AMD Fury X Reviews

[blastingcap]

If you reviewed 10 or 15 games then there would be no grounds for any "bias" accusations

Let's be real here, you know they would still be accused of bias by some people no matter what they did.

 

[moonbogg]

@ wand3r3r

I got both 980ti's at a super easy 1360 or so MHz with no voltage increase. The fans were loud, but the GPU's had no issues at all, so you can imagine how easy 1400 would be from here. You can then imagine how reachable 1500 is with a voltage bump.

 

[SexyK]

I guess I'll ask you this question over here, Brent: Why do you review such a small number of games? If you reviewed 10 or 15 games then there would be no grounds for any "bias" accusations...plus, you would likely have reached a somewhat different conclusion.

I believe earlier in the thread he stated that their methodology requires extensive playtime with each game, so there is simply not time to include more than 5-6 titles if they want to publish a review on the day the NDA is lifted.

 

[Jaydip]

Where are people getting the 980 TI mythical overclock numbers from?

According to hwbot (~250 submissions) it reaches 1294 MHz.
The way people keep repeating 1400-1500 MHz is strange if those numbers are true. Please correct me if they are incorrect.

The Titan X does OC very well at 1410 MHz. (2300 submissions!)

It seems like a perpetuated forum myth to me. I would venture a guess that the 980 TI is intentionally limited or binned with worse chips, therefore ensuring that the Titan X does indeed remain on top (even OC vs. OC).

Now my take on the Fury...

AMD deserves props for a few things:
Noise
Heat
AIO cooler. (A considerable expense/bonus for most people)
Power consumption. (Not my thing, but the forum seems infatuated with it)
Essentially matching the best.. (Might not much OC headroom but stock vs. stock is very close)
SFF. (Again an infatuation lately)
4k Crossfire looks hot (figuratively speaking, need FCAT)
HBM (partially created by them)
If you don't overclock and want reference (say multi card) I'd take it over the ti reference

Where it messed up/lacks:
Performance isn't high enough to decisively win anything (until 4k)
OC is very limited, at least without voltage. Let's see where this shakes out. It won't be thermally limited.
No HDMI 2 (doesn't matter to me, but I can see why it should have it)

And most importantly, price. A little less would have potentially made it an absolute must buy. Either way if it's such of a limited edition it will sell out regardless (apparently).

The Fury will be the card to watch. If it's 95% of the X then it'll be a lot easier decision. The nano will also be interesting, depending on pricing and exact performance.

Luckily I am not in the market at this very moment since I am too busy with other things for a while.

I believe when people are talking about oced 980 Ti they are talking about Gigabyte 980 Ti G1 which is a very fast chip and even faster than Tx. It oces very well(check the TPU review of it). I belive the card is good but it seems AMD was not prepared for 980Ti at all.

 

[Azix]

@ wand3r3r

I got both 980ti's at a super easy 1360 or so MHz with no voltage increase. The fans were loud, but the GPU's had no issues at all, so you can imagine how easy 1400 would be from here. You can then imagine how reachable 1500 is with a voltage bump.

Last I understood voltage didn't give much of a bump with maxwell. Though that was when I was looking at 970s. People weren't even sure they could up the voltage and when they thought they did, progress was minimal. Is the 980Ti Different?

 

[Alatar]

Where are people getting the 980 TI mythical overclock numbers from?

According to hwbot (~250 submissions) it reaches 1294 MHz.
The way people keep repeating 1400-1500 MHz is strange if those numbers are true. Please correct me if they are incorrect.

A lot of people on hwbot report the non boosted clocks of the cards instead of the actual running clocks. There's no real rules that dictate that you have to go one way or the other.

So using hwbot for average NV OCs doesn't really work. For AMD yes.

 

[blastingcap]

Here you go.. this will help ease the pain of no review to read, but instead, look at the bar charts.

http://www.anandtech.com/bench/product/1442?vs=1513 Fury vs 980
http://www.anandtech.com/bench/product/1496?vs=1513 Fury vs 980 Ti

So basically:

At 1080p, Fury X = GTX 980

At 4K, Fury X = a half step behind GTX 980 Ti (but varies depending on game)

We had people complaining about HardOCP comparing the Fury X to the GTX 980 but that is in fact the more apt comparison at 1080p if these Anandtech numbers are right. Especially looking at MINIMUM frame rates which matter far more than average frame rates. And the Fury X is watercooled. If you max OC a Fury X and a watercooled GTX 980 Ti I think the latter would easily outpace the Fury X at pretty much every game, sometimes by a quite a lot.

 

[moonbogg]

Last I understood voltage didn't give much of a bump with maxwell. Though that was when I was looking at 970s. People weren't even sure they could up the voltage and when they thought they did, progress was minimal. Is the 980Ti Different?

I have no idea about the 970. I'm personally expecting a solid 1400 under water. I'd be happy with that. It seems reasonable. No need to push for 1500 when 1400 is already well up there. I bet most cards if not all can hit 1400 easily. Maybe Fury can clock high as well, but it looks like it will take a lot more to get there.

 

[wand3r3r]

@ wand3r3r

I got both 980ti's at a super easy 1360 or so MHz with no voltage increase. The fans were loud, but the GPU's had no issues at all, so you can imagine how easy 1400 would be from here. You can then imagine how reachable 1500 is with a voltage bump.

Yeah one can imagine but then again I've had a couple lightnings, one responded well with voltage and the other couldn't do anything with more voltage. It's always down to the silicon lottery in the end.

The average oc numbers from a considerable amount of submissions >>> forum speak for numbers. Of course there will be some higher and some lower.

I believe when people are talking about oced 980 Ti they are talking about Gigabyte 980 Ti G1 which is a very fast chip and even faster than Tx. It oces very well(check the TPU review of it). I belive the card is good but it seems AMD was not prepared for 980Ti at all.

• The gaming (default) mode base clock is ticking over at 1152 MHz, while there is boost allowance up-to 1241 MHz.
• However if you fire up Gigabyte's OC Guru software and hit the OC mode button, you card will run at default at 1291 MHz (on the Boost clock -- baseclock is 1190 MHz).

It does look like a nice card, guaranteed to run at 1190 MHz. The average OC is still 1290 tho, not 1400-1500 which some claim. (Anyways at almost the same price it's a no brainer to get the G1, I just don't agree with myths being spread when the data demonstrates otherwise)

I have no idea about the 970. I'm personally expecting a solid 1400 under water. I'd be happy with that. It seems reasonable. No need to push for 1500 when 1400 is already well up there. I bet most cards if not all can hit 1400 easily. Maybe Fury can clock high as well, but it looks like it will take a lot more to get there.

Except they clearly can't. If 1290 is average, far less then half will reach 1350+. Unless you have evidence showing otherwise?

 

[x3sphere]

My EVGA ACX does 1400 MHz with no voltage bump. It actually boosted to nearly 1300 MHz out of the box.

 

[flopper]

seems amd is doing things right, beating everything nvidia has with the 390 series and now have the awesome card at entusiast level and soon windows 10 is about to come down and unleash the fury.

couldnt be happier

 

[wand3r3r]

A lot of people on hwbot report the non boosted clocks of the cards instead of the actual running clocks. There's no real rules that dictate that you have to go one way or the other.

So using hwbot for average NV OCs doesn't really work. For AMD yes.

Where do people report their clocks? AFAIK they seem to be polling benchmarks. Nobody would be "entering" their clocks. Sounds like another forum myth, unless you have evidence to the contrary.

Are you claiming the 1400 average OC on the TX is the base clock? :awe:

 

[AtenRa]

Ofcourse you can. Coolers going beyond the PCB isnt new. What do you think the non X will be?

Im talking about keeping the same small size as Fury-X with Water. You cannot do that with AIR cooler unless your cooler thickness is double of what your ordinary cards coolers have.

 

[SolMiester]

Yeah one can imagine but then again I've had a couple lightnings, one responded well with voltage and the other couldn't do anything with more voltage. It's always down to the silicon lottery in the end.

The average oc numbers from a considerable amount of submissions >>> forum speak for numbers. Of course there will be some higher and some lower.

• The gaming (default) mode base clock is ticking over at 1152 MHz, while there is boost allowance up-to 1241 MHz.
• However if you fire up Gigabyte's OC Guru software and hit the OC mode button, you card will run at default at 1291 MHz (on the Boost clock -- baseclock is 1190 MHz).

It does look like a nice card, guaranteed to run at 1190 MHz. The average OC is still 1290 tho, not 1400-1500 which some claim. (Anyways at almost the same price it's a no brainer to get the G1, I just don't agree with myths being spread when the data demonstrates otherwise)



Except they clearly can't. If 1290 is average, far less then half will reach 1350+. Unless you have evidence showing otherwise?

Perhaps you should get one before claiming they dont! You have just been advised that hwbot has the set OC not the boost!

 

[Erenhardt]

My EVGA ACX does 1400 MHz with no voltage bump. It actually boosted to nearly 1300 MHz out of the box.

@ wand3r3r

I got both 980ti's at a super easy 1360 or so MHz with no voltage increase. The fans were loud, but the GPU's had no issues at all, so you can imagine how easy 1400 would be from here. You can then imagine how reachable 1500 is with a voltage bump.

Not true. When you overclock the maxwell gpu it automatically adjusts the voltages, that is where the overclocking headroom comes from.

Even if you don't touch the voltage, the cards bios do that for you. Here is example:

With amd cards, the voltage is fixed and is altered separately from the core frequency which gives more flexibility for the user.
Maxwell overclocking is kind of automated in regards to voltage.

 

[moonbogg]

Except they clearly can't. If 1290 is average, far less then half will reach 1350+. Unless you have evidence showing otherwise?

Evidence? Of what? Like a door to door survey of what 980ti owners are doing with their cards? Nothing short of that would be evidence. You want real, hard evidence, like the kind that could hold up in court. That's impossible to produce, sorry.
I can tell you about my cards and they boost to 1360 with a simple slide of the dial and the ram went right up +300 also with no issue, no extra voltage and no tweaking. Everywhere I read people have great OC results with zero effort. That's evidence enough for me. The 980ti: Buy it with confidence.

@Erenhardt,

That's interesting. I didn't know that. I wonder what sliding the voltage dial does in precision x?

 

[wand3r3r]

Perhaps you should get one before claiming they dont! You have just been advised that hwbot has the set OC not the boost!

"Advised" by yet another forum myth?

Right. So I should buy one to demonstrate that the average of 250 submissions is wrong. So you think the Titan X average of 1400 is the base clock? ()

/why do I bother

 

[SolMiester]

"Advised" by yet another forum myth?

Right. So I should buy one to demonstrate that the average of 250 submissions is wrong. So you think the Titan X average of 1400 is the base clock? ()

/why do I bother

Just saying you could relax calling people liars regarding their OC!..cause thats what you are doing..

 

[wand3r3r]

Evidence? Of what? Like a door to door survey of what 980ti owners are doing with their cards? Nothing short of that would be evidence. You want real, hard evidence, like the kind that could hold up in court. That's impossible to produce, sorry.
I can tell you about my cards and they boost to 1360 with a simple slide of the dial and the ram went right up +300 also with no issue, no extra voltage and no tweaking. Everywhere I read people have great OC results with zero effort. That's evidence enough for me. The 980ti: Buy it with confidence.

Personally I'll take a large data source over hearsay. HWBot is hard to refute. You can buy the 980 ti with confidence, just be aware that no OC is guaranteed. It doesn't need much to distance itself from the fury in any case, my only concern is passing off hearsay as fact.

Just saying you could relax calling people liars regarding their OC!..cause thats what you are doing..

Don't put words in my mouth. If you want debate stick to the facts and you don't need to twist anything.

 

[Alatar]

Where do people report their clocks? AFAIK they seem to be polling benchmarks. Nobody would be "entering" their clocks. Sounds like another forum myth, unless you have evidence to the contrary.

Are you claiming the 1400 average OC on the TX is the base clock? :awe:

hwbot takes the average clocks from the benchmark submissions that people send in.

And yes you have to manually enter the clock speeds with every submission which is why some people use the GPU-Z reported Nvidia clocks (not actual clock speed). Same goes for the scores, hardware, details etc. everything is manually added and that just means that average clocks reported for NV cards are going to be useless because people don't use the same standards for reporting clocks.

source: I've done a bunch of LN2, DICE, Phase change etc. benching on hwbot. Both 2d and 3d. I know how the submission system works.

 

[sze5003]

Looks like I will either get a 980ti or wait and see what happens with the drivers of the fury as time passes. My card is decent now but getting a bit old so I'm unsure if I should even spend that kind of money if the performance won't be major.

 

[LTC8K6]

You cannot have an AIR cooler to fit in such a small pcb and dissipate 275W TDP, unless it would be 2x times as thick as triple-slot coolers.

Well, I was thinking that we have an R9-285 in a short air cooled version. We have an R9-380 also.

I didn't think Fury was that much higher in power use.

 

[wand3r3r]

hwbot takes the average clocks from the benchmark submissions that people send in.

And yes you have to manually enter the clock speeds with every submission which is why some people use the GPU-Z reported Nvidia clocks (not actual clock speed). Same goes for the scores, hardware, details etc. everything is manually added and that just means that average clocks reported for NV cards are going to be useless because people don't use the same standards for reporting clocks.

source: I've done a bunch of LN2, DICE, Phase change etc. benching on hwbot. Both 2d and 3d. I know how the submission system works.

I wasn't aware of the manual part. It seems like it would still be accurate since you have to submit proof (of the clocks)? Do you have a source for gpuz reporting the wrong clocks?

(Example of a TX, quite the suicide OC )

I'll try avoid derailing the thread, my point was that fury x has to compete with 980 TI at about 1300 MHz to compete OC vs OC. Anything more is simply winning the silicon lottery (again, unless gpu-z actually does lie).

Maybe I should make another thread to get the bottom of it completely.

 

[wand3r3r]

Ok but when I go out to buy a new card, I see a $650 Fury X and a $690 Gigabyte G1 that's 25% faster out of the box with 0 overclocking required. What am I going to buy/recommend? Overclocking is not guaranteed but that 25% faster performance is warrantied by Gigabyte. That's not even the fastest 980Ti as Zotac AMP! 980Ti should have 1355mhz Boost out of the box.

RS I have to give you credit. Your stance is exactly the same as always. :thumbsup:

It's a no brainer situation again if you value money over "brand".

 

[AtenRa]

Well, I was thinking that we have an R9-285 in a short air cooled version. We have an R9-380 also.

I didn't think Fury was that much higher in power use.

Its about TDP, Fury Nano at 175W TDP can use a small PCB and aircooling. But Fury-X is a 275W TDP GPU, you will need a very large aircooler making the small PCB card impossible to create.

So either you create a large AirCooler or you use WaterCooling and keep the overall Graphics Card size as small as possible.