• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Navi21 XT/6800XT(?) allegedly scores more than 10,000 in Fire Strike Ultra

Recently leaked benchmarks for the RTX3070 have released showing a score of 8749 in Fire Strike Ultra settings.

In response to this, known leaker Patrick Schur with confirmed access to AIB models of Radeon 6800XT stated that the 6800XT reached > 10,000 in Fire Strike Ultra.



This was also backed up by famous leaker kopite7kimi who stated the following regarding the Navi 21 die in general:



Although this information is obviously unconfirmed officially, these are two very trustworthy sources.

This should for those paying attention, put the final nail in the narrative we see constantly spouted: "Navi 21 will only be around 3070 performance".

Granted, we already knew that the card AMD previewed on stage was a Navi 21 based card that was way out of the 3070's league so this is not exactly shocking information for anyone paying attention, but the narrative still seemed to prevail for some unknown reason. This should hopefully put it to bed for good.

For reference, the 3080 FE has a Fire Strike Ultra score of: 10667 - 10780ish

Exciting times ahead, not long now until we see the official reveal of these cards on the 28th of October.

EDIT: Can a mod fix the tags for this thread? I accidentally chose "Review" instead of "Rumour", thanks! Mod of War Mod of War , DGrayson DGrayson
 
Last edited:

GHG

Gold Member
Unfortunately we play games not benchmarks. It's easy for Nvidia/AMD to optimise their drivers for particular benchmarks if they wish.

So until I see real world game performance that is consistently good (as it is with Nvidia) I wont take any notice.
 

CrysisFreak

Banned
Unfortunately we play games not benchmarks. It's easy for Nvidia/AMD to optimise their drivers for particular benchmarks if they wish.

So until I see real world game performance that is consistently good (as it is with Nvidia) I wont take any notice.
Keep coping kiddo Su Bae gonna crush Nvidiass like there's no tomorrow.
Get ready and prepare uranus because team read finna unleash hell.
 
Unfortunately we play games not benchmarks. It's easy for Nvidia/AMD to optimise their drivers for particular benchmarks if they wish.

So until I see real world game performance that is consistently good (as it is with Nvidia) I wont take any notice.

While I do understand where you are coming from to a degree, in the sense that it is better to expect nothing and be pleasantly surprised than expect to reach the stars and only get to the moon.

I think you are also being a little dismissive of the information above and what we have heard about these cards so far. While I'll be the first one to tell you to always wait for reviews/real benchmarks from reputable places like Hardware Unboxed and Gamer's Nexus, which I still think is true, I also know that these synthetic benchmarks such as Firestrike are a rough ballpark representation of general performance tier.

Obviously these types of benchmarks are not the be all, end all when it comes to gaming performance they do give a general sense of the performance tier of a card. In the above case we can see a card that is clearly somewhere in the 3080 tier of performance, well beyond what a 3070 could achieve.

I mean given all of the information we have so far such as clocks, infinity cache, number of CUs etc... not to mention the tease AMD gave us at their Ryzen conference all of the evidence seems to point towards a 3080 competitor, in other words something well out of the reach of a 3070 and clearly a tier above it.

Can we at least agree on that and put the "Hurr durr, AMD can only compete with a 3070! lololol~" narrative to sleep once and for all?
 
Last edited:

Senua

Member
voejf1D.gif
 
Last edited:

dave_d

Member
When it comes to these cards the only 3 questions I really care about are

Can I actually get it?
How good is it?
How much is it going to cost?

Until I know the answers to those questions I have no idea which one anybody, myself included, to get.
 

ZywyPL

Banned
Unfortunately we play games not benchmarks. It's easy for Nvidia/AMD to optimise their drivers for particular benchmarks if they wish.

So until I see real world game performance that is consistently good (as it is with Nvidia) I wont take any notice.


Exactly. AMD is leading the synthetic benchmarks since I can only remember for both CPU and GPU, I've been hearing/reading for well over a decade than NV and Intel are packing their bags, and I just cannot believe people still fell for that. If anything, their products are the best examples how useless all those benchmarks are and how little they have in common with the actual games.
 

Black_Stride

do not tempt fate do not contrain Wonder Woman's thighs do not do not
if amd manage to make a gpu with performance between 3700 and 3800 with more vram and a price tag of 600 or less im in

If they can price it at 3070 levels then thyve got me interested.

Otherwise 100+ dollars for a bit more performance, but losing out on DLSS and DediRT hardware doesnt sound like a good deal.
Because once I reach 600 dollars I could stretch my budget to 700 and get a 3080.

Theyve got to do what Ryzen 3000 did in terms of pricing, basically have much better bang for buck.
Because, beating the RTX 3080 in bang for buck is going to be hard with a card thats marginally faster than a 3070 but 100+ dollars more expensive.
 

Croatoan

They/Them A-10 Warthog
Umm in Firestrike Ultra?

index.php





Your analogy is also really poignant.
5xLdbzq.jpg
How about a 2020 Ford GT (new gt40)? ;)

Also which mustang is that? Couple different models that have different horsepower ratings. I could see a tuned up rice burner beating a stock GT for sure.

Also this.

"Car & Driver testing showed the Mustang GT racking up a 12.6 second quarter mile at 115 mph. "

Looks like mustang losing in this particular test is a bit like MSM polling. Ineptitude or just outright fabrication.

Oh and how about this one.
But of coarse some Aussies can go fuck up a race and claim otherwise.

Child, your gotcha game is in need of work.
 
Last edited:

CrustyBritches

Gold Member
Nice! Hopefully we'll see some RT performance numbers soon. That's really what it's going to come down to for me.

Do keep in mind that the 3070 is basically a $499 2080ti. If Navi 21 XT is $499 then this would be a valid comparison. Otherwise, it doesn't make much sense.
 

Black_Stride

do not tempt fate do not contrain Wonder Woman's thighs do not do not
How about a gt40 ;)
The GT40 does the quarter mile in 13.2.
And if you meant the barely worth being called homologated GT MKII, then im embarrassed for you.

Mustang 5.0 vs Stinger GT S.
Both stock both driven by a racing driver.
So both the range topper cars and the Mustang lost.

Not to derail the thread but yeah AMD have certainly pulled their socks up for this one.


10K score in FireStrike Ultra is a really good score....granted its relatively easy to get an RTX 3080 well into the 12 - 13000 range.
But its something for them to show off in their spreadsheets during the presentation.
And im not sure why Twitter is upset at the RTX 3070 score.....we knew the RTX 3070 was at RTX 2080Ti levels of performance.....and the leaked scores pretty much confirm that.
 
RTX 3070 is currently the weakest 3000 series GPU, while 21 XT might be the most powerful big navi, we will see...
why not compare it to the 3080?

Its like AMD doesn't want to compete with the big dog.

In other news...

"Kia confirms new car will SMOKE a ford focus in the quarter mile!"

It seems some people didn't bother reading the thread. Up to now there has been a running narrative that whatever the top card AMD can produce will only be 3070 level...or MAYBE slightly better than 3070. We see this mentioned a lot in drive by posts in Radeon related threads regarding these cards. Despite the mountains of evidence to the contrary we tend to still see it pop up again and again.

Once the Fire Strike Ultra scores for the 3070 were revealed, someone well connected on Twitter mentioned that the 6800XT (There is also a 6900XT above this card) reached more than 10,000 in Fire Strike, thus dispelling that silly narrative (on top of the other mountains of evidence and common sense).

I quoted that person and also mentioned the 3070 scores for the context of what the quote was in relation to. I'll quote where I mention the relevant parts here.

benchmarks for the RTX3070 have released showing a score of 8749 in Fire Strike Ultra settings.
...
.... 6800XT reached > 10,000 in Fire Strike Ultra.
...
This should put the final nail in the narrative we see constantly spouted: "Navi 21 will only be around 3070 performance".
...
For reference, the 3080 FE has a Fire Strike Ultra score of: 10667 - 10780ish

This means that both the 3080 and 6800XT are both scoring > 10,000 in Fire Strike Ultra. Meaning that the 6800XT should be (roughly speaking) 3080 tier and compete with it, probably +/-5% vs 3080 depending on the title.

Granted a synthetic benchmark is not the be all, end all for performance but it is a very good ballpark indicator for performance.
 
If they can price it at 3070 levels then thyve got me interested.

Why on earth would AMD price a 3080 level performance card at the price of a 3070? That doesn't make any sense.

If that is your criteria then you will be sorely disappointed with pricing, but then your expectation was never realistic anyway.
 
If they can price it at 3070 levels then thyve got me interested.

Otherwise 100+ dollars for a bit more performance, but losing out on DLSS and DediRT hardware doesnt sound like a good deal.
Because once I reach 600 dollars I could stretch my budget to 700 and get a 3080.

Theyve got to do what Ryzen 3000 did in terms of pricing, basically have much better bang for buck.
Because, beating the RTX 3080 in bang for buck is going to be hard with a card thats marginally faster than a 3070 but 100+ dollars more expensive.

I imagine a scenario similar to Vega or RDNA 1.0 to be honest. Great cards, but not exactly the performance/price ratio people were expecting. I expect cards performing similarly to Nvidia equivalent, and priced accordingly.
 
Last edited:
How about a 2020 Ford GT (new gt40)? ;)

Also which mustang is that? Couple different models that have different horsepower ratings. I could see a tuned up rice burner beating a stock GT for sure.

Also this.

"Car & Driver testing showed the Mustang GT racking up a 12.6 second quarter mile at 115 mph. "

Looks like mustang losing in this particular test is a bit like MSM polling. Ineptitude or just outright fabrication.

Oh and how about this one.
But of coarse some Aussies can go fuck up a race and claim otherwise.

Child, your gotcha game is in need of work.

"Rice burners" are good cars, reliable and good economy vs performance, unlike most American cars, which have lot of faults, are expensive to maintenance, are laughably big + have big motors that are weak + drink way too much. (When you need huge V8 to get same amount of HP as tuned 2.0 engine, they are weak)

At least that is how it is seen outside of US.

Maybe AMD cards do The same now, less tflops but similar performance.
 

Croatoan

They/Them A-10 Warthog
The GT40 does the quarter mile in 13.2.
And if you meant the barely worth being called homologated GT MKII, then im embarrassed for you.

Mustang 5.0 vs Stinger GT S.
Both stock both driven by a racing driver.
So both the range topper cars and the Mustang lost.

Not to derail the thread but yeah AMD have certainly pulled their socks up for this one.


10K score in FireStrike Ultra is a really good score....granted its relatively easy to get an RTX 3080 well into the 12 - 13000 range.
But its something for them to show off in their spreadsheets during the presentation.
And im not sure why Twitter is upset at the RTX 3070 score.....we knew the RTX 3070 was at RTX 2080Ti levels of performance.....and the leaked scores pretty much confirm that.
Then why did Car and Driver get the Mustang into the 12s? Oh, and the 2018 Mustang is a 11 sec car. Hmm, the shitty aussie test seems like an outlier here... Maybe they had an axe to grind?!?!

Also the quarter mile for the ford gt (new gt40) is 10.8 seconds...Jesus you are wrong a lot!

2019 Ford GT

11 sec mustang
 
Last edited:

Black_Stride

do not tempt fate do not contrain Wonder Woman's thighs do not do not
Then why did Car and Driver get the Mustang into the 12s? Oh, and the 2018 Mustang is a 11 sec car. Hmm, the shitty aussie test seems like an outlier here... Maybe they had an axe to grind?!?!

Also the quarter mile for the ford gt (new gt40) is 10.8 seconds...Jesus you are wrong a lot!

2019 Ford GT

11 sec mustang

You are putting way too much work into this I dont give a fuck about the drag times of a fucking "DRAG MODE" Mustang or a damn Kia that im never going to buy, the GT vs GT video is just something I remembered when you made the analogy.........I already own both my dream cars. None of them are Fords or Kias.
Chill the fuck out.



This is a Navi21 Thread and my post was a direct response to this:
why not compare it to the 3080?

Its like AMD doesn't want to compete with the big dog.

In other news...

"Kia confirms new car will SMOKE a ford focus in the quarter mile!"

Navi21XT scored around 10K which is at or about RTX 3080 levels.
They are competing with the big dog.

The four door family sedan car from Kia, keeps up with Fords 5.0 Pony car........in both cases it looks good for the underdog.
 

OutRun88

Member
I love seeing Red and Green throwing blows at one another. Reminds me of the 00s.

Hopefully these are valid, and that the power translates to real world performance.
 

RoboFu

One of the green rats
Cool the closer or is to the 3080 the better. Only thing to do at that point is go into a price war.
 
Last edited:

RedVIper

Banned
Why on earth would AMD price a 3080 level performance card at the price of a 3070? That doesn't make any sense.

If that is your criteria then you will be sorely disappointed with pricing, but then your expectation was never realistic anyway.

Because they need to beat Nvidia on the perf/$ ratio. Because they lack DLSS. Because people who buy nvidia aren't going to switch for something just as good.
 
Top Bottom