Radeon RX Vega to be Air and Liquid Cooled - XL XT and XTX models

The power usage is important when in context to the rest of the chip. If RX Vega performs 20-30% faster than 1080Ti across the board, I don't think anyone would care about the power usage. But so far all FE testing indicates Vega will perform roughly between a 1070/1080 and using almost double the power to get that. And what's even more troubling is that Nvidia claims Volta is getting an additional 50% perf/w increase. Which means a GV104 with the same specs as a 1080 would be like 120w, vs the 350w of the Vega FE WC that only hits 1080 performance in like 1-2 games.

And I get obviously that a lot of this is subject to change with driver updates, bios changes, potential hardware changes with RX Vega - but so far it doesn't look good for the architecture. It also doesn't look good for AMD's margins - having to drop a liquid cooler onto a card you can only sell for ~$500 at the most while also having more expensive HBM2, definitely isn't ideal.

 

You´re right! I completely forgot about them... Feel free to mock me for this stupid mistake...:bang: Still i dislike the naming scheme and i wish they used something better.
Also if they are found of the past, them bring back the ATI name instead of AMD graphics...

 

OK, I am now going back to elementary to see why I do not think that this statement is from our universe.

And just fyi, there wes report claiming that liquid itself reached 100 degrees Celsius. IIRC we laughed at that too.

 

I don't think anyone really cares about power usage alone.

If it uses more power than Ti but is significantly slower, that's the issue.

Poor perf/watt

 

Yep, some of these were kings of the hill, where Nvidias own flagships could not touch (9800XT and x1950XTX). I can understand AMD trying to recapture the old ATI glory days with the new naming scheme.

 

Would you like to refute the statement that the VRM's got to excess of 90°C? Or are you going to start another sh*tposting sh*t storm again to derail the topic at hand in another one of your strawman troll fests?

 

The problem was Certainly not going to be fast flowing water, but rather bad contact of the pipe to the VRMs

 

The reason why I care about TDP is pretty simple. I hate noise and excessive heat coming from my PC.

 

I excuse you lack of elementary physics knowledge, but will advice you to read this.
http://www.guru3d.com/articles-pages/amd-radeon-r9-fury-x-review,12.html

I get that that thermodynamics missed you. But please, consider that HH could not measure high temperatures on back plate as fact, that this back plate works as heat shield.
Therefore temperature every Fury X with back plate reaches in given area of board should be considerably higher than those from your source.

And think about heat transfer, if back of PCB had 104 degrees Celsius, then VRMs on other side of it have much more. Every Fury X owner can tell you their VRMs temperatures under load... Or you can go through owners' thread and find them yourself.

And I wonder if ocn forum sourced those images from that amateurish French site. (Really can't check it easily on cellphone.)

 

It was poor design however. Though moving to fast may have been an incorrect reason and more on not enough surface area. It still doesn't take away from the VRM's were not adequately cooled.

With Vega appearing to consume 350W+ in the XTX variety I hope they have a better solution for cooling the VRM.

@Fox the image was the first FLIR image I saw on Google image that had the back plate removed.

Here are some more.

This is just the front cover taken off so the heat from the VRM is saturating the coolant tube over the VRM so poor contact can be ruled out. I had not seen anyone claim coolant hitting 100°C (that would cause pump failure no).

 

Please, heat complaining again...the event is labeled "Capsaicin". Absolutely no deceptive marketing involved. Plus there are stars and there are planets. Stars are hotter than volcanoes. When they get to a naming convention with ice and subzero temperature references we can then complain about how slow it is. Heat is all about motion and this thing moves. Does it matter it is not faster or as fast as a 1080ti...no. Those that wanted one of those already have it.

 

Review of the LC Vega FE:

https://www.pcper.com/reviews/Graph...ga-Frontier-Edition-16GB-Liquid-Cooled-Review

 

Nothing about this post makes any sense.

"you can't complain about perf/w because AMD marketed it's event/architecture names towards how poor it's perf/w is" - Raja/Slides say otherwise but clearly AMD thinks it's better to get their marketing across via metaphor in naming.

"you can't complain about temps because temps is 'motion'".. even though the entire argument is that the 'motion' is worse for the heat that's outputted.

"you can't complain about the performance relative to competition because 'no'" - I don't even know how to respond to this.

"Those that wanted one of those already have it." - despite the fact that I've personally seen hundreds of posts of people saying "i'm going to wait to see if Vega is better"

 

Obviously something was not right in a number of cases (I've not heard about this issue being anywhere near universal), from the look of the design if mounted correctly it should have had good contact area and plenty of ability to dissipate enough heat.
So must have been a mounting/TIM issue ?

 

Congrats, You found it. That's very image showing tubes carrying liquid reaching close to 100°C.

Guess what will be GPU temperature if its cooling medium reaches 100°C. Or if it reaches even 60°C.
YES! It will be Higher than 50°C HH measured. Yes, it will be higher than 52°C I have with radiator fan speed reduced to 11~14% (Auto).

And guess what? If cooling medium has such high temperature, Radiator itself will reach pretty close to that.
But HH measured 45°C on radiator, and up to 42°C on tubes (because they are insulated by plastic mesh).
And that corresponds with my readings done by IR thermometer.

So congratulations once more. Things you posted are nothing less than blatant lies.

 

If AMD is for poor people how do they afford the gigantic power bills caused by AMD cards? LOL AMD's TDP has never affected my power bill in any significant way.
The AMD enthusiasts could care less about TDP.
I've owned the 7970, 290X and 295x2. 300 watts on a card far more powerful than those cards is great to me.

 

Terrific review.

1500$ GPU is only able to beat Nvidia 1080 by a slim margin in a couple benchs...and lost in the rest by not so slim margin.

AMD gaming GPU drivers team must be praying in all know religions right now asking for miracles.

 

Considering the picture is showing the coolant pipe itself and not the backside of the PCB reaching 100°C that would indicate good contact. Not too sure what is going on here other than the current solution is not keeping up with the VRM's.

Yes it will if it is transferring the heat. I'm saying it's not transferring the heat adequately. Is the too hard for you to understand?

If the coolant pipe is hitting 100°C but the coolant temps is staying in the mid 40's then heat transfer is not happening.

 

That's same I want to know if cards are identical except cooler so we can simply remove and replace it with a full cover water block.

 

Vega has completely different cooling compared to Fury and looking at disassembled pictures, the block is covering VRMs as well, at least on water cooled Vega FE.
On top of that, you can check PCB analysis done on Vega FE and find out that VRM's on Vega are a litte bit over kill, we can just hope RX Vega will keep same VRM design.