• 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.
Would people be happy with 1080 performance at 1080 prices out of curiosity?
Nope. That would be fail after being one year late and on top of that more power hungry. It needs to bring that level of performance for £400 or less.

Also like I have been saying for a year now, it also needs to bring Titan XP performance for £500-600.
 
I cant see it being below 1080 performance and realistically 1080ti for the gaming version unless something major is wrong

Judging from what we've seen thus far it seem Vega is either a horrible flop with drastically reduced IPC over Fiji, or there's something else going on.

It's hard to imagine Vega is barely above 1070 aka a 980Ti, with a nearly 30%-55% clock speed boost over a Fury X. A die shrunk Fiji from 28nm to 14nm should be faster.

In fact an Normal overclocked Fury X is beating the Frontier Edition in 3D Mark. Something is really going on here.
http://www.3dmark.com/compare/fs/12987144/fs/11047656/fs/6657103#

YEGkMUM.jpg

This is the issue with not sampling well known and respected reviewers with the cards. Now we're getting average joes get ting their hands on it putting out numbers.
 
Last edited:
There must be something else going on, if not they would have been better pulling the gaming card version all together. If they bring out something that is so sub par (even at a reasonable price) surely its going to damage their rep and the share holders are going to walk
 
Do we know that the card is constantly running at 1600MHz, maybe there is thermal throttling happening lower clocks. Otherwise some bug that is sapping performance, that can hopefully be fixed.
That to me looks like constant 1600mhz at 100% load.

04102c1a4794a9cdd6da90cfa729bb13de3caf30265f3d5727c33a382e3fb60e.jpg


3c676b7f57ebba5da9b3bf3250fb2b6bf8b11a9d5c8cd9b687305d71784c8d8b.png
 
Judging from what we've seen thus far it seem Vega is either a horrible flop with drastically reduced IPC over Fiji, or there's something else going on.

It's hard to imagine Vega is barely above 1070 aka a 980Ti, with a nearly 30%-55% clock speed boost over a Fury X. A die shrunk Fiji from 28nm to 14nm should be faster.

In fact an Noel like overclocked Fury X is beating the Frontier Edition in 3D Mark. Something is really going on here.
http://www.3dmark.com/compare/fs/12987144/fs/11047656/fs/6657103#


This is the issue with not sampling well known and respected reviewers with the cards. Now we're getting average joes get ting their hands on it putting out numbers.

Pretty much where my thoughts are at. If the RX Vega gaming card is faster i would want to know whats so different as the specs are basically the same except for double the memory on the FE. Something atm has to be off in the drivers or it's the design.
 
Judging from what we've seen thus far it seem Vega is either a horrible flop with drastically reduced IPC over Fiji, or there's something else going on.

It's hard to imagine Vega is barely above 1070 aka a 980Ti, with a nearly 30%-55% clock speed boost over a Fury X. A die shrunk Fiji from 28nm to 14nm should be faster.

This is the issue with not sampling well known and respected reviewers with the cards. Now we're getting average joes get ting their hands on it putting out numbers.
If it turns out to be a major flop, I will just wait for Volta. I actually do no mind buying it even if it is not as power efficient as Pascal as long as it brings at least 1080 performance for under £400. Pricing will be key, if they mess that up then I will likely not buy it.
 
Pretty much where my thoughts are at.

If it turns out to be a major flop, I will just wait for Volta. I actually do no mind buying it even if it is not as power efficient as Pascal as long as it brings at least 1080 performance for under £400. Pricing will be key, if they mess that up then I will likely not buy it.

Something is really up, an overclocked Fury X on normal "stock" cooling is beating the Frontier Edition that's at 1600mhz

How the hell does Vega at 1600Mhz lose to Fiji at 1125Mhz

YEGkMUM.jpg
 
Something is really up, an overclocked Fury X on normal "stock" cooling is beating the Frontier Edition that's at 1600mhz

How the hell does Vega at 1600Mhz lose to Fiji at 1125Mhz

YEGkMUM.jpg
Well end of the day this is not the gaming card or the proper drivers. No point getting all worked up until proper reviews come out in August. I will reserve judgement until then.
 
Pretty much where my thoughts are at.
Something is really up, an overclocked Fury X on normal "stock" cooling is beating the Frontier Edition that's at 1600mhz

How the hell does Vega at 1600Mhz lose to Fiji at 1125Mhz

YEGkMUM.jpg
Those Fury clocks are being reported wrong. That looks to be an Ln2 score at probably around 1400mhz. Here's what AMD Matt scores at 1172 core and his graphics score is much lower. Note in Graphics test one the Vega is also around 18% faster but loses in test 2.
  1. Score 15986, GPU Duo(1) @1172/600, GFX Score 18064, Physics Score 19411, Combined Score 7516, CPU 5960X @4.0, AMDMatt, Post No.0909 - Link Drivers 16.5.3
 
Those Fury clocks are being reported wrong. That looks to be an Ln2 score at probably around 1400mhz. Here's what AMD Matt scores at 1172 core and his graphics score is much lower. Note in Graphics test one the Vega is also around 18% faster but loses in test 2.
  1. Score 15986, GPU Duo(1) @1172/600, GFX Score 18064, Physics Score 19411, Combined Score 7516, CPU 5960X @4.0, AMDMatt, Post No.0909 - Link Drivers 16.5.3

Still not good at all. Shouldn't be losing any graphics tests to Fiji when clocked higher. Not even if that's is a mis reported 1400mhz Fiji in the database. That's still means Vega stepped back on IPC to lose when clocked much higher.
 
Pretty much where my thoughts are at. If the RX Vega gaming card is faster i would want to know whats so different as the specs are basically the same except for double the memory on the FE. Something atm has to be off in the drivers or it's the design.

Also, doesn't the FE have better Mem bandwidth too due to the 2x8GB stacks?
 
AMD Vega Frontier Edition Live Bench Testing


Just been watched first 10 mins with Metro Redux 4K benchmark, Vega ran with 1650MHz core, 100% fan average over 75C but performance is below stock 1080.

Timespy graphics score 6785 with 8 core Ryzen 7 1800X CPU, slower than stock 1080 and as fast as OC 1070.

Firestrike Ultra graphics score 5091, slower than stock 1080!

DOOM 4K performance is slower than stock 1080.

Mining hashing around 30MH/s.

The Witcher 3: Wild Hunt 4K Ultra performance is slower than stock 1080.

PC system with Vega power consumption average around 800W in gaming. :eek:
 
Last edited:
Still not good at all. Shouldn't be losing any graphics tests to Fiji when clocked higher. Not even if that's is a mis reported 1400mhz Fiji in the database. That's still means Vega stepped back on IPC to lose when clocked much higher.

Ohh i agree. Just got the feeling that something is wrong as it should be faster than Fury X at higher clocks in every test and by a decent margin. If the RX card is faster and where it should be then AMD are again shooting themselves in the foot by releasing this card first as it's about to turn the gaming market off. It's only saving grace is it beats the Titan Xp out in pro work.
 
Also, doesn't the FE have better Mem bandwidth too due to the 2x8GB stacks?

Not that we know of.

Maybe they've managed to get stacks at half the height at the full 2 GHz, or maybe lower stacks have lower latency I don't know.

But it'll definitely have the same bus-width.

EDIT
: Also if you're referring to Vega-FE vs the FuryX, it actually has worse bandwidth due to not running at the full 2 GHz. It's 1.88 GHz at 2048-bit, vs 1 GHz at 4096-bit. So about 484 GB/s vs 512 GB/s
 
Something is really up, an overclocked Fury X on normal "stock" cooling is beating the Frontier Edition that's at 1600mhz

How the hell does Vega at 1600Mhz lose to Fiji at 1125Mhz

YEGkMUM.jpg

I'd suggest most of it is drivers maybe a few other hardwar tricks the Gaming Vega has aswell but i cant see it being too different
 
Ohh i agree. Just got the feeling that something is wrong as it should be faster than Fury X at higher clocks in every test and by a decent margin. If the RX card is faster and where it should be then AMD are again shooting themselves in the foot by releasing this card first as it's about to turn the gaming market off. It's only saving grace is it beats the Titan Xp out in pro work.

My goodness, if the the talk from Videocardz is correct about a critical flaw in a Vega stepping, and them re manufacturing it all for Radeon RX; then selling off the defects as Frontier Edition for semi workstation.... My goodness that means something should really be fired for such a mistake.
 
Status
Not open for further replies.
Back
Top Bottom