• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

ASUS Radeon RX 5700 XT STRIX OC

For example we have Ace Combat 7, which is a non-mainstream game and the performance of AMD cards there is pathetic (RX5700 is 10% behind a 1660Ti, WTF), while a more mainstream racing game, Forza is not included in the test (where AMD cards lead NV counterparts by nearly the same percentage as in AC7)
I wish @W1zzard would elaborate on the former. (not including something is understandable, including non-mainstream with harsh perf penalty is not).
 
I wish @W1zzard would elaborate on the former. (not including something is understandable, including non-mainstream with harsh perf penalty is not).
Ace Combat 7 was released, giving us a "flight sim" option for our benchmarks. So I added it, didn't know anything about performance numbers until I retested it on all cards. Now it sucks on AMD, do I remove it for that reason?

Forza is only available on the horrible Microsoft Store, which I refuse to use
 
But now you can overclock custom designs as well ;)


While that is true, Navi has closed some of the gap between AMD and Nvidia: 5700 is slightly more efficient than the 2060, while 5700XT is within spitting distance of 2070S that sells for $100 more (it's not as efficient as its sibling, but that's another story).
Navi doesn't put on Nvidia the pressure we need, but other than that, it turned out better than I was expecting.


Barely... Proper 2060 Super and 2070 Super custom cards does ~10% on top of the ~5% out of the box compared to Founders Edition.

Tbh I'm not impressed with any of the these cards and I'm not buying anything before next year. My 1080 Ti still holds up and performs like a RTX 2080 or so.

Ampere 7nm is going to be a big leap.
 
Now it sucks on AMD, do I remove it for that reason?
It depends on how you chose games for the set.
My understanding was that they should be representative of what typical gamers plays.
Regardless, 2070S being 50% faster in that bench should be raising eyebrows.

Forza is only available on the horrible Microsoft Store, which I refuse to use
:(

Barely... Proper 2060 Super and 2070 Super custom cards does ~10% on top of the ~5% out of the box compared to Founders Edition.

Which one of them?
TPU saw 4% GPU overclock on 2070S ASUS Strix OC.
 
It depends on how you chose games for the set.
My understanding was that they should be representative of what typical gamers plays.
Regardless, 2070S being 50% faster in that bench should be raising eyebrows.


:(



Which one of them?
TPU saw 4% GPU overclock on 2070S ASUS Strix OC.

All of them. Look at the Overclocking section and compare the OC number with the stock number.
It's 12-15% on pretty much all cards tested, compared to founders.

Now do the same with the two custom 5700 XT cards tested here.
5% perf gained compared to reference.
 
One of the things that whelm me about comparing GPUs in forums is the lack in discussions of price factor. Below are 2 examples

https://www.newegg.ca/sapphire-rade...5700XT&cm_re=RX_5700XT-_-14-202-341-_-Product



There are no AIB cards available yet but I have heard $10 to $50 premiums, even then they are a great deal

I looked for the cheapest RTX 2070 Super I could find. The Founders edition is over $1000.00 at 50% off


This is why for me anyway that the 5700 and 5700XT are absolutely fantastic cards. Just as fast as the Vega 7 at 1/2 the cost!!?? What is not to like. As much as people like to these cards are supposed to take on the the 2060 to 2080 non super cards.
 
TPU saw 4% GPU overclock on 2070S ASUS Strix OC.
All of them. Look at the Overclocking section and compare the OC number with the stock number.
It's 12-15% on pretty much all cards tested, compared to founders.

giphy.gif


Let me repeat for particularly slow: TPU saw 4% GPU overclock on 2070S ASUS Strix OC.

See that 4%? That's a figure. That is by how much GPU was overclocked.
See that 4% has color different than the rest of the text? That is because it is a link (to the respective article).

It ain't hard, is it?

Now, the MSI X Trio: TPU saw 3% GPU overclock on it.
See that 3%? That's a figure. That is by how much GPU was overclocked.
See that 3% has color different than the rest of the text? That is because it is a link (to the respective article)

Let me know if you have further questions.
 
giphy.gif


Let me repeat for particularly slow: TPU saw 4% GPU overclock on 2070S ASUS Strix OC.

See that 4%? That's a figure. That is by how much GPU was overclocked.
See that 4% has color different than the rest of the text? That is because it is a link (to the respective article).

It ain't hard, is it?
Let me know if you have further questions.


I'm talking about PERFORMANCE, not GPU OC percent, because WHO CARES. You can't use that for ANYTHING because of GPU Boost and power limiters. You new to this or something?

Performance GAINED is 14.2% compared to founders edition.

Oh yeah... You are absolutely clueless I see.
 
I'm talking about PERFORMANCE (joking, I actually have no freaking idea what I'm rambling about)

giphy.gif


No problem dude. See this picture?

relative-performance_2560-1440.png


See the blue line in it? That's how FE card performs. See the green/blue ones? Those are Strix OC ones. If you substract figures that are to the left of % sign, you'd get rough idea of the performance difference.

And one more time, MSI:

relative-performance_2560-1440.png



See that blue bar? That is how FE version performs.
See the green one? That's MSI Tri X something something performance. Again, to the left of % sign, you can see figures, If you substract figures that are to the left of % sign, you'd get rough idea of the performance difference.

Let me know if you have more questions!
 

I'm talking about PERFORMANCE, not GPU OC percent, because WHO CARES. You can't use that for ANYTHING because of GPU Boost and power limiters. You new to this or something?

Performance GAINED is 14.2% compared to founders edition.

Isn't the 2070 Super on the same Turing die as the 2080? Comparing a full Turing card to a cut down one would at least yield what you are talking about. I notice that you seem to be not impressed with Navi. I expect that if Navi were to give the 2080TI performance for the price that you might be mildly impressed.
 
Regional pricing variations aside, the 5700XT exists in its market segment below the 2070S at $500 MSRP.

No amount of overclocking, overvolting, and ambitions cooling is going to make a 5700XT a threat to even the base-model 2070S cards. Even if on paper, the 5700XT only needs a 15% overclock to match a 2070S, that'll be at the screaming, near-death limit of even this overkill cooler and VRMs of this STRIX model. And that's a bone-stock, reference/entry 2070S model before its own overclocking - the very worst of which can still overclock by another 150MHz (better/luckier samples are seeing 250MHz+ overclocks).

Some people want to pay a premium for flashy lighting and factory overclocked performance but that premium is worthless if it costs more than an objectively better product that's higher up the food chain. If I can use a car analogy, there's no point buying $20K of aftermarket bolt-ons and cosmetics for a cheap car when you could just buy a car that's $20K more expensive in the first place with a better engine, chassis and handling right out of the factory. Not only is it nicer to look at, drive, and own - it also is a better starting point if you want to start modifying it down the line.
 
Regional pricing variations aside, the 5700XT exists in its market segment below the 2070S at $500 MSRP.

No amount of overclocking, overvolting, and ambitions cooling is going to make a 5700XT a threat to even the base-model 2070S cards. Even if on paper, the 5700XT only needs a 15% overclock to match a 2070S, that'll be at the screaming, near-death limit of even this overkill cooler and VRMs of this STRIX model. And that's a bone-stock, reference/entry 2070S model before its own overclocking - the very worst of which can still overclock by another 150MHz (better/luckier samples are seeing 250MHz+ overclocks).

Some people want to pay a premium for flashy lighting and factory overclocked performance but that premium is worthless if it costs more than an objectively better product that's higher up the food chain. If I can use a car analogy, there's no point buying $20K of aftermarket bolt-ons and cosmetics for a cheap car when you could just buy a car that's $20K more expensive in the first place with a better engine, chassis and handling right out of the factory. Not only is it nicer to look at, drive, and own - it also is a better starting point if you want to start modifying it down the line.

In my opinion the best thing to do with Navi is get a reference card (when they drop in price) and put a water block on it. You may not see an improvement in overall speed but the noise issue would be gone and the "heat" issue would become a non factor.
 
giphy.gif


No problem dude. See this picture?

relative-performance_2560-1440.png


See the blue line in it? That's how FE card performs. See the green/blue ones? Those are Strix OC ones. If you substract figures that are to the left of % sign, you'd get rough idea of the performance difference.

And one more time, MSI:

relative-performance_2560-1440.png



See that blue bar? That is how FE version performs.
See the green one? That's MSI Tri X something something performance. Again, to the left of % sign, you can see figures, If you substract figures that are to the left of % sign, you'd get rough idea of the performance difference.

Let me know if you have more questions!

Hahah. You know there's a difference between out of the box performance and OC'ed performance right?

That's what the Overclocking section is for. Less memes, more thinking.


Actual 3D performance gained from overclocking is 0.7%.
 
Regional pricing variations aside, the 5700XT exists in its market segment below the 2070S at $500 MSRP.

No amount of overclocking, overvolting, and ambitions cooling is going to make a 5700XT a threat to even the base-model 2070S cards.

It can't beat or come close because... it is pricier?

giphy.gif


Are you from planet Earth, son?

relative-performance_2560-1440.png


Hahah. You know there's a difference between out of the box performance and OC'ed performance right?

That's what the Overclocking section is for
Yep. That's what 4% and 3% post was about.
 
In my opinion the best thing to do with Navi is get a reference card (when they drop in price) and put a water block on it. You may not see an improvement in overall speed but the noise issue would be gone and the "heat" issue would become a non factor.

And waste money on a waterblock, for a mid-end GPU..?

giphy.gif


relative-performance_2560-1440.png



Yep. That's what 4% and 3% post was about.

...and 2070 Super Founders/Stock is still 5% faster than custom 5700 XT (adding in the 0.7% perf gained from additional OC on the 5700 XT - An Overclockers Dream).
 
And waste money on a waterblock, for a mid-end GPU..?

You are calling a card that is the fastest card that AMD has released Mid range.....vs let me guess the 2070S, 2080, 2080TI cards that are all more expensive. Water blocks were released for Navi as soon as they were released we had to wait a year to get them with Vega so the manufacturers of those parts knew something.

You may think that it is a waste of money but it is not that expensive and does lead to noticeable improvement in terms of voltage, fan noise and cooling potential. If you already have a loop and could get a block for $100 and a reference 5700 or 5700XT for $300-$350 it would be the same as getting an AIB card in cost.
 
..and 2070 Super Founders/Stock is still
Nearly 100 bucks more expensive and yada yada, bu tit has RTX!

Yeah. Good that you've figured OC-ing on 2070s was way below 10-15% that your imagination has drawn, (assuming 3-4% qualifies as "well below 10-15%", right?)

Forza is only available on the horrible Microsoft Store, which I refuse to use
I shouldn't have checked test results for Forza...
And when people think you are biased in green favour, just know that you give good reasons to think so. (and no, I'm not buying "I refuse to use Microsoft Store (on a bloody TEST MACHINE)")
 
Last edited:
It can't beat or come close because... it is pricier?
Are you from planet Earth, son?
Your post makes no sense. The previous four pages of this thread are discussing the potential price of this STRIX OC model.
I'm saying that no 5700XT can be worth over $500 because at that point it is outclassed by the 2070S, which starts at $500. What's hard to understand about that?

Anyway, that TPU chart you linked shows that the slowest possible 2070S you can buy is still 6% faster than this extra-bling Asus 5700XT STRIX.

TPU have reviewed six 2070S cards, the WORST of which can be made 8.4% faster on top of that. That's not me claiming that, just read the TPU 2070S reviews - they all show the overclocking results ranging from 8.4% to 14.2% faster than a stock, reference 2070S. I know overclocking is a crapshoot depending on the silicon lottery and all that, but taking the worst possible result means that the cheapest 2070S is going to be 106% x 108.4% = 15% faster than the 5700XT STRIX OC.

If you weren't fussed about the RGBLED, that means the $500 2070S is worth 15% more than the performance the 5700XT STRIX OC offers, putting the value of the 5700XT STRIX OC at $435 at best. If Asus are charging more than that once the price is announced, people will need to decide how much the Asus brand name and RGBLED bling are worth for themselves.
 
Last edited:
I'm saying that no 5700XT can be worth over $500 because at that point it is outclassed by the 2070S, which starts at $500. What's hard to understand about that?
Oh, mea culpa, I actually am guilty of not reading 4 pages, and agree with your assessment. (qualitatively, as far as figures go you are off in green camps favour, but it doesn't matter)
 
I wasn't intentionally trying to bias towards the green team, I just pulled the numbers from TPU's reviews.

Value or performance/dollar is always better the lower down the food chain you go, so if I compare against a $149 RX 570, everything looks bad. Obviously, the problem with that is that an RX 570 doesn't really cut it above 1080p and therefore not really as relevant to this discussion!

I'm also trying to avoid comparing to Vega, and the non-super cards these days as stocks are vanishing so they'll soon be irrelevant in a purchasing context.

What will really matter is the street price of these cards once initial availability and demand balance out.
 
I wasn't intentionally trying to bias towards the green team, I just pulled the numbers from TPU's reviews.

Value or performance/dollar is always better the lower down the for chain you go, so if I compare against a $149 RX 570, everything looks bad. Obviously, the problem with that is that an RX 570 doesn't really cut it above 1080p and therefore not really as relevant to the discussion.

I'm also trying to avoid comparing to Vega, and the non-super cards these days as stocks are vanishing so they'll soon be irrelevant in a purchasing context.
You need to read more. He wasn't saying you were biased towards the green team, he's saying TPU is. It surfaced somewhere that TPU doesn't benchmark Forza (which presumably runs better on AMD) so now they have a new argument for their conspiracy theory.
 
Ahhh, okay. I don't worry too much about specific titles swaying the numbers a few percent either way. If you play one specific game exclusively, then look at the best card for that one game. Otherwise the averages are just a snapshot of some current (ish) titles that cover a range of engines and APIs. For every game that runs better on Nvidia there will be a different game that runs better on AMD.

Personally, if one particular game sucks on one particular card, I have the luxury of just switching to the other team by moving to the other room in my house. Destiny 2 ran like crap on my old Vega, for instance - but the gtx 970 was smooth as butter...
 
  • Like
Reactions: bug
hah I refused to play Forza because of the exact same reason as W1zzard. MS store needs to die.
 
  • Like
Reactions: las
Ahhh, okay. I don't worry too much about specific titles swaying the numbers a few percent either way. If you play one specific game exclusively, then look at the best card for that one game. Otherwise the averages are just a snapshot of some current (ish) titles that cover a range of engines and APIs. For every game that runs better on Nvidia there will be a different game that runs better on AMD.

Personally, if one particular game sucks on one particular card, I have the luxury of just switching to the other team by moving to the other room in my house. Destiny 2 ran like crap on my old Vega, for instance - but the gtx 970 was smooth as butter...

Jup generally if you play any indie game that happen to be good (Ace Combat 7, Frostpunk, Road to Eden, Original Sin 2, etc...) best to go with Nvidia, chances that those game are running DX11 and don't even need Nvidia optimized driver to run well. If you only play shitty loot box ridden AAA games that EA and Activision pump out every year (Battlefield, Star Wars, Call of duty, etc...) then Navi is a fine choice.

There are plenty of games that sit at the top of steam players count and are hardware killers but not often show up in reviews:
ARK: Survival Evolved, No Man's Sky, Arma 3, Rust, etc...guess which one favor Nvidia ?

Also with the current rise of Epic game store, there are more and more AAA titles running UE4 and Nvidia just owns those games. New UE4 games coming out are Borderlands 3, The Outer Worlds, FF VII Remake, S.T.A.L.K.E.R 2, Outriders, System Shock, Vampire: The Masquerade - Bloodlines 2, etc...
 
Last edited:
  • Like
Reactions: las
Back
Top