• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Poll: ** The AMD VEGA Thread **

On or off the hype train?

  • (off) Train has derailed

    Votes: 207 39.2%
  • (on) Overcrowding, standing room only

    Votes: 100 18.9%
  • (never ever got on) Chinese escalator

    Votes: 221 41.9%

  • Total voters
    528
Status
Not open for further replies.
I've been saying for a long time in these forums that gaming-wise we will get the short end of the stick with Vega because AMD is NOT gaming-focused right now. Vega is part of the 'attack on compute front' strategy. It has the TFLOPs to serve that market, a nice programming model, a full suite of software tools in MIopen and most of all: it is absolutely on time! Yes! Once you realise that it was not being built to compete for gaming but for the data centre, you will understand why it was important that it gets launched together with EPYC. You will also understand why AMD doesn't care if the gaming drivers are ready as long a MIopen is ready! In that context, Vega FE is a card you buy to evaluate the tech at home. Deep learning developers (like all developers) want to be able to prototype and work on 'pet projects' out side the office and that's the card for them.

Yep, I agree with you on that, it seems to be the case, they are bringing out some disruptive products when it comes to the "real" professional stuff. On a business standpoint the route they are taking makes sense, that's where all the dough is.
I so wish I had have bought some shares back at the below 2$ days when the rumor of a purchase by Samsung was going around (and I had a decent amount of money back then to play around with as well), but I got talked out of it -_- . That being said if you can spare the money for a long period of time they are still worth buying today for the futur
 
Regardless if Vega is a hit or miss, who is looking forward to just an end of all this speculation and unknowns (until Volta or Navi)? I certainly am. A good 700 pages between these two threads, without about 1 page of facts :p
 
Regardless if Vega is a hit or miss, who is looking forward to just an end of all this speculation and unknowns (until Volta or Navi)? I certainly am. A good 700 pages between these two threads, without about 1 page of facts :p

Hey where's the wait for Navi thread by the way? Has nobody started it already?
 
Interesting


RIP AMD
I am an investor (I own stock).

I am happy that MI25 is ready and offered with EPYC as a bundle. As long as the compute power is there (and it is) I'm happy.

Would I be happier if they beat the 1080ti in gaming? Sure, but I always thought it was a long shot.

However, delaying any compute-oriented products just because because the gaming card is not ready would've been 10 times worse.


Ahh yeah, the MI25, could have just released that then to meet the H1 2017.
 
Ahh yeah, the MI25, could have just released that then to meet the H1 2017.

They could've, but I guess they also need a semi-pro card to put in the hands of developers that want to use MIOpen to target the MI25.

Even the WX9100 would be prohibitively expensive for the would-be developer who wants to get their hands dirty...
 
should wait for a game version reviews, nvidia quadros dont game partularly well either.

Mate why you wrote this? Don't you see the trash posts in this discussion about this subject? All the trolls came out. Just write nothing until the lady sings at the end of the month.

And then all of them will go back to their caves.
 
Did nobody actually notice the image i posted yesterday, it pretty much shows that the current Vega driver branch is months old, older than drivers seen at the end of march. The driver may have the latest game profiles from the main driver branch in it, but that doesn't mean the vega driver itself is current. [..]

Assuming the image is true, there would be a reason why AMD would ship Vega with drivers that are 3 months old when they had more recent drivers. Note that Vega FE is not using certified drivers, which would be expected to be older and give lower performance because they emphasise absolute reliability. It's unsubstantiated speculation that AMD have for some unknown reason shipped Vega with obsolete half-finished drivers that ignore key features of Vega despite having drivers that don't. Speculation that AMD have publically denied (an official statement was made saying that they had not gimped Vega FE and it was ridiculous a ridiculous claim).

Refinements in drivers are not likely to give a >20% increase in performance across the board while reducing power consumption by 40%, nor will they reduce production costs.
 
Enabling tile based rendering alone, based on the 780>980 performance jump *should* provide around 15-20% all on its own, while lowering power consumption at the same time.

That's assuming that AMD released Vega without bothering to enable tile based rendering at all. Why would they do that? Why would they then state that they have not gimped Vega FE, when deliberately disabling a key feature is exactly that?
 
Except it's based on assumptions not facts.

What evidence do you have that Vega RX will use a different and better GPU than Vega FE?
What evidence do you have that Vega RX will use different and better memory than Vega FE?
What evidence do you have that Vega RX will have more memory than Vega FE?
What evidence do you have that Vega RX will not be much cheaper than Vega FE?
What evidence do you have that Vega FE drivers are so seriously defective despite at least a year of development that huge gains are possible from drivers alone in a month?

If you think what I said is based on assumptions that are wrong, give a counter-argument.
 
Is the 1080 twice as fast as the 580? Because looking at the tflops I'm assuming it should be roughly twice as fast as a 580.

1080 is 8.2 and 580 is 6 I think where Frontier shows 13 FP32 and ti 11.3?

So the omly way it can be different is if it's cut down.

So rx vega will very similar price, say 800 + vat
very similar performance, say 1080
Its not making a lot of sense but would be funny to see. AMD thinks the two are distinct enough even sharing a chip

I've been saying for a long time in these forums that gaming-wise we will get the short end of the stick with Vega because AMD is NOT gaming-focused right now. Vega is part of the 'attack on compute front' strategy. It has the TFLOPs to serve that market, a nice programming model, a full suite of software tools in MIopen and most of all: it is absolutely on time! Yes! Once you realise that it was not being built to compete for gaming but for the data centre, you will understand why it was important that it gets launched together with EPYC. You will also understand why AMD doesn't care if the gaming drivers are ready as long a MIopen is ready! In that context, Vega FE is a card you buy to evaluate the tech at home. Deep learning developers (like all developers) want to be able to prototype and work on 'pet projects' out side the office and that's the card for them.

I think this guy gets the gold star and it mirrors the best of commentary Ive seen elsewhere. Rx Vega is just a small part of a bigger picture, patience padawan.
Rather then ignore the forum should have a highlight function
 
What evidence do you have that Vega RX will use a different and better GPU than Vega FE?
What evidence do you have that Vega RX will use different and better memory than Vega FE?
What evidence do you have that Vega RX will have more memory than Vega FE?
What evidence do you have that Vega RX will not be much cheaper than Vega FE?
What evidence do you have that Vega FE drivers are so seriously defective despite at least a year of development that huge gains are possible from drivers alone in a month?

If you think what I said is based on assumptions that are wrong, give a counter-argument.

I can't square the lack of FE performance in gaming with the claims from Koduri about bringing the cost of 4k gaming down to $1000 in 2017. If FE is the pinnacle for 2017 its all bluster as AMD won't have any card that can run 4k, never mind at an all in price of $1k.
 
That's assuming that AMD released Vega without bothering to enable tile based rendering at all. Why would they do that? Why would they then state that they have not gimped Vega FE, when deliberately disabling a key feature is exactly that?

PC Perspective ran a test showing clearly that it was no different from Fury or Polaris in it's rendering techniques.
 
Just to add this. Go back 100 or so pages and you would see a number of people moaning about how "late" Vega is. Whether releasing early was the right choice, I don't know. But i do know that people would have complained either way and this would have caused problems for them either way.
Mostly the moaning was about continous lack of any kind of information. Including any release schedule. Or expected performance.

You can notice that we compare FE doom performanc to benchmarks showed last year. Because there was nothing useful in between.
 
I'm wondering if they're not enabling due to the fear of getting sued the **** out of by nVidia!!
Looking at Wikipedia Nvidia don't own the concept, they also weren't the first to do it. https://en.wikipedia.org/wiki/Tiled_rendering

Mostly the moaning was about continous lack of any kind of information. Including any release schedule. Or expected performance.

You can notice that we compare FE doom performanc to benchmarks showed last year. Because there was nothing useful in between.

Which is even stupider. Who expects such information 3+ months in advance. I can't think of any industry that does this. I expect the volta thread is full of people bringing up these same concerns then.
 
Status
Not open for further replies.
Back
Top Bottom