• 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.
Panos said:
Actually not.... A 1600mhz FuryX is faster than the GTX1080, let alone a brand new core design.
Click to expand...


If we take one of the new custom 1080s we can see the FuryX is about 67% of the performance at 1050MHZ. 1600MHZ is a 52.3% overclock, so 1.523X67 = 102%. The regular 1080 is at 93% for comparison. Therefore, best case scenario for a FuryX overclocked to 1600Mhz puts it at a very similar performance level, although performance would never ever scale linearly like that . To close the gap to the 1080Ti requires the architectural improvements.

This should all be possible, but does indicate that best case for Vega wont be absolutely humiliating the 1080ti, maybe 10-15% faster? that is good but does mean if you were to buy a 1080 or 1080Ti now you don;t had a lot to worry about. maybe you could wait a fe months to get a Vega 10-15% faster at similar prices, its hardly game changing.
 
silversurfer said:
What cost is likely on Frontier, is that known, how many digits even

Is the xbox details being released relatable or its all last gen despite being 'new'
The GPU photo was an actual enchanced take of the GPU die apparently. Its standard practise to overlay colours, etc on top of the real photo hence dimensions speculated on re HBM2 could be correct info
Click to expand...




The FE cards will be 4 digits prices. I would assume AMD would want to charge $2.5-3.5k USD, can;t see less than $1500 at the low end. The fastest Polaris pro radeon is $750
 
D.P. said:
If we take one of the new custom 1080s we can see the FuryX is about 67% of the performance at 1050MHZ. 1600MHZ is a 52.3% overclock, so 1.523X67 = 102%. The regular 1080 is at 93% for comparison. Therefore, best case scenario for a FuryX overclocked to 1600Mhz puts it at a very similar performance level, although performance would never ever scale linearly like that . To close the gap to the 1080Ti requires the architectural improvements.

This should all be possible, but does indicate that best case for Vega wont be absolutely humiliating the 1080ti, maybe 10-15% faster?
Click to expand...

For once, I agree with D.P. on something........
 
BriT said:
Come on the average guy wouldn't be using LN2 for cooling.
Click to expand...

It's not about the average guy. It's theoretical performance numbers confirmed with practical testing. We know the RX Vega to have clocks at 1600mhz, we know the performance of an older (usually implies less efficient) architecture at 1450mhz - so there are some guestimations you can make on the unknown performance of the Vega from that. Even if we assume no architectural improvements, we can expect close to Ti performance levels based on frequency alone, the unknown element is how much the IPC has improved / declined.

Strangely, I don't think DP is too far wrong with ~10% faster - but with AMD, who ******* knows. They don't have a consistent track record when it comes to performance.
 
Howling said:
https://www.pcper.com/news/Graphics-Cards/AMD-Radeon-R9-Fury-Unlocked-Fury-X-Overclocked-1-GHz-HBM

That's a Fury at 1450mhz and it's a little above 1080 Founders edition in Firestrike extreme. It's not much of a stretch to conclude that at 1600mhz it would definitely be more than a match for the custom 1080s.
Click to expand...

It is a stretch, as I linked to, a 1600MHz FuryX will be similar to a good 108 when you considering average benchmarks. Performance will scale sub-linearly with clocks, and that is if you are optimistic ignoring the bandwidth limitations.

What you failed to point out in that link is memory speed was double as well as the core clock increase. The reference FuryX has 512GB of memory, the best case scenario for vega is the same 512GB/s, but more realistic given recent HBM2 rumours would be in the 410-480GB/s range. As soon as memory bandwidth becomes a problem Vega is going to have to rely on architecture improvements.

What would be more useful is if a FuryX owner managed a small overclock to say 1200MHz and down-clocked the memory to get 480GB/s. This will give you a kind of base level scaling effect for the Fiji architecture. You will find that without the extra memory bandwidth FuryX at 1600MHz just wouldn't be able to keep up with the 1080, but wont be too far behind.

IMO: i would be surprised if the fastest Vega was worse than 10% slower than a 1080ti and surprised if its more than 10-15% faster than a stock ti. AMD and Nvidia are fond of saying stuff like 2.3x performance per watt ut when you actually benchmark a game you see 50-70% tops unless a game makes heavy use a of a feature that the old architecture sucked at. With the changes to the tessellation engine one might see Vega do best in games with lots of tessellation which includes some games-works effects
 
Last edited:
Panos said:
Actually not.... A 1600mhz FuryX is faster than the GTX1080, let alone a brand new core design.
Click to expand...

Don't forget about the architectural differences and the longer pipeline, AMD are hoping that the faster clockspeed will help with its efficiency to make up that shortfall (as Pascal does currently).
 
JediFragger said:
Don't forget about the architectural differences and the longer pipeline, AMD are hoping that the faster clockspeed will help with its efficiency to make up that shortfall (as Pascal does currently).
Click to expand...

They have already said that the IPC is higher, so the longer pipeline won't be detrimental to that. better IPC and much higher clocks should definitely put Vega comfortably past a 1080 and in my opinion probably a bit above a 1080Ti.
 
Spotted Vega cards, and engineering samples on Videocards.

https://videocardz.com/69760/engineering-sample-update
7VNSOqpaTTywR5y6p7skdg.png
 
Sargatanas2511 said:
They have already said that the IPC is higher, so the longer pipeline won't be detrimental to that. better IPC and much higher clocks should definitely put Vega comfortably past a 1080 and in my opinion probably a bit above a 1080Ti.
Click to expand...

If Vega is equal to (or bit better than) a Ti in Vulkan/DX12 then I will be happy with that for one, as I know their will be more in the tank as time goes on :)
 
D.P. said:
If we take one of the new custom 1080s we can see the FuryX is about 67% of the performance at 1050MHZ. 1600MHZ is a 52.3% overclock, so 1.523X67 = 102%. The regular 1080 is at 93% for comparison. Therefore, best case scenario for a FuryX overclocked to 1600Mhz puts it at a very similar performance level, although performance would never ever scale linearly like that . To close the gap to the 1080Ti requires the architectural improvements.

This should all be possible, but does indicate that best case for Vega wont be absolutely humiliating the 1080ti, maybe 10-15% faster? that is good but does mean if you were to buy a 1080 or 1080Ti now you don;t had a lot to worry about. maybe you could wait a fe months to get a Vega 10-15% faster at similar prices, its hardly game changing.
Click to expand...

Its hard to give an accurate overall representation based purely on clocks alone. For example a 2100mhz 980Ti is less than 10% behind a 1080Ti at 2100mhz.
 
Status
Not open for further replies.
Back
Top Bottom