Anandtech's AMD Fury X review

[Z15CAM]

I run CF XSPC Hawaii 290X RAZOR's and will consider 2nd Gen AMD Radeon FURY X Figi's in the future. Love AMD's Figi design.

By then I'll require a new Processor and MB ;o)

 

[96Firebird]

oh, what exactly do you mean by boost and base clock?

I'm sure someone else can explain it better, as I don't even know what all the terms mean. From what I understand, Nvidia has 3 clockspeeds to note; base, boost, and actual. Base is the guaranteed minimum for the card, it should never fall below that. Boost is the average clockspeed during gaming, however the actual speeds are typically over the boost clock. As you can see in the Anandtech 980 Ti review, actual clockspeeds averaged around 1150Mhz during gaming. If he is getting actual clockspeeds of 1492MHz during gaming, that is a 30% overclock. Which is really good, don't get me wrong, but as I showed above, is pretty typical of those GPUs with better cooling.

 

[myocardia]

oh, what exactly do you mean by boost and base clock?

It is more or less the identical thing as Intel's turbo and base clocks, for their CPUs.

 

[Silverforce11]

if all 980 ti can oc to 50% over reference, there is no competition. it would be so worth it to get the 980 ti evga hybrid.

but that is a big if.

is there a dedicated website with user submitted oc numbers + proof?

oh, what exactly do you mean by boost and base clock?

It's not 50% OC over reference because that's misleading.

Out of the box, 980Ti reference reach 1.2ghz actual boost clocks from various sites I've seen, including AT's review.

If they OC to 1.45ghz, that's 250mhz on top. ~20% OC. If they reach 1.5ghz, that's a 25% OC.

Custom 980Ti models reach >1.3ghz boost out of the box, that's why they are so much faster than Titan X.

http://www.guru3d.com/articles-pages/gigabyte-geforce-gtx-980-ti-g1-gaming-soc-review,1.html

An example here, its set for nearly 1.3ghz default boost, which usually translate to higher actual in-game boost.

They can be OC further, there's usually 10-15% performance left, depending on the models.

http://www.techpowerup.com/reviews/EVGA/GTX_980_Ti_SC_Plus/33.html
http://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_G1_Gaming/33.html

It's why Fury X looks crap compared to these custom 980Ti offer great bang for buck comparatively and has excellent OC potential eventhough they already are highly clocked out of the box!

For Fury X to reach 20% OC, it needs to hit ~1.25ghz which is just not possible without vcore tools. 25% OC on Fury X is 1,312Mhz, which is unlikely to happen even with vcore. GCN peaks around 1.25ghz when we examine 7970, R290X. Without major driver improvements, OC v OC, Fury X still loses with vcore tools.

 

[amenx]

AT's 2015 GPU benchmark charts now include FuryX.

vs 980ti:

http://anandtech.com/bench/product/1496?vs=1513

 

[railven]

It's not 50% OC over reference because that's misleading.

Out of the box, 980Ti reference reach 1.2ghz actual boost clocks from various sites I've seen, including AT's review.

If they OC to 1.45ghz, that's 250mhz on top. ~20% OC. If they reach 1.5ghz, that's a 25% OC.

Custom 980Ti models reach >1.3ghz boost out of the box, that's why they are so much faster than Titan X.

http://www.guru3d.com/articles-pages/gigabyte-geforce-gtx-980-ti-g1-gaming-soc-review,1.html

An example here, its set for nearly 1.3ghz default boost, which usually translate to higher actual in-game boost.

They can be OC further, there's usually 10-15% performance left, depending on the models.

http://www.techpowerup.com/reviews/EVGA/GTX_980_Ti_SC_Plus/33.html
http://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_G1_Gaming/33.html

It's why Fury X looks crap compared to these custom 980Ti offer great bang for buck comparatively and has excellent OC potential eventhough they already are highly clocked out of the box!

For Fury X to reach 20% OC, it needs to hit ~1.25ghz which is just not possible without vcore tools. 25% OC on Fury X is 1,312Mhz, which is unlikely to happen even with vcore. GCN peaks around 1.25ghz when we examine 7970, R290X. Without major driver improvements, OC v OC, Fury X still loses with vcore tools.

This is pretty much what sold me to Nvidia for another round. Looking around for $680+ some change, I can get a Hybrid setup, keep it clocked to around 1350+ core, and just leave a Fury X behind. Icing is more VRAM, less power (assuming an OC/OV'd Fury X), and probably still cooler.

Just all win.

 

[boozzer]

It's not 50% OC over reference because that's misleading.

Out of the box, 980Ti reference reach 1.2ghz actual boost clocks from various sites I've seen, including AT's review.

If they OC to 1.45ghz, that's 250mhz on top. ~20% OC. If they reach 1.5ghz, that's a 25% OC.

Custom 980Ti models reach >1.3ghz boost out of the box, that's why they are so much faster than Titan X.

http://www.guru3d.com/articles-pages/gigabyte-geforce-gtx-980-ti-g1-gaming-soc-review,1.html

An example here, its set for nearly 1.3ghz default boost, which usually translate to higher actual in-game boost.

They can be OC further, there's usually 10-15% performance left, depending on the models.

http://www.techpowerup.com/reviews/EVGA/GTX_980_Ti_SC_Plus/33.html
http://www.techpowerup.com/reviews/Gigabyte/GTX_980_Ti_G1_Gaming/33.html

It's why Fury X looks crap compared to these custom 980Ti offer great bang for buck comparatively and has excellent OC potential eventhough they already are highly clocked out of the box!

For Fury X to reach 20% OC, it needs to hit ~1.25ghz which is just not possible without vcore tools. 25% OC on Fury X is 1,312Mhz, which is unlikely to happen even with vcore. GCN peaks around 1.25ghz when we examine 7970, R290X. Without major driver improvements, OC v OC, Fury X still loses with vcore tools.

but that is ok as long as it is close enough in performance since I get water cooling out of it for the same price. the evga hybrid is 750$. as long as performance is within 5% I honestly don't give a damn if I can get super nice clc cooling.

Noise is a deal breaker for me. 980 ti would not be able to do the oc without the clc hybrid. that makes fury x a very good buy if it can oc just 20% I am basically waiting for factory oc 3rd party cards to be release to make my decision.

 

[tential]

Has anyone seen the results from ixbt's review? That had a later driver revision than other review sites apparently. So maybe, because anandtech's review is later they'll use a later revision and be able to confirm ixbt's results?

 

[Pandora's Box]

It's not 50% OC over reference because that's misleading.

Out of the box, 980Ti reference reach 1.2ghz actual boost clocks from various sites I've seen, including AT's review.

If they OC to 1.45ghz, that's 250mhz on top. ~20% OC. If they reach 1.5ghz, that's a 25% OC.

Custom 980Ti models reach >1.3ghz boost out of the box, that's why they are so much faster than Titan X.
...

You make it sound like the Titan X can't overclock. (Mine's running at 1450MHz). 980Ti can't touch a Titan X when they are both at the same clock speeds. Takes about 100Mhz more on a 980Ti to match a Titan X.

 

[Subyman]

Has anyone seen the results from ixbt's review? That had a later driver revision than other review sites apparently. So maybe, because anandtech's review is later they'll use a later revision and be able to confirm ixbt's results?

That "newer" driver has already been debunked by AMDMatt.

http://forums.anandtech.com/showthread.php?p=37512868&highlight=#post37512868

 

[5150Joker]

1492 oc number. can we all get his luck with our gpus?

I hit 1500+ with both my Titan X's. 1400+ is very common with GM200.

 

[tviceman]

but that is ok as long as it is close enough in performance since I get water cooling out of it for the same price. the evga hybrid is 750$. as long as performance is within 5% I honestly don't give a damn if I can get super nice clc cooling.

Noise is a deal breaker for me. 980 ti would not be able to do the oc without the clc hybrid. that makes fury x a very good buy if it can oc just 20% I am basically waiting for factory oc 3rd party cards to be release to make my decision.

Open air 980's can do 20+% OC with quiet noise levels and adequate cooling. It won't be Fury X temps, but the difference between 76 C and 60 C is almost entirely e-peen talking points. The power draw between a potential OV Fury X and OC 980 TI is going to be way more noticeable on your electric bill and room temperature than the noise levels between aftermarket 980 TI's and Fury X CLC water.

If you want AMD or you absolutely must have CLC then that's fine. But as it stands right now there are just as many cons, if not more, going with Fury X even if it CAN overvolt and get 20% OC.... which is still a major, major IF right now. Tied at 4k (still not fast enough IMO), 5-10% slower at 1440p, and 15% slower at 1080p 120/144hz all before OCing is taken into account.

 

[boozzer]

Open air 980's can do 20+% OC with quiet noise levels and adequate cooling. It won't be Fury X temps, but the difference between 76 C and 60 C is almost entirely e-peen talking points. The power draw between a potential OV Fury X and OC 980 TI is going to be way more noticeable on your electric bill and room temperature than the noise levels between aftermarket 980 TI's and Fury X CLC water.

If you want AMD or you absolutely must have CLC then that's fine. But as it stands right now there are just as many cons, if not more, going with Fury X even if it CAN overvolt and get 20% OC.... which is still a major, major IF right now. Tied at 4k (still not fast enough IMO), 5-10% slower at 1440p, and 15% slower at 1080p 120/144hz all before OCing is taken into account.

ok, tell me which is the best 980 ti air cooled factory oc gpu right now? the G1? I want to check it's temp and especially noise. and oc 980 ti uses a ton of wattage too, please don't bring it up anymore. any talk about raising room temperature is just ridiculous. please leave that kind of talk out of any discussion with me.

 

[tential]

That "newer" driver has already been debunked by AMDMatt.

http://forums.anandtech.com/showthread.php?p=37512868&highlight=#post37512868

Lol... I hoped oh well. I guess AMD Fury X really is a failed launch. To think people purchased the stock too after the presentation. Those sales numbers won't be too good nor their stock price a couple months from now.

 

[3DVagabond]

Lol... I hoped oh well. I guess AMD Fury X really is a failed launch. To think people purchased the stock too after the presentation. Those sales numbers won't be too good nor their stock price a couple months from now.

We'll see. The card isn't a bad card at all. Consider that it's $100 cheaper than a hybrid 980 ti and it's 100% water cooled. It's cooler and quieter than any comparable performing air cooled card out there and we still have the air cooled versions to come. They're likely to be ~$100 cheaper. There's still the non X and Nano versions too that will cost even less.

 

[Subyman]

I'm still waiting to see what unlocked voltage does. Someone on OCN was able to push the memory to 600mhz and get a 3k score increase in firestrike, so I'm still keeping an open mind about the card. If it hits 1300mhz with a voltage bump, can overclock the memory consistently, and still stay under 65C while being quiet then it will be a decent alternative to the competition. Then we may get a trickle of driver improvements. Not the run away success that we were hoping for, but it may turn into a nice option.

Edit: Here's a link to the men oc:
http://www.overclock.net/t/1562593/...n-be-overclocked-after-all/0_50#post_24105109

 

[tviceman]

ok, tell me which is the best 980 ti air cooled factory oc gpu right now? the G1? I want to check it's temp and especially noise. and oc 980 ti uses a ton of wattage too, please don't bring it up anymore. any talk about raising room temperature is just ridiculous. please leave that kind of talk out of any discussion with me.

I think the EVGA SC+ reviewed extremely well with techpowerup. The G1 is less efficient and louder than the EVGA SC+. The MSI gaming cards have always been a personal favorite of mine, but there aren't any reviews online yet of the 980 TI MSI gaming.

The EVGA SC+ is 35 db under load (3 db louder than Fury X)
The EVGA SC+ is actually slightly more efficient than the 980 TI, which was already more efficient than the Fury X. The card draws slightly less power than the Fury X.
The EVGA SC+ is 12-13% faster than a reference 980 TI out of the box, which means it's close to 25% faster than Fury X at 1440p and close to 15% faster at 4k.
http://www.techpowerup.com/reviews/EVGA/GTX_980_Ti_SC_Plus/

The 980 TI doesn't use a "ton" of wattage when OC'd. Anandtech showed a 30 watt increase in Crysis 3 while performance increased over 20%. That is a massive gain in performance and a very minimal gain in power consumption. So check your facts next time.

Believe what you want about power usage and room temperature. Of course the size of the room definitely mitigates (or amplifies) the heat dump from the case, but an hour or two of gaming on my old OC'd GTX 465 card in a "normal" sized bedroom definitely raised the temp in the room by a few degrees.

 

[boozzer]

I think the EVGA SC+ reviewed extremely well with techpowerup. The G1 is less efficient and louder than the EVGA SC+. The MSI gaming cards have always been a personal favorite of mine, but there aren't any reviews online yet of the 980 TI MSI gaming.

The EVGA SC+ is 35 db under load (3 db louder than Fury X)
The EVGA SC+ is actually slightly more efficient than the 980 TI, which was already more efficient than the Fury X. The card draws slightly less power than the Fury X.
The EVGA SC+ is 12-13% faster than a reference 980 TI out of the box, which means it's close to 25% faster than Fury X at 1440p and close to 15% faster at 4k.
http://www.techpowerup.com/reviews/EVGA/GTX_980_Ti_SC_Plus/

The 980 TI doesn't use a "ton" of wattage when OC'd. Anandtech showed a 30 watt increase in Crysis 3 while performance increased over 20%. That is a massive gain in performance and a very minimal gain in power consumption. So check your facts next time.

Believe what you want about power usage and room temperature. Of course the size of the room definitely mitigates (or amplifies) the heat dump from the case, but an hour or two of gaming on my old OC'd GTX 465 card in a "normal" sized bedroom definitely raised the temp in the room by a few degrees.

that 12% is not good enough. I will search on my own.

 

[exar333]

At this point, AT should just wait to do their review when voltage tools are available for the Fury. No point in releasing a week late with no new info.

 

[Silverforce11]

I'm still waiting to see what unlocked voltage does. Someone on OCN was able to push the memory to 600mhz and get a 3k score increase in firestrike, so I'm still keeping an open mind about the card. If it hits 1300mhz with a voltage bump, can overclock the memory consistently, and still stay under 65C while being quiet then it will be a decent alternative to the competition. Then we may get a trickle of driver improvements. Not the run away success that we were hoping for, but it may turn into a nice option.

Edit: Here's a link to the men oc:
http://www.overclock.net/t/1562593/...n-be-overclocked-after-all/0_50#post_24105109

AMD can't even do vcore lock right in their drivers.

That is a huge gain just for a memory OC, we all thought it wouldn't be bandwidth limited, must be a latency thing.

 

[tviceman]

that 12% is not good enough. I will search on my own.

Crazy! 12% faster than a reference 980 TI, which means it's 20-25% faster at 1440p than Fury X for only $30 more than Fury X. If that isn't good enough, then suite yourself. AMD needs more customers like you.

 

[therealnickdanger]

That is a huge gain just for a memory OC, we all thought it wouldn't be bandwidth limited, must be a latency thing.

When AMD was touting the overclocking potential of Fury X, it sounded completely believable. I can only hope that AMD is cooking up a BIOS/driver package that will let people push these cards well beyond current limitations, limitations that appear completely artificial.

 

[Agent-A01]

AMD can't even do vcore lock right in their drivers.

That is a huge gain just for a memory OC, we all thought it wouldn't be bandwidth limited, must be a latency thing.

Take that with a grain of salt..

Ive seen users reporting gains with tessellation disabled to artificially increase their score

 

[exar333]

I'm still waiting to see what unlocked voltage does. Someone on OCN was able to push the memory to 600mhz and get a 3k score increase in firestrike, so I'm still keeping an open mind about the card. If it hits 1300mhz with a voltage bump, can overclock the memory consistently, and still stay under 65C while being quiet then it will be a decent alternative to the competition. Then we may get a trickle of driver improvements. Not the run away success that we were hoping for, but it may turn into a nice option.

Edit: Here's a link to the men oc:
http://www.overclock.net/t/1562593/...n-be-overclocked-after-all/0_50#post_24105109

Firestrike score improvement is great, but would we really see much of a real-world performance increase in fps? Asking because I would be surprised if the Fury X is anywhere close to bandwidth constrained at stock settings. Could be nice in multi-GPU maybe?

 

[sm625]

I'm still waiting to see what unlocked voltage does. Someone on OCN was able to push the memory to 600mhz and get a 3k score increase in firestrike, so I'm still keeping an open mind about the card. If it hits 1300mhz with a voltage bump, can overclock the memory consistently, and still stay under 65C while being quiet then it will be a decent alternative to the competition. Then we may get a trickle of driver improvements. Not the run away success that we were hoping for, but it may turn into a nice option.

Edit: Here's a link to the men oc:
http://www.overclock.net/t/1562593/...n-be-overclocked-after-all/0_50#post_24105109

This is really irritating. They also overclocked the cpu from 4.0GHz to 4.5GHz. So how are we supposed to know what effect the CPU overclock had on these results?

I also found this from page 7 of that thread:

A 1% increase in score for a 10% increase in mem clock. Seems kinda useless and pointless to me.