU name!!!

X800 XT PE BABY!!!!

Anything would be better than that geforce FU edition.

 

[linkgoron]

I think that people are desperate for some competition. I think that people have given up on Polaris, and are now moving to the next thing - Vega.

An October "launch" with an actual November release? Maybe.

 

[MrTeal]

Seems very optimistic given previous information. Assuming it were true it probably means a paper launch and an extra $100 for the privilege of purchasing a Founder^H^H^H^H^H^H^H reference card.

Something big would had to have happened since the Capsaicin event in March for the timeline to shift that much.

I'm still expecting an early 2017 launch, especially for the larger die. October seems extremely early, but hope springs eternal.

AMD isn't that tasteless. They'd call it the XT PE version. DO IT AMD! Bring back suffixes that ran longer than the GPU name!!!

X800 XT PE BABY!!!!

Knowing AMD, they'd launch the PE with a platinum cooler to one up nVidia's FE vapor chamber. Given Pt's thermal conductivity being 18% of that of copper the fan will run at 15kRPM, but that seems a sacrifice they're willing to make.

 

[maddie]

It should be noted that the credibility of the original (German) 3Dcenter source is more than a little suspect, as they seem to place Vega 11 higher in the product stack than Vega 10.

When Koduri explained the Polaris 10 & 11 names he said it was the order in which the designs started and that any additional ones would simply be a higher #.

I mentioned a few times before that this allows the Vega chips to be the opposite of Polaris. Namely Vega 10 could be the smaller one if the design work for it preceded the larger one. We can't assume one way or the other without direct confirmation.

 

[piesquared]

I think that people are desperate for some competition. I think that people have given up on Polaris, and are now moving to the next thing - Vega.

An October "launch" with an actual November release? Maybe.

I don't think so. I think people are considering it a forgone conclusion that Polaris is going to rule the mainstream market. There's about 1% left will be looking for the fastest at any cost.

 

[MrTeal]

When Koduri explained the Polaris 10 & 11 names he said it was the order in which the designs started and that any additional ones would simply be a higher #.

I mentioned a few times before that this allows the Vega chips to be the opposite of Polaris. Namely Vega 10 could be the smaller one if the design work for it preceded the larger one. We can't assume one way or the other without direct confirmation.

Really? I'm hesitant to call that retarded given the societal trend to move away from that epitaph, but c'mon man.

 

[maddie]

AMD isn't that tasteless. They'd call it the XT PE version. DO IT AMD! Bring back suffixes that ran longer than the GPU name!!!

X800 XT PE BABY!!!!

EDIT:



Somethine tells me this trend won't continue. Not if the node helps reign in their heat issue.

A bit mystified by this statement.

Watts are watts. The new node allows more performance for a given power consumption. Does not mean you won't hit a given wattage level, just that you have much more performance if you do.

All power ends up as heat. Basic thermodynamics seeing that no potential energy is generated.

 

[maddie]

Really? I'm hesitant to call that retarded given the societal trend to move away from that epitaph, but c'mon man.

???????? Did you quote the right post? Please explain.

 

[linkgoron]

I don't think so. I think people are considering it a forgone conclusion that Polaris is going to rule the mainstream market. There's about 1% left will be looking for the fastest at any cost.

Oh, I'm not talking about it beating the 1070, but about beating Grenada handily, and at least getting somewhat close to fiji/980ti (stock).

 

[Ancalagon44]

Not impossible, but I consider it unlikely. If they move it forward, it would only be because their original estimates for launch were too conservative, and things are moving faster than they anticipated. It's unlikely to be a response to Nvidia, because there is only so much you can do to make a product ship faster. Especially 3 months faster, which is quite a lot.

The thing is, if AMD feels a bit of competitive pressure from GP104, there are other ways that they can respond. They can add game bundles to Polaris, and/or tweak the price of Polaris. Even if GP104 outguns them, doesn't necessarily mean they have to launch VEGA right that second.

That being said, the danger is that a lot of people have postponed GPU purchases until the 14nm generation. If Nvidia is the only one with an enthusiast GPU and AMD only has Polaris 10, a lot of customers will buy Pascal rather than wait for Vega. And once they have bought Pascal, those are lost sales to AMD.

 

[poofyhairguy]

Anything would be better than that geforce FU edition.

A GeForce Fury would be funny.

 

[MrTeal]

???????? Did you quote the right post? Please explain.

AMD's naming scheme. It's a preposterous system to use to describe your products, even internally. It'd be like Chevy calling the Volt "Coupe 10", the Corvette "Coupe 11", and the Spark "Coupe 12".

 

[maddie]

AMD's naming scheme. It's a preposterous system to use to describe your products, even internally. It'd be like Chevy calling the Volt "Coupe 10", the Corvette "Coupe 11", and the Spark "Coupe 12".

I don't agree. Within the specific design parameters, why not. Polaris is low/main GDDR memory. Vega is perf/enthusiast HBM2 memory.

I really don't see why they have to be in performance order. Polaris itself is a reversed order and no one seems to be complaining. If we try to be overly ordered, the next Polaris, if it came into existence would be lower that Polaris 11 in performance to keep the sequence.

Corvette, Volt and Spark are different categories and markets, so can't be used for an analogy.

 

[MrTeal]

I don't agree. Within the specific design parameters, why not. Polaris is low/main GDDR memory. Vega is perf/enthusiast HBM2 memory.

I really don't see why they have to be in performance order. Polaris itself is a reversed order and no one seems to be complaining. If we try to be overly ordered, the next Polaris, if it came into existence would be lower that Polaris 11 in performance to keep the sequence.

Corvette, Volt and Spark are different categories and markets, so can't be used for an analogy.

It's not that they are in reverse performance order, that obviously works fine for nVidia and isn't counter-intuitive. What is extremely strange is that the order varies generation to generation based on the start date. It's bad enough when the order gets inverted, but it could be even sillier in generations where there are three or more dies where the code name conveys no real information. It would be the same as if AMD had run with SI10 for Tahiti, SI11 for Cape Verde, and SI12 for Pitcairn.

 

[piesquared]

Oh, I'm not talking about it beating the 1070, but about beating Grenada handily, and at least getting somewhat close to fiji/980ti (stock).

What will qualify it as a success, is winning in the price bracket it is targeting. It's pretty much a done deal as far as perf/watt goes too, and that sells a lot of product in the category it is going into.

 

[maddie]

It's not that they are in reverse performance order, that obviously works fine for nVidia and isn't counter-intuitive. What is extremely strange is that the order varies generation to generation based on the start date. It's bad enough when the order gets inverted, but it could be even sillier in generations where there are three or more dies where the code name conveys no real information. It would be the same as if AMD had run with SI10 for Tahiti, SI11 for Cape Verde, and SI12 for Pitcairn.

We'll have to disagree. I don't see any problem as to why should a name convey any other info except a unique call sign.