Poll: ** The AMD VEGA Thread **

[N19h7m4r3]

yeah also
https://images.discordapp.net/attac...8049550073857/image.png?width=1230&height=229

So yeah it beats an xp but loses to a $900 professional card if the Xp beats it in games then there's an obvious trade-off tit for tat. If it beats the xp in games though

Kinda old news, I even compared it to their own Radeon Pro cards.
https://forums.overclockers.co.uk/posts/30889664/

FE is sitting at a great spot, specially if you also need and want FP16 performance; where NVIDIA only have two cards that can offer it. Telsa P100 and Quadro GP100

Vega FE
Catia = 135.75
CREO = 83.94
SolidWorks = 114.88
Cinebench OpenGL = 183.28

Radeon Pro WX 7100 ( £660)
Catia = 104.34
CREO = 82.08
SolidWorks = 111.55
Cinebench OpenGL = 197.11

Titan Xp
Catia = 107.29
CREO = 65.2
SolidWorks = 67.75
Cinebench OpenGL =169.72

 

[D.P.]

For gaming off course but these cards are supposed to be aimed at middle ground. Gaming plus light work then it's looking like Vega is better if the gaming performance matches up. Titan is aimed at a similar market so get 4 ordered and show us what's what

Titan is not aimed at a similar market at all, you would be better off with a Quadro card that can game but also has certified professional drivers and better performance for half the price.

 

[TwsT]

EDIT: AMD have also been very sly with their marketing slides. They have been listing peak FP32/16 performance which might not be that sustainable at all, going by their listed clock speeds the Vega FE air cooled would average 11.1TFLOP FP32, while the 1080ti at base clocks averages 11.1 but since they nearly always boost higher giving typically around 13TFLOP. Now the theoretical FP performance is not directly related to gaming, Nvidia cards have always performed better at a lower FP32 performance than AMD cards. IF AMD have done a lot of improvements we may see the tables turned.

Can't wait to see the reviews.

1080 Ti Base = 11.1 TFLOP 1080 Typical = 13 TFLOP
Vega FE Max = 12.5/13 TFLOPS Vega FE Typical = 11.1 TFLOP

Is that what you're saying?

 

[TNA]

No gaming benches yet?

Someone let me know when there is so I put it on the front page.

 

[TwsT]

Kinda old news, I even compared it to their own Radeon Pro cards.
https://forums.overclockers.co.uk/posts/30889664/

FE is sitting at a great spot, specially if you also need and want FP16 performance; where NVIDIA only have two cards that can offer it. Telsa P100 and Quadro GP100

Vega FE
Catia = 135.75
CREO = 83.94
SolidWorks = 114.88
Cinebench OpenGL = 183.28

Radeon Pro WX 7100 ( £660)
Catia = 104.34
CREO = 82.08
SolidWorks = 111.55
Cinebench OpenGL = 197.11

Titan Xp
Catia = 107.29
CREO = 65.2
SolidWorks = 67.75
Cinebench OpenGL =169.72

So assuming it's not junk at games it's a great $1000 jack of all trades.

It's a pro card so i'm not gonna hold my breath on gaming being good i'm just trying to understand where vega gaming offering might sit

I suspect it will be like the ti in that it offers less in precision work but has a a better sustained clock. and ofc it will be priced around $500-600


hmmm.

 

[D.P.]

Calling me a fanboy now really? I don't even own AMD hardware at the moment, and in the last two years have own GTX 980, Titan X, and two 980Tis. There's only one person here coming off as a Fanboy D.P

NVIDIA Clearly told this reviewer that they wouldn't supply him the cards because they're not for gaming; and list it for their neural network cards, and deep learning.

https://youtu.be/EYlMrxIgoME?t=5m44s

I'm not calling you a fanboy, I am saying that anyone who makes up blatant lies about a company will appear to be a fanboy, so please just stick to the facts.


Deep learning and neural networks have absolutely nothing to do with professional software.


I wouldn't trust what that reviewer says at all, seems like he has a chip on his shoulders. Lets take a look at the official Nvidia page:
https://www.nvidia.com/en-us/geforce/products/10series/titan-xp/

Nope, not seeing anything about professional use or that is not for gaming.



PS: Sicne you seem to think what card you own is some how important to the discussion, my main computer is a 2011 27" iMac with an AMD Radeon 6970M. I have owned far more AMD cards in my lifetime than NV, I buy whatever suits my needs at the time. I have absolutely zero brand preference.

 

[Nexus18]

And people still think the ignore function is pointless

 

[D.P.]

And people still think the ignore function is pointless

With junk like this posted, i can start to see the value.

 

[shankly1985]

I'm not calling you a fanboy, I am saying that anyone who makes up blatant lies about a company will appear to be a fanboy, so please just stick to the facts.


Deep learning and neural networks have absolutely nothing to do with professional software.


I wouldn't trust what that reviewer says at all, seems like he has a chip on his shoulders. Lets take a look at the official Nvidia page:
https://www.nvidia.com/en-us/geforce/products/10series/titan-xp/

Nope, not seeing anything about professional use or that is not for gaming.

Why do you waste so much of your free time on a product/thread for something you will never buy even if it was so much better than the competition? We all know you just here for Damage control Am sure they is some excellent Summer steam sales your GPU will love to be playing right now.

 

[N19h7m4r3]

So assuming it's not junk at games it's a great $1000 jack of all trades.

It's a pro card so i'm not gonna hold my breath on gaming being good i'm just trying to understand where vega gaming offering might sit

I suspect it will be like the ti in that it offers less in precision work but has a a better sustained clock. and ofc it will be priced around $500-600


hmmm.

Yup, I myself stated a few times it's a real jack of all trades master of none. The main thing missing as you mentioned is gaming performance.
If it's performance there can get close to it's current shown numbers it'll be a fantastic card.

AMD's slogan was "For the creators who game, and the gamers who create" a while back. We'll see if it can suit it all.

Although, for A.I development, I can see some places pick up some Frontier Editions.
Peak 26 TFLOPs of FP16 compute; that's more than the Instinct MI25's 24 TFLOPs; and a lot cheaper than buying some Tesla P100's or Quadro GP100s.

It's a funky niché card that'll have to certainly earn it's spot. Either way, we need to wait for gaming results, and for the Radeon RX Vega card that doesn't have these hybrid drivers.

 

[N19h7m4r3]

I'm not calling you a fanboy, I am saying that anyone who makes up blatant lies about a company will appear to be a fanboy, so please just stick to the facts.


Deep learning and neural networks have absolutely nothing to do with professional software.


I wouldn't trust what that reviewer says at all, seems like he has a chip on his shoulders. Lets take a look at the official Nvidia page:
https://www.nvidia.com/en-us/geforce/products/10series/titan-xp/

Nope, not seeing anything about professional use or that is not for gaming.

As posted before, NVIDIA launched it at an A.I conference, and have it listed for their deep learning cards.
Nothing about gaming here.
https://www.nvidia.com/en-us/deep-learning-ai/developer/

The Titan Xp is a Prosumer card according to what NVIDIA advertise and show it as; just like the Frontier Edition is.

• NVIDIA TITAN Xp - The fastest accelerator for deep neural network training on a desktop PC based on the revolutionary NVIDIA Pascal™ architecture

 

[D.P.]

Why do you waste so much of your free time on a product/thread for something you will never buy even if it was so much better than the competition? We all know you just here for Damage control Am sure they is some excellent Summer steam sales your GPU will love to be playing right now.

Because I am interested in the technology. I would absolutely buy Vega if it suited my needs, it might well end up ina one of our servers at work if the FP16 performance is up to to scratch.

We all know you are just here to bait.

 

[D.P.]

As posted before, NVIDIA launched it at an A.I conference, and have it listed for their deep learning cards.
Nothing about gaming here.
https://www.nvidia.com/en-us/deep-learning-ai/developer/

The Titan Xp is a Prosumer card according to what NVIDIA advertise and show it as; just like the Frontier Edition is.

Are you unable to understand plain English? Deep learning has absolutely nothing to do with professional software. Repeat after me, "Deep learning has nothing to do with professional software"



It doesn't matter how many times you repeat a lie, it doesn't become fact. It is bad another wit fake news and alt-facts taking over the political media, why the heck do people try and do the same in a discussion about graphics cards I have no idea.

 

[shankly1985]

Because I am interested in the technology. I would absolutely buy Vega if it suited my needs, it might well end up ina one of our servers at work if the FP16 performance is up to to scratch.

We all know you are just here to bait.

Not at all I just noticed you keep posting the same rubbish over and over again even though you have been corrected. You will never buy an AMD GPU I know that and everyone else knows that. You just wasting your time. This isn't about Nvidia I dont care what Nvidia can and cant do I come in here to read about Vega not Nvidia! So go and post that rubbish elsewhere.
Thank you

 

[Sargatanas2511]

Can I ask how many people that do deep learning aren't professionals?

 

[Nexus18]

Shankly, you pretty much said it yourself, damage control..... he has even said before that the company he works for is invested in nvidia hardware....

Honestly, just add people like that to the ignore list, it makes reading this forum so much better and quicker/easier to get factual information rather than a load of verbal diarrhea, there is no point trying to change their minds even if you have all the hard evidence you need like the above image as they have an emotional/personal/financial investment in such companies and will do or/and say whatever there is to put out the fires so to speak.... It's always the same old on here, big up a certain company when they do really well at something, come new gen, the other company does even better and then suddenly it is not such a big deal and find something else to nit pick at zzzzzzzzzzzzzzzzzzzzzzzzzzzzz

Better of putting that wasted time bickering on an internet forum into a hobby you enjoy such as gaming

And +1, couldn't care less about professional/deep learning and all that..... I imagine that 95% of the people on here only care about gaming performance..... With 4% of those people only wanting internet epeen points and the 1% left having a genuine interest in it.....

 

[N19h7m4r3]

Are you unable to understand plain English? Deep learning has absolutely nothing to do with professional software. Repeat after me, "Deep learning has nothing to do with professional software"



It doesn't matter how many times you repeat a lie, it doesn't become fact. It is bad another wit fake news and alt-facts taking over the political media, why the heck do people try and do the same in a discussion about graphics cards I have no idea.

What lie? What are you on about D.P?

Neither the FE nor Xp are "Professional" cards like the Quadros or Radeon Pros.

The only one lying ( at best being confusing in their marketing ) here is NVIDIA. On one hand stating it's not for gamers; and publicly doing so for reviewers, and then listing gaming performance on their site as well.

Also can you tell me which games and consumer uses there are for Deep Learning, Neural Network Training; along with A.I development? Since it's clearly not something a professional will want to use.

 

[D.P.]

Not at all I just noticed you keep posting the same rubbish over and over again even though you have been corrected. You will never buy an AMD GPU I know that and everyone else knows that. You just wasting your time. This isn't about Nvidia I dont care what Nvidia can and cant do I come in here to read about Vega not Nvidia! So go and post that rubbish elsewhere.
Thank you

What rubbish have I posted, where have I been corrected with actual facts? Please show me, I am all for learning. Lets see this evidence? Quite the opposite, I have been presenting facts that have supported what I have been saying. For example, I gave caution to people about the clock speeds and then AMD themselves have collaborated the fact that the typical clock speed is going to be less. A recent example where I did make a mistake in assuming Vega would have have full speed FP64 support somebody showed me I was wrong and I readily accepted it and apologized.


I don't really care about the rubbish you post. So stand up and show me where I have posted rubbish and been proved wrong, or move on. What I wont do is agree to somebody else's lie.

 

[D.P.]

Can I ask how many people that do deep learning aren't professionals?

Just because one is a professional does not mean they use what is referred to as professional software within the industry.


Deep learning is part of the whole HPC and data center product line.

Professional software refers to things like AutoCAD, Solid Works, 3DSmax etc and other visualization software. It has nothing to do with deep learning or HPC.

That might well be where some posters are getting confused though.

 

[Samwell]

Can I ask how many people that do deep learning aren't professionals?

Professional is the wrong word here, as both are professionals, but they do totally different work. Workstation/Cad people don't use Titan X and Deap Learning People don't use Quadros. Both are totally different market segments. You won't find many Workstation People which ever worked with DL. The Vega FE is also beating the Titan in Deap Learning workloads, which is a good comparison, but cad work isn't done on titans and it's a useless comparison by amd. Especially as even a Quadro4000 seems to beat it.