I don't think HBM2 is that expensive any more. No doubt it has more cost associated with it. But there will be different models here that we see from. They have lot of slots to fill above 300 mark.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Radeon RX Vega Discussion Thread
- Thread starter DPI
- Start date
I don't think HBM2 is that expensive any more. No doubt it has more cost associated with it. But there will be different models here that we see from. They have lot of slots to fill above 300 mark.
cybereality
[H]F Junkie
- Joined
- Mar 22, 2008
- Messages
- 8,789
I have no clue what HBM2 costs. My feeling is AMD can't release a card that is only just as good as Nvidia's from last year at the same price. It won't sell. So either is has to be cheaper, or it has to be better. If they release multiple SKUs and can hit both targets, then it's a home run.
It will sell. If AMd can have gtx 1080 performance at 1080 price or less(knowing amd it likely will be less) it will sell. Why? Because who lot of people have freesync monitor. I will likely sell my 1080 and buy it so I can leverage freesync monitor I have. and don't have to spend another 2-300 on a g-sync monitor. They won't price it high if it doesn't perform. We will likely see 3 cards. 1070 competition, 1080 and close to 1080ti ( probably the water cooled special edition model) each priced 50 or so less than competition.
cybereality
[H]F Junkie
- Joined
- Mar 22, 2008
- Messages
- 8,789
That sounds about right. At least with FreeSync there is a tangible reason for people to choose AMD over Nvidia.
But if it were only about performance, the people that wanted a 1080-class card could have purchased it any time in the past year (especially in the last month since the price drop). In which case those sales are already lost (assuming Vega was exactly as good as 1080 for exactly the same price). Something has to give.
But if it were only about performance, the people that wanted a 1080-class card could have purchased it any time in the past year (especially in the last month since the price drop). In which case those sales are already lost (assuming Vega was exactly as good as 1080 for exactly the same price). Something has to give.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
The difference between DX11 and DX12 between IHVs is gone.
JUST NOT POSSIBLE YOU SAY, we been down this rodeo before yet you bring up TPU, what's wrong with the site you're on data? This was pointed out to you before. I've shown you that does not have to be the case yet you bring up this same crap again. I've also shown actual game play showing the Nano can maintain FuryX speeds. If Vega speed is indeed 1500+ it will perform way better then a 1080, if it OC to 1700+ 1080Ti territory. Once again, Nano is a very small video card, one fan, relatively small cooler (AMD KICKED ASS ON THIS DESIGN)!Nano has to throttle because its rating is too similar to Fury/Fury X.
It is impossible for it to sustain such rating without something giving; it has a TDP of 175W vs 275W for Fury/X and yet officially has very similar clocks and identical FP32 TFLOPs.
Just not possible with that TDP rating at sustained levels.
Here are Tom's results.
TPU:
This will vary game-to-game though but far from consistent even to reach 950Mhz.
And going forward this will become even tricker for AMD with the more dynamic Boost-power algorithm they use (just look at the difficulty they had with official spec for Polaris models in this regard).
Fiji was pushed just has hard to its limits as Polaris was by AMD for a launch product.
Cheers
HardOCP testing (real hard data, real testing in a real case and frankly really knowing what they are doing):
They were able to get the Nano to maintain speed virtually at 1050mhz or at FuryX speeds which they didn't have the new bios yet and more clock friendly drivers:
Here are the game tests: Note some are faster than the FuryX: (If AMD can do this with the Nano what make anyone think they cannot with Vega?)
Now I wonder why you are not at TPU in their forums vice here. If you just take a FuryX performance at 1050mhz and increase the clock speed to 1550mhz it would slaughter a 1080. Only reason Nvidia would release Volta early would be that they are getting their asses handed to them. Enough with the non-sense that a Nano cannot maintain a high clock speed, if you know what you are doing you can. I designed my whole case around this review and the case has a 140mm fan blowing straight into the Nano fan - as simple as that, fan speed rarely goes above 60% temp usually less than 76c maxed out.
Last edited:
SO now your saying they overspecc'ed all the Furys with triple full length fans and a power management system that is over 50% better than the Nano...
There is a reason AMD gave it a TDP 100W lower than those models.
The review conclusion with HardOCP overclock:
In other words as I said it is impossible to do this without something giving for sustained performance (longevity obviously is also part of this equation), this card is NOT designed to work at the same rating as a Fury and Fury X.
And Battlefield 4 is one of the friendlier games to test in this regard as can be seen when compared to Witcher 3.
That is ignoring the requirement of running the fan permanently at 100% and the additional thermal hotspots the Nano will suffer due to such excessive strain, let alone the leakage this has compared to the better rated and cooled designed Fury and Fury X.
Cheers
Last edited:
So it is possible but not possible you're saying - more BS. I do not run my fan at 100% did you not read that - are you the same person I posted before? Your acting like this is totally new to you.
Here you go once again, look at gpu temp, gpu speed, fan speed (is it at 100%). Doom running 2560x1440 in window on a 4K resolution desktop with the Nano recording at the same time:
So I say your statement is false, incorrect or wrong in the end. Let's get back to Vega, but I want to let the folks know who has credibility here.
You quoted HardOCP.
They STATE they needed 100% fan.
So if you want to use them as fact, stay with it.
Again it is about you being different, now your different to HardOCP.
I see a trend here
Another snippet about fans in the HardOCP overclocking of Nano:
So we have 100% fans by them, they also state they are not convinced about longevity.
Sorry but what you linked from HardOCP does agrees with what I said.
It is impossible for Nano to sustain a Fury or Fury X performance envelope/TDP without something giving.
Can you try Witcher 3 please.
Cheers
Noko,
I think you are missing the point with regards to thermal hotspots with the actual Nano card.
My concern in heat is regarding the VRM/power stage that will not be cooled as well as the full blown Fury that also has a higher rated phase/VRM.
The 100% fan that HardOCP had to use will help with the GPU itself clocking behaviour, but not fully help cooling the power stage nor do we see any analysis regarding power leakage (critical point) in most of this including your Youtube vid.
You are basically taking a Fury and reducing by a lot the size/power stage/ICs design/cooling.
It is a Fiji GPU, but that on its own does not make a GPU card.
Also worth noting the Fury was a cut down core so had less TDP demand, while the Nano is the full die.
So another factor to consider on the reduction in size-spec-rating from Fury to Nano.
Cheers
I think you are missing the point with regards to thermal hotspots with the actual Nano card.
My concern in heat is regarding the VRM/power stage that will not be cooled as well as the full blown Fury that also has a higher rated phase/VRM.
The 100% fan that HardOCP had to use will help with the GPU itself clocking behaviour, but not fully help cooling the power stage nor do we see any analysis regarding power leakage (critical point) in most of this including your Youtube vid.
You are basically taking a Fury and reducing by a lot the size/power stage/ICs design/cooling.
It is a Fiji GPU, but that on its own does not make a GPU card.
Also worth noting the Fury was a cut down core so had less TDP demand, while the Nano is the full die.
So another factor to consider on the reduction in size-spec-rating from Fury to Nano.
Cheers
HBM2 is very expensive, Raja wouldn't say majority of the costs of the graphics card is the ram if it wasn't. And he also stated over a year ago, setting up production for HBM is very expensive too. So going lower than 500 bucks is probably not going to be an option for full Vega (8GB)
Gideon
2[H]4U
- Joined
- Apr 13, 2006
- Messages
- 3,548
We have no idea what HBM or HBM2 costs, we only know its more expensive then GDDR.
Anarchist4000
[H]ard|Gawd
- Joined
- Jun 10, 2001
- Messages
- 1,659
Under what conditions though? Ambient temperatures, case design, and airflow will make a difference. Smack a water block on that little card and the difference largely goes away. They are the same chips after all. Bigger VRMs is nice, but only does so much.
100% fans that made little actual difference. Cooling, from everything I've seen, wasn't the primary problem on the Furies. The LN2 overclocks managed ~50% core, 100% memory, but tend to use various power mods in the process. That's well beyond what most water cooled chips were achieving. HBM saves die space by not needing the large transistors with lots of voltage/current to drive external memory like DDR/GDDR/etc. On the flip side that really wide IO was causing a lot of ripple which is likely where the problem lies. I can't recall where I read it, but AMD or one of the memory manufacturers mentioned that as an issue with HBM. Correctable, but a consideration nonetheless. Faulting a Nano for it's size and cooler is a poor example. They're small cards with better perf/watt than most designed for smaller cases.
The Nano chips were better binned than the ones in Fury X's which made the Nano the better card if you could take care of the thermals and power limit. AMD's only real fault with the Nano was selling it at the same price as the Fury X at a perceived negative performance delta.
What do you mean made no difference?
a) If fans did not make a difference it would not throttle to below 900MHz as per the frequency behaviour I posted that is now on the other page nor would I have to repeatedly quote HardOCP lol, it is more than just power but then power is also part of it and this card does NOT have the same level of power stage and ICs as the full size triple slot cut down die Fury.
b) Just in my last few pages I mentioned twice how HardOCP said they NEEDED 100% fans for consistent frequency clocks for 1000MHz.
Maybe a 3rd time will be the charm
Here is another section they mention fan speed:
So yeah might be better binned, but that is quite a few reviews where it is not....
And I am amazed that those who cannot see the issue with a short card with a single fan when it his hitting over 250W once you OC it to try and behave like a Fury X....
Nvidia should had made all 1080s a short card single fan, what could possibly go wrong.
Here is the power behaviour BEFORE you start to push it to behave like a Fury/Fury X, meaning this scope measurement is it throttling down to 850MHz to 900MHz depending upon game, think what that looks like when forcing the Nano to behave like a Fury/Fury X.
And here is the frequency behaviour to go with that power performance, ignore the red line as we are interested in the Blue Gaming loop measurement.
Here is a thermal image, again in normal behaviour BEFORE (ok if kept within SKU spec) forcing it to try and behave like a Fury/Fury X, and comes back to my point about thermal hot spots on the GPU/card when stressed and potential impact on components and compounding the power leakage.
But this thread is about Vega, maybe those AMD fans (really feels like it with the effort to defend how 'great' the Nano is when pushed beyond its rating) who really feel there is something to defend with a single fan short card Nano and making it behave like a Fury/Fury X should start a new thread.
So yes in engineering, something will have to give if one wants to turn a 175W official TDP/TBP spec-rated model card into the 275W Fury/Fury X with sustained and permanent gaming.
The Fury X is a much better designed-rated card with the component-IC and power stage implemented, if you check teardowns of Fury X/Fury/Nano.
Cheers
Last edited:
Gideon
2[H]4U
- Joined
- Apr 13, 2006
- Messages
- 3,548
Just so you know Furmark is throttled by drivers for a reason, it's a unrealistic load on the video card. A better way is to run a game in a loop and then check the thermals, Furmark should not be used for that.
Yeah and all GPUs throttle Furmark (or should but some in the past did not do a good job if I remember) as it is a power virus.
That image I provided shows a game (blue line) as well that is associated to the thermal and power images I also linked, which is my context.
Cheers
Anarchist4000
[H]ard|Gawd
- Joined
- Jun 10, 2001
- Messages
- 1,659
So when I water cool the card and the fan is sitting in my closet that fan is still going to slow down my performance? A Polaris 10 has a power stage capable of over 600W. I haven't seen very many people pushing that limit on the cards. Fiji isn't that different. This performance still depends on ambient temperatures, games being played, and a host of other factors. Cooling is a choice.
Who cares about a consistent 1000MHz though? If I'm playing benchmarks all day then sure. If I'm playing games then would I really notice the couple percentage points difference in fps?
What's wrong with a single fan short card? It's an option that works well for people wanting power efficient cards in a small form factor. I'm at a point where I'd rather have something small and compact. No more dual socket CPUs, extensive water cooling, RAID arrays, etc. Hardware is still fun, but after a while the giant cases become an eyesore. For what it is, a Nano is a nice little card, and probably not that different from what a Vega 11 would look like.
Bottom line, put a water cooler on each and I wouldn't expect much of a difference. In fact the Nano might perform better as it is binned higher. The size is a feature, not a problem.
So water cool the thing an suddenly there is very little difference between the cards. While that bigger power stage is nice, it still only does so much. Frankly it's not all that different from audiophile hardware where the specs are "better" and people claim to "hear the difference" despite being well beyond what the average human can perceive for significantly more cost.
You really should had read what my posts was in context of before defending the Nano and arguing.
The discussion was around myself and Noko arguing about that it is not possible to run the Nano sustained and permanently as a Fury/Fury X without something giving.
That is why the 1000MHz matters and the discussion around the throttling and limitations of the IC and power stage/etc....
And all the data I provided backing this up.
There is nothing wrong with the Nano, but it is not designed to be a 275W Fury/Fury X.
THAT was what the discussion and context.
I got to say to it is hard work dealing with AMD threads
And BTW there is more to the power stage in context of GPU and card than just summing the VRMs, again there is a big difference in design between the Nano and Fury X in this regard if you strip down both - and this does fit in with my context that unfortunately you missed and are arguing about something else.
Cheers
Last edited:
AMD built the Nano tough enough to handle the OCs, you really don't hear too much about Nano's failing. Sounds like more made up fud - time to move on. Basically AMD does things that others think are impossible, over and over again.You really should had read what my posts was in context of before defending the Nano and arguing.
The discussion was around myself and Noko arguing about that it is not possible to run the Nano sustained and permanently as a Fury/Fury X without something giving.
That is why the 1000MHz matters and the discussion around the throttling and limitations of the IC and power stage/etc....
And all the data I provided backing this up.
There is nothing wrong with the Nano, but it is not designed to be a 275W Fury/Fury X.
THAT was what the discussion and context.
I got to say to it is hard work dealing with AMD threads
And BTW there is more to the power stage in context of GPU and card than just summing the VRMs, again there is a big difference in design between the Nano and Fury X in this regard if you strip down both - and this does fit in with my context that unfortunately you missed and are arguing about something else.
Cheers
Vega at NAB
http://www.pcworld.com/article/3192...gas-ability-to-handle-8k-graphics-at-nab.html
I am liking the 3 8 pin power connectors and the water cooling tubes on the rear end of the card
.
Last edited:
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
Basically AMD does things that others think are impossible, over and over again.
View attachment 23086
Vega at NAB
http://www.pcworld.com/article/3192...gas-ability-to-handle-8k-graphics-at-nab.html
I am liking the 3 8 pin power connectors and the water cooling tubes on the rear end of the card.
Yes, AMD have repeatedly done the impossible. Like claiming a multiple quarter lead on NV (with regards to maxwell successor) then have NV release their whole product stack from top to bottom before AMD even sorts out inventory for Polaris.
What else? Raja koduri claiming two RX480 are both faster and more efficient than a 1080,while costing less.
4GB HBM > 12GB Gddr5
'Overclockers dream'
The list goes on, and I appreciate noko pointing this out. Nothing is impossible when it comes to AMD. Life is like a box of chocolates, and so are AMD product launches
AMD built the Nano tough enough to handle the OCs, you really don't hear too much about Nano's failing. Sounds like more made up fud - time to move on. Basically AMD does things that others think are impossible, over and over again.
View attachment 23086
Vega at NAB
http://www.pcworld.com/article/3192...gas-ability-to-handle-8k-graphics-at-nab.html
I am liking the 3 8 pin power connectors and the water cooling tubes on the rear end of the card
3 8 pin power connectors and water cooling that better be a dual vega lol but the board looks too small for that
You quoted HardOCP.
They STATE they needed 100% fan.
So if you want to use them as fact, stay with it.
Again it is about you being different, now your different to HardOCP.
I see a trend here
Another snippet about fans in the HardOCP overclocking of Nano:
So we have 100% fans by them, they also state they are not convinced about longevity.
Sorry but what you linked from HardOCP does agrees with what I said.
It is impossible for Nano to sustain a Fury or Fury X performance envelope/TDP without something giving.
Can you try Witcher 3 please.
Cheers
Some Nanos can sustain full clocks with a voltage drop. I have a Nano at home that I could prove it to you. I picked up a brand new Asus white nano for $250 shipped + 8% ebay bucks back the other day
I can certainly run it through its paces for some giggles.1050mhz Nano with power slider maxed. Temps maxed 79c - Auto fan
http://www.3dmark.com/fs/12448825
Stock Fury X http://www.3dmark.com/fs/10506294
Its close!
EDIT: My old Nano was able to sustain the full 1050mhz. It was a good and maybe rare chip! http://www.3dmark.com/fs/8159552
Last edited:
Still too early for reviewers to have in hand I believe. Brent's been awol working on a pretty big review, I remember seeing Kyle make some posts on all the systems he was sending out to Brent. Probably a big article comparing Ryzen/Intel with various GPU's and scenarios, I imagine would keep him pretty busy.
My FuryX sustains 1100, but that's under water of course.
I was debating on going 1080ti this round, but with TV's likely to be supporting freesync soon, I'm gonna stay locked in to AMD I guess. I'll be waiting to replace my 60 inch living room tv with a 4k equivalent that has some adaptive sync support. If Vega ends up being shit, I'll just have to wait until its a reasonable price to upgrade or wait it out until Navi and work on my backlog for a year or 2.
Some Nanos can sustain full clocks with a voltage drop. I have a Nano at home that I could prove it to you. I picked up a brand new Asus white nano for $250 shipped + 8% ebay bucks back the other day
1050mhz Nano with power slider maxed. Temps maxed 79c - Auto fan
http://www.3dmark.com/fs/12448825
Stock Fury X http://www.3dmark.com/fs/10506294
Its close!
EDIT: My old Nano was able to sustain the full 1050mhz. It was a good and maybe rare chip! http://www.3dmark.com/fs/8159552
What are the fans %, the sustained frequency and power demand in say a game like Witcher 3 or Metro Last Light - as you mention maximum power setting and I touched upon that several times now, and showed an accurate thermal image of Nano card at normal operation even before this.
Worth remember frequency will change depending upon how pushed by the game (rather than 3dmark).
But my point still stands, from an engineering perspective the Nano is not built to the same spec and rating as the Fury X or Fury that is cut down to be used on air and triple fan and still with design for 275W TDP/TBP, the Nano is designed around its 175W TDP and even operating at 840-900MHz it is exceeding this.
Yes Nano may be binned but that only gets you so far and it seems a lottery, even then I would be leery about pushing it as it is just not rated to push like the other two Fiji models (remember the Fury X in most cases only OCs to 1100-1140MHz at best)
And that binning, well it is not showing up in the TDP and frequency-power behaviour of Tom's Nano review.
Undervolting should be reserved as a more unique 'OC' situation as it does not guarantee stability in all situations, and tbh it is applicable to all GPUs included Nvidia's Pascal but we never bother discussing that or applying it to how it improves them.
But then AMD buyers were happy to OC and push their 480's waaay beyond the 66W 12V motherboard limit so the SAS motto is alive and well with these owners and the 'Who Dares, Wins'
Anyway I agree the product is good, but it is not up to the build-spec-rating of a Fury X.
Cheers
Last edited:
Anarchist4000
[H]ard|Gawd
- Joined
- Jun 10, 2001
- Messages
- 1,659
The difference isn't actually that big. It's always based on assumed ambient temperatures. Power ratings of VRMs change based on thermals. So it is within the context and limiting comparisons to cooler performance isn't justified.
Not designed and not performing similarly are separate issues. By the same argument no chips are designed to be overclocked. If proper cooling is applied to the chip and VRMs, there isn't much of a difference between designs. Water cooled VRMs quite frankly could push well over 1kw. Is a Fury/X more robust? Yes, but is that difference meaningful to actual performance?
While I haven't specifically looked into the power stages of a nano, I wouldn't be surprised if the components were superior to the Fury/x given adequate cooling. 100W difference is nothing.
And they still burnt out fewer motherboards than 1070s (EVGA)!But then AMD buyers were happy to OC and push their 480's waaay beyond the 75W motherboard limit so I should not be surprised
The EVGA issues turned out not to be proved and in fact some were fake or stupid mistakes; one was so bad even der8aur commented in the Youtube vid of the alleged bad 'Nvidia 1070', and then went on to do a whole video himself to explain the nonsense behind this and whatever was happening was not related to the power stage design and components used by Nvidia.
The issue was EVGA's implementation of thermal pads and thermal connectivity and did not affect motherboards apart from an idiot and that was his fault as he powered up with the GPU in the motherboard without the auxiliary connections plugged in (bad idea).
Nothing to do with Nvidia nor anything to do with the design.
If the power stages/IC in Nano were better rated (it is not as I have seen both stripped), it would not had been a bloody 175W TDP part (made to operate at most 190W and quite down from 1000MHz) AMD would had gone with a similar performance behaviour as the Fury X
And please do not tell me they did this to be efficient product compared to the Fury X.
Look I have provided multiple professional review sources (and one that actually has lab level and support for power measurement with scopes) where it needed 100% fan, where it throttles, where even in normal operation the card itself is hitting just under 100 degrees in spots, and a power draw and frequency behaviour that must be 840-900MHz to stay under 190W.
The Nano was a niche product designed to be compact as its primary design focus, if you were an engineer you would know just how freaking hard it is to take a larger design (Fury X) with better cooling and shrink it down while also reducing the cooling to a single fan and reduced thermal transfer capabilities.
All designs allow for some leeway, and if you take the fact Nano has 1x8pin to Fury and Fury X with 2x8pin that gives you some ideas about reduced rating focus and primary focus on compact size.
The certified standards gives a total of 225W with motherboard+1x8pin, and Tom measured 190W in-game with frequency of 840-890MHz.
But also one has to apply temps to power leakage as I mentioned before, and part of what helps the Fury X relative to the Fury and Nano.
That said I accept some on here are happily running their Nano's at such overclocks, while professional reviews I mentioned had limitations and reservations for the reasons I have to keep outlining.
And to reiterate I do think it is a good product for what it is designed as, relative to the Fury and Fury X.
Cheers
Last edited:
Anarchist4000
[H]ard|Gawd
- Joined
- Jun 10, 2001
- Messages
- 1,659
I've designed power supplies and amps to do just this. I'm well aware of the requirements. It's the thermals that are the concern here. The assumption is that cooling won't change. The nano was designed for 175W provided the cooling off a single fan. A water block, excluding some extreme overclock, changes that.
Remember that's 175W at probably 90C or other high temperature. Actively cooled that rating changes substantially. Spec sheets normally have a graph showing output by temperature with stated min/max. In fact 250W with better cooling is probably more sustainable than 175W on air.If the power stages/IC in Nano were better rated (it is not as I have seen both stripped), it would not had been a bloody 175W TDP part (made to operate at most 190W and quite down from 1000MHz) AMD would had gone with a similar performance behaviour as the Fury X
And please do not tell me they did this to be efficient product compared to the Fury X.
We're getting off topic, but just want to be clear about my point here. There are enough variables that a Nano could perform similarly to larger cards.
Yeah, found out later PCWorld stuck in a Pro Duo Fiji while talking about Vega. Blew my bubble gum all over my face in fact.
So any real info leaking out about Vega is about the same as light leaking out of a black hole. Speculation is the best that anyone has at the moment - all is good and bad and someone might have a lucky guess. Now the updated Polaris chip can hit 1500mhz and that chip was not designed right to hit those speeds (even though I think AMD really wanted a faster chip, lower power to begin with). Vega if it was designed right should be hitting at least 1500+, the MI25 has to go that fast to meet ratings/specs and those are 24/7 cards at 100% capacity designs. The gaming card can relax some of the long term 24/7 requirements and I would think will be able to run faster, 1600+. Just clock speed alone will bring up performance significantly so I see a potential for a great performer.
Now the 1080Ti at 4K is, if I remember right 70% faster than a FuryX - that is a lot of performance to make up. I would be really surprised Vega will hit that out of the box. Now if it has a great OC range or headroom but not used due to power reasons (you just can't put out a 350w rated card) then Vega could be a true racer and when pushed to the max beats out everything. Only a few select few really know the answer to this.
It is not just thermals that are a concern Anarchist.
If you have designed power supplies and amps you also know reducing the IC and power stage and making it more simple is also one of the more challenging aspects to do and impacts stability/leakage/ripples/droop/etc and all these must then be designed around a performance behaviour envelope.
But if you think they called it a 175W TDP to actually do 275W sustained with a basic single fan short card design single 8-pin carry on.
BTW you were the one arguing earlier that it did not matter it did not do 1000MHz
And well I totally disagree with your last post, as an engineer you would also know what was wrong with the 480 and its power distribution and why it was a concern with custom/OCing but again never spoke out against or it or agreed even when I mentioned it some posts ago.
But then this talk about Nano and putting aside those factors I keep touching upon such ; never considered the 100C the Nano card is hitting, how the circuitry is much more simple in design,a basic air fan-cooler implementation, how you feel a 175W TDP just means they designed it to be same as as the 275W Fury X because they 'build all these cards to overclock or all designs to handle substantially higher TDP and power than rated', gave it 1x8pin but hey it is still designed to be a Fury X that Joe Macri mentioning is a 275W product in normal operation design hence their TDP-TBP but given cooling of 500W (bit of artistic license by AMD) and power design over 350W....It was built to an excessive standard beyond the 275W TDP while Nano was not even to its 175W TDP.
Importantly and most obvious the AMD design for Nano is limited to max 225W with the power connectors (notice much less than the AMD engineers position of normal operation for the Fury X at 275W - AMD engineers stated envelope design), now I am not saying it is an issue if goes over this on the auxiliary (issue was motherboard slot on 480) but puts it into perspective that AMD engineers focus was NOT to provide comparative normal operation/performance envelope of the Fury X.
Meaning the cooling and IC/power stage/etc is designed around a reduced performance envelope because the engineering priority was a small form factor card.
Or the fact the data I provided in terms of measurement behaviour did back up what I said; the Nano is not designed to operate like the Fury X at sustained 1000MHz-1050MHz and TDP of 275W without something giving from an engineering perspective.
But yes lets drop this, another lets agree to disagree.
We can agree it is a nice card once its price dropped and relative to the purpose/positioning relative to Fury and Fury X.
Cheers
Last edited:
N4CR
Supreme [H]ardness
- Joined
- Oct 17, 2011
- Messages
- 4,947
I had solid info months before that Polaris was going to be at C-tex. It appears that them not listing as exhibitors/vendors is not unusual.
All we know is that the interposer process costs more if all other things are equal. Hence Fury-X was more of an interposer production test than anything else. Really makes me wonder (and hope) they have something special up their sleeves if that much fore-planning went in.. or is it all for Navi? I hear people mention Vega is Rajas' first GPU, when I've seen Navi mentioned in this regard earlier.
Bit like the 7970. It did great in stock form but when you OC'd it, was when it shined. Right bios and decent sillicon (not mega lottery but just decent) and you had a 1.25-1.3GHz card which was about as fast as the next generation from Nvidia. A GCN card. In 2011...
>Ryzen
Only a fool writes off AMD.
Ieldra, although we disagree on many things knows this well.
We'll find out in due time.
Well all I can say my Nano hasn't exploded, melted nor degraded so far - so all is good in Nano Nano land.It is not just thermals that are a concern Anarchist.
If you have designed power supplies and amps you also know reducing the IC and power stage and making it more simple is also one of the more challenging aspects to do and impacts stability/leakage/ripples/droop/etc and all these must then be designed around a performance behaviour envelope.
But if you think they called it a 175W TDP to actually do 275W sustained with a basic single fan short card design single 8-pin carry on.
BTW you were the one arguing earlier that it did not matter it did not do 1000MHz
And well I totally disagree with your last post, as an engineer you would also know what was wrong with the 480 and its power distribution and why it was a concern with custom/OCing but again never spoke out against or it or agreed even when I mentioned it some posts ago.
But then this talk about Nano and putting aside those factors I keep touching upon such ; never considered the 100C the Nano card is hitting, how the circuitry is much more simple in design,a basic air fan-cooler implementation, how you feel a 175W TDP just means they designed it to be same as as the 275W Fury X because they 'build all these cards to overclock or all designs to handle substantially higher TDP and power than rated', gave it 1x8pin but hey it is still designed to be a Fury X that Joe Macri mentioning is a 275W product in normal operation design hence their TDP-TBP but given cooling of 500W (bit of artistic license by AMD) and power design over 350W....It was built to an excessive standard beyond the 275W TDP while Nano was not even to its 175W TDP.
Importantly and most obvious the AMD design for Nano is limited to max 225W with the power connectors (notice much less than the AMD engineers position of normal operation for the Fury X at 275W - AMD engineers stated envelope design), now I am not saying it is an issue if goes over this on the auxiliary (issue was motherboard slot on 480) but puts it into perspective that AMD engineers focus was NOT to provide comparative normal operation/performance envelope of the Fury X.
Meaning the cooling and IC/power stage/etc is designed around a reduced performance envelope because the engineering priority was a small form factor card.
Or the fact the data I provided in terms of measurement behaviour did back up what I said; the Nano is not designed to operate like the Fury X at sustained 1000MHz-1050MHz and TDP of 275W without something giving from an engineering perspective.
But yes lets drop this, another lets agree to disagree.
We can agree it is a nice card once its price dropped and relative to the purpose/positioning relative to Fury and Fury X.
Cheers
7970 was a killer of a card and mine would do 1320, a far cry from 920mhz. For a long time just kept it at 1200 without issue and eventually put the ghz edition bios in it. I had an A/C cooled case so cooling was not an issue nor was noise.