980 TI stock and OC vs. Fury X stock and OC [techspot]

[tviceman]

http://www.techspot.com/review/1329-buying-gpu-radeon-fury-geforce-980/

980 TI, even with the terrible reference model blower, wins all but 1 benchmark when both cards are overclocked. OC Fury X didn't win a single benchmark, but did manage a tie. Furthermore, the 980 TI power consumption when OC'd stays in line with its performance increase while the Fury X power consumption takes a dump on efficiency.

 

[Face2Face]

GTX 980Ti OC performance in Gears of War DX12 is nasty...just nasty.

 

[Head1985]

Yeah reference 980TI runs at very low clocks.Once OC even with super crap reference cooler and low TDP limits it is faster than aftermarket 1070.
Aftermarket 980TI is even 10% faster than this(with no throttling and no TDP limits)

 

[IEC]

Are either of the cards worth buying right now? Probably not, barring an awesome deal on a used card.

Was the Fury Nitro worth buying when it was on sale as low as $215AR and more regularly $230-240AR for almost the past ten months? Absolutely, especially if you have a monitor with Freesync.

 

[n0x1ous]

GTX 980Ti OC performance in Gears of War DX12 is nasty...just nasty.

this one surprised me. I figured that would be the one Fury X wins with it being a pure DX12/Async game.

 

[tamz_msc]

this one surprised me. I figured that would be the one Fury X wins with it being a pure DX12/Async game.

It's an Unreal Engine game, and I don't think it implements DX12 in a way that can really take advantage of AMD cards.

 

[Headfoot]

980 Ti was a great overclocker. Between 7900 series and 9xx series we have some monster overclockers on 28nm. Fury X definitely holding up better than 4GB would have under normal circumstances, but there's no doubt that when both were $650, aftermarket 980 Ti was the better buy. Today - I wouldn't buy either since we're so close to Vega and probably a 1080 Ti

 

[crisium]

In line with what I expected. Even when the trend of Nvidia aging poorly started to become accepted, I stated that Fury X will never match 980 Ti when both are max OC. The 980 Ti OC lead was so massively large, and it has more VRAM to boot, which in the future would still help. 980 Ti is the gem of late 28nm OCing (overall OC king is 7950 though). Fury X is likely the card closest to its limit out of box.

Fury X has closed the gap though. From slower at stock:
https://tpucdn.com/reviews/AMD/R9_Fury_X/images/perfrel_2560.gif

To faster now in this review.

With an OC it's still clearly in the 980 Ti's favour. But that max OC lead was bigger before too.

Are they worth buying today? Max OC 980 Ti can still beat Max OC 1070. But with a newer architecture, lower power consumption, and more VRAM, the 1070 is preferential to me especially since even if the max OC 980 TI leads it's usually not by a significant amount. So the 980 Ti would have to undercut the 1070 to be worth buying. A used market shopper with a hard limit below the 1070's price would look for a 980 Ti.

Fury X hasn't really had a stellar discount. I've seen $300, but it's generally a clearly slower card than the 1070 so it's iffy there. $215-$240 Fury air's made sense since they beat the 1060/480. Fury X would have to get almost that low imo to be recommended. Cheapest Fury X right now is $370, so that's a no-go. Fury air at $266 is a maybe.

 

[vissarix]

The overclockers dreams strikes again

Seriuosly tho i think fury x is one of the worst gpu ever launched..$650, 300W, no OC headroom despite being advertised as overclockers dream and faster then a Titan X..

If they launched it at $550 then it would have made sense...

 

[ElFenix]

i wonder how they do their power chart? do they plug in a killawatt and play for 20 minutes, record the kwh and then do the math to convert to watts?

 

[Bacon1]

Fury X hasn't really had a stellar discount. I've seen $300, but it's generally a clearly slower card than the 1070 so it's iffy there. $215-$240 Fury air's made sense since they beat the 1060/480. Fury X would have to get almost that low imo to be recommended. Cheapest Fury X right now is $370, so that's a no-go.

Exactly. The X was just if you wanted a WC card. The air was always the best bang/buck and only 5% slower than the X. The reason the power consumption goes crazy is because most reviewers just up the voltage to highest they can and set +50% power while the card is clearly held back by its design / engine design and not clocks (hence 10% OC giving 3% perf increase in game for instance). Would be nice to see them test the Fury Nitro instead, since that's the one that has been on sale monthly for under $250. The max power draw on a Fury non-X is also much smaller and very close to Maxwell: http://www.hardocp.com/article/2016/05/24/xfx_radeon_r9_fury_triple_dissipation_video_card_review/10

 

[tamz_msc]

The overclockers dreams strikes again

Seriuosly tho i think fury x is one of the worst gpu ever launched..$650, 300W, no OC headroom despite being advertised as overclockers dream and faster then a Titan X..

If they launched it at $550 then it would have made sense...

I don't think that the Fury X was ever advertised as an overclocker's dream. Especially when it was known that the R9 390/X did not overclock that well. It was advertised as a Titan X competitor with cherry-picked benchmarks, but then the GTX 980Ti happened.

The Fury/X were never intended to sell in volumes like the GTX 980Ti. It was more likely an experiment to test HBM.

 

[AtenRa]

i wonder how they do their power chart? do they plug in a killawatt and play for 20 minutes, record the kwh and then do the math to convert to watts?

Seems like peak power (Watt) to me from that graph.

 

[ElFenix]

Seems like peak power (Watt) to me from that graph.

It says average so I don't know why it would be peak

 

[AtenRa]

It says average so I don't know why it would be peak

You are right I didnt see it says average, well i have no idea how they measured this then.

 

[pj-]

Thank god there were almost no Fury X's available on launch day. I tried several times to buy one that day because of all the "it will age better due to dx12 and VR" talk.

Maybe in another 2 years the Fury X will start beating the 980ti in new releases, but by then I will likely be using a card that replaced the card that replaced my 980ti.

 

[n0x1ous]

Thank god there were almost no Fury X's available on launch day. I tried several times to buy one that day because of all the "it will age better due to dx12 and VR" talk.

Maybe in another 2 years the Fury X will start beating the 980ti in new releases, but by then I will likely be using a card that replaced the card that replaced my 980ti.

It is already catching up and beating it at stock. I think in Vulcan Doom it even beats max OC 980ti. Its been an interesting card, that I don't regret buying. The lack of OC headroom is a bummer as is the 4GB vram. On the positive, its the best reference designed card ever (imo) and cool and quiet under load and looks very sexy. Its has been a good fit for me for couch pc gaming on 1080p TV where the VRAM hasn't hurt me at that resolution

 

[Bacon1]

If they launched it at $550 then it would have made sense...

Guess like most people you missed the Fury air, which was a much better buy price/perf and perf/watt. Only ~5% slower than the X but $550 vs $650. The X cost more because of the water cooler. Cheapest WC'd 980 Ti was $750.

 

[Carfax83]

It's an Unreal Engine game, and I don't think it implements DX12 in a way that can really take advantage of AMD cards.

I don't think this is true. Gears of War 4's asynchronous compute implementation has a significant performance increase on AMD hardware, more than what you see in Sniper Elite 4.

The reason for NVidia's exceptional performance in this game is due to the engine being very well optimized for NVAPI.

 

[pj-]

It is already catching up and beating it at stock. I think in Vulcan Doom it even beats max OC 980ti. Its been an interesting card, that I don't regret buying. The lack of OC headroom is a bummer as is the 4GB vram. On the positive, its the best reference designed card ever (imo) and cool and quiet under load and looks very sexy. Its has been a good fit for me for couch pc gaming on 1080p TV where the VRAM hasn't hurt me at that resolution

Isn't the gap between 980ti and fury x usually larger at lower resolutions?

I'd like to see the max OC 980ti being beaten in Vulkan Doom. That would be impressive if it's running in the 1475-1500mhz range.

 

[Snarf Snarf]

This review makes me sad, not because of the results but the fact that the 980ti is still biting at the heels of Pascal. My 1500 MHz MSI Gaming card is begging to be upgraded but no one wants to give me 50% more performance at $650

 

[casiofx]

Yeah reference 980TI runs at very low clocks.Once OC even with super crap reference cooler and low TDP limits it is faster than aftermarket 1070.
Aftermarket 980TI is even 10% faster than this(with no throttling and no TDP limits)

Again with your lies.

It has been your daily routine to spread lies on GTX1070 and GTX980Ti

GTX980Ti hardly exceeds 1500Mhz unless you have golden sample. In the link posted by OP, 1.48Ghz 980Ti is faster than stock Gigabyte GTX1070 G1 on average on 7%. Did you forgot that the Gigabyte GTX1070 G1 can overclock as well?

http://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_1070_g1_gaming_review,29.html

Overclocked Gigabyte GTX1070 G1 on average are 9% faster than stock Gigabyte GTX1070 G1.

So we can see that overclocked Gigabyte GTX1070 G1 is slightly faster than 1.48Ghz GTX980TI.

1-2% variation is too small. GTX1070 and GTX980Ti is essentially equal in performance.

So stop spreading your lies.

i wonder how they do their power chart? do they plug in a killawatt and play for 20 minutes, record the kwh and then do the math to convert to watts?

The power consumption chart on the review site is not that accurate.
They measured the power consumption while running games in 1080p. For GPU power consumption they should run it at 1440p or 4K since they may hit the CPU bottleneck first on 1080p, leaving the GPUs not running 100%. 4K would best illustrates GPU power consumption since they would stress the GPUs a lot, especially GPUs that are heavily overclocked.

 

[tamz_msc]

I don't think this is true. Gears of War 4's asynchronous compute implementation has a significant performance increase on AMD hardware, more than what you see in Sniper Elite 4.

The reason for NVidia's exceptional performance in this game is due to the engine being very well optimized for NVAPI.

That's still 10 percent at most in the best-case scenario. Nothing like the 30 percent gains in Doom Vulkan with Shader Intrinsics and Async Compute on GCN.

Doom Vulkan remains the best implementation of these low-level APIs till date.

Moreover, I think that Async compute only alleviates the CPU overhead issue on GCN cards in Gears of War 4.

 

[Shmee]

Ya the Fury nitro is a great card at $250 or less. Doesn't OC so well, but keep in mind some of them unlock. 980Ti though, has some massive OC potential. It is a pity that Nvidia has artificially limited OC tweaks on Pascal. That said, they still OC pretty well, albeit with pretty much identical boost ceilings.

 

[Jaydip]

I don't think that the Fury X was ever advertised as an overclocker's dream. Especially when it was known that the R9 390/X did not overclock that well. It was advertised as a Titan X competitor with cherry-picked benchmarks, but then the GTX 980Ti happened.

The Fury/X were never intended to sell in volumes like the GTX 980Ti. It was more likely an experiment to test HBM.

It was indeed advertised as an Overclocker's dream by their own management team.