Poll: ** The AMD VEGA Thread **

[nashathedog]

I blame Ryzen.

That has been AMD's focus over the last 4 or 5 years which is why RTG have lacked the funding needed to keep up with Nvidia. The last 3 Titans have yet to be beaten and Maxwell's Titan X to Pascal's Titan X held a significant performance leap along with a price leap, With the high end 10 series cards it's felt like we were forced to pay extra for every frame it was faster than it's relevant predecessor, which was why I didn't buy one.

 

[muon]

GCN is just inefficient.

Has been mentioned time and time again but people refused to believe Vega would have similar IPC.

 

[D.P.]

Well they've just released a new card with most of the new performance features disabled, and not said why. We've also seen AMD hold their cards very close to their chests for a few years now, with no leaks, and a tight hold on any meaningful information. I think Adored is just wrong, and he makes a lot of assumptions on zero evidence.

Where is the evidence that features are diswbled?

 

[Hedge]

It's really frustrating because I wanted to try Vega in a second PC since I've had a Gtx 1080 for so long. However it's looking like a complete waste of time.

Amd shouldn't bother making products priced at the high end if they can't compete with a card that's over a year old.

 

[D.P.]

Yep. And many have forgotten already two days later that where the Vega FE card matters, machine learning etc, is faster than the TXp by a hefty margin. That shows the the grunt power is there, and that piece of software doesn't need gaming optimisations.

QUOTE


Can you show me a single machine learning benchmark of Vega. No one has benchmarked Vega for DL yet afaik

 

[muon]

I wonder if the people who got angry at people that supposedly posted "fake" benchmarks even though they appeared to be genuine, will eat their words.

 

[CAT-THE-FIFTH]

That has been AMD's focus over the last 4 or 5 years which is why RTG have lacked the funding needed to keep up with Nvidia. The last 3 Titans have yet to be beaten and Maxwell's Titan X to Pascal's Titan X held a significant performance leap along with a price leap, With the high end 10 series cards it's felt like we were forced to pay extra for every frame it was faster than it's relevant predecessor, which was why I didn't buy one.

That maybe the case but it still does not account for AMD making errors when launching the cards in the first place - they seem to not launch cards with their best foot forward.

 

[Shaz12]

Why even bother releasing a product which barely beats a 1070? None of this makes any sense

 

[akarypid]

That maybe the case but it still does not account for AMD making errors when launching the cards in the first place - they seem to not launch cards with their best foot forward.

In all fairness, if you have a dog of a card, what would a 'successful' launch look like?

Add to that the fact that even when you have a decent performer (RX 480) people are still ready to jump on you and scream foul (power consumption non-issue).

 

[r7slayer]

Well FuryX Bearly beat a 980 last gen and this gen Vega can't even beat a 1070 lol. Gone backwards after all the supposed IPC improvements. The arch improvements and die shrink etc etc.
Poor Volta lol. Id hate to be AMD right now.

 

[JediFragger]

Bearly

Beerly?

 

[zornyan]

You do realise that essentially every high end card went up by $50 between the 9xx and 10xx series, right? That's not including the extra cost of the "founders edition" cards. Then you have the Titan branded cards which are basically a way of saying "**** our cards are now too fast to just sell at the prices we normally sell them at, let's whack on a few hundred dollars and flog the cut-down ones in our normal price bracket". The Titan series would not exist if AMD was competitive in that arena, at least not at anywhere near their current prices. nVidia are kind of at the point Intel were back in 2011 when they released Sandy Bridge. They're not entirely sure they have no competition yet so are just "testing the waters". I have no doubt that soon nVidia will become like Intel and push marginal improvements each year if they no longer see any threat from AMD. If Vega fails as hard as the FE card indicates it might, that point will have been reached.

you forgot brexit effecting prices, even inflation in general.

pack of fags cost me £4 when I started smoking, now it's £10 quid.

but anyway, forgetting that, the 980 launched with a higher msrp then had a price cut when the 980ti dropped, just like the 1080 had.

 

[r7slayer]

Beerly?

barely **
Spell checkers out full force.

 

[D.P.]

can't wait to see what excuses for amd people make when rx vega launches "yeah well finewine!"

gonna be a lot of people on suicide watch...

The excuses coming out are already hilarious. "Vega is not a gaming card" well tell that to AMD marketing, the AMD website, the AMD drivers,the Vega control panel, etc.

"It is faster than a TitanXP in pro stuff" organising Apple's and oranges for drivers. Normalize those drivers and it is slower than a 1080, as shown by the P5000.

"But it's way faster in machine learning" there isn't a single published benchmark released, so why make up such lies.

" it doesn't have all the features enabled" how do you know? Have you reverse engineered the drivers? The rest for tile based rendering is not known to work for Vega so it is entirely pointless. We have no way of knowing what is it isn't working. And if major features aren't working, what kind of excuse is that.

"The drivers are Fiji drivers from 17.1.1" no they aren't, GPU-Z just can't read the drivers properly. The actual creation dates and driver oF of the actual DX and OpenGL library files within the driver are newer than the latest 17.6.2 crimson driver. Some kid the files are dated June 24th.I

"Its running Fiji drivers" only way that would be possible is if the architecture and IS A was the same as Fiji , so why would you expect any performance difference clock for clock if it's identical? Fact is it isn't the same and it isn't running Fiji drivers.




Performance may very well change but all this fantasy and straight out lying to try and soothe their egos is just not needed.

 

[CAT-THE-FIFTH]

In all fairness, if you have a dog of a card, what would a 'successful' launch look like?

Add to that the fact that even when you have a decent performer (RX 480) people are still ready to jump on you and scream foul (power consumption non-issue).

Well even that was an issue since AMD made some weird design decisions on the reference model:
1.)Make the VRM handle more than what the power connectors could supply by a decent amount,so probably was an expensive PCB for that class of card
2.)Underspec the heatsink on the reference cooler so the card ran hot
3.)Underspec the PCI-E power connector so the card drew more power than required through the PCI-E power slot,or simply made it power starved

Then you had reviewers showing the card was both downclocking due to power and temperature and hence underperforming.

It would have made more sense to perhaps over-engineer the VRMs a bit less,and put that money into a heatsink with more metal in it and an 8 pin PCI-E power plug.

 

[muon]

The stupid thing was that the signs were all there.

Read the following thread from around the end of April onwards when leaked benchmarks started coming out.

https://forums.overclockers.co.uk/threads/amd-vega-confirmed-for-2017-h1.18746880/

I think my frustration at the level of optimism culminated in the following posts.

https://forums.overclockers.co.uk/posts/30749565

Why are people so obsessed with dismissing this. 'It can't be true, Vega will have magical improvements'.

For all the marketing, even Pascal was pretty much just Maxwell overclocked. The IPC improvement from Fiji to Polaris 10/20 was minimal if any (r9 390 went toe to toe with rx 480). Vega is still on the same process as Polaris.

People are setting themselves up for massive disappointment.

Who knows, maybe the top end Vega will havr more than 4096 shaders.

I don't think it is useless unless someone can show this informatiom is fake.

Which nobody can. Only that it doesn't line up with people's prior expectations of a massive IPC improvement. In fact I would bet against one given all the information from the last few months. I can be hopeful there will be one, but that is all it is.

 

[CAT-THE-FIFTH]

This could have been solved if AMD just locked down gaming drivers on the FE in the first place,and said gaming ready drivers would be released when the RX Vega was launched.

Edit!!

Look even Nvidia could not get away from people doing gaming tests on the Titan Z,so did they honestly think people wouldn't try to do the same with the Vega FE??

 

[shankly1985]

You guys still going on? FE Vega isn't a gaming GPU so why the Doom and gloom based on that?
Why is this GPU a GPU not aimed at gaming becoming the number one source for Vega gaming failure?

Honestly just wait for RX results otherwise you might just make yourself look stupid.

 

[Nexus18]

I am bookmarking a lot of posts just in case vega isn't a flop, if it isn't... it will be very amusing reading them again

 

[nashathedog]

It's really frustrating because I wanted to try Vega in a second PC since I've had a Gtx 1080 for so long. However it's looking like a complete waste of time.

Amd shouldn't bother making products priced at the high end if they can't compete with a card that's over a year old.

We don't know for sure what it can do, It may turn out to be tit for tat with the 1080 and it'll be great if it's the right price regardless of being late to market compared to the 1080 etc.
Price will be the deciding factor. If they come out with a 1080ti priced card that can't compete with the 1080ti or the 1080 at times it'll be a shame.