• 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.
My spelling is pretty bad at times, i can't spell and i'm lazy with it to make it worse.

I'm also Dyslexic
 
My spelling is pretty bad at times, i can't spell and i'm lazy with it to make it worse.

I'm also Dyslexic

At least you have a partially good excuse.

Me im just lazy when it comes to typing and hit the keys willy nilly closest to the key i really want and hope for the best.
 
Last edited:
Two-thirds of overclock scaling comes from the GPU, one-third comes from the VRam.

If you overclock the GPU by 10%, say from 1000Mhz to 1100Mhz, you get a 6% performance boost. if you overclock the VRam by 10%, say from 1250Mhz to 1375Mhz you get a 3% performance boost.

Overclock the GPU and Memory by 10% each you get about 9% extra performance.

That applies to Shader and Memory Bandwidth scaling.

If its 2/3 and 1/3 then why is the answer not 10% ? 6.6 and 3.3. ? Which makes the whole calculation pointless if you just split the performance boost into a ratio and added the ratios back together?

Still very confused by whats actually happening with this :(
 
If its 2/3 and 1/3 then why is the answer not 10% ? 6.6 and 3.3. ? Which makes the whole calculation pointless if you just split the performance boost into a ratio and added the ratios back together?

Still very confused by whats actually happening with this :(

I have never been able to quite get 100% scaling by overclocking the GPU and VRam by equal amounts, its always around 90% scaling.

To confuse it even more sometimes you get more from the GPU and less from the VRam, or the VRam scales a little better while the GPU less so, it depends on the game, even what IQ settings you run.

But mostly its about as i explained.

Fire Strike is very GPU scaling, almost nothing on the VRam, while Alien likes Memory Bandwidth, for example.
 
Last edited:
I have never been able to quite get 100% scaling by overclocking the GPU and VRam by equal amounts, its always around 90% scaling.

to confuse it even more sometimes you get more from the GPU and less from the VRam, or the VRam scales a little better while the GPU less so, it depends on the game, even what IQ settings you run.

But mostly its about as i explained.

Ah makes bit more sense now .
 
Two-thirds of overclock scaling comes from the GPU, one-third comes from the VRam.

If you overclock the GPU by 10%, say from 1000Mhz to 1100Mhz, you get a 6% performance boost. if you overclock the VRam by 10%, say from 1250Mhz to 1375Mhz you get a 3% performance boost.

Overclock the GPU and Memory by 10% each you get about 9% extra performance.

That applies to Shader and Memory Bandwidth scaling.

now that explanation is much better, pretty much how I took when you first posted it up, only after thinking about it too much did it start to make less and less sense. :)
 
I seem to spend more time correcting the errors in my posts than I do actually typing them out to begin with. :D
 

Go AMD !!

1GB – 32GB capacity

390X with 8GB HBM or more and I'll get one at launch.

Imagine what kind of performance will a GPU with 4096 stream processors, 256 texture units and up to 640GB/s memory bandwidth provide in 4K (3840*2160) ultra-high-definition resolution!

aH2WRkQ.gif
 
Last edited:
Pretty sure HBM1 is limited to 1GB per stack and 4 stacks but if not it would be the slam dunk to end all slam dunks. It would be the first true 4K card.
 
Pretty sure HBM1 is limited to 1GB per stack and 4 stacks but if not it would be the slam dunk to end all slam dunks. It would be the first true 4K card.

Recently SK Hynix announced that it would start mass production of commercial HBM DRAM devices in the first quarter of 2015. The first commercial HBM chip is expected to feature 8Gb capacity (eight 1Gb 128-bit channels, four 2Gb DRAM dies), 1024-bit wide interface, 1Gb/s per pin data-rate and 1.2V voltage.
 
Status
Not open for further replies.
Back
Top Bottom