• 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.
humbug said:
Unknown, its all just speculation unlike some people i'm not saying anything as fact.

The 285 is 22% faster than the 270X
The 280X is 3% faster than the 285.
The 290 is 30% faster than the 285.

270X 1280 Shaders @ 1050Mhz, 256Bit, 1500Mhz
285: 1796 shaders @ 918Mhz, 256Bit, 1375Mhz
280X: 2048 Shaders @ 1000Mhz, 384Bit, 1500Mhz
290: 2560 Shaders @ 947Mhz, 512Bit, 1250Mhz

Despite having only 85% the shaders @ 8% lower Mhz the 285 has 97% the performance of the 280X

15% + 8% = 24%, 7/10 scaling = about 18%

At this point a 2048 Shader 1000Mhz 256Bit Tonga would be 88% the performance of a 290.

Add +100% Bus width, add + 100% ROP's at 4/10 scaling.

Now at 92% of a 290.

Add 512 Shaders (25%) 7/10 Scaling (18%) + 92% = 110% (290X)

Add 20% Memory Speed 1250 - 1500Mhz @ 3/10 scaling (6%) + 110% = 116% (3% faster than a GTX 970 RE: TPU)

Add +5% GPU (1000 - 1050Mhz) 7/10 Scaling = 3% + 119%

IMO a 290P spec 1050Mhz Tonga would be a bit faster than a GTX 970.
Click to expand...

Totally fine humbug I'm not holding you to it at all, I appeciate your maths lmao I need about 2 hours to check them, My prediction was a tonga 2048 would need to be 1300mhz to show a good gain over a 1-1.1ghz 2560 hawaii,
So a theoretical 2304 tonga would need to be 1100-1200? so a 2560 tonga 1050- 1100mhz sounds about right
 
Davedree said:
Totally fine humbug I'm not holding you to it at all, I appeciate your maths lmao I need about 2 hours to check them, My prediction was a tonga 2048 would need to be 1300mhz to show a good gain over a 1-1.1ghz 2560 hawaii,
So a theoretical 2304 tonga would need to be 1100-1200? so a 2560 tonga 1050- 1100mhz sounds about right
Click to expand...

The thing is the 290 doesn't run at 1-1.1Ghz, it runs significantly less than that, 947Mhz.

If you take the 918Mhz 285 and bring it up to 1050Mhz your adding 15% to the clock rate, that would destroy its big brother 280X, it would start eating into the 290's performance, thats why AMD clocked it so pathetically low even with its gimped 256Bit Bus, GCN 1.2 is a good chunk faster than GCN 1.1.

Edit: a 290 is about 30% faster than a 285, if overclocking the GPU adds 7/10 performance scaling then you would need to add about 40% to the 285's clock rate to match the 290. So 918Mhz +40% = 1285Mhz, clock the VRam up abit to bring the overall scalling up to around 9/10 or 1/1 then a 1250/1500 Mhz 285 is about as fast as a 290, a 2048 Shader Tonga may need to be around 1150/1500 Mhz
 
Last edited:
384bit Tonga.
A bit of an unknown, but probably quite close to humbugs example ( now quite as fast as the 970 in my opinion), that would explain why we haven't had the 960ti yet, as they want to set the clock speeds on that to be a suitable challenge but not to be faster than the 970.
 
bru said:
384bit Tonga.
A bit of an unknown, but probably quite close to humbugs example ( now quite as fast as the 970 in my opinion), that would explain why we haven't had the 960ti yet, as they want to set the clock speeds on that to be a suitable challenge but not to be faster than the 970.
Click to expand...

With 2048 Shaders (What we know to be true) certainly not as fast as a 970, but probably have a stock 290P trembling looking at its heels.
 
Last edited:
humbug said:
The thing is the 290 doesn't run at 1-1.1Ghz, it runs significantly less than that, 947Mhz.

If you take the 918Mhz 285 and bring it up to 1050Mhz your adding 15% to the clock rate, that would destroy its big brother 280X, it would start eating into the 290's performance, thats why AMD clocked it so pathetically low even with its gimped 256Bit Bus, GCN 1.2 is a good chunk faster than GCN 1.1.

Edit: a 290 is about 30% faster than a 285, if overclocking the GPU adds 7/10 performance scaling then you would need to add about 40% to the 285's clock rate to match the 290. So 918Mhz +40% = 1285Mhz, clock the VRam up abit to bring the overall scalling up to around 9/10 or 1/1 then a 1250/1500 Mhz 285 is about as fast as a 290, a 2048 Shader Tonga may need to be around 1150/1500 Mhz
Click to expand...

A throttled 50mhz less at 947 isn't too much from a 1ghz, I know it doesn't run at 1ghz-1.1 ghz but most can and do.

Old bench but 1792 285 overclocked 1020/1500 256 bit put's it just above a stock 280x sometimes quite a bit below . Don't think a 384 bit tonga 2048 will scale that well at 1150/1500 like you say, dependent on game engine of course.
http://www.guru3d.com/articles_pages/amd_radeon_r9_285_review,20.html
 
Davedree said:
A throttled 50mhz less at 947 isn't too much from a 1ghz, I know it doesn't run at 1ghz-1.1 ghz but most can and do.

Old bench but 1792 285 overclocked 1020/1500 256 bit put's it just above a stock 280x sometimes quite a bit below . Don't think a 384 bit tonga 2048 will scale that well at 1150/1500 like you say, dependent on game engine of course.
http://www.guru3d.com/articles_pages/amd_radeon_r9_285_review,20.html
Click to expand...


On the contrary

If scaling is as bad as their results imply, looks like about 3/10 < 'n that is catastrophic, then the 256Bit Bus is strangling the crap out of it.

A 384Bit Bus will free it. hell just adding a 384Bit bus would probably give it a boost.
 
humbug said:
On the contrary

If scaling is as bad as their results imply, looks like about 3/10 < 'n that is catastrophic, then the 256Bit Bus is strangling the crap out of it.

A 384Bit Bus will free it. hell just adding a 384Bit bus would probably give it a boost.
Click to expand...

Agreed on the bus 100%, those benches were old drivers too.
We need Kaap/you or someone with a spare 285 to do some clocking and benches.;)
 
If I do pick up the 390X or whatever, I hope to god it solves the popping/flickering on my monitor over DP. Only affected 290/x as far as I know, so fingers crossed.

Have to run at 48hz or it pops like mad..
 
Davedree said:
A throttled 50mhz less at 947 isn't too much from a 1ghz, I know it doesn't run at 1ghz-1.1 ghz but most can and do.

Old bench but 1792 285 overclocked 1020/1500 256 bit put's it just above a stock 280x sometimes quite a bit below . Don't think a 384 bit tonga 2048 will scale that well at 1150/1500 like you say, dependent on game engine of course.
http://www.guru3d.com/articles_pages/amd_radeon_r9_285_review,20.html
Click to expand...


yeah I'd be most surprised if any enthuiast is running their 290 at stock

very low stock clocks for what these cards are capable - mine I'd say is a relative "poor" clocker and it still does 1115
 
1115 is @ + 0.75V

mine doesn't like mem clocks much though - anything over 1450 and everything AOK - but when I come out of standby/hibernate I get black screen - so I just leave at 1400

I normally run at 1080 @ + 0.38V as nice and quiet and quick (Tri-X)
 
well to be fair they are rather well clocked out of the box

what I'd meant originally is that I thought the base 950 or whatever it is seemed very conservative to me
 
http://wccftech.com/amd-fury-series-lineup-flavors-fury-nano-xt-pro/

AMD Fury to have 3 flavours?

Probably accurate, maybe AMD wanted 3 flavours like Nvidia had with GTX 980, 970 and 960. Now it over a week left for final adjustment on Fury, the flagship Fury XT will cost $899 to compete with 980 Ti but slight slower than 980 Ti.

My guess below what Fury price range and performance look like.

$899 Fury XT = $649 GTX 980 Ti
$799 Fury PRO = Custom OC GTX 980
$699 Fury Nano = 390X & GTX 980
 
Status
Not open for further replies.
Back
Top Bottom