PCPer AMD R9 Nano review

[zMeul]

you do realize that most GPU's on the market run hotter then that right? so if this card will die then 980ti, titans, 290x, 390x all have the same problem.

as I said, you ignore the fact of this being a mini-ITX oriented product with total volume at about 10 liters shared by a lot of heat generating components

 

you pay 650$ for a product that thermal throttles?! I'm quite curious why are you defending this

[zMeul]

I want to avoid the flame war here but zMeul you do understand that VRM running that hot is kinda expected for air cooled cards? VRM gets hawt. 

I was not talking about VRMs temp but about the thermal stress put on the soldering - it will crack; is there a close relation, yes there is

[HalGameGuru]

@zMeul That's a little disingenuous. nVidia cards using GPU Boost are ALL thermal throttling. The question is do the thermals in this case negatively effect the system or performance?

[zMeul]

@zMeul That's a little disingenuous. nVidia cards using GPU Boost are ALL thermal throttling. The question is do the thermals in this case negatively effect the system or performance?

want to talk about nVidia, go ahead .. make a different topic

 

this isn't about nVidia vs AMD, it's about what AMD is selling

[HalGameGuru]

want to talk about nVidia, go ahead .. make a different topic

 

this isn't about nVidia vs AMD, it's about what AMD is selling

Any card clocking itself based on temps is thermal throttling, there is a difference between thermal throttling as a mechanism and throttling as an unexpected shortfall.

 

If it is not negatively impacting the system or performance, or it is not causing any more stress than any similarly performing card this is just negatively biased pedantry at best.

 

If it is able to keep its target clocks and performance under load its not an issue of thermal throttling.

[zMeul]

Any card clocking itself based on temps is thermal throttling, there is a difference between thermal throttling as a mechanism and throttling as an unexpected shortfall.

 

If it is not negatively impacting the system or performance, or it is not causing any more stress than any similarly performing card this is just negatively biased pedantry at best.

 

If it is able to keep its target clocks and performance under load its not an issue of thermal throttling.

O'really? it's keeping it's clocks?

 

how about this:

 

 

how much would you say it stayed at 1000Mhz? 1% of the time? lower?

it didn't even reached 1000Mhz! 

 

they're selling a card for 650$ that averages at around 870Mhz and I'm the only one seeing this?! come on! how much were you laughing when you saw Titan X thermal throttling?

[xAcid9]

O'really? it's keeping it's clocks?

 

how about this:

 

 

how much would you say it stayed at 1000Mhz? 1% of the time? lower?

it didn't even reached 1000Mhz! 

 

they're selling a card for 650$ that averages at around 870Mhz and I'm the only one seeing this?! come on! how much were you laughing when you saw Titan X thermal throttling?

 

That is not thermal throttled.

[HalGameGuru]

I'm confused, is 1GHz being labeled the base clock now? It was always labeled as a max clock when I read up on it, I don't even know what the base clock is.

It's a pretty predictable conservative temp and fan profile.

 

I didn't really have any feelings one way or the other when thermal throttling has been brought up. Where it happens by design I look for improvements, where it happens unexpectedly I look for the errors. Whether it was Titan or 200 series reference, I don't have a dog in this fight, I just want to see the why and what is being done to improve the situation.

 

For the Nano the why is a conservative profile, and acceptable performance at stock. There is headroom for stable high clocks or OC in proper use cases. 

 

 

 

And seeing as how it is capable of stable overclocking...

 

http://www.guru3d.com/articles_pages/amd_radeon_r9_nano_review,36.html

 

Your own picture shows the plateau at 75 degrees. Conservative, what you want out of the box for an ITX build.

 

 

 

As an aside ITX or ATX you still need airflow.

[The Benjamins]

I was not talking about VRMs temp but about the thermal stress put on the soldering - it will crack; is there a close relation, yes there is

 

It depends on the solder used, the units I design can operate at 120C with with out running into solder issues. if it was a issue you would not see almost every high end GPU do the same thing, or even worse. as stated before Titans have higher thermal stress then the nano.

[Phillyphries]

The card just released and it looks quite awesome for a card THAT small. From what I'm getting, it's more expensive than I'm comfortable with, but smaller and newer things are always more expensive at first.

 

The coil whine though... Card could use a revision which might happen at some point? Personally I've never experienced coil whining on any of the AMD/ATi cards I've ever owned.​

[Briggsy]

O'really? it's keeping it's clocks?

 

how about this:

 

how much would you say it stayed at 1000Mhz? 1% of the time? lower?

it didn't even reached 1000Mhz! 

 

they're selling a card for 650$ that averages at around 870Mhz and I'm the only one seeing this?! come on! how much were you laughing when you saw Titan X thermal throttling?

 

You're whining about something that was stated up front by AMD in an honest manner, months before the card launched.

 

If you compare the Nano to the FuryX:

870Mhz = 82.8% of 1050Mhz

175Watt = 63.6% of 275Watt

So for 63.6% of the target power draw, you are getting 83% of the performance on average, depending on the severity of the load. That's a pretty big win.

 

Oh, and for the record in case someone says the actual power draw is higher than 275Watts on the FuryX - the FuryX is in fact soft Capped at 275 Watts, much like the Nano is soft capped at 175 Watts, so my comparison above is fair.

[Prysin]

O'really? it's keeping it's clocks?

 

how about this:

 

 

how much would you say it stayed at 1000Mhz? 1% of the time? lower?

it didn't even reached 1000Mhz! 

 

they're selling a card for 650$ that averages at around 870Mhz and I'm the only one seeing this?! come on! how much were you laughing when you saw Titan X thermal throttling?

i dont know with you. But i never really liked the Furmak game.....Just too damn repetitive and boring... Character development and even the artwork was just MEH...

 

As for Metro... amd DID state "up to 1000MHz".... Basically that means if the R9 Nano reaches any clock speed between 0MHz and 1000MHz it will be operating within promised and advertised factory specs.

[Misanthrope]

From what I see you could potentially get this as a one slot card, that'd be a lot more useful if it was cool-able as a 1 slot card.

[The Benjamins]

From what I see you could potentially get this as a one slot card, that'd be a lot more useful if it was cool-able as a 1 slot card.

 

EKWB makes 1 slot blocks for the Fury, FuryX. and I think they come with 1 slot IO shields too. If you could find the right MOBO you can have 4 furyX with water blocks on one mATX.

 

I know you can get a mATX X99 MOBO with 3 PCIE 3 x16 

 

I would assume EKWB will release a block for the nano.

 

 

Enclosed:

- EK-FC R9 Fury X series water block

- single-slot I/O bracket

- mounting mechanism with screw-in brass standoffs

- thermal pads

- thermal grease EK-TIM Ectotherm (1g)

 

 

[zMeul]

You're whining about something that was stated up front by AMD in an honest manner, months before the card launched.

 

If you compare the Nano to the FuryX:

870Mhz = 82.8% of 1050Mhz

175Watt = 63.6% of 275Watt

So for 63.6% of the target power draw, you are getting 83% of the performance on average, depending on the severity of the load. That's a pretty big win.

 

Oh, and for the record in case someone says the actual power draw is higher than 275Watts on the FuryX - the FuryX is in fact soft Capped at 275 Watts, much like the Nano is soft capped at 175 Watts, so my comparison above is fair.

funny how this little detail is ignored - it has the same price point!!!!!! and perf wise is under R9 Fury

[zMeul]

It depends on the solder used, the units I design can operate at 120C with with out running into solder issues. if it was a issue you would not see almost every high end GPU do the same thing, or even worse. as stated before Titans have higher thermal stress then the nano.

Titans were not designed to operate in cramped cases - and if you do put one in such a closed environment, you should expect it to have a shorter life span

if memory serves, from Anand's review, the Titan X throttled on a open bench

[zMeul]

That is not thermal throttled.

funny how DF said the same thing:

2 minutes in

---

or Guru3D's review: http://www.guru3d.com/articles_pages/amd_radeon_r9_nano_review,12.html

 

Result: 78 Degrees C - Throttling base clock at 850~880 MHz.

[The Benjamins]

Titans were not designed to operate in cramped cases - and if you do put one in such a closed environment, you should expect it to have a shorter life span

 

how dose tight space shorten life span? GPU are not claustrophobic.

 

If you are talking about thermal issues then that is up to the system builder to keep in mind how to keep the system cool. its not impossible and that picture you originally link was a case were it was running with in its capacity. so their will only be issues IF the person who put the system together mess up.

[SuperShires]

I thought the nano was going to be cheaper than the fury... but it performs the same and is $100 more... 

 

 

idc how small it is. That isn't really what I call "value"

Then this card clearly isnt for you so you dont have to worry about it.

[Briggsy]

funny how this little detail is ignored - it has the same price point!!!!!! and perf wise is under R9 Fury

 

No one is ignoring the price, it's just irrelevant. Let's compare the Nano's performance to the TitanZ.

[Misanthrope]

EKWB makes 1 slot blocks for the Fury, FuryX. and I think they come with 1 slot IO shields too. If you could find the right MOBO you can have 4 furyX with water blocks on one mATX.

 

I know you can get a mATX X99 MOBO with 3 PCIE 3 x16 

 

I would assume EKWB will release a block for the nano.

 

 

Enclosed:

- EK-FC R9 Fury X series water block

- single-slot I/O bracket

- mounting mechanism with screw-in brass standoffs

- thermal pads

- thermal grease EK-TIM Ectotherm (1g)

 

 

Well yes but that requires an additional radiator somewhere. What I was thinking by single slot card would be to sit it on top of the cpu, some cases do this:

The M300

Though air cooling only single slot solution is probably too much to ask for the nano, they should release something like a Fury Micron that cuts down not only the tpd but also number of cores and performs on par or a tad worst than the 390 for something like 300 bucks, now that'd be something and it would really kill the otherwise superior choice of a 970 itx card.

[xAcid9]

funny how DF said the same thing:

2 minutes in

Power limited, Ryan at PCPER  simply increased the power limit slider in CCC and the card reach it's theoretical speed of 1 jigahertz almost all the time.

 

 

 

 

http://www.pcper.com/reviews/Graphics-Cards/AMD-Radeon-R9-Nano-Review/Power-Consumption-and-Overclocking

[zMeul]

how dose tight space shorten life span? GPU are not claustrophobic.

yes, they kinda' are - it's about the operating ambient temperatures in the volume of air it operates in

[zMeul]

No one is ignoring the price, it's just irrelevant. Let's compare the Nano's performance to the TitanZ.

it's irrelevant!? and yet yet you bring into question against Titan Z - HA!

please make up your mind  :lol:

[The Benjamins]

Well yes but that requires an additional radiator somewhere. What I was thinking by single slot card would be to sit it on top of the cpu, some cases do this:

The M300

Though air cooling only single slot solution is probably too much to ask for the nano, they should release something like a Fury Micron that cuts down not only the tpd but also number of cores and performs on par or a tad worst than the 390 for something like 300 bucks, now that'd be something and it would really kill the otherwise superior choice of a 970 itx card.

 

AMD has made 1 slot coolers for servers, but they sure will not be quiet. you have to find a trade off somewere.

 

they could try and engineer this for the extra 25W and make it shorter

http://www.amd.com/en-us/products/graphics/workstation/firepro-3d/7100