AMD Radeon RX Vega 64 and 56 Reviews [*UPDATED* Aug 28]

[PeterScott]

Is it possible for you and others to understand amd can not decide the price?
That the gpu market isnt the former Soviet union.

This amd bashing is hillarious. Its a very small player on the market yet all those conspiracy theories is pasted all over. Get realistic.

AMD can force pricing up, by charging a unreasonable wholesale that doesn't allow for a reasonable retail profit when selling for the $499 "MSRP".

There is every indication that this IS the case, and that AMD was giving retailers rebates to offer a small batch of "MSRP" cards.

The rebates went away, and so did the "MSRP" cards.

 

[tential]

Tpu took power saving mode on their entire bm suite.
You gain 30% perf/w efficiency. And get below 980ti fe power usage.
At a loss of meager 4% perf.
Fact.

And its litteraly a push with a button in driver setup with the std bios. Every one buying this card have the ability to do it. If power is an issue for you its simply in all practicallity solved by the cost of 4% perf.

And at 4k that this card is meant to be played it still does it good deal faster in some titles than the 1080 even in this mode. And excuse me we both knows it performs in the most important engines.
The market reaction and reviews shows the doom and trashtalk is way overboard.

Its eg a far faster card where it matters and a gazilion times more future proof than the 6970 vs fermi. And futureproof like we probably get 4-5% perf uplift with wolfenstein via fp16 alone. So its tomorrows card.

We also have to add if you share pro and consumer market you also share cost. So this mm2 trashtalk is also way over board. Amd aparently thought the saved dev cost outweight the mm2 and efficiency drawbacks. I have meant for 5 years its a stupid decision having all that compute build in so its good people start to think the same but in all fairness we dont have access to remotely the info amd have. All logic says they are right so its all asumptions from here.

How is AMD unable to see that this BIOS is the bios that shows their card in the best light? Why would you do this for 4% performance...

 

[Tup3x]

How is AMD unable to see that this BIOS is the bios that shows their card in the best light? Why would you do this for 4% performance...

They have huge obsession to beat (or match) GTX 1080 FE, no matter how much power it draws. So, they can say that their product is "faster" or trade blows with the competitor's product. They obviously think that most people do not value perf/W (or noise & heat). I personally would sacrifice few % of performance if that would mean massive gains in power usage, heat and noise reduction. That's one reason why I do not ever up the voltage when overclocking GPUs.

 

[tviceman]

Actually, V56 is faster than the 1070 pretty much across the board. V64 trades blows with the 1080.

This is why most review sites gave V56 "editor's choice" or "gold" or whatever else they want to call it.


The only thing it has against it is that power draw of 45-75W more than the 1070, and that isn't really on the minds of most people when they are going out looking for video cards. They just want something that is best bang for the $$$, and at $399, the V56 was.

Think about that. 15 months later, 5% faster, 35% higher power draw, worse perf/$ at MSRP prices... GOLD AWARD.

Classic.

 

[french toast]

Every review i have read, VEGA 56 is faster than GTX1070.

http://www.anandtech.com/show/11717/the-amd-radeon-rx-vega-64-and-56-review

http://www.eurogamer.net/articles/digitalfoundry-2017-amd-radeon-rx-vega-56-review

https://www.techpowerup.com/reviews/AMD/Radeon_RX_Vega_56/31.html

https://www.techspot.com/review/1468-amd-radeon-rx-vega-56/

And power is only 40-50W more (Chill enabled) for a desktop GPU which many will not even care. Im not saying power consumption doesnt matter, im saying that most people doesnt care for 50W more on a desktop gaming system to be the first criteria of purchase.

With chill enabled? What does that do to the performance?
Should be compared like for like, in which case it's about 90w!

I just hope AMD learns from this and fixes the bottleneck ready for Vega 11, I hope they get the right CU to frequency ratio on point, and I hope they give it enough rops and bandwidth, even if that means gddr5x.

 

[EXCellR8]

Reference coolers are usually considerably worse than aftermarket, usually even just due to blower designs. Case exhaust designs are more efficient.

that i know, trust me. I've owned plenty of reference AMD cards including the 290X but I've never had any issues with NVIDIA's reference design and I actually like that style a lot. just not sure if Vega is worth another additional upcharge yet... plus I do like the simple and clean design of the LE

 

[krumme]

The thing is... In "real world" you got extremely lucky and that doesn't apply to 98 % of potential buyers.

This is the reality here in Finland:

Zero cards available and prices are beyond ridiculous. Cheapest available GTX 1070 costs 439,90 € in the same shop (all prices have 24% VAT included).

What shop can you get 1070 for 439 euro in finland?
My card was send from norway.

 

[krumme]

How is AMD unable to see that this BIOS is the bios that shows their card in the best light? Why would you do this for 4% performance...

Its not even a bios change but merely a driver setting.
My take on it is amd is less concerned about long term vrand value than profit for the next q. What else could it be?

 

[krumme]

AMD can force pricing up, by charging a unreasonable wholesale that doesn't allow for a reasonable retail profit when selling for the $499 "MSRP".

There is every indication that this IS the case, and that AMD was giving retailers rebates to offer a small batch of "MSRP" cards.

The rebates went away, and so did the "MSRP" cards.

The market happens to work the way amd will not sell cards if the retailers have no profit. Simple as that. Msrp price is just a guideline. If the demand outsttrips supply price will go up otherwise it goes down.
I dont care what amd msrp price is. What matters is price.
The cards offers tons of compute for the price so that alone will unfortunately give a solid demand. Sorry for most of us but good for amd.

 

[AtenRa]

With chill enabled? What does that do to the performance?
Should be compared like for like, in which case it's about 90w!

According to Techpowerup, VEGA 56 is 6% faster than GTX 1070

Also according to Techpowerup enabling Chill will decrease performance by 8% and reduce power by 33,33% on the VEGA 64 edition.

So if we apply the same to VEGA 56 then it will be 2% slower than GTX 1070 and only consume 153W (229W - 33,33%) vs 145W for the GTX 1070.

 

[krumme]

With chill enabled? What does that do to the performance?
Should be compared like for like, in which case it's about 90w!

I just hope AMD learns from this and fixes the bottleneck ready for Vega 11, I hope they get the right CU to frequency ratio on point, and I hope they give it enough rops and bandwidth, even if that means gddr5x.

The basic shaders needs a redo.
Lets see how drivers for the primitive shaders pan out the next half year but imo the challenge for efficiency goes beyond high level stuff.

 

[krumme]

According to Techpowerup, VEGA 56 is 6% faster than GTX 1070

Also according to Techpowerup enabling Chill will decrease performance by 8% and reduce power by 33,33% on the VEGA 64 edition.

So if we apply the same to VEGA 56 then it will be 2% slower than GTX 1070 and only consume 153W (229W - 33,33%) vs 145W for the GTX 1070.

Rx56 is another point on the power curve so you can not use same results. We can asume it will be less as its less agressively clocked.
But notice for std bios power save rx64 is only 4% slower for the aprox 30% powersave. Its absurd.

 

[AtenRa]

Rx56 is another point on the power curve so you can not use same results. We can asume it will be less as its less agressively clocked.
But notice for std bios power save rx64 is only 4% slower for the aprox 30% powersave. Its absurd.

Somebody has to tell them they dont need to beat the competition no matter what
Also I believe the reviews only tell half the story, i want to see chill performance and power consumption per game.

 

[IRobot23]

We should also know that this is gaming efficiency (current games - up to 4-5 year old games).

Lookslike AMD will have Far Cry 5 bundle.

 

[PeterScott]

Somebody has to tell them they dont need to beat the competition no matter what
Also I believe the reviews only tell half the story, i want to see chill performance and power consumption per game.

By clocking the cards up enough to trade blows with 1070/1080, AMD hopes to command 1070/1080 pricing. And perhaps thinks they have less egg on their faces than if they came in clearly under-performing 1070/1080.

So it is pretty obvious, why they pushed it so close to the limits.

Tied performance + consumes more power, is easier to sell than:
Lesser performance, but still consumes more power (but slightly less than the matching performance case)

No amount of convolutions will bring equal performance at anywhere near equal power.

 

[krumme]

By clocking the cards up enough to trade blows with 1070/1080, AMD hopes to command 1070/1080 pricing. And perhaps thinks they have less egg on their faces than if they came in clearly under-performing 1070/1080.

So it is pretty obvious, why they pushed it so close to the limits.

Tied performance + consumes more power, is easier to sell than:
Lesser performance, but still consumes more power (but slightly less than the matching performance case)

No amount of convolutions will bring equal performance at anywhere near equal power.

Meh. Ofcource for rx64 its nuts to go for 4% perf for 30% eff because its still comfortably faster where it matters. In 4k and the new engines. Then only a few cares that it uses 38w more. Really who cares then? Its lost cause for amd to fight in old dx11 games or dx 12 labelled games that is even more dx9 than a proper dx11 implementation as nv dx11 driver support is better. I would have set them to avg 4% slower gained the 30% efficiency and agressively pitched them for 4k and new and comming aaa dx12/vulcan games. And clearly also stated what they were not good at to control anticipation instead of more or less indirectly hyping it with all the smoke. Segmenting with sense instead of trying to sell a nuclear reactor to all.

 

[richaron]

No amount of convolutions will bring equal performance at anywhere near equal power.

Dude I can easily provide multiple benchmarks which show how wrong this statement is; Although they will obviously be about compute performance or ideal games. What you are talking about is gaming performance, and an average of current gaming performance. But what you said is obviously wrong.

Technical people reading these posts will expect you to know what you're talking about before making massive generalisations. Sometimes I know what posters are trying to say, and they may have a point. But they make themselves look stupid with sweeping generalisations.

 

[PeterScott]

Dude I can easily provide multiple benchmarks which show how wrong this statement is; Although they will obviously be about compute performance or ideal games. What you are talking about is gaming performance, and an average of current gaming performance. But what you said is obviously wrong.

Technical people reading these posts will expect you to know what you're talking about before making massive generalisations. Sometimes I know what posters are trying to say, and they may have a point. But they make themselves look stupid with sweeping generalisations.

Of course I am talking about the average of gaming performance. Which is the only fair way to consider gaming cards.

If you are going to cherry pick exceptions, that is just your bias showing.

 

[krumme]

Of course I am talking about the average of gaming performance. Which is the only fair way to consider gaming cards.

If you are going to cherry pick exceptions, that is just your bias showing.

In tpu gaming suite rx56 is like 6% faster than 1070. But in reality that gaming suite is far more tilted to amd than all games out there is on average. I dont game my old steam libary. Heck there are not even local. Others might have it different.
Tpu gaming suite is actually quite new. The former years it was really not the most progressive but anyway how a card perform is about the games you play and your future games. Thereby also how long time you keep the card.

 

[french toast]

I thought power saving mode and radeon chill were 2 different things?
I knew the impact of power saving mode already, thought you were talking of chill which restricts frame rate at certain points or something?

Nonetheless, it's AMDs fault for not tuning the cards properly, in standard mode it consumes vastly more power than gtx 1070, overclocking really compounds the issue, this is what reviews must show and base their opinions on.
Any power saving modes should be judged in isolation.

 

[Unreal123]

For the people who say that RX VEGA 56 will sold at $399 can stop posting now.

http://hexus.net/tech/news/chassis/109171-amd-radeon-rx-vega-56-8gb-listed-uk-retailer/

It is priced at $520 and same price as GTX 1080 in UK.

 

[krumme]

For the people who say that RX VEGA 56 will sold at $399 can stop posting now.

http://hexus.net/tech/news/chassis/109171-amd-radeon-rx-vega-56-8gb-listed-uk-retailer/

It is priced at $520 and same price as GTX 1080 in UK.

I hate that exageration. At least read your own link.
Its 400 pound. The 1080 is 450 pound.
But anyway if demand is higher prices go up. Bad luck. Dont buy it.

 

[coercitiv]

I hate that exageration. At least read your own link.
Its 400 pound. The 1080 is 450 pound.
But anyway if demand is higher prices go up. Bad luck.

The same retailer sells RX 580 8GB for £330. People really need to adjust to the current market realities.

 

[krumme]

I thought power saving mode and radeon chill were 2 different things?
I knew the impact of power saving mode already, thought you were talking of chill which restricts frame rate at certain points or something?

Nonetheless, it's AMDs fault for not tuning the cards properly, in standard mode it consumes vastly more power than gtx 1070, overclocking really compounds the issue, this is what reviews must show and base their opinions on.
Any power saving modes should be judged in isolation.

I dont know what this chill is.
Clearly its amd responsibility how they run the card.
But why this obsession with reviews and how its reviewed? Each can make his own evaluation for personal use and everyone can change a simple powerprofile. This is not bios change or similar but a simple setup in the driver settings.

 

[Unreal123]

Jaycents will never do a AMD review sample of there GPU because they lied about prices on their faces and it makes people miss guide. It only made reviews look stupid and miss guided.
https://twitter.com/JayzTwoCents/st...ault#version=efca37d5b932a816b8b4b6753d8e3a4e
https://twitter.com/JayzTwoCents/st...ault#version=efca37d5b932a816b8b4b6753d8e3a4e