• 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.
As I said we're all entitled to an opinion and I struggle to understand why someone who could pick any game would pick the same game as AMD have showed us a couple of times already, Doom on Vulcan may run well on both brands but it's clearly been running better on AMD hardware. If I was given the chance to do some testing with no conditions on the game I use it would have been with a DX11 title not more of the same. It doesn't sit right that a tech reviewer/journalist would pick Doom when all things are considered.


Maybe because its a twitch type shooter that smooth framerates play a big part in? They showed doom in January i think and since then its been battlefield 1. Had they been given more time they possibly could have tested other games but the card was only there a matter of hours.
 
i heard they produce like 1 per wafer

I reckon they must be running off only a small number of the big die alongside other products or something - by my estimates depending on layout and other challenges you can fit around 60 V100 dies on a wafer and even at that size yields shouldn't be as low as 1.

EDIT: Some really rough maths based on what I can find out from other companies would put that about 8-10 working dies per wafer mind which isn't great. I can see how they might manage only 1 in early sampling though.
 
Last edited:
I reckon they must be running off only a small number of the big die alongside other products or something - by my estimates depending on layout and other challenges you can fit around 60 V100 dies on a wafer and even at that size yields shouldn't be as low as 1.

EDIT: Some really rough maths based on what I can find out from other companies would put that about 8-10 working dies per wafer mind which isn't great. I can see how they might manage only 1 in early sampling though.


A lot of people have not understood what Nvidia said about yields. The reality is something along the lines that if they only get 1 or 2 working per a wafer then it wouldn't be economically feasible, but they get much more (maybe 6-8), this breaks even but gains them the market share. Then as yields improve they might be getting 20 etc per wafer. At that point, they rake in a load of money selling them at 12K a pop. Nvidia's point was that since they are sold at such a premium they can afford terrible yields, and obviously yields are not great with such a chip, but it just doesn't matter.

Also the 12nm process is just an optimization of the 16nm, so it should be relatively mature anyway
 
You guys really believe it's going to cost over £600 for big Vega? I just can't see it tbh

In reality I doubt it. However, like the Fury X AMD have made themselves a noose by using a big die with HBM2. They need to charge a certain price for it to recoup the costs of the materials, and I fear like Fury X it's simply going to be far too much. Especially as it will be a hot pig and chow down on voltage. And probably throttle too.
 
A lot of people have not understood what Nvidia said about yields. The reality is something along the lines that if they only get 1 or 2 working per a wafer then it wouldn't be economically feasible, but they get much more (maybe 6-8), this breaks even but gains them the market share. Then as yields improve they might be getting 20 etc per wafer. At that point, they rake in a load of money selling them at 12K a pop. Nvidia's point was that since they are sold at such a premium they can afford terrible yields, and obviously yields are not great with such a chip, but it just doesn't matter.

Also the 12nm process is just an optimization of the 16nm, so it should be relatively mature anyway

Won't lie I wasn't really paying attention when he was talking about it - was kind of tired at the time being in a different timezone.
 
I know Lisa Su said she no longer wanted AMD to look like the Primark brand but charging Armani price would be madness :D
Their RED rebellion marketing against obscene nvidia prices just doesn't fit. AMD is a mess. :D If the RX Vega costs more than the gtx 1080 and performs worse I seriously think raja should be fired.
 
I know Lisa Su said she no longer wanted AMD to look like the Primark brand but charging Armani price would be madness :D

Yeah I have no problem AMD trying to increase their margins (although as I consumer I'd rather they didn't of course), but the problem is if you want to charge Armani prices you need to have Armani performance.
 
Their RED rebellion marketing against obscene nvidia prices just doesn't fit. AMD is a mess. :D If the RX Vega costs more than the gtx 1080 and performs worse I seriously think raja should be fired.

I guess it's better to sell 1000 at profit than 100000 at loss, considering the costs are sunk either way.
 
Status
Not open for further replies.
Back
Top Bottom