AMD Radeon VII Will Ship Without Double-Precision

[Guest]

AMD’s Director of Product Marketing confirms that double-precision is disabled on its upcoming Radeon VII graphics card.

AMD Radeon VII Will Ship Without Double-Precision : Read more

 

[JamesSneed]

Can we assume this was done to make it less attractive to miners?

 

[redgarl]

^ Simply... no...

I kind of understand AMD on this, if they didn't, the MI50 Instinct would have been outsold by the VII on price alone by lower quality silicon.

Still, I believe AMD could have left it there if they could. It would have been a superb selling point for professional.

 

[thernn8891]

They should have left it intact and marketed it as a professional card that can also game. Now the only option for gamers is the Titan V or some ancient hardware off of EBay. There are plenty of ways to segregate the VII from the instincts without lopping off the FP64. There are many amateur professionals and students for which the Instincts are not a good fit beyond just price considerations and that has left an unfulfilled market which is hungry for a FP64 capable card that can also game but just not at the outrageous prices that Nvidia wants for the Titan V.

Personally, FP64 was the primary driver of my interest in the card but now I couldn't care less about it. I knew many people that were excited also but now not one of them plan to get one. The extra ram is simply not enough to make it a compelling purchase.

 

[arielvtpma]

Can we assume this was done to make it less attractive to miners?

Yes.

 

[thernn8891]

They could make a Radeon VII Pro with FP64 and sell it for $999 and it would sell like hotcakes in the community.

Market it like we know our customers need to work but they also like to relax. Now you can do both with one card! The Radeon VII pro provides a 4k gaming experience and a similar compute value compared to a Titan V for a third of the price. Buy two and double your compute and it will still cost you less!

Instant hit.

 

[shabbo]

If you're into FP64 compute then you should stick with the dedicated hardware and software that supports it. Most likely you'll be needing more than one GPU for those kind of tasks including ECC memory, and also affiliated with a university or a scientific organization that affords it. There's no reason for AMD to include FP64 and then have to support it at a higher-cost for everyone. It's perfectly fine the way it is.

Maybe AMD can release drivers/software for $300 that you guys can purchase?

 

[rabbit4me1]

sorry AMD...crash and burner..and away you go

 

[derekullo]

Can we assume this was done to make it less attractive to miners?

Yes.

Miners don't even use the double precision feature.

If every 100,000th calculation was wrong for a miner they wouldn't care. (Although they should probably dial back their overclock slightly.)

They would just keep going with more calculations.

Double precision is literally doing every calculation twice to make sure the math was done right.

The chart on

https://www.anandtech.com/show/12673/titan-v-deep-learning-deep-dive

illustrates this beautifully.

 

[TJ Hooker]

Double precision is literally doing every calculation twice to make sure the math was done right.

No, not at all. It refers to how many bits of precision you have. Storing an arbitrarily long number in a fixed number of bits results in rounding/truncation error, by doubling the number of bits you reduce this error. Double precision (FP64) FLOPS refers to the rate at which operations involving 64 bit floating point numbers can be performed.

You're right that miners probably don't care about FP64 performance though. But that's because most popular algorithms for GPU mining these days are more memory-bound than compute-bound.

 

[mihen]

The AMD Vega Frontier Edition still exists if you wanted something to compete with the Titan V. It will get similar performance to this Radeon VII but consume more power.

 

[thanhxle]

It is absolutely not similar performance. The FP64 performance is only 1/10 of the titan V which was the entire appeal of the VII for scientists - which was also the point of this article.

 

[tikal]

AMD should have released gamer edetion wiht 8 Gb HBM2 and sell it cheaper, it would be an instant success with gamers. No they have a product almost no one wants.

 

[elroy.coltof]

And that's where they lost me. I'll just keep my RX580 in my development workstation, radeon instinct cards are simply way out of budget for that, so I wouldn't have canabalized that either.

"If you're into FP64 compute then you should stick with the dedicated hardware and software that supports it."
For production, sure. But not all of us work at places where money grows on trees. There simply isnt budget to outfit every dev with $ 9000+ dedicated hardware.

 

[systemBuilder_49]

Typical brain-dead NVidia-fanboy article. AMD provides higher FP64 than ANY NVidia card and believes PCIe v3 is good enough to support it; Toms headline reads "AMD Disables FP64 on AMD cards!" - morons.

 

[systemBuilder_49]

Typical brain-dead NVidia-Fanboy headlline. Most gamer cards provide FP64 at 1/16th or 1/32 of the FP32. AMD does this and Tom's declares, "AMD DISABLES FP64!" which is wrong, as usual !!

 

[aragingmonk]

This was done because business's would buy Radeon VII instead of the card they need, MI50 to save money. But the Radeon VII should be on a cheaper PCB and not designed for 24/7 use in data centers, etc... The RMA numbers would skyrocket within a year and AMD would lose more money because business's were to cheap to buy the right product.
Don't like it, buy Nvidia. They are always willing to rip you off.

 

[alextheblue]

Headline is wrong. It has double precision, they just gimped the rate so it wouldn't murder Instinct sales. It's a consumer variant, this shouldn't be a shock. Nvidia and AMD have both done this quite a few times.

For that matter I can't think of a card that pushes out more FP64 performance at this price point.

 

[superkoizumi]

I lost the interest. It could have been the best GPU for me but I guess not......

 

[passivecool]

durring the presentation, they mentioned "content creators" about a hundred times.
Must be selective attention.
The adobe suite - needing no DP, should perform on this card like on a quadro 4000, but for a few thousand less. or like on a Radeon Pro WX 8200, for half the price. We content creaters are poor MF who need bang for the buck.
I'll wait for the TH review (IGOR?!) ... but it's my baby.

 

[alextheblue]

durring the presentation, they mentioned "content creators" about a hundred times.
Must be selective attention.
The adobe suite - needing no DP, should perform on this card like on a quadro 4000, but for a few thousand less. or like on a Radeon Pro WX 8200, for half the price. We content creaters are poor MF who need bang for the buck.
I'll wait for the TH review (IGOR?!) ... but it's my baby.

The story keeps getting better. Ryan Smith at AT says he contacted AMD and the FP64 rate is 1/8 (vs 1/2 native on Vega 20). That's actually less gimped than claimed here, so the FP64 performance should be around DOUBLE what this article claims. 862 x 2 = ~1.7 TFlops! On a consumer card... at consumer card prices. That's substantially more than anything in this price range.

Still no correction here at Brave New Tom's Hardware. Nobody reads comments but the users anyway.

 

[TJ Hooker]

@alextheblue according to Ryan Smith's tweet from about a week ago, Vega VII actually has full FP64 performance of the M150, so 6.7 TFLOPs.

https://mobile.twitter.com/RyanSmithAT/status/1083959608371175424?s=20

 

[alextheblue]

@alextheblue according to Ryan Smith's tweet from about a week ago, Vega VII actually has full FP64 performance of the M150, so 6.7 TFLOPs.

He was mistaken and he corrected himself. I was referring to his more recent statement, which included confirmation from AMD.

https://twitter.com/RyanSmithAT/status/1085680805802733568

Either way it CERTAINLY ships with double-precision no matter what the outrageous headline claims, and the actual performance is double what is claimed in the article itself. 1/8 works out to 1/4 of what the MI50 puts out, which is impressive for a consumer card these days.