• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Possible Radeon 390X / 390 and 380X Spec / Benchmark (do not hotlink images!!!!!!)

Status
Not open for further replies.
Boomstick777 said:
What you're comparing there is the 290X a near 2 year old design (GCN 1.0) VS Nvidia's latest architecture (Maxwell). It's not just the spec but the architecture used as well, comes with inherent improvements.

I.e new generation Maxwell cores VS GCN 1.0. Even then the near 2 year old 290X does pretty well against the GTX 980 especially at higher res.

So with Fiji and the new architecture not only do you have better spec in every metric (Except volume of memory) but also GCN 1.2 'newer architecture'.

Expect it to be quite a leap. Whether it can beat Titan X we will find out very soon. My money is on 10% - 15% faster than Titan X in all but Gameworks titles.
Click to expand...

Gcn 1.1 not 1.0 cough cough.
;)
 
There was a rumour kicking around a while ago that it was as fast as a 295x2, so 15% faster than a Titan X only with no driver reliance like the X2.

Honestly it's quite amazing just how little respect people in this thread are giving HBM. From what I have been reading it absolutely slaughters GDDR. Not only that, but I have reliable sources telling me that 4gb is absolutely plenty. Maybe it's because it's stupid fast, so can flush and replace very quickly?

Don't know, but if it is what it's cracked up to be it will make Titan X and the 980ti look like old tech.

I've been saying it for ages, but I'll say it again. There's a reason Nvidia rushed out the TI and it's not because they wanted to release first. All they've done is pretty much cease the sale of the Titan X overnight, shaving 40% off of the price of each card sold and Nvidia are not known to undercut themselves.
 
Davedree said:
But equally so is gm200, if these results are truthful
Click to expand...

Yeah but for instance in Witcher 3 it would just work without loads of crashing and the need for multiple new unstable drivers before you get a working one.

Not even close to the first time Nvidia's new game support is unstable as crap.

CAT-THE-FIFTH said:
Yeah,but the GPU is smaller with Fiji it appears and if AMD have not massively reduced DP performance too,then its a big achievement over Hawaii,considering it is only 27% bigger.
Click to expand...

If AMD manage to get that much performance and the DP performance, well, they'll continue to be sold in every Apple product ahead of Nvidia cards for the foreseeable future.


The most impressive thing is the tech though. Pretty much every move to bring things on package/die has been a fairly big technological leap but HBM and in the not too distance future HMC are monumental leaps forward in memory technology. Brilliant for GPU, soon to be brilliant for APU both for gpu and cpu cache duties. Iris Pro shows what some on package memory can do but because of the way it's made and the limit of it's size there will still be a big move forwards going with HBM/HMC for a huge amount of memory on die.

HBM is the start of the way forward for memory techs in all types of chips, it's a huge first step, as are interposers. This is tech that for the mainstream AMD is pioneering for frankly the entire industry.
 
Last edited:
Boomstick777 said:
About 10% - 15% faster (;

BuIWQ1n.jpg

Titan X doing better in Metro, in all seriousness I have no idea if this is legit..
Click to expand...

Very unlikely

Best test to see if any benchmarks are genuine is to compare them to the 295X2

Forget about memory speeds and just base it on the shaders and work out the % difference.

295X2 = 5632

Fiji = 4096 rumoured

which means the 295X2 should score about 37.5% more than Fiji

This all goes out the window if the Fiji architecture is very different to Hawaii

Using the table that has been posted on batman for example

Fury scored 65
295X2 should have scored 89.375

A couple of my old 290Xs scored 74 on the Batman AO bench maxed out including all the AA lol.
 
drunkenmaster said:
Yeah but for instance in Witcher 3 it would just work without loads of crashing and the need for multiple new unstable drivers before you get a working one.

Not even close to the first time Nvidia's new game support is unstable as crap.



If AMD manage to get that much performance and the DP performance, well, they'll continue to be sold in every Apple product ahead of Nvidia cards for the foreseeable future.
Click to expand...

I can't speak from experience with gm200 in games, but I imagine both parties have history of bad support in games.
 
Kaapstad said:
Just noticed something else -

Why is the TX scoring less than a 980 Ti in some of the tests lol.
Click to expand...

Newer drivers when Ti was tested? Better cooling on Ti so boosted higher? Ti and TX are so close in performance to be within the margin of error, so not entirely plausible to see it trade blows sometimes?

I mean I agree this may not be legit, but not because the 980 Ti trades blows. TX and Ti are very close..
 
Kaapstad said:
Very unlikely

Best test to see if any benchmarks are genuine is to compare them to the 295X2

Forget about memory speeds and just base it on the shaders and work out the % difference.

295X2 = 5632

Fiji = 4096 rumoured

which means the 295X2 should score about 37.5% more than Fiji

This all goes out the window if the Fiji architecture is very different to Hawaii

Using the table that has been posted on batman for example

Fury scored 65
295X2 should have scored 89.375

A couple of my old 290Xs scored 74 on the Batman AO bench maxed out including all the AA lol.
Click to expand...

290x cf with higher clocks than a 295x2 scored a bit more, then this 72fps seems ok
 
Kaapstad said:
Very unlikely

Best test to see if any benchmarks are genuine is to compare them to the 295X2

Forget about memory speeds and just base it on the shaders and work out the % difference.

295X2 = 5632

Fiji = 4096 rumoured

which means the 295X2 should score about 37.5% more than Fiji

This all goes out the window if the Fiji architecture is very different to Hawaii

Using the table that has been posted on batman for example

Fury scored 65
295X2 should have scored 89.375

A couple of my old 290Xs scored 74 on the Batman AO bench maxed out including all the AA lol.
Click to expand...

The 295X2 is GCN 1.1 with GDDR5.

This is GCN 1.3? with HBM.

Its far to easy to say Fiji will scale the same as Hawaii when you ignore the differences in technology.

PS: i'm more inclined to beelieve a slide that has the 980TI and TX trading blows than one that has the TX always ahead, thats not realistic especially given how close they are and that the TX ref cooler can and does throttle it a little.
 
Last edited:
humbug said:
The 295X2 is GCN 1.1 with GDDR5.

This is GCN 1.3? with HBM.

Its far to easy to say Fiji will scale the same as Hawaii when you ignore the differences in technology.
Click to expand...

I don't want to spoil the party but I just ran a single TX using the same settings as they did lol.

Be prepared for a shock

2160p
Max settings
No AA

GB240Vx.jpg

Even allowing for the overclock I was using it is no contest.

The Fiji Fury is going to be a great card but I would rather judge it on real benchmarks.:)
 
humbug said:
The 295X2 is GCN 1.1 with GDDR5.

This is GCN 1.3? with HBM.

Its far to easy to say Fiji will scale the same as Hawaii when you ignore the differences in technology.
Click to expand...

He is also assuming that crossfire is scaling 100% which is not always the case although at 4k it's more likely to be on the high side if cf is working well.

As above though those benchmarks are a to be taken with a big bucket of salt.
 
That wccftech article is all well and good and the specs may very well be accurate. but the bit that does it for me is.

Fury X is faster than the R9 290X by a minimum of 54%

What utter worthless tosh.

I mean they cannot be meaning just on the chart listed in the article because furry isn't not 54% faster in for example clock speed. So do they mean in all benchmarks, well obviously not, as only an imbecile would believe that furry is 54% faster in every conceivable benchmark.
So we can only assume they mean out of the ones they ran, which they conveniently don't mention what they were. Making that figure completely pointless.

If the specs are right though it does look like a powerhouse for sure.
 
Status
Not open for further replies.
Back
Top Bottom