• 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.
No, based on the video where the price was mentioned.

But if it turns out to be significantly faster than a 1080 why is it too expensive at $599? Bearing in mind there's not really any evidence the price is actually correct, but even if it is surely it's performance will determine whether it's too expensive or not?
 
They can only charge $599 if it matches the 1080 Ti.

And by all reasonable signs, it won't.

1080 Ti performance would require something like 1700 MHz rock stable clocks, and ~25% performance increase from drivers.
 
They can only charge $599 if it matches the 1080 Ti.

And by all reasonable signs, it won't.

1080 Ti performance would require something like 1700 MHz rock stable clocks, and ~25% performance increase from drivers.

I would agree with you, but there's clearly something amiss with the FE, and we have nothing except a couple of people talking about the prices. I don't think it's exactly worth saying that it's too expensive already being that we still know not a great deal.
 
Are we in the 1950's?
steps to launch a graphic card in the 21st century:
1. Send samples to reviewers in all major territories under NDA [day 0]
2. Jointly make a press release regarding availability, one youtube video with reviewer benchmarks [day 3]
3. Start shipping [day 4]

But then there are no stocks... hoping that reference cards will be available on the shelf by 31 July with pre-orders starting tomorrow :)
 
But if it turns out to be significantly faster than a 1080 why is it too expensive at $599? Bearing in mind there's not really any evidence the price is actually correct, but even if it is surely it's performance will determine whether it's too expensive or not?
As I said it was based on the price and performance in the video.

Personally, based on the FE and the various leaks I think the RX performance will similar to the faster custom 1080s, maybe slightly faster in DX12 and slower in DX11 so the good custom 1080Tis will be significantly faster.

If it's released at $599 / £599 and there are shortages it could end being £650 which is the current price of a reference 1080Ti and as I previously said the 1080 / 1080Ti prices will lowered according to the RX price.

It's a difficult sell for AMD. They've stated numerous times they're not a budget brand now but the performance has to be there to charge the top prices.
 
Vega needs to be way cheaper than $599, as that was how much that performance was over a year ago, prices go down, not up over time, it will need to match, or be lower than the 1080s new price.
 
Vega needs to be way cheaper than $599, as that was how much that performance was over a year ago, prices go down, not up over time, it will need to match, or be lower than the 1080s new price.

If Vega is $599 then it's going to be faster than we think it is imo. As you say gtx1080 performance is not enough at this stage in the game especially at $599. Any how this is all just based on a video so nothing confirmed.
 
Are we in the 1950's?
steps to launch a graphic card in the 21st century:
1. Send samples to reviewers in all major territories under NDA [day 0]
2. Jointly make a press release regarding availability, one youtube video with reviewer benchmarks [day 3]
3. Start shipping [day 4]

But then there are no stocks... hoping that reference cards will be available on the shelf by 31 July with pre-orders starting tomorrow :)
The launch date is confirmed as 30th July. The release date isn't confirmed yet, but should be stated at launch or maybe leaked prior to launch.
 
Last time I ended up with the cut down Fury pro because AMD messed up and the water pumps on the Fury X, They were whistlers, 3 to 4 weeks later I gave up waiting for fixed ones and bought a cut down Fiji card. I want the full chip this time.

Fair point i guess.
 
They can only charge $599 if it matches the 1080 Ti.

And by all reasonable signs, it won't.

1080 Ti performance would require something like 1700 MHz rock stable clocks, and ~25% performance increase from drivers.

IMO judging by what they said during the financial analysis day concerning price to performance ratio (and we can ignore current prices and assume they were talking about MSRP prices coz this was before the mining craze that bought up prices), there are only 2 ways it can go, either they were aiming for 1080 performance priced as a 1070, or aiming for Ti performance at 1080 price, now considering the size of the die (even though we haven't got an exact size yet) which is bigger than GP102 no matter the exact dimensions, HBM2 being more expensive, and a beefy power delivery system being necessary (and they uses quality components on their vanilla cards), it seems highly unlikely that it will be priced at 1070 MSRP unless they have managed to use some magic pixie dust in order to reduce costs of a monolithic die (highly unlikely, and to my knowledge unheard of on a brand new die), so I can only deduce that they were at that time aiming for Ti performance while being priced as a 1080.

IMO for the RX vega to be priced the same as a 1070 it needs to have either a modular die design which would highly surprise me even though Vega is infinity fabric capable, or a smaller die with less transistors, but nothing seems to indicate that except that there seems to be a ******** of transistors on the Vega FE doing **** all at the moment when it comes to gaming (or only being useful for productive loads) and using a hell of a lot of power to power nothing lol!
 
Last edited:

that's just the pcie cables, dont forget the 75w from the pcie slot, buildzoid got 400w total from lower voltage in his testing.

so basically AIB cards could he pushing 500w with a hefty overclock....kinda makes the price even worse if you need to spend £100+ on a pay upgrade, you could SLI 1080tis and have the same power draw.
 
Which is ironic considering how much the HBCC has been advertised as a major improvement. I wonder if something's not working correctly there of if it's HBM2 related, or if something's just disabled (or enabled which would not be in the gaming card)...

You could reasonably hypothesize that a lot of the hardware in a sophisticated controller such as HBCC isn't required for a consumer graphics product and could actually be an obstacle in routing data. I think its plausible it could hinder performance in graphics workloads as a unnecessary or more complex intermediary.

Just throwing it out there; the move to making the shaders and rasterisers clients of the L2 cache could be solely about increasing data locality to mitigate added latency from HBCC, and not about implementing an Nvidia-like tile approach to rasterisation? It's essentially just holding more data closer to execution engines using even bigger caches, rather than taking it a step farther and implementing a more efficient way to use the L2 to pursue additional benefits from reduced memory bandwidth use and an increased ROP throughput. Nvidia's TBR strives to keep data on chip to actively minimize costly (and avoidable) memory operations and data movement. No reason to think AMD took it that far yet.
 
Status
Not open for further replies.
Back
Top Bottom