FineWine===>FineVinegar FuryX Aging Horrible in 2017

[Fouquin]

Definitely not just me regretting the decision of buying Fury/X

https://www.reddit.com/r/Amd/comments/771nrf/r9_fury_fury_x_nano_owners_have_you_benefited/


Oh and in case people didn't know. AMD intentionally locked up HBM1 overclocking via driver update right around Vega launch. So as of right now neither BIOS mods nor afterburner can OC HBM1 on Fury/X/Nano

Grasping at straws. Overclocking HBM on Fiji provides such minuscule, insignificant increases in performance that it is simply not worth the stress to the relatively fragile IMC in Fiji.

Also they locked that well before Vega. 16.12.2 was the last to officially allow it, with the accidental inclusion of the slider in 17.4.3 before the newer revision of ReLive.

 

[MuhammedAbdo]

With latest drivers. https://www.techpowerup.com/reviews/ASUS/GTX_1070_Ti_STRIX/30.html

That review is useless with respect to the current discussion, none of the late 2017 games are included in the review, which is the point of the OP, FuryX is aging badly in these games.

 

[Vya Domus]

So those newer games are guaranteed to run terrible because they're not included.

Flawless logic.

 

[MuhammedAbdo]

So those newer games are guaranteed to run terrible because they're not included.

Because these newer games have already been tested and have shown the FuryX dropping hard behind. See the dozens of links in the Reddit thread!

 

[fullinfusion]

Definitely not just me regretting the decision of buying Fury/X

https://www.reddit.com/r/Amd/comments/771nrf/r9_fury_fury_x_nano_owners_have_you_benefited/


Oh and in case people didn't know. AMD intentionally locked up HBM1 overclocking via driver update right around Vega launch. So as of right now neither BIOS mods nor afterburner can OC HBM1 on Fury/X/Nano

I wouldn't go as far to say that AMD lockdown hbm 1 near Vega launch. I think it has more to do with their driver optimizations that are ongoing right now and different fixes coming down the pipeline which is quite possible why it's locked down at the moment but I could be wrong. And Fury still has fixes coming. There's so much code involved with driver optimizations to make all cards work and to fix current bugs and to enable future features it's a definite Balancing Act for sure.

 

[ikeke]

FuryX vs 980Ti
http://www.guru3d.com/articles_pages/destiny_2_pc_graphics_performance_benchmark_review,5.html
@1440p it's AVG 58 vs 56
http://www.guru3d.com/articles_page..._graphics_performance_benchmark_review,5.html
@1440p it's AVG 44 vs 50
http://www.guru3d.com/articles_page..._pc_graphics_analysis_benchmark_review,5.html
@1440p it's (against GTX1070, 980ti errored out) AVG 73 vs 79
http://www.guru3d.com/articles_page..._pc_graphics_analysis_benchmark_review,5.html
@1440p memory limited - AVG 40 vs 48
http://www.guru3d.com/articles_pages/f1_2017_pc_graphics_performance_benchmark_review,6.html
@1440p it's AVG 71 vs 77
http://www.guru3d.com/articles_pages/prey_pc_graphics_performance_benchmark_review,7.html
@1440p it's AVG 91 vs 93

So yeah...about that...

I can't feel the difference if it's below 10fps, can you?

 

[EarthDog]

Id feel 40 v 48.....thats over a 16% difference.

 

[MuhammedAbdo]

http://www.guru3d.com/articles_page..._graphics_performance_benchmark_review,5.html
@1440p it's AVG 44 vs 50

Yet here it's over 20% difference from 4 different sites:
http://gamegpu.com/action-/-fps-/-tps/middle-earth-shadow-of-war-test-gpu-cpu
http://www.pcgameshardware.de/Mitte...erde-Schatten-des-Krieges-Benchmarks-1240756/
https://www.computerbase.de/2017-10...amm-mittelerde-schatten-des-krieges-2560-1440
https://www.purepc.pl/karty_graficz...th_shadow_of_war_srodziemie_w_ogniu?page=0,10

http://www.guru3d.com/articles_page..._pc_graphics_analysis_benchmark_review,5.html
@1440p memory limited - AVG 40 vs 48

Here it's 50% faster than FuryX @1080p and 1440p from 3 different sites:
http://www.pcgameshardware.de/Wolfe...us-im-Technik-Test-Benchmarks-Vulkan-1242138/
https://www.computerbase.de/2017-10...nstein-2-1920-1080-anspruchsvolle-testsequenz
http://gamegpu.com/action-/-fps-/-tps/wolfenstein-ii-the-new-colossus-test-gpu-cpu
http://gamegpu.com/action-/-fps-/-tps/wolfenstein-ii-the-new-colossus-test-gpu-cpu

http://www.guru3d.com/articles_pages/destiny_2_pc_graphics_performance_benchmark_review,5.html
@1440p it's AVG 58 vs 56

Here, it's 25% faster than FuryX @1080p from 5 different sites, and a new driver just launched with 50% uplift for all NVIDIA GPUs, taking care of 1440p and 2160p results
http://gamegpu.com/action-/-fps-/-tps/destiny-2-test-gpu-cpu
https://www.computerbase.de/2017-10/destiny-2-benchmark/2/#diagramm-destiny-2-1920-1080
http://www.pcgameshardware.de/Desti...Destiny-2-PC-Technik-Test-Benchmarks-1241945/
http://www.guru3d.com/articles_pages/destiny_2_pc_graphics_performance_benchmark_review,4.html
https://overclock3d.net/reviews/software/destiny_2_pc_performance_review/8

http://www.guru3d.com/articles_pages/f1_2017_pc_graphics_performance_benchmark_review,6.html
@1440p it's AVG 71 vs 77

F1 2017 not tested at max quality.

 

[Fouquin]

Id feel 40 v 48.....thats over a 16% difference.

That's a difference of 4.17ms per frame. Without bionic implants I'm not so sure you would be noticing that.

F1 2017 not tested at max quality.

Ah so the tests are only legit if we peg the resource limits and remove the GPU shader performance as the bottleneck.

 

[EarthDog]

Lol, bologna... i can notice that in FPS..

 

[Fouquin]

Lol, bologna... i can notice that in FPS..

Not bologna. 48FPS is 20.833ms between displayed frames, 40FPS is 25ms. In a blind test I doubt anyone would guess the difference between those two.

 

[MuhammedAbdo]

Ah so the tests are only legit if we peg the resource limits and remove the GPU shader performance as the bottleneck.

Testing high end cards with the max settings is the standard procedure in benchmarking. We are not going to invent the wheel here, or pander to specific GPUs. Kepler was announced as a bad aging architecture based on the same merits.

 

[EarthDog]

Not bologna. 48FPS is 20.833ms between displayed frames, 40FPS is 25ms. In a blind test I doubt anyone would guess the difference between those two.

In some titles, especially FPS, oh yes.

 

[cdawall]

In some titles, especially FPS, oh yes.

I don't know what this argument is about. The fury sucks. It was a 290X/390X with HBM and a prayer glued to it. The card actually gets outperformed by the 290X/390X depending on title. I really wish the 290X/390X drew less power because everything after it has been crap. Even the miners don't want furys.

 

[ikeke]

I've had these working side-by-side. Nano/Fury/FuryX is faster. (edit: unless you intentionally go over 4GB memory usage and compare to 8GB Hawaii)

 

[cdawall]

I've had these working side-by-side. Nano/Fury/FuryX is faster. (edit: unless you intentionally go over 4GB memory usage and compare to 8GB Hawaii)

I said specific titles for a reason there are games that exceed 4GB of memory. Hence why the 980Ti is aging so much better.

 

[Fouquin]

Testing high end cards with the max settings is the standard procedure in benchmarking. We are not going to invent the wheel here, or pander to specific GPUs. Kepler was announced as a bad aging architecture based on the same merits.

Except Kepler hasn't aged poorly and the people proposing that idea were proven wrong multiple times. Also, what is the time constraint on considering hardware "high end"? 2 years, 5, 10? A lot of publications retire GPUs from testing at 3 years, and Fiji is not far from that mark. Do we expect GPUs to be topping performance metrics in applications designed to utilize exceptionally more resources?

Even the miners don't want furys.

If you're mining ZCash the Fury X is actually quite good. 550 Sol/s at 145W is better perf/watt than a GTX 1070.

 

[cdawall]

If you're mining ZCash the Fury X is actually quite good. 550 Sol/s at 145W is better perf/watt than a GTX 1070.

I have never seen someone push over 500-530 sol at any less than a real 220w at the wall. Myself included.

 

[Fouquin]

I have never seen someone push over 500-530 sol at any less than a real 220w at the wall. Myself included.

Huh weird I must have some magical space-dust mining rig then because that's what I've been running for 6 months. 1.1v core, 544.4 Sol/s average with 552.2 Sol/s recorded peak. Meanwhile the 1070 AMP Extreme is sucking down 156W and getting 466.3 Sol/s.

 

[ikeke]

I said specific titles for a reason there are games that exceed 4GB of memory. Hence why the 980Ti is aging so much better.

Dunno, when i game i set the in-game quality settings so that i don't trip the limit. Usually means losing some AA, but i really dont find much value in 4-8x AA at 1440p. If you intentionally trip the limit and the card faceplants then i can't blame it [the card], really.

WRT newer games, i've lost faith in developers long time ago, so i usually buy the games ~6-12 months after release, when they are patched and drivers have matured. I don't like being a beta tester anymore, i value my time and want the experience.

 

[cdawall]

Huh weird I must have some magical space-dust mining rig then because that's what I've been running for 6 months. 1.1v core, 544.4 Sol/s average with 552.2 Sol/s recorded peak. Meanwhile the 1070 AMP Extreme is sucking down 156W and getting 466.3 Sol/s.

I kind of want to play with them some more now, how are you measuring power consumption?

Dunno, when i game i set the in-game quality settings so that i don't trip the limit. Usually means losing some AA, but i really dont find much value in 4-8x AA at 1440p. If you intentionally trip the limit and the card faceplants then i can't blame it [the card], really.

WRT newer games, i've lost faith in developers long time ago, so i usually buy the games ~6-12 months after release, when they are patched and drivers have matured. I don't like being a beta tester anymore, i value my time and want the experience.

"play the game at lower settings" That is not a real answer.

 

[ikeke]

Reducing AA != lower settings.

 

[cdawall]

Reducing AA != lower settings.

Except for the whole lower settings part of it

 

[ikeke]

I guess being old and having poor eyesight helps, AA'ing at 1440p seems kinda conter-intuitive for me. For each it's own, though.

 

[xkm1948]

I guess being old and having poor eyesight helps, AA'ing at 1440p seems kinda conter-intuitive for me. For each it's own, though.

4K and AA maybe redundant, 1440P and AA is a must.