ASRock Radeon RX 6700 XT Phantom Gaming D

[W1zzard]

The ASRock RX 6700 XT Phantom Gaming D is the first factory overclocked card that's actually faster than the AMD reference. All other competitors failed to deliver any performance gains. Not only out-of-the-box performance is good, overclocking is better than with any other RX 6700 XT, too.

Show full review

 

[tvamos]

Any pictures with the card powered-on? Would like to see if rgb is too much.

 

[altermere]

$800 for a 6700xt is truly pathetic. but hey, it's a steal compared to $1200 2080ti! just pretend we're still in 2018.

 

[Super XP]

MOST GPUs are MSRP Overpriced by quite a huge percentage.
And even worse right now with the GPU mining craze, just compounding on an already overpriced market place.

 

[Searing]

Wow, I never thought I'd see ASROCK at the top. Nice.

 

[Caring1]

Any pictures with the card powered-on? Would like to see if rgb is too much.

ASRock Radeon RX 6700 XT Phantom Gaming D 12GB OC

<b>Clock: GPU / Memory</b>, Boost Clock: Up to 2622 MHz, Game Clock: Up to 2548 MHz, Base Clock: 2457 MHz<br /><b>Key Specifications</b>, 7nm AMD Radeon™ RX 6700 XT Graphics, 12GB 192-bit GDDR6, AMD RDNA™ 2 Architecture, DirectX 12 Ultimate, PCI® Express 4.0 Support, 2 x 8-pin Power Connectors...

www.asrock.com

It also has a LED switch so they can be turned off.

 

[cellar door]

@W1zzard - you show what the card power limits are for different Nvidia cards models/cards, even graph them and extrapolate performance - yet, you do not for the 6xxx AMD cards.

Instead you show in the notes that there is "poor" OC" and only Asus had good results up to now. While its clear from looking at the Asus bios that and your power charts, its way upped and shows in it's review, the power limit has been substantially increased.

 

[W1zzard]

@W1zzard - you show what the card power limits are for different Nvidia cards models/cards, even graph them and extrapolate performance - yet, you do not for the 6xxx AMD cards.

Instead you show in the notes that there is "poor" OC" and only Asus had good results up to now. While its clear from looking at the Asus bios that and your power charts, its way upped and shows in it's review, the power limit has been substantially increased.

NVIDIA makes it easy to read the actual limits. It's complicated or impossible on AMD, because they change their BIOS structure for every release.

No power limit increase = fail ?

 

[cueman]

well..its highest clocked 6700 xt model, but looks rtx 3070 FE edition still beat it.
so we can guess that example AIB version, RTX 3070 GAMING X TRIO leading more...

 

[cellar door]

NVIDIA makes it easy to read the actual limits. It's complicated or impossible on AMD, because they change their BIOS structure for every release.

No power limit increase = fail ?

MorePowerTool can easily read the power tables from the bios, which are all in Techpowerup database or can be saved with your GPUZ.

There clearly is a power limit increase, that is why ASUS and AsRock are getting better performance. The power consumption charts show this already.

 

[W1zzard]

MorePowerTool can easily read the power tables from the bios, which are all in Techpowerup database or can be saved with your GPUZ.

There clearly is a power limit increase, that is why ASUS and AsRock are getting better performance. The power consumption charts show this already.

ASUS STRIX and ASRock Phantom Gaming are both 211 W and other settings are the same. It's not the power limit

 

[lukart]

I'm not sure the extra power and noise makes for the additional 2 fps? (2-3%) performance?
Did you increase the power limit as well on wattman? would be interesting to see how far it goes with all maxed out.

But I like the card, looks pretty good, not the question is, will it be available?

 

[W1zzard]

Did you increase the power limit as well on wattman? would be interesting to see how far it goes with all maxed out.

Makes no difference, I tested it, the issue is that AMD limited the sliders in Wattman and I'm already at max

 

[Deleted member 205776]

Literal phantom gaming.

 

[Solid State Soul ( SSS )]

Just look at that heat sink at this rate, few years from now, we would see GPUs with heat sinks the size of an air conditioner radiator.

 

[Deleted member 205776]

Just look at that heat sink at this rate, few years from now, we would see GPUs with heat sinks the size of an air conditioner radiator.

It's still somehow smaller than my 3070 Gaming X Trio, despite looking chonkier. GPU database confirms that

 

[cellar door]

Makes no difference, I tested it, the issue is that AMD limited the sliders in Wattman and I'm already at max

@W1zzard
That is exactly what it is - I have modded the power limit on a reference 6700XT. If I open up wattman while in game, on a static scene, and increase the power limit - the FPS will proportionally scale up, while the card will boost higher.


It is the exact same boost mechanism that nvidia uses - more power allows the card to boost higher, as long as temps are in check.

Can explain what you mean by your comment - that it's not the power limit. You provide no explanation as to why you think it is NOT the power limit.

Ps. I currently hold the #2 spot in the timespy graphic score (for single 6700X). This is simply done with upping the power limit with MorePowerTool, on a stock reference card.

Result not found

www.3dmark.com

 

[W1zzard]

that it's not the power limit. You provide no explanation as to why you think it is NOT the power limit.

Ah there's two issues here. In the review I'm specifically talking about the lack of out-of-the-box performance gains from the factory overclock.

The ASRock card has the same power limit value as other reference designs, yet one card shows performance gains at stock, all others do not

In the comment above, Wattman sliders for GPU and memory already at max, power at +0 -> x FPS, set power slider to max -> still x FPS
It's certainly possible though that manually adjusting the OverDrive tables can yield additional performance, I didn't test that

 

[cellar door]

Ah there's two issues here. In the review I'm specifically talking about the lack of out-of-the-box performance gains from the factory overclock.

The ASRock card has the same power limit value as other reference designs, yet one card shows performance gains at stock, all others do not

In the comment above, Wattman sliders for GPU and memory already at max, power at +0 -> x FPS, set power slider to max -> still x FPS
It's certainly possible though that manually adjusting the OverDrive tables can yield additional performance, I didn't test that

AMD continues to apply high stock gpu voltages, perhaps their binning process is more generic vs nvidia. Hence why they offer an automatic undervolt feature in their driver.

At stock power limit, as per what you did during the test, I do see meaningful FPS increase - this is most likely due my asic having lower leakage and needing less power to scale frequency higher.

Thanks for the reply, I understand better your comment/test. In my findings, the reference 6700XT benefits immensely from ~220-230watts GPU power limit vs the stock 186watts. At least with my example, this allows for sustained 2.8-2.85Ghz clock in any game - just like your ASUS strix example.

This is what I meant - that it is the stock GPU power limit of the Strix, from the factory, that allows for better performance you noted in your review.

 

[W1zzard]

that it is the stock GPU power limit of the Strix, from the factory, that allows for better performance

The strix has no performance gain at factory out of the box settings

 

[cellar door]

The strix has no performance gain at factory out of the box settings

The stock asus boost clocks vs reference are not far off. 2581 vs 2629 = 48mhz difference

Strix might simply not be utilizing all the stock power limit available to achieve those clocks - which your graph for voltage vs frequency shows, that at stock settings it falls near the same as a reference. And if it is using all of the stock power, the pcb design might have higher losses, your particular Asus example might have higher leakage etc.

211watts +15% = 242.6watts
186watts +15% = 213.9watts

There is a clear 30watts power advantage once you max out the OC and sliders - which is significant.

This is just my guess. Like I mentioned, my reference example with 230watts limit is able to get #2 spot on Timespy. It's clear that the stock card is starved for power. Try it with yours.

 

[W1zzard]

The stock a boost clocks vs reference are not far off. 2581 vs 2629 = 48mhz difference

So 2%, yet there's 0% gain. and the same is happening on 6 custom designs I've tested, only the asrock, the 7th, shows gains

power

of course, no doubt. my concern is with the fact that there's the promise of a factory overclock, but it doesn't have any gains

 

[seoguru]

Its weird that even after 2 years Asrock is the only card manufacturer that can honestly claim their cards are really built to overclock over reference cards when the rest can only manage at best plus 3% which is close to margin of error yet they are still claiming it till this day, its just a scam to say one care is at best slightly faster and I know for a fact Gigabyte OC cards are slightly slower than the eagle cards they claim are faster out of the box!