AMD Radeon RX Vega Put Through 3DMark

[Fluffmeister]

980Ti vs Fury debate is awful, more so when you try and make the fury sound better than it is. The reference 980Ti is barely edged out by the fury, which sees no improvement in performance with AIB cards in normal situations.

The AIB model 980Ti's still consistently compete and often best 1070's

It's funny isn't it, I enjoy revisiting the comments section of the GTX 1080 review (among others), their main argument then was that was barely faster than AIB 980 Ti's.

Short memories I guess.

 

[cdawall]

It's funny isn't it, I enjoy revisiting the comments section of the GTX 1080 review (among others), their main argument then was that was barely faster than AIB 980 Ti's.

Short memories I guess.

Very short. I mean what do people expect when you further neuter Kepler...its also why I have stood behind 1070/1080 being midrange.

 

[Basard]

Que?

Well, that's how I understand it to be anyways. An Intel chip can run up to 95C and AMD's FX chips have something like 72C max. If both have a TDP of 125w you will have a hell of a time cooling your FX chip with a "125w" cooler because when they give a cooler a "wattage TDP" they are usually talking about "when used with Intel chips (because they can handle more heat)".... It's just marketing....

Or am I just unaware of some scientific way they use describe a cooler's cooling capacity?

 

[EarthDog]

Well, that's how I understand it to be anyways. An Intel chip can run up to 95C and AMD's FX chips have something like 72C max. If both have a TDP of 125w you will have a hell of a time cooling your FX chip with a "125w" cooler because when they give a cooler a "wattage TDP" they are usually talking about "when used with Intel chips (because they can handle more heat)".... It's just marketing....

Or am I just unaware of some scientific way they use describe a cooler's cooling capacity?

Noe i see what you were trying to get at!

Capacity of a rad/heatsink is its capacity, period.

Let me ask you this... which is hotter? A yellow flame from a lighter, or yellow flames from a bonfire? (A: both the same temperature)

Now let me ask..which has more energy that lighter or the bonfire? (A: The bonfire)

See where im going with this?

The temps are a product of that heatsink as well as many other variables...things like the core material, whats on the die making the heat, size of it, thermal paste under the ihs, the ihs, thermal paste above the ihs...now we get to the heatsink...then fans.

You cant really compare temps between amd and intel.

 

[Basard]

@EarthDog So, basically, if you put a 125 watt cooler on two different chips, then set up both systems to feed 125w exactly to each chip(FX vs Intel), the temps at load will be TMAX on both chips? How do they measure the capacity of any given cooler?

 

[EarthDog]

A heatplate with a specific load is how they are tested id assume. Not sure honestly.

As far as cpu temperature, that would vary by cpu and its tdp. In theory a processor with a lower tdp than the heatsink it shouldnt reach tjmax. And vice versa with higher. But again soooo many other variables...

Think of it the other way around...rad/heatsinks given capacity doesnt change (assuming no other variables change), yet, temperatures between the exact same model cpu vary, none the less completely different brands in amd and intel with two different processes/parameters for their substrate and getting the heat out to the ihs.

 

[N3M3515]

This is going to release over a year later than the 1080, at the same performance, but a lot worse perf/watt??
Damn AMD......... i expected over 1080Ti perf...

Price should be U$400 max.

 

[Footman]

Like Ryzen, AMD will have to come in at an attractive price point to sell Vega.

Ryzen has not proven to be any faster than the equivalent clocked and cored Intel cpu's, however it is hugely cheaper. I finally upgraded my own personal pc to Ryzen R5 1600X.

I don't expect Vega to set any speed records, but it needs to be at least as fast as the 1080 with a price somewhere between the 1070 and 1080 for success.

Firestrike scores show a similarity to the 1080, so now we need to hope for a price at or around the $450-$499 mark.

I am planning on buying Vega myself, I need better performance than the RX 580 to drive my 2560x1440 IPS 144hz Freesync monitor. I will not be happy if performance is at 1080 levels with a higher cost of entry along with higher power requirements.

Just saying...

 

[Aenra]

that gotta hurt for ppl that waited this long

Not at all.
While decidedly an increasing minority, i don't have the mindset of a juvenile, nor their customs or habbits. "Above 'X' FPS" performance is more than good enough for me, leaving it to brand preference, price vs performance and/or voicing my arguments as a customer in the only i have available to me; by paying. And i can assure you i will be paying for one of these.

People are free to stick to their pixelated pew pew, RGB lightz and 'does it come from Asus' mentalities, they are entitled to them.
They should however refrain from generalising. Not all of us are and think like "gamerz", thank God

 

[Brusfantomet]

And here I was hoping that the RX vega would be really good, guess it is down to pricing now.

Capacity of a rad/heatsink is its capacity, period.

At a given ΔT for the system, so the same cooler rated a 125 W for a chip with a ΔT of 80 degrees opposed to one with a ΔT at 60 degrees will be lower, it will actually be at 93,75 Watt. how you ask?
Look at the equation here. considering this is not LaTex we are using here i will simplify the Q-dot and m-dot signes with Q and m.

The equation then becomes this: Q =m*Cp*ΔT
where:
Q = cooling capacity [kW]
m = mass rate [kg/s]
Cp = specific heat capacity [kJ/kg K]
Δ T = the temperature change [K] from the cooled thing to the ambient air.

m and Cp are given by the cooler, in a normal tower cooler m can be increased with a fan blowing more air over the cooler, but for our experiment we keep the same fan on the same speed, this means that m*Cp is a constant.
ΔT for the first cooler will be 80 degrees (100 °C - 20 °C = 80 °C and °K) and Q for this ΔT is 0,125 kW, giving us:

0,125 = m*Cp*80

hence m*Cp = 0,125/80 = 0,0015625

now, using the lower ΔT (80 °C - 20 °C = 60 °C and °K) we get the flowing equation:

60 * 0,0015625 = 0,09375 and 0,09375 kW is 93,75 kW.

This means that a cooler rated at 125W for a ΔT of 80 °K will cool 93,75 W when the ΔT is 60 °K.

So no,

Capacity of a rad/heatsink is its capacity, period.

unless you have a fixed ΔT it is not.

Or am I just unaware of some scientific way they use describe a cooler's cooling capacity?

From Wikipedia: Cooling capacity

 

[EarthDog]

Yes, when you change variables its rating would change. Thats inferred, no? Oops, said... #131.

In other words, a heatsink rated at xxxW with a xxC delta will yield different temps on different cpus...but its the cpu (and all other variables) which is causing the difference in temperature between different cpus with the same heatload.

Heatsinks arent tested initially with chips, outside of computer modeling, its hot plates with specific heat loads. Its essentially saying, this chunk of metal can dissipate xxxW at a deltaT of xxC. When you put something under it at ths same heatload with different substrate materials, die sizes, paste, ihs, etc...its going to have a different temperature due to the OTHER variables. I think what you are trying to say is the amount of wattage a heatsink can handle will vary based on the max temp you want out of it... which is of course true...but not what im saying.

Am i missing something still?

 

[Basard]

@EarthDog I dunno... you started it lol....
@Brusfantomet Yeah! What you said, with the math numbers and stuff!

 

[EarthDog]

I wasnt asking you for this answer...lol!

 

[Th3pwn3r]

And here I was hoping that the RX vega would be really good, guess it is down to pricing now.




At a given ΔT for the system, so the same cooler rated a 125 W for a chip with a ΔT of 80 degrees opposed to one with a ΔT at 60 degrees will be lower, it will actually be at 93,75 Watt. how you ask?
Look at the equation here. considering this is not LaTex we are using here i will simplify the Q-dot and m-dot signes with Q and m.

The equation then becomes this: Q =m*Cp*ΔT
where:
Q = cooling capacity [kW]
m = mass rate [kg/s]
Cp = specific heat capacity [kJ/kg K]
Δ T = the temperature change [K] from the cooled thing to the ambient air.

m and Cp are given by the cooler, in a normal tower cooler m can be increased with a fan blowing more air over the cooler, but for our experiment we keep the same fan on the same speed, this means that m*Cp is a constant.
ΔT for the first cooler will be 80 degrees (100 °C - 20 °C = 80 °C and °K) and Q for this ΔT is 0,125 kW, giving us:

0,125 = m*Cp*80

hence m*Cp = 0,125/80 = 0,0015625

now, using the lower ΔT (80 °C - 20 °C = 60 °C and °K) we get the flowing equation:

60 * 0,0015625 = 0,09375 and 0,09375 kW is 93,75 kW.

This means that a cooler rated at 125W for a ΔT of 80 °K will cool 93,75 W when the ΔT is 60 °K.

So no, unless you have a fixed ΔT it is not.




From Wikipedia: Cooling capacity

What about constant temps versus peak? Cpu don't run at the same temperature forever is what I'm getting at. Ability to cool xxx watts of heat for eternity or for an hour? My wording isn't great but maybe you understand what I mean. Being able to sustain the cooling capacity of its maximum ability basically.

 

[Prima.Vera]

Price this to 1060 levels and they will sale like hotcakes.

 

[cadaveca]

The equation then becomes this: Q =m*Cp*ΔT
where:
Q = cooling capacity [kW]
m = mass rate [kg/s]
Cp = specific heat capacity [kJ/kg K]
Δ T = the temperature change [K] from the cooled thing to the ambient air.

Woah, buddy.

Your calculations are incorrect to this scenario, as they are for a closed-loop refrigerant-based cooling system. I did go to school for this stuff, so I saw this, I nearly spit my tea all over my desk seeing refrigerant calculations in this thread. You need a far more complicated equation; mass in your equation refers to the refrigerant flow, not airflow. You cannot increase "m" with a higher fan; that's not the purpose of this equation. "m" is the flow rate of the refrigerant, in this case, water, not the air across the rad (which is why it is rated in Kg/s, and not CFM). The Cp is actually the specific heat of the refrigerant, or the water in the loop, not the cooler. We use various refrigerants, which is why it works this way. Delta-T is the change across the evaporator (ie, from the inlet to the outlet), not the cooler's difference from ambient. Delta-T refers to how the refrigerant changes, and is not TD, which is what you at referring to. Delta-T always refers to temperature changes within the same media, not differences between two media. Many people get this bit wrong,


It's not often the stuff I learned in school I get to put to practice... but thanks.

 

[the54thvoid]

Woah, buddy.

Your calculations are incorrect to this scenario, as they are for a closed-loop refrigerant-based cooling system. I did go to school for this stuff, so I saw this, I nearly spit my tea all over my desk seeing refrigerant calculations in this thread. You need a far more complicated equation; mass in your equation refers to the refrigerant flow, not airflow. You cannot increase "m" with a higher fan; that's not the purpose of this equation. "m" is the flow rate of the refrigerant, in this case, water, not the air across the rad (which is why it is rated in Kg/s, and not CFM). The Cp is actually the specific heat of the refrigerant, or the water in the loop, not the cooler. We use various refrigerants, which is why it works this way. Delta-T is the change across the evaporator (ie, from the inlet to the outlet), not the cooler's difference from ambient. Delta-T refers to how the refrigerant changes, and is not TD, which is what you at referring to. Delta-T always refers to temperature changes within the same media, not differences between two media. Many people get this bit wrong,


It's not often the stuff I learned in school I get to put to practice... but thanks.

Thanks teach! The glaring thing to the layperson would be the use of Delta T. I hadn't read the initial post (saw equations and bugged out) but i read your critique of it and it was that misuse of Delta T that struck a chord. Think I've known since I bought my first cooler that it meant the change/difference in temp of the chip (or point of measurement on chip) from idle to full load.

 

[ratirt]

I see not much has changed in regards of posting. Always trying to prove who's right and who said it first and who predicted anything. kinda boring i'd say. I was hoping for a knowledge ride from all the people here who conside themselves as experts in all it's meaning. (some know more about a graphics chip than the company producing it does). Amusing predictions
it will be out soon than we can talk about it. 1080 was the Vega's perf point it always has been. Some even say they predicted that funny. How can you predict something which was stated by the company producing the chip and announced.
in my opinion the confusion is at the highest state now.
Vega was hitting on 1080 and it did. ( well from what we know so far). Maybe the power consumption isn't satisfying but well AMD was never great at that.
THE DELAY.
Well it is delayed and there was so many assumptions.
First HBM is so expensive and that might cause a delay in delivering the memory or simply the demand was big and lack of resources caused that.
Second driver issues. Well might be true since AMD is hiring people for the driver development but keep in mind that console market is in AMD's hands now. It's almost twice as much customers than PC market.
And third. They are tweaking Vega to boost performance to match or come close to 1080 TI. I'm sure AMD is at least trying to accomplish that. if they succeed I don't know but I think Vega has the potential and they can squeeze more outta it.

As for me all above can be true but also none can be right. There's so many things you people may not be aware of what's going on in AMD and that's the one fact here that's correct. Soon it will be out and it will all come clear

 

[las]

What denial? LMAO!

Can you read? Most of my posts here center around Vega being a big disappointment so far (And likely overall). But everything I said still stands:

• Vega Frontier's drivers ARE terrible. Learn how to read some reviews, there are bugs everywhere. This is a fact that the drivers are bad, what is opinion is if better drivers will improve performance.

• Although none of us are fortune tellers, it would be idiotic to think performance won't increase by a decent margin considering how bad they are now. In fact GCN 1.0 had gained so much performance from it's 12.11 drivers that TechPowerUp said "The 7870 felt like an entirely different card", and GCN had nowhere near the issues Vega clearly has buddy:

https://www.techpowerup.com/reviews/AMD/Catalyst_12.11_Performance/

• If GCN 1.0 could gain 10-20% in the first year, it is not insane to think Vega could gain the same or even more performance considering how big a departure this architecture is. Again, I am not saying Vega will become substantially stronger, but it is not at all crazy to think it could.

• The 980 Ti vs Fury debate is a dead horse. Stop beating it. The only thing I will say is that the Fury X is currently trading blows with the 1070 while the 980 Ti is treading water above he 390X. If you call that a victory, congratulations.

Custom 980 Ti beats Fury X with ease, out of the box. Custom 980 Ti is ~20% faster than reference, which Fury X is being compared to in 99% of tests. Fury X can't OC. You might gain 5% if lucky, with huge increase in power.

Fully clocked custom 980 Ti is pretty much on par with 1080 FE. 30-40% faster than reference here. Go see TPU's 980 Ti reviews for proof.

I had 1080 shortly, but returned it because it felt like a side-grade coming from 980 Ti @ 1.5+ GHz.

 

[Frick]

Just bringing my popcorn to see how the local AMD Defense Force will defend this.

Fisticuffs, obviously.

 

[I No]

Price this to 1060 levels and they will sale like hotcakes.

Hell let's hope they price it @ 1030's levels. Why not? that 1080 perf for $75 /s. All that HBM and cooling is gonna be a tad steep from a production's point of view.
Hard to think AMD will make any money off these. After all this time and they can't even match Maxwell's efficiency. Gotta hand it to them that PR got everyone hyped up and they delivered a lemon, same performance you could've gotten an year ago. Fun stuff.

 

[nemesis.ie]

Still don't quite understand why haven't they just slammed two Polaris GPU's on a single card with internal CrossfireX. Or just shrunk the Fury X and clock it higher and call it a day.

HBM is a waste of money a fat polaris card with GDDR5X would have happily competed with the 1080.

What about Hawaii?

Hawaii wiped the floor with Kepler and Grenada even managed to stay competitive with Maxwell.

Something is really strange here, if the performance doesn't get an uplift (or the intent of the card (chip) is really to compete in another market (AI etc.)) it's pointless (edit: for gaming, other than freesync/if you put in more than one, but then the potential heat issue gets worse).

My 2 x 290X get roughly the same FireStrike score, 2 x almost 3 year old cards at 600MHz less (each, that's 60% uplift on Vega's clock) and only 25% more power consumption for the same result and Vega does not have to use crossfire? Weird stuff. Although the die area is not much larger than Hawaii the compute performance is supposed to be > 2x ...

As mentioned 2 x RX580 would use the same or less power and produce a similar score for ~€500-600 at non-gouged RRP.

I can't believe all this money, R&D time/effort etc. would be poured into the new architecture to effectively end up with 2 x 290x/RX580 in one slot.

There must be more to this or gaming is just an afterthought "we can do it and it'll be good with freesync and no crossfire issues, but the main focus of Vega is pro/AI/compute".

Anyway we should know more over the next few days.

I'm still a little hopeful the numbers posted (if real) are for the lowest tier card/older API/old drivers.

 

[Brusfantomet]

Yes, when you change variables its rating would change. Thats inferred, no? Oops, said... #131.

In other words, a heatsink rated at xxxW with a xxC delta will yield different temps on different cpus...but its the cpu (and all other variables) which is causing the difference in temperature between different cpus with the same heatload.

Heatsinks arent tested initially with chips, outside of computer modeling, its hot plates with specific heat loads. Its essentially saying, this chunk of metal can dissipate xxxW at a deltaT of xxC. When you put something under it at ths same heatload with different substrate materials, die sizes, paste, ihs, etc...its going to have a different temperature due to the OTHER variables. I think what you are trying to say is the amount of wattage a heatsink can handle will vary based on the max temp you want out of it... which is of course true...but not what im saying.

Am i missing something still?

I think we are almost in agreement, my point was that if chip A i designed for a Tj of 95 °C while another is designed for 70 °C, if the power consumed by both chips are the same, say 125 W a cooler that is capable of cooling chip A at 125 W will exceed the Tj of chip B, therefore, your line in post #129:

Capacity of a rad/heatsink is its capacity, period.

becomes a bit wrong.

Then again, as cadaveca points out I am not completely correct myself. closes ting i get to cooling is on hobby basis with computers.

Also, so that not all of this post is OT, if the new RX vega can handle a higher Tj a single 120 mm rad could be enough.

 

[Slizzo]

And third. They are tweaking Vega to boost performance to match or come close to 1080 TI. I'm sure AMD is at least trying to accomplish that. if they succeed I don't know but I think Vega has the potential and they can squeeze more outta it.

Again, from what we've seen of Vega Frontier Edition, if you expect drivers to make up a 30%+ difference in performance you guys need to adjust your thinking.

 

[EarthDog]

I think we are almost in agreement, my point was that if chip A i designed for a Tj of 95 °C while another is designed for 70 °C, if the power consumed by both chips are the same, say 125 W a cooler that is capable of cooling chip A at 125 W will exceed the Tj of chip B, therefore, your line in post #129:

becomes a bit wrong.

Then again, as cadaveca points out I am not completely correct myself. closes ting i get to cooling is on hobby basis with computers.

Also, so that not all of this post is OT, if the new RX vega can handle a higher Tj a single 120 mm rad could be enough.

buuuut, the heatsink's ability doesnt change. Its the product underneath it which has the limits ARTIFICIALLY lowering its effectiveness. Again, the HS properties do not change all otber variables remaining tbe same. Heatsinks arent measured by what temps it can achieve on cpus, but on the amount of heat it can dissipate and the delta over ambient.

I digress as well.. OT.