Radeon RX Vega Discussion Thread

well it looks like they got 5% more on clocks for this rebrand, so yep, not much change there.

respins aren't going to do miracles ;)

Something has to give, and its going to need more power, but at least its got the power delivery in the reference design to where it should be.
 
whateverer said:
Well, that jump to 8-pin screams factory-overvolted.

Guess AMD doesn't like being no faster than the factory-overclcoked GTX 1060 cards. We'll have to see how much higher power consumption this costs them.
Click to expand...

I don't think these are overvolted. Polaris at launch was over volted. I had one but now I have a 1080. Mine was overvolted like crazy. I could run 1350 at 1.10v. But factory was over 1.2v. I think they might have fixed the variation in voltage, you might actually get around 1400-1450 on these hopefully with similar voltage. I am actually keeping it on the low side. I think Polaris was not ready when it was released and it was more like a test subject on the new 14nm finfet.

This is like what polaris should have been a year ago when it was launched.
 
What's the prices? If AMD knocks down prices I could see them doing nicely. Rebrands or not, price/performance is the question.
 
razor1 said:
Interesting enough when talking about packed math they are showing tessellation demos, I think that is where their tessellation through put increases will come from.
Click to expand...
Do you mean TressFX demos? Because it is not a tesselation demo but a pure compute one.
 
razor1 said:
https://videocardz.com/67693/amd-radeon-rx-570-and-580-exposed

Not vega but rx 570 and 580 pics,

580 now has a 8 pin connector

AMD-Radeon-RX-570-Power-Connector-1000x668.jpg


rx 570 w/ cooler off

AMD-Radeon-RX-570-GPU-1000x537.jpg
Click to expand...

Power consumption increase in exchange of 50-60Mhz. At least the 580 is still within PCIe specs ;)
 
Reminds me very much of Hawaii. Slightly better chip/thermals and a slight clock bump.
 
lolfail9001 said:
Do you mean TressFX demos? Because it is not a tesselation demo but a pure compute one.
Click to expand...


if its pure compute that is definitely equal to the programmable geometry pipeline increases cause they mention in the beginning of the video geometry through put increases and they will talk about it later which the only thing they showed was the tressfx.

So by that it means Vega has very little front end changes over Polaris.
 
SighTurtle said:
What's the prices? If AMD knocks down prices I could see them doing nicely. Rebrands or not, price/performance is the question.
Click to expand...


How much can they knock down? They are already at 30% margins with 4xx line of pricing. I think the rx 5xx line is going to just replace the rx 4xx line and maintain prices.
 
razor1 said:
if its pure compute that is definitely equal to the programmable geometry pipeline increases cause they mention in the beginning of the video geometry through put increases and they will talk about it later which the only thing they showed was the tressfx.

So by that it means Vega has very little front end changes over Polaris.
Click to expand...
Fairly positive they meant "later" as in "not at this event but at another closer to launch". Because slide deck on arch teaser did mention something about having 11 polygons per geometry engine per clock against 4 of Fiji.
 
lolfail9001 said:
Fairly positive they meant "later" as in "not at this event but at another closer to launch". Because slide deck on arch teaser did mention something about having 11 polygons per geometry engine per clock against 4 of Fiji.
Click to expand...


just rewatched the presentation, he says you will see some examples later on trough out today.

They need to use RPM and the programmable geometry pipeline to get more than Fiji per clock, otherwise it seems like they are stuck with what they had before with Polaris's front end changes added to the frequency increases.
 
Last edited:
razor1 said:
How much can they knock down? They are already at 30% margins with 4xx line of pricing. I think the rx 5xx line is going to just replace the rx 4xx line and maintain prices.
Click to expand...

I dunno, with Nvidia lowering the prices of the 1070, 1080 with the ti, and putting out new models of the 1050 and 1060, i feel that puts more pressure on AMD's current lineup.
 
1070 hasn't been lowered to my knowledge, just the 1080, and that was to make space for the 1080ti.
 
About the 580: price matters,as freesync monitors are cheaper already. around that price poitn i find very hard recommend anything that is not a C24FG70 monitor + 480/580 GPU.

On the upper echelons however, there is no great Freesync monitor to pair Vega with. The new Agon 35 GSync offers a good combination of price, features and size, currently unmatched by their Freesync counterparts. Vega is becoming a GPU in search of a market niche: it must beat NVIDIA in value or performance, and the buyer must also pick a Monitor that beats Gsync offers.
 
SighTurtle said:
I dunno, with Nvidia lowering the prices of the 1070, 1080 with the ti, and putting out new models of the 1050 and 1060, i feel that puts more pressure on AMD's current lineup.
Click to expand...

Aren't the new models of the 1050 and 1060 just rebrands with a slight bump in bandwidth?
 
I do wonder what AMD will do with price. Will 570/580 be slightly more expensive than 400 series or will it be the same price but with a price drop on the 400s? The later seems like a better play.

I do like GSync/FreeSync but after getting a 40" 4K TV to use as a monitor, it's hard to go back to the smaller sizes. In a perfect world there would large FreeSync 4K TVs and (while I'm wishing) 144Hz and HDR. I guy can dream, right?
 
geok1ng said:
. Vega is becoming a GPU in search of a market niche: it must beat NVIDIA in value or performance, and the buyer must also pick a Monitor that beats Gsync offers.
Click to expand...

It needs to beat a 1080 and cost around 500. This isn't really new info. Nothing has changed.
 
Rizen said:
When is this GPU actually dropping? So tired of hearing about it and seeing nothing.
Click to expand...

It is surprising as to no leaks on it yet and there is bigger silence about it now than there has been in recently. Looks AMD is shutting up about it until launch.
 
Peppercorn said:
Aren't the new models of the 1050 and 1060 just rebrands with a slight bump in bandwidth?
Click to expand...
How can it be a rebrand if it is still call GTX 1060? Also, I believed the memory refresh only applies to GTX 1060 and 1080.

Edit, change 1050 to 1060.
 
Peppercorn said:
Aren't the new models of the 1050 and 1060 just rebrands with a slight bump in bandwidth?
Click to expand...

selling a different sku same card name, no rebrand, they aren't a different generation of the same old thing, the last time nV did that was before Fermi right?

Nothing wrong with rebrands as long as price goes down with them, but I don't think that is the case with the 5xx line, I think they will have the exact same price as the 4xx line.
 
Early May was the most reliable timeframe I've seen but they have until June 30th and still be compliant with their stated first half of 2017.
 
Chimpee said:
How can it be a rebrand if it is still call GTX 1060? Also, I believed the memory refresh only applies to GTX 1060 and 1080.

Edit, change 1050 to 1060.
Click to expand...

OMG don't tell me they are going to pull the same con job that they did with the Titan X????? Seriously? So more abuse from NV then, although not unexpected. Regurgitating card names so they can fool people into buying the lower specced card for the same or more money is probably illegal, but i'm sure NV includes in enough 'fine print' that allows their lawyers to WIN. There was a time that reviewers used to frown on such behavior, now they seem to embrace it.

It's because of shenanigans like that I wiill continue to refuse to purchase any Nvidiia product. That and the fact that they have known AMD's strategy ever since they have had access to those 100,000 documents that were stolen from AMD and the thief hired by NV.

I'll be handing my excellent 470 on down to my child and replacing it with Vega soon! :)
 
Peppercorn said:
OMG don't tell me they are going to pull the same con job that they did with the Titan X????? Seriously? So more abuse from NV then, although not unexpected. Regurgitating card names so they can fool people into buying the lower specced card for the same or more money is probably illegal, but i'm sure NV includes in enough 'fine print' that allows their lawyers to WIN. There was a time that reviewers used to frown on such behavior, now they seem to embrace it.

It's because of shenanigans like that I wiill continue to refuse to purchase any Nvidiia product. That and the fact that they have known AMD's strategy ever since they have had access to those 100,000 documents that were stolen from AMD and the thief hired by NV.

I'll be handing my excellent 470 on down to my child and replacing it with Vega soon! :)
Click to expand...


I think a person that has 1k to spend on graphics cards are well in a position to know where to spend their money on lol. They know what they are buying and that is why you don't see them unhappy about the TI's. Cause they know they will come out several months after the Titan versions and be close in performance, but those several months they have the best card on the market.

The only reason nV can charge so much is because there is NO COMPETITION from AMD. Out side of the Titan series, all other cards are priced well so don't see why you are complaining about it, because it sure looks like 1k for a graphics card for you is out of range for what ever reasons, maybe your wife will get mad lol, it happens. But for others it might just be a normal day for them.

Before you complain about nV and its pricing of their titan products, complain than AMD hasn't come to the plate to hit the ball, they haven't even got out of the dug out yet and its already a year.
 
Peppercorn said:
OMG don't tell me they are going to pull the same con job that they did with the Titan X????? Seriously? So more abuse from NV then, although not unexpected. Regurgitating card names so they can fool people into buying the lower specced card for the same or more money is probably illegal, but i'm sure NV includes in enough 'fine print' that allows their lawyers to WIN. There was a time that reviewers used to frown on such behavior, now they seem to embrace it.

It's because of shenanigans like that I wiill continue to refuse to purchase any Nvidiia product. That and the fact that they have known AMD's strategy ever since they have had access to those 100,000 documents that were stolen from AMD and the thief hired by NV.

I'll be handing my excellent 470 on down to my child and replacing it with Vega soon! :)
Click to expand...

Talk about deflecting, razor1 and I corrected you to say GTX 1060 memory refresh is not a rebrand and this is your response. Do you just go into a anti-nVidia fit when someone corrects you?7

May I ask what con-job are you referring in regards to Titan X? And are you complaining since the refresh is still called GTX 1060 or 1080, that it is illegal to do so? If that is the case, the it must be illegal to AMD to rename RX 480 to RX 580, since it fools people to believe that the 580 must be substantially better than the 480. /s

I get it, you don't like nVidia, but I am not even sure why you bring up a court case when there is still no ruling of said court case dated back January 2013.

And good for you, your child will most likely enjoy the RX 470.
 
Peppercorn said:
Whoooooosshhhh. First off, associating yourself with razor sort of kills your chances of being taken seriously. With that out of the way, the con job by nvidia was regurgitating the Titanx brand name over 2 generations, with zero difference in the branding. It's a con job. Now they are trying again with a rebranded 1060? Your shill partner thinks that if anyone buys a $1000 graphics card and gets TitanX instead of TitanX then it's their own fault. Let's assume that flawed and failed logic is in any way favorable to consumers, it doesn't hold for a 1060 vs a 1060. But like i said i'm sure their lawyers have enough fine print located somewhere to cover there asses.

As far as those 100,000 documents that were stolen by an ex AMD employee and then hired by NV are concerned, it's pretty clear how that would negatively impact AMD's competitive position until those documents have expired. In fact Rory Reed said almost exactly that, that it would hurt their competitive position for future products.
Click to expand...


Wow nice to see how you think in plain view, didn't need to, because the way you post it was obvious. You have been reported.

Now back to topic shell we?

All the stuff you mentioned is BS btw, that has nothing to do with what you talked about with Titan X, you just threw all the things you can think off out there to make a mess of this thread in the past three posts, so just leave or post respectfully back on the topic.

Vega is no where, the rx 5x0 are rebrands without price drops from what is rumored, interetsting you can't stick to the what seems to be right there in front of you.

Just like where is Vega? wasn't it supposed to have come out already?

You stated 2 weeks right? that was 3 weeks ago? Should have came out last week?

What is your recommendation? People should still wait and not get the TI? yeah gotta wait another 2 more months probably.
 
razor1 said:
Wow nice to see how you think in plain view, didn't need to, because the way you post it was obvious. You have been reported.

Now back to topic shell we?

All the stuff you mentioned is BS btw, that has nothing to do with what you talked about with Titan X, you just threw all the things you can think off out there to make a mess of this thread in the past three posts, so just leave or post respectfully back on the topic.

Vega is no where, the rx 5x0 are rebrands without price drops from what is rumored, interetsting you can't stick to the what seems to be right there in front of you.

Just like where is Vega? wasn't it supposed to have come out already?

You stated 2 weeks right? that was 3 weeks ago? Should have came out last week?

What is your recommendation? People should still wait and not get the TI? yeah gotta wait another 2 more months probably.
Click to expand...

There's a lot of "seems" and "maybes" and "probablys" and "rumors" in this post.

Yeah i tell people to wait a few weeks for Vega and possibly save some money. Seems like sound advice. Oh, i also forgot about the anti consumer 1060 3GB version that isn't even a 1060! Now if you could get back on topic that would be great.
 
I hope this doesn't turn out to be a deflated launch like the Fury was....I really hope not.
 
Whach said:
I hope this doesn't turn out to be a deflated launch like the Fury was....I really hope not.
Click to expand...

I think Vega will be quite successful regardless of the consumer version and gaming performance, with Radeon Pro. Only a few weeks until we find out.
 
Peppercorn said:
There's a lot of "seems" and "maybes" and "probablys" and "rumors" in this post.

Yeah i tell people to wait a few weeks for Vega and possibly save some money. Seems like sound advice. Oh, i also forgot about the anti consumer 1060 3GB version that isn't even a 1060! Now if you could get back on topic that would be great.
Click to expand...


Oh I see you want me to go back to topic after your flame post ok, here ya go then.

These are things from what AMD is saying and showing, so if anything take it to the worse and you will get reality.

So what AMD has shown so far, lets see, 1080 like performance maybe 10% faster, so lets say 1080 performance.

AMD showing x2 geometry through put performance when using primitive shaders, that goes with what they had in their word cloud and their presentation (max geometry throughput increases), they don't take about when not using primitive shaders so that speaks to no changes over Polaris, which means 4 tris per clock, that is 4 times less then the current Pascal offerings in the same performance bracket Vega should be in.

HBCC performance will be noticeable when hobbled to 2 gb in current games, that means any card above that with HBCC will not show that dramatic performance improvements most likely will show negligible performance increases (hence why they didn't try to show 4 gb).

AMD reps also stated tiled rendered performance/power consumption benefits they can't say exactly how much it will benefit for today's games, take that as in you won't see much.....

AMD vega power consumption looks to be at around 225-250 watts, so 1080ti power consumption at gtx 1080 performance? OK that is worse then Fiji, at least Fiji got to 980ti performance. So........

Clocks look to be at 1300mhz - 1500mhz boost, fits in with how they have done their boost clocks and tflops before.

AMD isn't even portraying this card to any degree you are trying to lol. Wonder why?
 
Last edited:
Peppercorn said:
I think Vega will be quite successful regardless of the consumer version and gaming performance, with Radeon Pro. Only a few weeks until we find out.
Click to expand...


How so, if they don't have software for their insight cards, they don't move into any DL supercomputers. HPC they have the same problems there as they have in DL computers, software is a big crutch for them.

So they are kinda left with gaming. If you are talking about streaming and gaming, yeah great they have one contract, but that is a different market, a market that nV already has been in for a couple of years, so its competition of liquid sky vs nV's geforce now, we don't know what kind of market uptake will happen during the next few years, but its safe to say, the hardware side isn't going to matter much, but nV has already built a consumer base with shield so that is in their favor for the time being. But down side is these things might go the way of the Butterfly grid for MMO's, the idea sounds great of streaming games, but we just don't know how the traditional gaming market will impact these at this point.

Lets see these are the same problems they had with Fiji, with polaris, worse yet they are much later than Fiji this time, as of right now 3 months later and that could be extended to 6 months later by the looks of it, Fiji was 9 months out. Vega already 12 months out and its still around the corner. Yep great they are executing better right? Not, this is by far the most delayed GPU in comparison to competitor's products, ever in the GPU industry,

There is only two ways you can look at this, since Insight versions of this GPU is already out. AMD doesn't care about gaming and they are not worried about being competitive, or something went wrong for a myriad of reasons, performance, financial, can be anything.

Don't think its first reason ;).

So the second reason, causes if we go through them

1) Financial

2) HBM 2 ramp ups

3) Performance

I think its a mixture of all three, all stemming from HBM 2. So its not AMD's fault directly, but its their fault by the direction they took, poor estimations on what HBM 2 will do for them and when it will ramp up. Does that sound like what happened with Fiji? Guess what Raja was the lead on Fiji too even though it wasn't his baby, he was running the show in the final 2 years of its development through production. Maybe it was too late for Vega to make any drastic changes once they found out HBM wasn't going to keep FIji competitive, and there HBM Hynix exclusivity contract screwed them over? Yeah looks that way. Otherwise they could have gone to Samsung right?
 
Last edited:
razor1 said:
Oh I see you want me to go back to topic after your flame post ok, here ya go then.

These are things from what AMD is saying and showing, so if anything take it to the worse and you will get reality.

So what AMD has shown so far, lets see, 1080 like performance maybe 10% faster, so lets say 1080 performance.

AMD showing x2 geometry through put performance when using primitive shaders, that goes with what they had in their word cloud and their presentation (max geometry throughput increases), they don't take about when not using primitive shaders so that speaks to no changes over Polaris, which means 4 tris per clock, that is 4 times less then the current Pascal offerings in the same performance bracket Vega should be in.

HBCC performance will be noticeable when hobbled to 2 gb in current games, that means any card above that with HBCC will not show that dramatic performance improvements most likely will show negligible performance increases (hence why they didn't try to show 4 gb).

AMD reps also stated tiled rendered performance/power consumption benefits they can't say exactly how much it will benefit for today's games, take that as in you won't see much.....

AMD vega power consumption looks to be at around 225-250 watts, so 1080ti power consumption at gtx 1080 performance? OK that is worse then Fiji, at least Fiji got to 980ti performance. So........

Clocks look to be at 1300mhz - 1500mhz boost, fits in with how they have done their boost clocks and tflops before.

AMD isn't even portraying this card to any degree you are trying to lol. Wonder why?
Click to expand...

Seems like you have it all figured out, we don't even need reviews! What's amusing is pulling out this line:
AMD isn't even portraying this card to any degree you are trying to lol. Wonder why?
Click to expand...
.
Now, everyone that has been following will know that if AMD were to say the opposite, so would you, and there is no doubt you would be criticizing for overhyping. Clearly you have it all nicely formulated into a no win situation. :)
 
  • Like
Reactions: Ronin
like this
jamesgalb said:
Back to my old-school predictions...

Vega 10
-Upgraded Fury/X (4096 & 3584 CU versions)
-8GB vram
-HBM2
-128 ROPs
-5-15% higher CU performance
-25% higher clocks
-20% less power


Vega 10 Small - $550
Vega 10 Large - $650

Clocks, CU performance, and power savings are calculated based on the performance and power improvements when comparing 28nm 380x to the 14nm 470...
Click to expand...


*cough*

Guess the clocks were more like 50% increases... :)
 
Peppercorn said:
Seems like you have it all figured out, we don't even need reviews! What's amusing is pulling out this line:
.
Now, everyone that has been following will know that if AMD were to say the opposite, so would you, and there is no doubt you would be criticizing for overhyping. Clearly you have it all nicely formulated into a no win situation. :)
Click to expand...


I am only going by what AMD has been saying or showing, and what has happened timing wise, so, its all based on real events, unlike what you have been saying. Whats the use of making things up, when AMD is showing and saying the best they can do?
 
razor1 said:
Oh I see you want me to go back to topic after your flame post ok, here ya go then.

These are things from what AMD is saying and showing, so if anything take it to the worse and you will get reality.

So what AMD has shown so far, lets see, 1080 like performance maybe 10% faster, so lets say 1080 performance.

AMD showing x2 geometry through put performance when using primitive shaders, that goes with what they had in their word cloud and their presentation (max geometry throughput increases), they don't take about when not using primitive shaders so that speaks to no changes over Polaris, which means 4 tris per clock, that is 4 times less then the current Pascal offerings in the same performance bracket Vega should be in.

HBCC performance will be noticeable when hobbled to 2 gb in current games, that means any card above that with HBCC will not show that dramatic performance improvements most likely will show negligible performance increases (hence why they didn't try to show 4 gb).

AMD reps also stated tiled rendered performance/power consumption benefits they can't say exactly how much it will benefit for today's games, take that as in you won't see much.....

AMD vega power consumption looks to be at around 225-250 watts, so 1080ti power consumption at gtx 1080 performance? OK that is worse then Fiji, at least Fiji got to 980ti performance. So........

Clocks look to be at 1300mhz - 1500mhz boost, fits in with how they have done their boost clocks and tflops before.

AMD isn't even portraying this card to any degree you are trying to lol. Wonder why?
Click to expand...

Until very recently leaks of two new Vega cards not shown by AMD came out, there is decent expectation given how mum and avoiding a certain person in AMD was about the subject. I do think AMD is hiding bigger Vega GPU's.

Other than that nobody really knows what Vega does, the Entry part looks like a 1080 type competitor, the others have not been seen.

I am somewhat interested in where Vega lies, not likely to be able to buy one anytime soon but I do like market players.
 
OrangeKhrush said:
Until very recently leaks of two new Vega cards not shown by AMD came out, there is decent expectation given how mum and avoiding a certain person in AMD was about the subject. I do think AMD is hiding bigger Vega GPU's.

Other than that nobody really knows what Vega does, the Entry part looks like a 1080 type competitor, the others have not been seen.

I am somewhat interested in where Vega lies, not likely to be able to buy one anytime soon but I do like market players.
Click to expand...

I know they aren't hiding anything, cause we know what the id board numbers are and what they mean.....

There board numbers haven't changed for years, and they definitely coincide with two types of Vega boards, and the one AMD showed is the top end one C1 which we saw come out mid last year right after Raja stated Vega's been tapped out.

So people that are guessing are going against what we have know for many years now and how AMD has structured their board numbers and ES.
 
Last edited:
razor1 said:
How so, if they don't have software for their insight cards, they don't move into any DL supercomputers. HPC they have the same problems there as they have in DL computers, software is a big crutch for them.

So they are kinda left with gaming. If you are talking about streaming and gaming, yeah great they have one contract, but that is a different market, a market that nV already has been in for a couple of years, so its competition of liquid sky vs nV's geforce now, we don't know what kind of market uptake will happen during the next few years, but its safe to say, the hardware side isn't going to matter much, but nV has already built a consumer base with shield so that is in their favor for the time being. But down side is these things might go the way of the Butterfly grid for MMO's, the idea sounds great of streaming games, but we just don't know how the traditional gaming market will impact these at this point.

Lets see these are the same problems they had with Fiji, with polaris, worse yet they are much later than Fiji this time, as of right now 3 months later and that could be extended to 6 months later by the looks of it, Fiji was 9 months out. Vega already 12 months out and its still around the corner. Yep great they are executing better right? Not, this is by far the most delayed GPU in comparison to competitor's products, ever in the GPU industry,

There is only two ways you can look at this, since Insight versions of this GPU is already out. AMD doesn't care about gaming and they are not worried about being competitive, or something went wrong for a myriad of reasons, performance, financial, can be anything.

Don't think its first reason ;).

So the second reason, causes if we go through them

1) Financial

2) HBM 2 ramp ups

3) Performance

I think its a mixture of all three, all stemming from HBM 2. So its not AMD's fault directly, but its their fault by the direction they took, poor estimations on what HBM 2 will do for them and when it will ramp up. Does that sound like what happened with Fiji? Guess what Raja was the lead on Fiji too even though it wasn't his baby, he was running the show in the final 2 years of its development through production. Maybe it was too late for Vega to make any drastic changes once they found out HBM wasn't going to keep FIji competitive, and there HBM Hynix exclusivity contract screwed them over? Yeah looks that way. Otherwise they could have gone to Samsung right?
Click to expand...

The Gefarce Now alternative to Liquid Sky is a disastrous business model. $2.50/hour?? Can't be serious. Liquid Sky has the customer base already with a much much better pricing structure, and Vega is already deployed there.

You missed the third reason for Vega launch window, which is the right answer: This timeline for product release has always been the plan. Unless you can show me a roadmap that says otherwise? No need to answer that rhetorical question, we both know you can't.
 
Peppercorn said:
The Gefarce Now alternative to Liquid Sky is a disastrous business model. $2.50/hour?? Can't be serious. Liquid Sky has the customer base already with a much much better pricing structure, and Vega is already deployed there.

You missed the third reason for Vega launch window, which is the right answer: This timeline for product release has always been the plan. Unless you can show me a roadmap that says otherwise? No need to answer that rhetorical question, we both know you can't.
Click to expand...


Again we don't know anything about the future of Liquid Sky or Geforce now and how they will impact gaming trends so to make any type of hype out of that for AMD, just doesn't work at the moment too early to say. They both can flop for that matter, just like the Butterfly MMO grid. Great on paper and inception doesn't mean great in the marketplace. The only farce is what you are trying to make out of something that is really an unknown.

I stated in comparative to competitor's products, I didn't say internal timelines lol, again don't try to put things into my mouth that I didn't say.

Fiji was the worst delayed product compared to competitor's products to date in the GPU marketplace, Vega just took that crown, not a good crown to have.
 
You must log in or register to reply here.
Back
Top