Anandtech's AMD Fury X review

[Owls]

wow still no review. safe to say this isn't happening.

 

[n0x1ous]

I don't know why Ryan Smith keeps saying tomorrow, the next day etc etc.....either hit the date you say or don't say a date at all

Will probably end up like the GTX 960 review we were promised and never got.

 

[ShintaiDK]

wow still no review. safe to say this isn't happening.

The numbers are in the anandtech bench database. I guess thats the closest we get for now.

 

[Keysplayr]

you get towers with 4x 120mm top and + rear. Like the Corsair Obsidian 900D etc. so no problem.

You get? You mean it comes with the FuryX? Because that is a north of 300.00 dollar case. Most cases I've used and purchase, have room for probably only one of those radiators at the back of the case. Two if I splurge on a higher end case. I guess it's par for the course though. Spend this much on GPUs and a 300 to 400 dollar case becomes a moot point.

 

[Keysplayr]

I don't know why Ryan Smith keeps saying tomorrow, the next day etc etc.....either hit the date you say or don't say a date at all

Will probably end up like the GTX 960 review we were promised and never got.

Perhaps AT isn't as reliant on review hits revenue as we believe they are?
Maybe they can afford to take a chill on new hardware releases now and then.

 

[n0x1ous]

Perhaps AT isn't as reliant on review hits revenue as we believe they are?
Maybe they can afford to take a chill on new hardware releases now and then.

Perhaps that's true. I'm more frustrated with the promised dates not being met again and again. Don't give us a date if it can't be met.

 

[ShintaiDK]

Or who knows if they lost their sample and rights.

 

[Sabrewings]

you get towers with 4x 120mm top and + rear. Like the Corsair Obsidian 900D etc. so no problem.

From what I've seen of the hose lengths on the Fury X, I don't know even that'll work out. I have a 900D and unless the hoses can be extended most of the roof ports are out of reach. Also, unless there's zero encroachment on the next fan location (very uncommon with radiators) you can't put them next to eachother in the top because it uses normal fan spacing. The lower compartment locations might be an option or the fonts if you keep the drive cages in the bottom.

I can see potentially getting all four mounted, but it wouldn't be pretty and would probably strain hoses. Rear mount, rear most top mount, front mount, and lower window side. Personally, I'd rather have my own cooling loop than deal with that mess.

 

[Headfoot]

From what I've seen of the hose lengths on the Fury X, I don't know even that'll work out. I have a 900D and unless the hoses can be extended most of the roof ports are out of reach. Also, unless there's zero encroachment on the next fan location (very uncommon with radiators) you can't put them next to eachother in the top because it uses normal fan spacing. The lower compartment locations might be an option or the fonts if you keep the drive cages in the bottom.

I can see potentially getting all four mounted, but it wouldn't be pretty and would probably strain hoses. Rear mount, rear most top mount, front mount, and lower window side. Personally, I'd rather have my own cooling loop than deal with that mess.

Yeah, if you're buying 4 Fury X and you already have a $1000 cpu, might as well go full custom

 

[imported_Tridam]

No, we're not saying that. Only hw.fr reviewer who has no grasps of water cooling claims that and a few silly people on here who jump onboard the negative hype train.

Not to mention other sites find it running much cooler, even under furmark. There's no way the GPU is reaching those temps unless they disabled temp throttling (not sure how), turn off the pump & fan on the rad and let it go into meltdown.

Its a case of a reviewer who has an uncalibrated FLIR camera and the lack of common sense to question his data, against every other site & end users who claims it runs cool.

As anyone who has exp with water will tell you, the loop is cooler than the CPU/GPU/Heat-source. If your GPU is reading 60C, there's no way your loop coolant/tubes are 100C. You have to be a complete idiot to publish that data.

Hi,

I usually stay away from answering back to such kind words; everybody is free to have an opinion about our work. However in this particular case, an answer appears to be required as it seems the exact same bullshit about our thermal tests is spreading on many forums (which I'd say is probably not a random coincidence).

First I've never written or implied in any way that the coolant reached 100 °C. That would be ridiculous. What our tests showed and what we wrote is that some components of the GPU power stage reached 104 °C. The thermal imagery shows the coolant tubes are at 54-56 °C and the internal sensor of the GPU is reporting 64 °C. All those numbers make perfect sense and the cooling system is just not doing a great job cooling down those components of the power stage, even though they're rated for such high temperatures. The conclusion is I wouldn't recommend playing with significant vmod with the Fury X as is.

Then there is no malicious testing with power virus or anything like that. In a closed case with proper ventilation, I use a 45min loop of 3DMark 11 scene 1. It pushes a similar power load as demanding games such as Anno 2070 : ~285W on the board sampled by AMD and tested here. I got my hands on a second sample for which power load was ~300W in the same tests but I didn't use if for thermal tests. Furmark for example would push the load to ~385W, but again I don’t use it for thermal tests.

Damien

 

[iiiankiii]

You get? You mean it comes with the FuryX? Because that is a north of 300.00 dollar case. Most cases I've used and purchase, have room for probably only one of those radiators at the back of the case. Two if I splurge on a higher end case. I guess it's par for the course though. Spend this much on GPUs and a 300 to 400 dollar case becomes a moot point.

Here's one for under $100:

Fractal Design S

Radiator Support

Front
360, 280, 240, 140 and 120mm radiators of all thicknesses

Rear
120 or 140mm radiator

Top
420*, 360, 280*, 240, 140* and 120mm radiators

Bottom
120mm radiator (limits PSU length to 165mm)

 

[Silverforce11]

Hi,

I usually stay away from answering back to such kind words; everybody is free to have an opinion about our work. However in this particular case, an answer appears to be required as it seems the exact same bullshit about our thermal tests is spreading on many forums (which I'd say is probably not a random coincidence).

First I've never written or implied in any way that the coolant reached 100 °C. That would be ridiculous. What our tests showed and what we wrote is that some components of the GPU power stage reached 104 °C. The thermal imagery shows the coolant tubes are at 54-56 °C and the internal sensor of the GPU is reporting 64 °C. All those numbers make perfect sense and the cooling system is just not doing a great job cooling down those components of the power stage, even though they're rated for such high temperatures. The conclusion is I wouldn't recommend playing with significant vmod with the Fury X as is.

Then there is no malicious testing with power virus or anything like that. In a closed case with proper ventilation, I use a 45min loop of 3DMark 11 scene 1. It pushes a similar power load as demanding games such as Anno 2070 : ~285W on the board sampled by AMD and tested here. I got my hands on a second sample for which power load was ~300W in the same tests but I didn't use if for thermal tests. Furmark for example would push the load to ~385W, but again I don’t use it for thermal tests.

Damien

Your own picture shows the loop/tubes are bright as your claimed 100C reading on the tube over the VRMs.

This is why people claim your results are rubbish. Either the loop is <64C (GPU temps), or the GPU temp is much much higher.

Take a few more FLIR shots and see if its repeatable and why you are getting results that don't match every other site.

ps. Take ones of the radiator too. In a closed loop, the coolant & rad should be lower temps than the heat source.

 

[imported_Tridam]

Are you really talking about my IR pictures ???

http://www.hardware.fr/articles/937-8/temperatures-nuisances-sonores.html

I know how to use my IR camera (we were actually the first to introduce IR imaging in graphic cards reviews in 2008 or 2009), how to make sure my results are correct and how a closed loop works, thanks for your patronizing concerns

Some websites rush to get those results and/or use a table bench instead of a real PC case, which can explain some differences. Some IR pictures can also be misleading if taken with a cheap IR camera or if not displayed with a proper scale.

 

[railven]

Are you really talking about my IR pictures ???

http://www.hardware.fr/articles/937-8/temperatures-nuisances-sonores.html

I know how to use my IR camera (we were actually the first to introduce IR imaging in graphic cards reviews in 2008 or 2009), how to make sure my results are correct and how a closed loop works, thanks for your patronizing concerns

Some websites rush to get those results and/or use a table bench instead of a real PC case, which can explain some differences. Some IR pictures can also be misleading if taken with a cheap IR camera or if not displayed with a proper scale.

Woof, that foot in mouth. I think he's referring to these IR pictures:

http://www.pcgameshardware.de/AMD-Radeon-Grafikkarte-255597/Tests/Radeon-R9-Fury-X-Test-1162693/

Which is clearly not your site. Haha. Welcome to ATF. Come for the drama, stay hopefully for the insight.

<popcorn.gif>

 

[Abwx]

Woof, that foot in mouth. I think he's referring to these IR pictures:

http://www.pcgameshardware.de/AMD-Radeon-Grafikkarte-255597/Tests/Radeon-R9-Fury-X-Test-1162693/

Which is clearly not your site. Haha. Welcome to ATF. Come for the drama, stay hopefully for the insight.

<popcorn.gif>

Both this site an Hardware.fr are wrong about VRM temps wich are likely much less than 100°C.

What reach this temp is the copper of the inductances, moreover thoses wires are soldered in both side of the PCB and copper being a good conductor, electricaly and thermicaly, it s logical that this temp can be measured on both sides of the PCB.

On the german site pic we clearely see that the hot temp point is between the tube that is over the VRMs and the square that contain the GPU cooler :

So temperatures that are not dangeourous at all, FTR copper is impregnated up to 240°C, and it require much more temp for the ferrites to lose their magnetical properties...


And Hardware.fr seen from the other side :

http://www.hardware.fr/articles/937-8/temperatures-nuisances-sonores.html

 

[railven]

Words

Add Tom's to the list, they too also claimed VRM's over 100c.

 

[Abwx]

Add Tom's to the list, they too also claimed VRM's over 100c.

Dont know for THG since i didnt read the review but with the exemples above things are clear, in HFR pic we can even see the 6 VRMs aligned at the right of the hot area, and wich are obviously at lower temp since they are cooled by a copper tube...

 

[Hitman928]

Add Tom's to the list, they too also claimed VRM's over 100c.

Tom's reported over 100 C on the VRMs only during Furmark which AMD doesn't throttle as much as Nvidia.

We see the consequences of that conservative fan setting in our infrared temperature measurement results. During gaming, the VRMs stay reasonably cool, even though they're only covered by a small heat sink that touches a heat pipe above it. The board hits 60 °C at the slot, meaning the VRM’s heat travels across the PCB under the rubberized back plate. . .

The story changes during our stress test. The water-cooling rule of thumb comes to mind right away: use one centimeter of radiator length per 10W of power. Almost 90 °C at the motherboard slot indicates that the VRM pins have passed 100 °C. This certainly isn’t a great way to run the card long-term, but then again, stress tests aren’t an everyday usage scenario. Still, it would have been nice to see some reserves for overclocking.

I'm not concluding one way or another, just that Tom's and hardware.fr don't agree.

 

[Hitman928]

edit: nvm, got a better picture of the card.

 

[railven]

Tom's reported over 100 C on the VRMs only during Furmark which AMD doesn't throttle as much as Nvidia.



I'm not concluding one way or another, just that Tom's and hardware.fr don't agree.

But they do agree. Both sites made their claim when stress testing the card. Both showed different temps during normal game play.

 

[Abwx]

I find that thoses temps issues are somewhat inflated to death, it s not the GPU, neither the VRMs but still, it s a concern for some people here, nevermind that HFR measured Nvidia VRMs on a card at 126°C, i didnt read in said site that it was problematic, here we have components that are not even semiconductors at much lower temp...

 

[railven]

I find that thoses temps issues are somewhat inflated to death, it s not the GPU, neither the VRMs but still, it s a concern for some people here, nevermind that HFR measured Nvidia VRMs on a card at 126°C, i didnt read in said site that it was problematic, here we have components that are not even semiconductors at much lower temp...

I don't recall anyone saying it was problematic. You probably should read the thread to find out what was "problematic". Hint: it wasn't the VRM temps.

 

[imported_Tridam]

Dont know for THG since i didnt read the review but with the exemples above things are clear, in HFR pic we can even see the 6 VRMs aligned at the right of the hot area, and wich are obviously at lower temp since they are cooled by a copper tube...

Well usually when we simply talk about VRM, we mean power stage as a whole, not really a specific component of the power stage. I guess it's the same for most tech medias.

I have to disagree about your analysis however. First what you think you identify as the 6 VRMs on the right of the hot area are actually just 6 batches of small capacitors with a different emissivity.

You should have a better look at the front side of the PCB and at the position of everything : http://www.hardware.fr/medias/photos_news/00/47/IMG0047713_1.jpg
On the front side, the MOSFETs (IRF6811 and IRF6894) are actually right in the middle of the hot area. Same for the drivers (CHL8510) on the back side.

 

[n0x1ous]

Well usually when we simply talk about VRM, we mean power stage as a whole, not really a specific component of the power stage. I guess it's the same for most tech medias.

I have to disagree about your analysis however. First what you think you identify as the 6 VRMs on the right of the hot area are actually just 6 batches of small capacitors with a different emissivity.

You should have a better look at the front side of the PCB and at the position of everything : http://www.hardware.fr/medias/photos_news/00/47/IMG0047713_1.jpg
The MOSFETs (IRF6811 and IRF6894) are actually right in the middle of the hot area.

How was the pump noise on your sample?

 

[imported_Tridam]

Very annoying on both samples.