Will AMD release any high performance GPU's in Q2 or Q3?

[Remilia]

Hmm, ok... 945MHz seems a bit low though.

They all work the same. Bandwidth is just clock speed * bus width. Divide by 8 to convert bit to byte.. For this case it'd be 945 * 2048 / 8

[mojo6912]

and can now be picked up for under $450...

Where? I've been looking.

- - - Updated - - -

And isn't this the non-gaming card? And it beats the 1080 in gaming, with beta drivers? There might be a bright side to this...

[Fascinate]

Calling it now, power consumption ~double gtx 1080 numbers.

[Life-Binder]

http://pro.radeon.com/en-us/product/...river-download

DRIVER DOWNLOAD
Unlock the full potential of the Radeon™ Vega Frontier Edition with Radeon™ Pro Software

The Radeon™ Vega Frontier Edition is designed to empower Pioneers. Those who are not afraid to lead. To properly enable you, we are providing more driver options than ever before. You can choose between compute specific drivers for ROCm, and graphics drivers that allow you to both create using Radeon™ Pro workstation capabilities and game.

I Am Focused on Compute Workloads
The ROCm driver will be available for users on June 29th.

I Am Focused on Workstation Workloads
If you are a workstation user on a Windows® platform, get your driver here.
If you are a workstation user on a Linux® platform, get your driver here.

I Am Focused on Game Development Workloads
Radeon™ Vega Frontier Edition introduces “Gaming Mode”. This allows you to switch from Radeon™ Pro Settings to Radeon™ Settings where you can access gaming features for playtesting and performance optimization.
If you are focused on game development on a Windows® platform, get your driver here and switch to “Gaming Mode.”

I Just Want to Game
If you want to game on a Windows® platform, get your driver here and switch to “Gaming Mode.”

this is a non-gaming card, but it has a gaming mode (which should be the same or similar to the gaming RX Vega) which this user enabled when running his benches .. he also made sure it stayed at 1600 mhz during his runs

the driver isnt beta http://support.amd.com/en-us/downloa...indows+10+-+64 its just the first inaugural driver for Frontier


the one thing to keep in mind is that the RX Vega comes out in 1 month, so they have 1 more month to put the last tweaks on the gaming drivers .. how much will that bring we will see

- - - Updated - - -

And it beats the 1080 in gaming

seems to be about equal to 1080

[Lathais]

http://pro.radeon.com/en-us/product/...river-download



this is a non-gaming card, but it has a gaming mode (which should be the same or similar to the gaming RX Vega) which this user enabled when running his benches .. he also made sure it stayed at 1600 mhz during his runs

the driver isnt beta http://support.amd.com/en-us/downloa...indows+10+-+64 its just the first inaugural driver for Frontier


the one thing to keep in mind is that the RX Vega comes out in 1 month, so they have 1 more month to put the last tweaks on the gaming drivers .. how much will that bring we will see

- - - Updated - - -

seems to be about equal to 1080

Just because it has gaming mode drivers does not mean the hardware is the same as the gaming cards. It just means the driver will be more efficient for games on that card. Nothing more. It doesn't magically change the hardware in to something that will run games better. Can almost guarantee that the gaming cards will have different specs and run games even better.

Also, I could the article said he was using beta drivers. Yes, there are launch drivers available now, but when he did this, he was using beta, the article clearly states as much:
http://www.guru3d.com/news-story/amd...benchmark.html

It seems that PC and the beta driver is responsible for the somewhat lackluster score, and the GPU frequency is all over the place he mentions. Firestrike scored 17313 points, however that is the generic P score. if you look at the graph then the actual graphics score is 22916 though, which is say 10% above the GeForce GTX 1080. A GeForce GTX 1080 Ti / Titan Xp (Pascal scores roughly 28K on that GPU score. I'll update this article once he has some more results, and setup the card better.

Again, from the very articles you linked, it says 10% better than a 1080. Not equal to, 10% better.

[Life-Binder]

Just because it has gaming mode drivers does not mean the hardware is the same as the gaming cards.

its the same hardware, only with 8GB HBM2 instead of 16 HBM2 (and of course cheaper)

RX Vega may be clocked higher then Frontier, but that remains to be seen

are you expecting RX Vega to have more shaders or something ? think they made another die just for the gaming Vega ?



https://www.reddit.com/r/Amd/comment...removing_this/

he used the latest drivers

https://uploads.disquscdn.com/images...ng?w=800&h=320

in fact, those are the only Frontier drivers available to the public, where would he even get any kind of beta drivers .. ?



there are plenty of FS results from users for stock 1080s that match this Vega result

[Gaidax]

Beta or no beta, clocks or not, it is clear as day Vega is between 1080 and 1080Ti, IMO it is a huge disappointment especially considering just how expensive and power hungry it is with HBM and its power envelope.

Better drivers and some meager clock boost won't gap it to 1080Ti.

I have no idea how they thought to take on Volta with that.

[Fascinate]

Beta or no beta, clocks or not, it is clear as day Vega is between 1080 and 1080Ti, IMO it is a huge disappointment especially considering just how expensive and power hungry it is with HBM and its power envelope.

Better drivers and some meager clock boost won't gap it to 1080Ti.

I have no idea how they thought to take on Volta with that.

Nvidia put billions into development for pascal, AMD is like a mom and pop shop compatively. If its faster than a 1080 for less money its a big time win, even if it inevitably consumes more power.

[Zenny]

Yeah, my previous statement might have been optimistic, in Firestrike at least it seems to roughly match a 1080, not really exceed it. AMD has to price this card for $400 at most to even be worth it.

[Amalaric]

They all work the same. Bandwidth is just clock speed * bus width. Divide by 8 to convert bit to byte.. For this case it'd be 945 * 2048 / 8

GDDR5X supports speeds of up to 12GHz and the upcoming GDDR6 will support speeds of up to 16GHz and here we have HBM2 with a speed of 945MHz... that just seems paltry to me.

[Zenny]

GDDR5X supports speeds of up to 12GHz and the upcoming GDDR6 will support speeds of up to 16GHz and here we have HBM2 with a speed of 945MHz... that just seems paltry to me.

As mentioned, clock speed is not the entire story, bus width matters as well. GDDR5X/6 usually have a bus width of 256bit-384bit while HBM2 has a bus width of 2048bit. Also I think I was wrong in my earlier assessment, you double the 945 to 1890 for the Mhz of HBM.

So a card with 1890Mhz HBM2 and a 2048bit bus would have bandwidth of 484GB/s. Another card with 11000Mhz GDDR5x and a 384bit bus would have 528GB/s.

[Amalaric]

As mentioned, clock speed is not the entire story, bus width matters as well. GDDR5X/6 usually have a bus width of 256bit-384bit while HBM2 has a bus width of 2048bit. Also I think I was wrong in my earlier assessment, you double the 945 to 1890 for the Mhz of HBM.

So a card with 1890Mhz HBM2 and a 2048bit bus would have bandwidth of 484GB/s. Another card with 11000Mhz GDDR5x and a 384bit bus would have 528GB/s.

GDDR6 at 16GHz with a 512-bit bus would be 1024GB/s or did I do the math wrong?

[Bryntrollian]

seems to be about equal to 1080

But what is the hashrate like ?

[Life-Binder]

I read something like 32-35, whatever that means

[Zenny]

GDDR6 at 16GHz with a 512-bit bus would be 1024GB/s or did I do the math wrong?

It would be yes, equal to 2Ghz HBM2 on a 4096bit bus. The only advantage HBM2 now offers is less space used and less power consumed, it's not the magical performance boost some have claimed it to be, at least not yet.

[Remilia]

GDDR5X supports speeds of up to 12GHz and the upcoming GDDR6 will support speeds of up to 16GHz and here we have HBM2 with a speed of 945MHz... that just seems paltry to me.

You're looking at one thing without the other as another has mentioned before. Again the formula is (Clock speed * Buswidth / 8). The difference between GDDR and HBM is that a stack of HBM is 1024bit buswidth where as 1 module of GDDR is 32bit. In order to get 512bit buswidth you need 16 GDDR5 modules. They both have their own pros and cons so it really depends on use case scenarios.

[Amalaric]

You're looking at one thing without the other as another has mentioned before. Again the formula is (Clock speed * Buswidth / 8). The difference between GDDR and HBM is that a stack of HBM is 1024bit buswidth where as 1 module of GDDR is 32bit. In order to get 512bit buswidth you need 16 GDDR5 modules. They both have their own pros and cons so it really depends on use case scenarios.

It is physically possible to place sixteen GDDR5 / GDDR5X / GDDR6 memory modules around a single GPU but how many HBM modules (or whatever they are called) can you stack?

[Remilia]

It is physically possible to place sixteen GDDR5 / GDDR5X / GDDR6 memory modules around a single GPU but how many HBM modules (or whatever they are called) can you stack?

So far in practice we've gotten in 4 stacks which means up to 1TB of bandwidth. In theory more could be added but we haven't gotten any product with more than 4 yet.

[Amalaric]

So far in practice we've gotten in 4 stacks which means up to 1TB of bandwidth. In theory more could be added but we haven't gotten any product with more than 4 yet.

Ah, I see.

[Fascinate]

pcper is testing card right now.

In 3dmark it scored like 1000 or 1500 points below gtx 1080....not gtx 1080 ti, 1080...