Vega/Navi Rumors (Updated)

[CatMerc]

So maybe performance between 1080 and Ti for the price of a 1080?

How to get absolutely murdered by Volta^

 

[Crumpet]

How to get absolutely murdered by Volta^

AMD's ad campaign had POOR VOLTA on it, so I hope they haven't written a cheque that their ass can't cash.

 

[itsmydamnation]

AMD's ad campaign had POOR VOLTA on it, so I hope they haven't written a cheque that their ass can't cash.

The first major revision of GCN
getting the rop caches global
improving rasterization
has massive bandwidth

front end/geometry was significantly improved with polaris, vega looks like it goes further.

We are talking something like near double of Polaris at a higher clock and probably ( if any of the patents form the last few years make it into NCU) higher utilization.

I wouldn't worry to much.

 

[Crumpet]

The first major revision of GCN
getting the rop caches global
improving rasterization
has massive bandwidth

front end/geometry was significantly improved with polaris, vega looks like it goes further.

We are talking something like near double of Polaris at a higher clock and probably ( if any of the patents form the last few years make it into NCU) higher utilization.

I wouldn't worry to much.

I'm looking forward to the launch, i'll be purchasing one anyway due to my freesync monitor.

 

[tviceman]

More likely is that RX Vega will be 95-105% of GTX 1080 Ti for 599-699$.

What was your prediction with Fiji when it was 4-6 weeks out?

 

[Glo.]

https://seekingalpha.com/article/4058480-amd-radeon-rx-vega-new-catalyst

Vega-based cards will outcompete even Nvidia's (NASDAQ:NVDA) most powerful graphics card for gaming, which is its Pascal-based GTX 1080 Ti with 11GB GDDR5X memory. Vega will also deliver smooth 4K display.

 

[Malogeek]

And they're basing that on.... voodoo?

 

[amenx]

"outcompete" can only mean close to ti performance at lesser price. If better than 1080 and slightly less in price, should turn out good for AMD. Would love to see AMD do for the high end what their 480 did to the mid end.

 

[IEC]

I think you guys expecting low prices are going to be disappointed. HBM2 ain't cheap.

 

[piesquared]

But, how many G5 chips for 8GB or 12GB? Sure the price is higher for HBM2 but the density is much higher so while more expensive, the difference between 8GB HBM2 and G5 is probably not huge. The packaging would be higher also, but then the BoM for the board is lower. I think the difference is exaggerated and while an HBM2 implementation is surely higher, i doubt that it is in the range that these types of comparisons tend to envision. Has there ever been an actual comparison between the 2? It would definitely be interesting to see.

 

[Crumpet]

I think you guys expecting low prices are going to be disappointed. HBM2 ain't cheap.

I'm willing to pay 1080ti prices for 1080ti performance.

Do I think it's too much money for a graphics card? Yes.

Would I rather give 800bucks to Radeon....yes.

 

[amenx]

I'm willing to pay 1080ti prices for 1080ti performance.

Do I think it's too much money for a graphics card? Yes.

Would I rather give 800bucks to Radeon....yes.

What if its a repeat of the FuryX, which performed slightly below 980ti but at same price (initially).

 

[unseenmorbidity]

What if its a repeat of the FuryX, which performed slightly below 980ti but at same price (initially).

Wasn't the 980ti considerably better with an oc? The furyx caught up at higher resolutions though.

oc speeds would matter to me.

For instance, the rx 470 is a lot better than it looks on paper because amd downclocked it.

 

[jrphoenix]

I'm just hoping for some updates next week after the refresh RX 500 hits. I'm guessing the cards will release at the end of May?

 

[Crumpet]

What if its a repeat of the FuryX, which performed slightly below 980ti but at same price (initially).

I would be slightly disappointed, but not enough to put me off the sale. Plus I have firm belief in AMD finewine anyway.. Which considering my upgrade habits would mean it would reach its full potential under somebody elses hands hahaha.

 

[Glo.]

I will give you guys a funny question.

What if, 687F:C1, 1.2 GHz GPU demoed by AMD, averaging 72(Exactly on par with GTX 1080 Ti) FPS in Doom 4K, Vulkan is not top of the line GPU?
What if there will be 1.5 GHz, water cooled version of the same GPU?

 

[Crumpet]

I will give you guys a funny question.

What if, 687F:C1, 1.2 GHz GPU demoed by AMD, averaging 72(Exactly on par with GTX 1080 Ti) FPS in Doom 4K, Vulkan is not top of the line GPU?
What if there will be 1.5 GHz, water cooled version of the same GPU?

I have already considered this myself, to be honest.

 

[Erithan13]

I'm just hoping for some updates next week after the refresh RX 500 hits. I'm guessing the cards will release at the end of May?

AMD have been doing some promotional stuff with Prey which releases on the 5th May. One would hope Vega is timed for that but who knows.

The complete silence on Vega is the most unnerving thing for me right now. Think of how much we knew about Polaris before release, both from official slides and all the leaks. I'm down to thinking Vega is either an absolute monster and AMD are keeping very tight-lipped deliberately, or there's something wrong and they're desperately slapping whatever band-aids they can on it before release. Either way I'll at least be glad to get the wait over with.

 

[linkgoron]

Anything less than 90% 1080ti would be a total failure, IMO, for a ~500mm^2 GPU. The Rx 480 is essentially 50% of the 1080ti at 1080p. The Rx Vega is (supposedly) more than double the die size, has much faster memory, is supposed to have an improved architecture and better clocks. Even having equal performance would be mediocre IMO.

 

[amenx]

Think of how much we knew about Polaris before release, both from official slides and all the leaks. I'm down to thinking Vega is either an absolute monster and AMD are keeping very tight-lipped deliberately, or there's something wrong and they're desperately slapping whatever band-aids they can on it before release.

The last time I recall them being so tight-lipped about any product was Phenom.

 

[french toast]

The last time I recall them being so tight-lipped about any product was Phenom.

Please dont say things like that.

 

[zinfamous]

https://seekingalpha.com/article/4058480-amd-radeon-rx-vega-new-catalyst

that dude's article is based on nothing. Honestly, I wouldn't trust an SA article as far as I could throw it--even the AMD bulls; much less Hibben, that asshat.

 

[CatMerc]

I will give you guys a funny question.

What if, 687F:C1, 1.2 GHz GPU demoed by AMD, averaging 72(Exactly on par with GTX 1080 Ti) FPS in Doom 4K, Vulkan is not top of the line GPU?
What if there will be 1.5 GHz, water cooled version of the same GPU?

Uhh, it was slightly above a 1080. Definitely not 1080 Ti lol

The last time I recall them being so tight-lipped about any product was Phenom.

Oh boy, if it has something like the TLB bug...

 

[alcoholbob]

Wasn't the 980ti considerably better with an oc? The furyx caught up at higher resolutions though.

oc speeds would matter to me.

For instance, the rx 470 is a lot better than it looks on paper because amd downclocked it.

Yep, 980 Ti were 1200MHz stock, but could OC to ~1500MHz, with some Classifieds/Kingpins able to hit nearly 1600MHz.

On the other hand, Fury X generally only had about (IIRC) about 50-75MHz in headroom.

 

[guachi]

Given how much more expensive a GSync monitor is (about 30% more) if Vega could actually reach 1080ti speeds, it would be the only real choice for 4k gaming.

I don't expect it to reach those speeds. But I suppose it might. Then all we need is someone to come out with a 4k Freesync screen that will do more than 60 Hz.

Even 1080 speeds for 1080 prices is still attractive if only, again, because Freesync is so much cheaper than GSync.