[updated] The Tech Report won't receive a R9 Fury Nano for review

[Notional]

UPDATE:

They aren't going to give it to Techpowerup... 

http://www.techpowerup.com/215776/amd-radeon-r9-nano-review-by-tpu-not.html

 

 

AFAIK HBM production has good yelds, I don't undestand why Fiji chips are so few

 

It's a 4096 stream processing GPU. NVidia has nothing even close to it in complexity. It is the most extreme in GPU world that can be made on 28nm right now. It doesn't take much for it to be binned a Fury instead. HBM does not make it easier of course.

[Agosto]

It's a 4096 stream processing GPU. NVidia has nothing even close to it in complexity. It is the most extreme in GPU world that can be made on 28nm right now. It doesn't take much for it to be binned a Fury instead. HBM does not make it easier of course.

 

I'm pretty sure it's because of the interposer, not because of Fiji and HBM chips themselves

[Notional]

I'm pretty sure it's because of the interposer, not because of Fiji and HBM chips themselves

 

Nah the interposer is 65nm and both cheap and easy to produce. Because it's only traces, it doesn't even have to be that precise either. They probably have a yield close to 100%. Of course putting it all together could be a problem. But I think it's partly HBM and mostly the complexity/yield of the full fat Fiji chip. After all, the Fury is not a problem, and everyone could get a review sample of that, HBM, interposer and all.

[typographie]

I agree, but AMD did send out cards for reviews. If they refused to send out any cards, I would be worried. But that is not the case.

 

I don't think the concern is simply that they're sending out only a few samples for review, it's the criteria by which they are choosing who gets one and who doesn't. The problem isn't that John Doe Tech Blog with 15 loyal viewers didn't get a sample—it's that AMD seems to be playing favorites among the biggest and best-regarded reviewers in the PC industry.

 

If they don't even have enough product that someone on the level of Tech Report can get one, I have to wonder if production is at a point where they are actually prepared for a release.

[Notional]

I don't think the concern is simply that they're sending out only a few samples for review, it's the criteria by which they are choosing who gets one and who doesn't. The problem isn't that John Doe Tech Blog with 15 loyal viewers didn't get a sample—it's that AMD seems to be playing favorites among the biggest and best-regarded reviewers in the PC industry.

 

If they don't even have enough product that someone on the level of Tech Report can't get one, I have to wonder if production is at a point where they are actually prepared for a release.

 

When you have limited supply, of course you're going to supply the reviewers who has the largest viewership. I generally like techpowerup. They seem very competent, but also slightly nvidia biased. Primarily with their choices of games, like Wolfenstein, which is an OpenGL game with proprietary nvidia shaders in it. Or broken gameworks games. But I'm sure those youtubers have a larger audience than TPU.

[ivan134]

I don't know, but this isn't that important after all, it's a marketing thing. Why some sites are getting one and some aren't, even if they used to receive cards from AMD before?

No, actually, it's very important if you people insist on throwing out wild and unsubstantiated Fox News-esque accusations.

[LinusTech]

There are logical leaps being taken in this thread..

Like if Nvidia sends a reviewer a sample of a card then they will likely give a glowing review because "now they are biased toward Nvidia"

But this ignores the fact that AMD also seeds graphics cards - in fact they also offered to seed us a GTX 970 mini when we said we didn't have one for direct comparison.

So by the above logic we are also based towards AMD.. and yet we criticize them both..

Here's a reality check: A graphics card means nothing to me. I have literally bins of them just like all the other hardware reviewers. They are just tools to do our jobs. It couldn't affect my opinion of the card in any meaningful way unless it was a super amazing card that lit my world on fire in which case there is nothing "biased" about giving it a glowing write up.

Do you see where I'm going with this?

Many other issues as well.. Especially calling out Scott as a sellout. He is anything but..

On my phone right now, but those two jumped out at me..

[marldorthegreat]

Sure the card is just on par with 290x/390. 

No, it should be closert to  fury (non-x) as it is just a slightly underclocked fiji xt card

[marldorthegreat]

There are logical leaps being taken in this thread..

Like if Nvidia sends a reviewer a sample of a card then they will likely give a glowing review because "now they are biased toward Nvidia"

But this ignores the fact that AMD also seeds graphics cards - in fact they also offered to seed us a GTX 970 mini when we said we didn't have one for direct comparison.

So by the above logic we are also based towards AMD.. and yet we criticize them both..

Here's a reality check: A graphics card means nothing to me. I have literally bins of them just like all the other hardware reviewers. They are just tools to do our jobs. It couldn't affect my opinion of the card in any meaningful way unless it was a super amazing card that lit my world on fire in which case there is nothing "biased" about giving it a glowing write up.

Do you see where I'm going with this?

Many other issues as well.. Especially calling out Scott as a sellout. He is anything but..

On my phone right now, but those two jumped out at me..

There probably is a perfectly good explanation behind this. I think its to save money as fiji xt is very limited. AMD knows they would get there hands on a unit, so why ship them a card if they can get one at no expense to AMD. Its probably nothing to do with scott. they didnt send you an r9 295x2, and you got the fury x (2 OF THEM!!!!)

[Agosto]

Nah the interposer is 65nm and both cheap and easy to produce. Because it's only traces, it doesn't even have to be that precise either. They probably have a yield close to 100%. Of course putting it all together could be a problem. But I think it's partly HBM and mostly the complexity/yield of the full fat Fiji chip. After all, the Fury is not a problem, and everyone could get a review sample of that, HBM, interposer and all.

 

I meant the whole put-stuff-on-interposer

 

No, actually, it's very important if you people insist on throwing out wild and unsubstantiated Fox News-esque accusations.

 

I'm just trying to figure out why AMD is doing this. It's a bad marketing move and since I don't dislike AMD (as many people do), I get disappointed by what I consider to be a bad behaviour.

[marldorthegreat]

I meant the whole put-stuff-on-interposer

 

 

 

Again, probably not as complicated. It will be the 8.9 billion transistor chip.

[Andrew Sh]

Honestly the argument of limited supply is bs because getting a popular reviewer a sample card to tell their viewers how the card performs ends up in losing 1 card which could potentially drive hundreds or thousands of sales.

For AMDs sake, I hope the nano performs well because if its a flop, then who the hell would buy it for such a high price. You'd be able to buy 2x gtx970s for this price so realistically it really needs to have killer performance.

If they are worried about bad reviews, and I'm not saying they are. IF they are, then perhaps they should have developed 1 retardedly beast card instead of wasting resources on design and production of 3. Look at the titan. Each generation, there is ONE card (excluding dual GPUs) and it dominates everything.

When will AMD learn that more is less when they need to split resources to develop so many.

They did a surprisingly good job at the 370, 380 and 390. But I bet if they made 370x, 380x, 385x etc etc at release, they'd flop.

The way I see it, if they made only a Fury X but threw the resources of Fury and Nano into it, it would have been a bloody killer. Where a, as it stands, it gets wrecked by the 980ti due to performance/potential loop issues as AIO tend to sometimes have after 2 years or so.

I sincerely hope that AMD starts to make smarter decisions.

/rantover.

[Agosto]

Again, probably not as complicated. It will be the 8.9 billion transistor chip.

 

nvidia has no issues with the GM200, which has 8 billion transistors and is made by TSMC just like Fiji. 

I still think it's because of the interposer, lots of lanes to connect, huge memory bus and new vram

[monstercameron]

There are logical leaps being taken in this thread..

Like if Nvidia sends a reviewer a sample of a card then they will likely give a glowing review because "now they are biased toward Nvidia"

But this ignores the fact that AMD also seeds graphics cards - in fact they also offered to seed us a GTX 970 mini when we said we didn't have one for direct comparison.

So by the above logic we are also based towards AMD.. and yet we criticize them both..

Here's a reality check: A graphics card means nothing to me. I have literally bins of them just like all the other hardware reviewers. They are just tools to do our jobs. It couldn't affect my opinion of the card in any meaningful way unless it was a super amazing card that lit my world on fire in which case there is nothing "biased" about giving it a glowing write up.

Do you see where I'm going with this?

Many other issues as well.. Especially calling out Scott as a sellout. He is anything but..

On my phone right now, but those two jumped out at me..

"I'm not a racist, I have plenty of black friends..."

Sounds eerily similar, methinks you doth protest too much.

[marldorthegreat]

nvidia has no issues with the GM200, which has 8 billion transistors and is made by TSMC just like Fiji. 

I still think it's because of the interposer, lots of lanes to connect, huge memory bus and new vram

who said they had no issues. And amd's chip is much denser than the nvidia chip.

[Notional]

nvidia has no issues with the GM200, which has 8 billion transistors and is made by TSMC just like Fiji. 

I still think it's because of the interposer, lots of lanes to connect, huge memory bus and new vram

 

Actually only the Titan X is full fat Maxwell. 980 ti are cut down, just like fury is. Titan X is a 1k$ GPU, so why are people whining about a 650$ GPU, that is the smallest on the market and outperforms GPU's 50% larger.

But Maxwell seems to be a simple architecture, which cannot do double precision compute to save its life, and is incapable of async compute.

The interposer is very simple, and it's not a problem to connect an interposer. Remember that every single pin on your 2011 CPU is connected to the silicon, which is not a problem. It's either HBM and/or the full fat Fiji.

[Slinkey]

HardOCP gave AMD's 390X, Fury, or Fury X cards a silver award or no award at all.

Not sure what they gave, but I do know they have a bit of traffic.

[LinusTech]

"I'm not a racist, I have plenty of black friends..."

Sounds eerily similar, methinks you doth protest too much.

Huh?

[MrRavens]

Thats where all my money comes from obviously, ill be ordering my pink lambo next week

 

Haha, that's what I'm talking about! xP

[pixeldensity]

Does TR have a youtube channel b/c if not, I'm thinking that what amd is choosing is to give the samples to reviewers who have a youtube channel and have a good amount of following on those channels. I maybe wrong but that what i'm thinking right now.

[RagnarokDel]

Contrary to what you may believe AMD, digging is not the best way to the promised land of China.

[shymi]

Doesn't mean it's AMD making the card. Sapphire has enough money to literally give away a card to every tenth person. AMD doesn't want to because AMD is afraid of what might be said about the card, not that it's a "limited supply."

I don't think, that Sapphire makes the HBM memory modules.

[Mahigan]

I feel like people have short term memories and are giving Scott W. a hard time with little reason to do so. Sure, his words are a little tangy stating he was not receiving a review sample, but this guys and his team IMO are at the top of tech jornalism. Scott could have done a lot of different things as a career, claiming he reviews PC hardware for the money is crazy (surely he makes a modest living).

 

 

Actually Scott W. questioned the 970 memory structure MONTHS before the big shit show, no one seemed to notice or care until it was popular to do so. He got no praise for bringing that up, but now gets crucified for being a little salty over not recieving a review smaple?

 

By the way, have you gotten any sleep in the last week Mahigan? I have seen you everywhere from reddit, [H], OCnet, B3D....and more . I have enjoyed reading the hundreds of pages your recent posts have stirred up this past week  B)

Barely any sleep...

 

I go to bed at 6am. I have a few advantages... My wife makes me sandwiches and coffee, so I can blog all day long and I'm currently living off the copious amounts of Bitcoins I mined a few years back waiting on immigration to accept my wife's request and head back to Canada.

 

I've stirred the pot a lot, probably too much, but I wanted nVIDIA to respond and they did

 

Now I can sit back and relax a bit.

[Mahigan]

I know these samples are scarce. But not giving it to HardOCP? Kinda weird.

I don't know what AMD's PR is thinking...

 

Going into DX12, you would think that they'd be on their best behavior. More open to supply samples to the tech reviewers. One thing that irritated me was how an editor at HardOCP summed up the R9 Nano as such, I'm paraphrasing, "I don't buy small cases where I can't fit a full length card, ergo a small card is of no use to me, ergo it's a useless product". I think statements like this leave a sour taste in AMDs mouth right now, not to mention their emphasis on "PAPER" launch.

 

AMD is struggling, perhaps even near bankruptcy. They can't afford bad press which doesn't use their Graphic cards for their intended purposes. The problem is, this is back firing on them. When they deny review samples... it creates bad press as well.

 

I'm looking forward to the R9 Nano reviews, curious to know the full function of those two HWS blocks (replacing four ACEs) in their block diagram.

[zMeul]

I missed one!

TechPowerUp are in the exact same boat: https://www.techpowerup.com/215776/amd-radeon-r9-nano-review-by-tpu-not.html

 

There won't be a Radeon R9 Nano review on TechPowerUp. AMD says that it has too few review samples for the press. When AMD first held up the Radeon R9 Nano at its "Fiji" GPU unveil, to us it came across as the most promising product based on the chip, even more than the R9 Fury series, its dual-GPU variant, and the food-processor-shaped SFF gaming desktop thing. The prospect of "faster than R9 290X at 175W" is what excited us the most, as that would disrupt NVIDIA's GM204 based products. Unfortunately, the most exciting product by AMD also has the least amount of excitement by AMD itself. 

 

The first signs of that are, AMD making it prohibitively expensive at $650, and not putting it in the hands of the press, for a launch-day review. We're not getting one, and nor do some of our friends on either sides of the Atlantic. AMD is making some of its tallest claims with this product, and it's important (for AMD) that some of those claims are put to the test. A validated product could maybe even convince some to reach for their wallets, to pull out $650.

at this point, I absolutely do not understand what AMD is doing ...

if they seriously don't have the press sample units, what the fuck are they actually launching? or are they precisely avoiding the press because of the claims they made are far fetched? only time will tell