Top Tier RX Vega card rumored to be more expensive then the GTX 1080TI

[pas008]

for all we know amd could have large performance drivers not publicly available

 

they did keep ryzen pretty wrapped up,

[PCGuy_5960]

Look man, if this is true, Vega is a failure. If they price it at $350-400 it's not going to fail. But a 1080 competitor for $900 is just a bad buy.

 

Ryzen is awesome, Vega, not so much. Let's hope that Navi is better.

[Fonzie92]

1 minute ago, PCGuy_5960 said:

Look man, if this is true, Vega is a failure. If they price it at $350-400 it's not going to fail. But a 1080 competitor for $900 is just a bad buy.

 

Ryzen is awesome, Vega, not so much. Let's hope that Navi is better.

Wait for Navi 

[Hunter259]

Love how no one is talking about just how incredibly expensive this card will be to produce. You have a huge die and a VERY new memory technology. This card is just a bad joke at this point and should have been canceled.

[tom_w141]

32 minutes ago, Fonzie92 said:

Wait for Navi 

Yay the next "Wait for X"

[tom_w141]

1 minute ago, valdyrgramr said:

The fact that AMD likes putting 2 GPUs on one?  They do it every generation.  It's a very small market, but that's AMD for ya.

2 gpus linked via IF is on the roadmap for navi not vega. If vega was a performer "leaks" would be appearing left right and centre.

[ApolloFury]

Pointless card.

This is literally Fury X 2.0.

I think AMD's strategy at this point is to use the Infinity Fabric from Ryzen to create their next-generation "Navi" GPUs which I can recall is a multi-die design (correct me if wrong).

I guess they hired Scott Wasson to study the negative impact of multi-GPU cards so they can improve upon it, it seems to appear that way.

[tom_w141]

Just now, valdyrgramr said:

Kinda hard to leak something when AMD isn't giving out review samples

notice I typed "leak" if performance was good AMD would leak results themselves. A lot of leaks are internal...

[Mira Yurizaki]

2 hours ago, BuckGup said:

Yes but what's performance like in DX12 titles and Vulkan. If a 480 gains 30 FPS from simply going to Vulkan imagine what a 1080 level Vega gains. My money is on AMD for the future if Volta is a Pascal refresh.

However, the GTX 1080 also receives a similar gain going from OpenGL to Vulkan. My money is on OpenGL is so inefficient that simply going to Vulkan let's GPUs flex their muscles.

 

DX12 really hasn't shown any real benefit on anything aside from Ashes.

[Masada02]

10 minutes ago, valdyrgramr said:

Well, I mean AMD has made the 7990, 295x2, and what was it call the Radeon Pro or something like that?

Technically, from start to finish: 3870x2, 4870x2, 6990, 7990, (8990 if you count OEMs), 295x2, Pro Duo (Fiji chips), Pro Duo (Polaris chips). 

[imreloadin]

1 hour ago, tom_w141 said:

2 gpus linked via IF is on the roadmap for navi not vega. If vega was a performer "leaks" would be appearing left right and centre.

You realize they've done this before IF existed right? He never said anything about a dual Vega card using IF. They'll most likely do it the old fashioned way they have done it several times in the past.

[Jito463]

2 hours ago, VagabondWraith said:

Typically G-Sync has a wider Hz range than FreeSync. Mine covers 30-165Hz and was one of the more "affordable" G-Sync monitors at $500.

My Viewsonic does 35-144 Hz (no higher, because it's a 144 Hz monitor).

[Jito463]

Just now, imreloadin said:

You realize they've done this before IF existed right? He never said anything about a dual Vega card using IF. They'll most likely do it the old fashioned way they have done it several times in the past.

Except Navi is supposed to have multiple dies that appear as one GPU to the system.  It's (theoretically) more about expandable dies than expandable GPUs.

[tom_w141]

Just now, imreloadin said:

You realize they've done this before IF existed right? He never said anything about a dual Vega card using IF. They'll most likely do it the old fashioned way they have done it several times in the past.

If they do then that works the same way as regular crossfire, i.e. shit. 2 gpus linked by IF will be interesting because the system will see it as a 1 and will utilise it far better. That is why Navi is exciting and why the previous way isn't.

[tom_w141]

1 minute ago, Jito463 said:

multiple dies that appear as one GPU to the system

Thats exactly it. The current 2 gpus 1 card? meh. 2+ being recognised as 1? Hell yes.

[Dylanc1500]

1 minute ago, tom_w141 said:

If they do then that works the same way as regular crossfire, i.e. shit. 2 gpus linked by IF will be interesting because the system will see it as a 1 and will utilise it far better. That is why Navi is exciting and why the previous way isn't.

So AMD is basically going to try, with Navi and IF, what NVidia is doing with NVlink?

 

I don't deal with AMD frequently as I typically deal with NVidia ,IBM and Intel.

[Juanitology]

3 hours ago, MilfShake said:

with that pricing and performance of a 1080 it will be DOA

A while back the VEGA FE was shown to perform the same clock-for-clock as the Fury X. I have two theories:

 

1. it's all bullshit and it's not a new architecture, just a Fury X with an optimized process and some extra gimmicks that allows for much higher clocks. 

 

2. Vega RX has 1080Ti-level performance in terms of FPS at the cost of very poor double-precision performance, which meant it wasn't suitable for the professional market. so they just took an OCed fury X and stuck HBM2 on it to sell it as the pro card and the consumer vega cards will be the true new architecture.

[Valentyn]

20 minutes ago, Terodius said:

A while back the VEGA FE was shown to perform the same clock-for-clock as the Fury X. I have two theories:

 

1. it's all bullshit and it's not a new architecture, just a Fury X with an optimized process and some extra gimmicks that allows for much higher clocks. 

 

2. Vega RX has 1080Ti-level performance in terms of FPS at the cost of very poor double-precision performance, which meant it wasn't suitable for the professional market. so they just took an OCed fury X and stuck HBM2 on it to sell it as the pro card and the consumer vega cards will be the true new architecture.

It's made worse when the world record for Fury X overclocking had it hit around 1450Mhz with 1000Mhz HBM, and it managed to hit similar performance to an GTX 1080.
https://www.pcper.com/news/Graphics-Cards/AMD-Radeon-R9-Fury-Unlocked-Fury-X-Overclocked-1-GHz-HBM

Now Vega with supposed improvements, higher clocks, and a die shrink can't manage it

[Morgan MLGman]

3 hours ago, VagabondWraith said:

Who in the hell is gonna buy a card that expensive except for the most die hard AMD fankids? I mean, you need a 1500w PSU for that thing.

How can someone who is a self-proclaimed Intel / Nvidia "enthusiast" that defended X299 in many threads before and bought one himself knowing it's just essentially bad and basically ocomparable with the R5 1600 both in games & professional workloads (talking about the 7800X here) even say those things?

 

Didn't you just buy a power-hungry, relatively bad-performing and overpriced platform yourself just because it's branded Intel? That seems like hypocrisy because from what you said, Vega looks exactly the same: power-hungry, relatively bad-performing and overpriced.

 

On topic: Vega leaks aren't impressive so far, if they are correct then I'm taking an educated guess here that AMD wouldn't price those cards that high. However, they might've priced Vega that high if they didn't show something cool about it yet that would justify the price tag. I doubt they priced a ~1080 level of performance card as high as top-end 1080Tis because it would just be a retarded business decision, and so far AMD has made the right ones with Ryzen. At least when it comes to pricing.

[Humbug]

4 hours ago, Bananasplit_00 said:

I find this to be quite an interesting turn, not compleatly unexpeced seeing as the top end card is expeced to have an AIO cooler but definatly not a good value card if this is the actiual pricetag. i just hope that the air cooled versions will be cheaper and that waterblocks will come soon so i could consider Vega for my current PC.

Well that depends on what the performance is like...

This top end SKU needs to be faster than the GTX 1080ti to be decent value at this price point.

[Humbug]

24 minutes ago, Terodius said:

A while back the VEGA FE was shown to perform the same clock-for-clock as the Fury X. I have two theories:

 

1. it's all bullshit and it's not a new architecture, just a Fury X with an optimized process and some extra gimmicks that allows for much higher clocks. 

 

2. Vega RX has 1080Ti-level performance in terms of FPS at the cost of very poor double-precision performance, which meant it wasn't suitable for the professional market. so they just took an OCed fury X and stuck HBM2 on it to sell it as the pro card and the consumer vega cards will be the true new architecture.

LOL. number 1 is impossible. It's obviously a radical new architecture.

 

I too saw the gamers Nexus tests which proved that in gaming it is no better than a Fury X clock for clock.

 

There are two possible explanations for this

(1) Gaming drivers are not ready yet

OR

(2) All the radical architecture changes which AMD made did not work out (for gaming workloads), so you end up with a highly innovative design which after years of R&D you realize does not perform great in practical applications (think Bulldozer CPUs).

[goodtofufriday]

Two questions 

 

Why are we taking a Singular rumor as fact. I feel like this is more to push AMD to definitely announce the pricing on friday.

 

And has anyone considered removing swedens VAT cost from that final price + tax?

[Notional]

I'm calling it now: There's something unique about VEGA. Based on the number of stream processors contra cuda cores in a 1080ti (even transistors), there is no way a Vega would ever beat a 1080ti. However! WIth the unique memory controller, we might see a card that is going to have a very low latency and a very high minimum framerate.

 

We are seeing the cards demoed against NVidia with no fps counter and with gsync/freesync, to test the gaming experience. People are reporting a smoother gameplay with the VEGA. So even though max fps will be worse than a 1080ti, the actual gaming experience might be the same or even better in smoothness. After all, there is a reason we left average/max fps and went on to lowest 1% and lowest 0.1% fps benchmarks.

[ravenshrike]

What's these guys' track record on rumors vs Bits and Chips? Cause according to B&C Vega will be very price competitive. At the price point quoted by the OP the only way that's true is if Vega's FPS band is all of +- 15fps at 1080p at GTX 1080 avg fps levels.

[Bananasplit_00]

1 hour ago, goodtofufriday said:

Two questions 

 

Why are we taking a Singular rumor as fact. I feel like this is more to push AMD to definitely announce the pricing on friday.

 

And has anyone considered removing swedens VAT cost from that final price + tax?

those are the EU prices including tax acording to Sweclockers sources.

 

30 minutes ago, ravenshrike said:

What's these guys' track record on rumors vs Bits and Chips? Cause according to B&C Vega will be very price competitive. At the price point quoted by the OP the only way that's true is if Vega's FPS band is all of +- 15fps at 1080p at GTX 1080 avg fps levels.

they got the prices of Ryzen 7 and 5 a few weeks before they launched, so thats something.