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

[Boomstick777]

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

Spoiler

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.

So you over-clocked your card, how is that a fair comparison

I agree we need to wait for legit reviews / benchmarks for a proper idea of how these cards stack up. These leaks / rumors is all we have until then lol.

 

[Kaapstad]

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.

Funny you should mention that with a single 290X using max settings including the AA I scored 39 compared to 74 for a pair running lower clocks so yes CF was getting pretty close to 100%.

 

[easyrider]

Do you think the green goblins are quaking in their boots ?

 

[Kaapstad]

So you over-clocked your card, how is that a fair comparison

I agree we need to wait for legit reviews / benchmarks for a proper idea of how these cards stack up. These leaks / rumors is all we have until then lol.

Because I would have need a 100% overclock to beat their figures if genuine lol.

I know Maxwell is good but not that good.

The point I am making is that table is total garbage and it would be unfair to judge the Fiji cards on it.

 

[Kaapstad]

Do you think the green goblins are quaking in their boots ?

Nope

I am getting some Fiji cards, are you ?

 

[humbug]

Kaap we are talking about single digit margins here, your CPU is far more powerful than the one they used, and then there are other differences between your rig and theirs.

72 at stock with an 8 thread i7 vs 81 overclocked with a 12 thread i7 + margins of error, i'm fine with that.

 

[Kaapstad]

Kaap we are talking about single digit margins here, your CPU is far more powerful than the one they used, and then there are other differences between your rig and theirs.

72 at stock with an 8 thread i7 vs 81 overclocked with a 12 thread i7 + margins of error, i'm fine with that.

Look at that score again, it is 109 average.

81 was the minimum.

5960X won't make much difference with Batman either and I was only running it @4.0

 

[TheRealDeal]

Kaap we are talking about single digit margins here, your CPU is far more powerful than the one they used, and then there are other differences between your rig and theirs.

72 at stock with an 8 thread i7 vs 81 overclocked with a 12 thread i7 + margins of error, i'm fine with that.

His average was 109 so way over what was in the table. Maybe they had aa on by mistake but if not the result is clearly wrong. Kaap give 4xaa and 8xaa a try and see if you are then in the ball park. That's if the benchmark is short and sweet.

 

[Boomstick777]

Because I would have need a 100% overclock to beat their figures if genuine lol.

I know Maxwell is good but not that good.

The point I am making is that table is total garbage and it would be unfair to judge the Fiji cards on it.

Haha, agreed. I think these dodgy leaks make a new launch a bit fun at least

Will you be picking up Fiji if performance is a good 10%+ faster?

I'm gonna have to try one of these for myself.

 

[Mayh3M]

Well there goes those benchmarks.

 

[humbug]

Look at that score again, it is 109 average.

81 was the minimum.

5960X won't make much difference with Batman either and I was only running it @4.0

Well then thats different, granted.

 

[humbug]

Well i may regret saying this but from guesstimate calculations i have made a few times in this thread my results are just about right there where these are.

I got the 290X bang on and i got my TX Tessellation sums to 0.2% of where you resulted did i not Kaap?

I think this is legit, wishfully i admit.

 

[Mayh3M]

To be fair these numbers aren't implausible or crazy at all. In fact they're kind of expected for those who thought Fiji was always going to be a little faster than the Titan X.

It really should be that much faster considering the use of HBM and years of development by AMD.

 

[TheRealDeal]

To be fair these numbers aren't implausible or crazy at all. In fact they're kind of expected for those who think Fury was always going to be a little faster than the Titan X.

It really should be that much faster considering the use of HBM and years of development by AMD.

AMD have pretty much had the same development time as Nvidia had with the Titan X and 980ti so if they are faster it's because they did a better job this time round.

 

[Kaapstad]

Well then thats different, granted.

Here is another one with my TX @stock
using the same settings as them
2160p max - no AA
5960X with no H/T
5960X using just 2 cores
This means most notebooks have better CPUs.

Spoiler

I just think that table is total garbage and we should wait for some real benches.

 

[Mayh3M]

AMD have pretty much had the same development time as Nvidia had with the Titan X and 980ti so if they are faster it's because they did a better job this time round.

Nvidia have released far more cards in the meantime. HBM is by all accounts supposed to be a game-changer - so if AMD only managed to only match Titan X GDDR5 performance that would be disappointing.

 

[Kaapstad]

His average was 109 so way over what was in the table. Maybe they had aa on by mistake but if not the result is clearly wrong. Kaap give 4xaa and 8xaa a try and see if you are then in the ball park. That's if the benchmark is short and sweet.

I already did 8XMSAA and got 51fps on an overclocked TX.

Even if they have made a mistake it does not say much for the accuracy of their data.

Also a 980 Ti should not beat a TX under any circumstances when they are both using the same clocks, drivers and PC again raising questions about the quality of data.

 

[TheRealDeal]

Kaap's 2 x 290x score from the benchmark thread.

74 - R9 290X x 2

This is in line with the 295x2 score but of course this is with 8 x aa. The graph most likely is fake but mistakes could have been made in the Batman test.

Only 6 days to go now and all will be revealed. You would think if someone was going to make up fake results they would do a better job than this if they are that far out.

 

[Kaapstad]

Kaap's 2 x 290x score from the benchmark thread.

74 - R9 290X x 2

This is in line with the 295x2 score but of course this is with 8 x aa. The graph most likely is fake but mistakes could have been made in the Batman test.

If you turn off AA it more than doubles the score.

 

[humbug]

Here is another one with my TX @stock
using the same settings as them
2160p max - no AA
5960X with no H/T
5960X using just 2 cores
This means most notebooks have better CPUs.

Spoiler

I just think that table is total garbage and we should wait for some real benches.

I think its more a compliment to your building skills, management and all the very high end components you run, you consistently top the leader boards.
To be fair i think you know your rig performs than bit better that your average TX/295X2 rig.

You have put some doubts in my mind but i'm not convinced enough to dismiss it.

Edit:

Kaap's 2 x 290x score from the benchmark thread.

74 - R9 290X x 2

This is in line with the 295x2 score but of course this is with 8 x aa. The graph most likely is fake but mistakes could have been made in the Batman test.

Only 6 days to go now and all will be revealed. You would think if someone was going to make up fake results they would do a better job than this if they are that far out.

just my point, Kaap can really push something incredible out his rigs when he really wants to, he's a pro...