KitGuru KitGuru.net – Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards
At a press conference at the Computex trade-show Advanced Micro Devices has officially demonstrated its first graphics processing unit with high bandwidth memory (HBM) code-named “Fiji”, which is set to power AMD’s next-generation flagship graphics card known unofficially as the Radeon Fury X. But while the company did show its “Fiji” chip, it did not want to show the “Fury” card, reports Leo Waldock, a KitGuru editor, from Taipei.
Leo arrived at Nangang at 9am and was told the floor was closed to the Press until 9:30am. He took the opportunity to head up to Sapphire, who were in a separate room, so he could say hello and grab a chair. At the Sapphire booth, Leo bumped into Ron Myers, Corporate Vice President of Corporate Marketing at AMD. Ron is a big guy in every sense of the word and only slightly intimidating.
Earlier AMD waved a “Fiji” chip around on stage. They refused to give its name but the unnamed chip had four stacks of HBM so it was “Fiji”, no question.
AMD is showing a running and apparently fully working version of “Fiji” to customers but not to the Press, so Leo asked Ron if he might relent. They both knew the other side of the Sapphire NDA door was the card Leo wanted to see, so Leo asked if he could see the card and take photos.
“How about static without performance?” Leo asked.
“No,” said Ron, and that was the end of that.
While it was impossible to take photos of the AMD Radeon Fury X, Leo tried to discuss some aspects of the flagship graphics card that only features 4GB of memory (compared to 12GB on Nvidia’s GeForce GTX Titan X). His question about the 4*1GB limit was side-stepped so it is unclear whether AMD could install 8GB of HBM if they choose. A natural guess is they cannot. Ron Myers assured that 4GB of high bandwidth memory is plenty for 4K gaming so it is not just a question of the amount of memory, but also the speed and bandwidth. Put it another way, the requirement for 8GB (or whatever) is in combination with the current, slower bandwidth.
Apparently, AMD wants to build the marketing impetus for the 16th of June, when the first reviews hit the Web, or the 24th, when the graphics cards hit the stores. The problem is, many aspects about the graphics board are already known: it has 4GB of memory, it has short printed circuit board, it requires two 8-pin PCI Express power connectors, it is power hungry (may consume up to 375W of power), it uses hybrid liquid cooling system, etc. The main intrigue about the AMD Radeon Fury X graphics board is its performance. Obviously, a static image of the product cannot reveal any performance details. Moreover, an image of the graphics card itself was recently published by Johan Andersson, technical director of Frostbite at Electronic Arts, in a Twitter post.
Discuss on our Facebook page, HERE.
KitGuru Says: It is hard to understand why AMD is not prepared to show “Fiji” to the Press. During the conversation Leo called the chip both “Fiji” and “Fury” and Ron did not blink, so we know that it is the correct name.
Looks like Fiji will go down as AMD’s Fermi, but worse, because they wouldn’t even show the card with wood screws.
u again ?
They probably want to find the balance between performance and power consumption or maybe have some problems with the drivers and need some time. I hope we will see reviews by the end of the month.
reviews will come sooner than that. I expect June 16 is going to be when the benchmarking sites will get their hands on a reference card. Then we’ll see posts by June 21. Just before official release… Or AMD will prevent any testing before release date.
In which case we can expect these cards to be absolute garbage. =
I am going to e3 so if no benchmarks I will steal one fly back home bench it and bring it back since I am not a real thief.
I just want 1440p 120fps.
How do you know its power hungry and could use up to 375W when AMD has said it shouldn’t use more than the 290x?
Why launch a premium gaming product to the press at Computex, when 2 weeks later, ALL the gaming press will be a captive audience at E3. Publicity = 10x greater.
Yeah, If the card is fast as Titan X. AMD probably had plans to use 20nm for Fiji for lower power consumption. In 2016 we will see 18nm and if nVidia releases the whole 9xx serie on it, I don’t know how AMD will catch up. They need new good mid-range GPU, not power hungry overpriced sh*t.
Anton, I appreciate the article. Why do you think Apple holds back on releasing info about their products until they are launched at a big media event? AMD is doing the same thing! E3 is the biggest gaming event in the largest market. AMD does not want to dilute the bang from its media event. So, the answer to your question is quite simple in my opinion.
They don’t know anything, and articles like these are one reason amd doesn’t show the press anything. Reading through this article you would think they did see fury and know everything about it already. Taking small jabs here and there because they’re butthurt they didn’t get some exclusive photos.
Oh God! AMD I used to like you, love you even, but this!? “8GB is because of the slow bandwidth.” Bull and you know it. HBM can’t solve storage speeds you spoon. “Up to 375W of power.” Well at least I can get use of the nuclear reactor I made, and the LN2 I bought, how considerate AMD…
You sir have more marketing knowledge than Ron there.
I like the straight forward article, though as Azix pointed out saying, “it is power hungry (may consume up to 375W of power)”, while the bracket part is technically apropos for the 2x8pin PCI-e connectors… the first part is complete speculation. It would have been more correct to have it written… 8-pin PCI Express power connectors, allow for up to 375W of power (power hungry?).
As to why AMD is waiting for E3? I say because Nvidia had to jump in with the 980Ti so abruptly, almost wanting to spoil AMD’s party, and AMD said fine we have a contingency for that.. When Nvidia appear set to move ahead, (originally Nvidia talk was July/Aug for a 980Ti) I think AMD said let’s see what they bring. Why compete head to head with the review juggernaut, and then have more control as folks watch that first trickle of reference cards go OoS for two or more weeks.
I really thought AMD was looking at Computex for a “paper launch” of Fiji with review cards released at that point, and NDA being lifted a couple of days ahead of E3. In this way allowing them ride the wave into E3, while starting to offer all the rest of their mainstream stack. I don’t know if AMD’s holding back offered any time to juggle the clocks with bois flash, perhaps. If AMD had held off packaging and everything was sitting at the ready, that’s only a day or so to have quantities out the door. Honestly, I don’t think they needed to that they knew from TitanX what they had. Sounds like they took the time to see how 980Ti dealt with some titles, and worked with drivers to pick up tighten up where they could, although more possibly tailoring back on the games they clearly had a lead, and hold that performance till later.
You ever think that maybe amd are watercooling because it can actually give them an advantage? How else can they separate their products from nvidia? If they built a similar reference fan and cooling unit as nvidia everyone would be crying that they’re just trying to copy nvidia.
nope. they are using it because they were already pushing 28nm with the 290X and its 95C reference load temps. and if it was an advantage, AMD wouldn’t be stalling until E3 to show it.
meanwhile, nvidia made the most efficient gpu ever with their third 28nm dGPU lineup…
another stalker….
Stalling? Pretty sure this e3 show was already planned several months ago with the intention of launching fiji. Who cares if titan x performs the same as fiji xt while using 30 watts less power. Nvidia fanboys had no idea what efficiency meant until recently. Where were all of you guys for the decade before nvidia started focusing on efficiency? Amd held that crown for over a decade, now that nvidia are more effecient you act like it is the single greatest achievement in their history.
if it was scheduled for E3 all along, why show anything at Computex?
they cant’, i want remember some GTX 980 CUSTOMS VERSIONS, have 2x 8 pin,, that is, i think, for tell again, amd is hot and is power hungry bla bla bla, but i don’t care from power consumption, i want REAL POWER, not gpu castrated how nvidia do, and poor chickens people go buy this gpus, because they tell, ohh is low pwr consumption, ohh is not hot ahha, 900 series is to hot, when one gpu go over 65° is HOT, and is remain only 2 weeks for see what furyx can do, and i’m sure one the best gpus we never see, and remember nvidia or amd brings innovations? for what i see in this last 6 years, AMD BRINGS INNOVATIONS MORE THAN NVIDIA, but nvidia marketing did brainwashed into believing that their products are innovative , just like apple hahaha, i hope with that series, people can’t stop go listen so much lie nvidia tell, because nvidia how many lie tell me in 10 years? SOOOOOOOOOOO MUCH, and naturally driver story, is a BULLSHIT, is a bullshit nvidia have better driver, actually drivers from amd is sometimes better than nvidia drivers, because 1 amd don’t block and castrated performance on older series, but nvidia do this, 2 amd don’t broken nvidia gpus performance when go put their effects on some games, but nvidia do this, and yeah sorry i just miss one, AMD DON’T MAKE ONE DRIVER BURN amd did the drivers that break and burn video cards, while nvidia not once, but several times, not to mention other problems which, believe me, the story that the nvidia driver works best is all a big jack ass set in motion by the nvidia fanboy, whatever, just released series FURY, I would bet that will be more powerful than the same titan x, because if so, I bet the nvidia fanboy just to say it a board of shit, will cling on consumption of the card, just wait, just missing a couple of weeks, anyway. I think the 300 series of AMD, will be a great series, ok I understand that some will use the GPU chip used on the 200 series, but it’s worth throwing away a chip that still offers very high performance, as hawaii of 290x? no, it’s worth it to discard, because in the end when one asks a PC to play and wants BRUTE POWER AND PURE in GPGPU and especially in gaming DOES NOT MATTER FOR NOTHING OF CONSUMPTION, in the end the difference on the bill in a ‘ year is 10 euro, is virtually NOTHING 🙂
omg…………yeah sure amd fermi, please stop tell so much BULLSHIT, ahhah another retard kid fanboy!!
hahahah yeah hahaha go buy you nvidia castrated gpu, haha maxwell on GPGPU, opencl performance is shit, and yeah i want remember nvidia disable FP64 on their gpus, that why have lower consumption, and come are you seriusly 95°? ahahah omfg yeah sure, ohhh how it’s possibile my 290x friend never go over 60° on full load? jimmy you are ridicolous, maxwell is not efficient is fake efficient, because when one gpu is EFFICIENT, HAVE ALL ENABLE AND IS NOT CASTRATED with low tdp and cosumption, and opsss enjoy your nvidia gpu castrated, and naturally suck on dx12 perfomance, and yeah is not fully support dx12 ahhaah, ejoy you overpriced fail gpu 😉
yeah i quote you, nvidia fanboys is full of retard, ignorance, and they are huge chickens, and nvidia like rape ass and money to chickens 😀
get help, k?
AMD is console. They have always been about PC games being console ports. They are all about console parity.
Only Nvidia is pushing new tech.
Fermi didn’t come close to putting out the kind of heat AMD does, kid.
They should call it the fiji fire starter pro.
Nvidia has been on efficiency since 2010. Perf per watt = perf.
The race to exascale computing. A race AMD wouldn’t take part in. Still no HPC product.
Since 2010? 2012 maybe when kepler launched.
Why did nvidia show titan x before the official reveal? If amd had shown nothing you fanboys would be crying all the same.
Same reason u show a trailer before the movie….
because E3 revolves around console gaming. your average E3 spectator can’t tell a graphic card from a dildo.
since 2010? where were you when nVidia reveled Fermi in 2010?