• 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.

Fury X Vs Titan X - Performance compared

And it runs much better at 4k, crossfired fury according to tweaktown is doing really well at 4k.

Crossfire 4k does seem to be Fury's saving grace, the X2 could be great. You are more likely to the VRAM limit in the future though.
 
The argument can be made that GCN 1.2 is hardly anything new to AMD either.

Indeed. Moreover, AMD have admitted that GCN in its current form is end of life,they said the limit was 4 shader cores and 1 geometry engine per shader engine with the ROPs and TMUs maxed out. Nvidia think maxwell is still in its infancy, pascal and even Volta are supposedly based I the maxwell concept.


Fiji is to Tonga architecture scaled up. The only new thing addition is HBM. If driver issues related to HBM were really the issues then the tale tale signs would be when bandwidth was most important, e.g. 4K and the most demanding games. Except that is when FuryX does best, when the game gets tough FuryX gets tougher. That is a pretty big indicator that the drivers are doing a reasonably good job. 4k and modern intensive games requires high bandwidth and high pixel processing, the only difference Fiji has improved upon significantly. At lower solutions architectural issues and other bottles necks like geometry likely become more important.

Fiji is an evolution of Hawaii and tonga. The drivers should be mature enough.
 
The fury is also Very large compared to Tonga. So it is a bit of a different situation. Inefficiencies in drivers and not being able to feed all the cores with enough data are what is holding it back at lower than 4k.

If you look at Gregs Thief video, where the FX is running on mantle. it beats the Titan x by a few fps at stock. Considering that Gregs TX is clocking itself to 1.3ghz. Although things could also go the other way if the TX is not being held back by API overhead.

And by that logic the TX is also a much bigger chip than the 970 so both should apply equally to AMD and nvidia.
Not being able to feed all the cores is more likely an architectural issue relating to the shader engines and controllers.

AMD have been making DX drivers for many years, they should be able to have an efficient driver by now, within the limits of the API.

Edit: I don't want to say there won't be any improvements with drivers, but I simply wouldn't best on future performance due to miracle drivers. Large boosts form drivers have only occurred when there has been a large change in architecture, the Fiji is certainly not a large change by historic standards apart from the addition of HBM. Where HBM and matters the FuryX shows it is performing as expected. In scenarios where you would expect the massive number of pixel shaders to excel the furyx does well. The synthetic benchmarks show the FuryX hitting the theoretical limits. There just isn't much evidence there for driver issues.
 
Last edited:
And by that logic the TX is also a much bigger chip than the 970 so both should apply equally to AMD and nvidia.
Not being able to feed all the cores is more likely an architectural issue relating to the shader engines and controllers.

AMD have been making DX drivers for many years, they should be able to have an efficient driver by now, within the limits of the API.

But sadly their drivers still have far more overhead than Nvidias. And the FX's shows this in Thief comparison at 1080p. Most review sites have the FX far behind the TX, yet the FX comes ahead in gregs video when running mantle.

But the next lot of drivers that hit should give everyone a nice performance boost in dx10-11. the 15.20 driver branch are awesome. i gained 34% in Draw call performance over the omega drivers. And compared to the 15.6 drivers, i still have 22% better draw call performance.
 
I have to say I think this card (Fury X) does extremely well given the cost difference compared to the Titan X. I only wish Gregster was able to borrow a 980Ti from someone so he could put it through its paces against the Fury X.
 
But sadly their drivers still have far more overhead than Nvidias. And the FX's shows this in Thief comparison at 1080p. Most review sites have the FX far behind the TX, yet the FX comes ahead in gregs video when running mantle.

But the next lot of drivers that hit should give everyone a nice performance boost in dx10-11. the 15.20 driver branch are awesome. i gained 34% in Draw call performance over the omega drivers. And compared to the 15.6 drivers, i still have 22% better draw call performance.

The mantle difference is largely due to API though. Also afaik mantle doesn't officially support FuryX because AMD has stopped Mantle development and specificaly requested reviewers not to use mantle, or at least warned that it may not work well.

I agree though that AMD does have poor driver overhead, the draw call results back that up. However, I don't see that as a sign that they will get a load of performance in the future that wont appear on other GCN cards, you findings support that.
 
I have to say I think this card (Fury X) does extremely well given the cost difference compared to the Titan X. I only wish Gregster was able to borrow a 980Ti from someone so he could put it through its paces against the Fury X.

Well the TC is just stupidly priced. No idea what planet nvidia is on. The 980ti is the only FuryX competitor (and 980/390x when run at 1080p).
 
The delta compression doesn't save memory usage but increases bandwidth. And bandwidth gains from HBM are also irrelevant, if a game requires more than 54GB of assets in vram then bandwidth wont change that requirement at all.

What AMD are doing with the FuryX is making the driver aggressively pull resources out of vram that they expect wont be needed as well as some kind of garbage collection. This is why they said the driver would have to be specially tuned for each game because they need to try and figure out what resources are safe to drop. this helps avoid the FuryX running in to vram limits.

And that would be my biggest concern for 4k gaming, How well will they keep up with the additional driver work. They don't do a good job keeping up with the dual card profiles and there adding another system that requires game for game driver development.
Are they really going to provide and use the resources needed to keep it working as it should do long term?
 
Last edited:
I would happily put a 980Ti through its paces if someone wants to send me one for testing, the same as I would any card :)
 
It's a whatever resolution you want it for gpu.

I imagine it will be bought to be used with 1440p just as often as 4k.

Probably more so. If it is only supposed to be a 4k card then it's a tiny market and not something a struggling company should or will be aiming for, imo it should be just as good at say 1080 120/144hz as it is at 4k.
 
Last edited:
Probably more so. If it is only supposed to be a 4k card then it's a tiny market and not something a struggling company should or will be aiming for, imo it should be just as good at say 1080 120/144hz as it is at 4k.


There is always a catch somewhere with them, or an excuse with some notion. The card should be competitive at all resolutions but the fact is it's simply not. Both companies have had their work cut out for them this time on this node, but maybe people can appreciate just how good Maxwell and the drivers are now. Both companies have made great progress.
 
Last edited:
Lets be fair, it mostly loses.

Personally I was hoping it would beat a Ti. Remains to be seen if it will ever even match one. However if they sort the pump issues I will still be getting one......:confused::p

:)

Everyone secretly wants AMD to smash nvidia but in business just competing is good enough. Something they haven't done across all models for some time. Lets just hope it's not 18 months till the next release. :cool:

2016 could be a good year for graphics with dx12 in full swing etc :)
 
:)

Everyone secretly wants AMD to smash nvidia but in business just competing is good enough. Something they haven't done across all models for some time. Lets just hope it's not 18 months till the next release. :cool:

2016 could be a good year for graphics with dx12 in full swing etc :)

No secret I would like the the tables to turn for a few years between the companies. AMD could use the cash, but ultimately I would want it to be 50-50 or thereabouts. This would mean better prices for us :)

I don't see the FX beating TX or Ti on DX11. Maybe on DX12 at 4K.

I would like to see some 4k comparisons from Greg, 1080p is like 10 years old now. I was gaming on 1600p 6-7 years ago and 1200p before that.

Only reason I would take a inferior tn panel at a lower resolution in order to get 144hz is if I primarily gamed on fps games online which I don't. 4k IPS lushness is where it's at :D
 
i think AMD did pretty well considering it was gregster doing the testing ^^;
can maybe take 10% off the tx scores bcos of the crazy loud fan speed that no one would use and greg blowing on it with icepops bcos jenhsun in that leather jacket makes his nipples hard
hehe <3
 
I could test a 4K if AMD allowed me to upscale that high as I also have a 4K capture card. Nvidia I can but AMD is 3200x1800 off the top of my head but does need to be UHD resolution for it to record.
 
Back
Top Bottom