HWZone Forum - Sapphire R9 Fury X pics and benchmarks!

[tolis626]

Well, seeing as the underlying architecture is practically the same GCN we know (and love?) I wouldn't expect Fury GPUs to overclock well. Sure, HBM frees up die space etc etc, but I guess the stream processors and the transistor configuration is largely the same as Hawaii and Tonga. I don't think the cards will be TDP limited but rather limited by the silicon lottery. Also I don't think there's any point at overclocking the memory. It already sits at over 500GB/s of bandwidth. Not being able to tinker with it would be a bummer, sure, but not a deal-breaker by any stretch of the imagination.

With all that said, and with me usually drooling over well-overclocking chips, I don't really think that the high-clocking Maxwell GPUs will have such a big advantage over the Fury, simply because Maxwell GPUs rely a lot on their high clockspeed potential to be as fast as they are. Reference cards already boost at 1300+MHz, so a jump to 1400-1500MHz isn't THAT huge. Still bigger than a possible 100MHz for the Fury, but nothing that creates a chaotic difference.

I say we wait to see what results proper overclocking gives. I couldn't care less how it overclocks at stock voltage, be it AMD or NVidia GPUs. Max the thing out and then we can talk.

 

[5150Joker]

Here's what I get with a Titan X in SoM:

1440p:

• 1215 MHz (like AT review stock boost clock): http://i.imgur.com/wYDJXfA.jpg
• 1408 MHz: http://i.imgur.com/rOu8mHR.jpg

150% scaling (3840 x 2160):

• 1215 MHz: http://i.imgur.com/hcfTubC.jpg
• 1408 MHz: http://i.imgur.com/BAqP22E.jpg
• 1498 MHz: http://i.imgur.com/pxVPXrB.jpg

Biggest question should be how far Fury X can overclock. If I can hit 1500 MHz on my air cooled reference Titan X's, then aftermarket 980 Ti's certainly should hit those speeds if not higher. If that's the case, Fury X is in trouble if it can't OC well. I did find Fury X's performance at 4K to be pretty close to the Titan X so it should be on par with the stock 980 Ti (but will certainly lose to the non-stock ones in this game). One interesting thing you will note is that the Titan X doesn't benefit much from higher clocks passed 1408 MHz in this game, could be a CPU limitation. If anyone ends up with a Fury X + 5820K I'd like to compare.

Edit: I just saw his overclocked results and they get close with avg fps to my 1408 MHz ones which is good news since he only had a 100 MHz OC (his max stable OC on stock voltage) BUT look at his min fps: http://i995.photobucket.com/albums/...Mordor 2015-06-24 01-09-27-54_zpsjor6x1j3.png 29.34 vs 45 fps (assuming his run was at 4k). His signature says he has a 4670K, does anyone know if SoM is CPU dependent for min fps? If the min. fps inconsistency isn't shown in official reviews, then AMD has a solid card on it's hands (at least at 4K) as long as it can keep up with Crossfire profiles and day 1 driver releases for AAA titles.

 

[Head1985]

He manage 1150Mhz(54fps).Thats only 10% oc.If he manage 1260mhz +20% he will have 59fps and if 1312Mhz +25% it will be 61fps
http://forums.hardwarezone.com.sg/94719998-post146.html

 

[Face2Face]

Here's what I get with a Titan X in SoM:

1440p:

• 1215 MHz (like AT review stock boost clock): http://i.imgur.com/wYDJXfA.jpg
• 1408 MHz: http://i.imgur.com/rOu8mHR.jpg

150% scaling (3840 x 2160):

• 1215 MHz: http://i.imgur.com/hcfTubC.jpg
• 1408 MHz: http://i.imgur.com/BAqP22E.jpg
• 1498 MHz: http://i.imgur.com/pxVPXrB.jpg

Biggest question should be how far Fury X can overclock. If I can hit 1500 MHz on my air cooled reference Titan X's, then aftermarket 980 Ti's certainly should hit those speeds if not higher. If that's the case, Fury X is in trouble if it can't OC well. I did find Fury X's performance at 4K to be pretty close to the Titan X so it should be on par with the stock 980 Ti (but will certainly lose to the non-stock ones in this game). One interesting thing you will note is that the Titan X doesn't benefit much from higher clocks passed 1408 MHz in this game, could be a CPU limitation. If anyone ends up with a Fury X + 5820K I'd like to compare.

Edit: I just saw his overclocked results and they get close with avg fps to my 1408 MHz ones which is good news since he only had a 100 MHz OC (his max stable OC on stock voltage) BUT look at his min fps: http://i995.photobucket.com/albums/...Mordor 2015-06-24 01-09-27-54_zpsjor6x1j3.png 29.34 vs 45 fps (assuming his run was at 4k). His signature says he has a 4670K, does anyone know if SoM is CPU dependent for min fps? If the min. fps inconsistency isn't shown in official reviews, then AMD has a solid card on it's hands (at least at 4K) as long as it can keep up with Crossfire profiles and day 1 driver releases for AAA titles.

Thanks for the comparison. Not sure about the minimum FPS anomaly?

 

[gamervivek]

The results for SoM look rather low, a 390X is like 10% behind a 980Ti in that game so Fury should be doing way better than that.

 

[sam_816]

The results for SoM look rather low, a 390X is like 10% behind a 980Ti in that game so Fury should be doing way better than that.

Can it be driver problem? May be AMD will launch new drivers(review sites might already have them) with the launch of fury..

 

[DiogoDX]

Same guy:

For VRAM on HBM is very interesting. SoM did not go over 3.8gb as per shown in GPUZ. It was running well and without any stuttering during my gaming experience.

Some facts that i know, most of the gaming benchmarks on stock settings (Out of box performance) was found to be on par with 980ti (reference) or slightly higher/faster. Will post all the results by this weekend.

http://www.overclock.net/t/1561745/...se-real-benchmarks-included/160#post_24081361

 

[DarkKnightDude]

Same guy:


http://www.overclock.net/t/1561745/...se-real-benchmarks-included/160#post_24081361

Huh, that's interesting. Doesn't the game use like 6 GB?

 

[LTC8K6]

Would limiting the frame rates result in less VRAM needed?

 

[DiogoDX]

Would limiting the frame rates result in less VRAM needed?

Nope.

 

[DarkKnightDude]

Would limiting the frame rates result in less VRAM needed?

Nope, only way to do that is to lower VRAM intensive settings, like textures or anti-aliasing.

 

[Silverforce11]

Same guy:


http://www.overclock.net/t/1561745/...se-real-benchmarks-included/160#post_24081361

Not new. 980SLI and R290X CF run SOM at 4K flawlessly. Only the 3.5GB 970 SLI have stutters.

Another case of vram allocated != vram needed.

 

[Majcric]

980ti SoM benchmark results @1440p, if anyone wants to see another comparsion .

Avg 88.58
Max 115.71
Min 54.33

 

[5150Joker]

Thanks for the comparison. Not sure about the minimum FPS anomaly?

Yeah that's what has me really curious to know why that happened. It's hard to draw a conclusion based on two very different configurations so we'll see what if anything the reviewers find.