The Fiji GPU: Go Big or Go Home

Now that we’ve had a chance to take a look at the architecture backing Fiji, let’s talk about the Fiji GPU itself.

Fiji’s inclusion of High Bandwidth Memory (HBM) technology complicates the picture somewhat when talking about GPUs. Whereas past GPUs were defined by the GPU die itself and then the organic substrate package it sits on, the inclusion of HBM requires a third layer, the silicon interposer. The job of the interposer is to sit between the package and the GPU, serving as the layer that connects the on-package HBM memory stacks with the GPU. Essentially a very large chip without any expensive logic on it, the silicon interposer allows for finer, denser signal routing than organic packaging is capable of, making the ultra-wide 4096-bit HBM bus viable for the first time.

We’ll get to HBM in detail in a bit, but it’s important to call out the impact of HBM and the interposer early, since they have a distinct impact on how Fiji was designed and what its capabilities are.

As for Fiji itself, Fiji is unlike any GPU built before by AMD, and not only due to the use of HBM. More than anything else, it’s simply huge, 596mm2 to be precise. As we mentioned in our introduction, AMD has traditionally shied away from big chips, even after the “small die” era ended, and for good reason. Big chips are expensive to develop, expensive to produce, take longer to develop, and yield worse than small chips (this being especially the case early-on for 40nm). Altogether they’re riskier than smaller chips, and while there are times where they are necessary, AMD has never reached this point until now.

The end result is that for the first time since the unified shader era began, AMD has gone toe-to-toe with NVIDIA on die size. Fiji’s 596mm2 die size is just 5mm2 (<1%) smaller than NVIDIA’s GM200, and more notably still hits TSMC’s 28nm reticle limit. TSMC can’t build chips any bigger than this; Fiji is as big a chip as AMD can order.

AMD Big GPUs
  Die Size Native FP64 Rate
Fiji (GCN 1.2) 596mm2 1/16
Hawaii (GCN 1.1) 438mm2 1/2
Tahiti (GCN 1.0) 352mm2 1/4
Cayman (VLIW4) 389mm2 1/4
Cypress (VLIW5) 334mm2 1/5
RV790 (VLIW5) 282mm2 N/A

Looking at Fiji relative to AMD’s other big GPUs, it becomes very clear very quickly just how significant this change is for AMD. When Hawaii was released in 2013 at 438mm2, it was already AMD’s biggest GPU ever for its time. And yet Fiji dwarfs it, coming in at 158mm2 (36%) larger. The fact that Fiji comes at the latter-half of the 28nm process’s life time means that such a large GPU is not nearly as risky now as it would have been in 2011/2012 (NVIDIA surely took some licks internally on GK110), but still, nothing else we can show you today can really sell the significance of Fiji to AMD as much as the die size can.

And the fun doesn’t stop there. Along with producing the biggest die they could, AMD has also more or less gone the direction of NVIDIA and Maxwell in the case of Fiji, building what is unambiguously the most gaming/FP32-centric GPU the company could build. With GCN supporting power-of-two FP64 rates between 1/2 and 1/16, AMD has gone for the bare minimum in FP64 performance that their architecture allows, leading to a 1/16 FP64 rate on Fiji. This is a significant departure from Hawaii, which implemented native support for ½ rate, and on consumer parts offered a handicapped 1/8 rate. Fiji will not be a FP64 powerhouse – its 4GB of VRAM is already perhaps too large of a handicap for the HPC market – so instead we get AMD’s best FP32 GPU going against NVIDIA’s best FP32 GPU.

AMD’s final ace up their sleeve on die size is HBM. Along with HBM’s bandwidth and power benefits, HBM is also much simpler to implement, requiring less GPU space for PHYs than GDDR5 does. This is in part due to the fact that HBM stacks have their own logic layer, distributing some of the logic on to each stack, and furthermore a benefit of the fact that the signaling logic that remains doesn’t have to be nearly as complex since the frequencies are so much lower. 4096-bits of HBM PHYs still takes up a fair bit of space – though AMD won’t tell us how much – but it’s notably lower than the amount of space AMD was losing to Hawaii’s GDDR5 memory controllers.

The end result is that not only has AMD built their biggest GPU ever, but they have done virtually everything they can to maximize the amount of die space they get to allocate to FP32 and rendering resources. Simply put, AMD has never reached so high and aimed for parity with NVIDIA in this manner.

Ultimately this puts Fiji’s transistor count at 8.9 billion transistors, even more than the 8 billion transistors found in NVIDIA’s GM200, and, as expected, significantly more than Hawaii’s 6.2 billion. Interestingly enough, on a relative basis this is almost exactly the same increase we saw with Hawaii; Fiji packs in 43.5% more transistors than Hawaii, and Hawaii packed in 43.9% more transistors than Tahiti. So going by transistors alone, Fiji is very much to Hawaii what Hawaii was to Tahiti.

Finally, as large as the Fiji GPU is, the silicon interposer it sits on is even larger. The interposer measures 1011mm2, nearly twice the size of Fiji. Since Fiji and its HBM stacks need to fit on top of it, the interposer must be very large to do its job, and in the process it pushes its own limits. The actual interposer die is believed to exceed the reticle limit of the 65nm process AMD is using to have it built, and as a result the interposer is carefully constructed so that only the areas that need connectivity receive metal layers. This allows AMD to put down such a large interposer without actually needing a fab capable of reaching such a large reticle limit.

What’s interesting from a design perspective is that the interposer and everything on it is essentially the heart and soul of the GPU. There is plenty of power regulation circuitry on the organic package and even more on the board itself, but within the 1011mm2 floorplan of the interposer, all of Fiji’s logic and memory is located. By mobile standards it’s very nearly an SoC in and of itself; it needs little more than external power and I/O to operate.

Fiji’s Architecture: The Grandest of GCN 1.2 Fiji’s Layout
Comments Locked

458 Comments

View All Comments

  • e36Jeff - Thursday, July 2, 2015 - link

    Do you think that there is some more optimizations in the drivers that would increase the performance of the fury x vs 980 Ti? I understand that its a bit like peering into a crystal ball to know what kind of performance driver updates will bring, but I'm thinking your guesses would be more educated than most peoples.
  • Ryan Smith - Thursday, July 2, 2015 - link

    I am going to decline to peer into the crystal ball at this time.
  • cruzinbill - Thursday, July 2, 2015 - link

    I still don't understand how these numbers are correct. The 290X performance is completely wrong on the Beyond Earth test. Where you are pulling only 86.3 avg fps on a standard 290X I am pulling 110.44. I noticed the same when you showed Star Swarm test before. I can supply proof if need be.
  • Ryan Smith - Thursday, July 2, 2015 - link

    Are you using MSAA? We're not doing anything special here on Civ, and these numbers are consistent and repeatable.
  • cruzinbill - Thursday, July 2, 2015 - link

    Yes same MSAA level. If I disable MSAA I get 118.74 avg fps.
  • Oxford Guy - Friday, July 3, 2015 - link

    Theirs is probably throttling.
  • Ryan Smith - Friday, July 3, 2015 - link

    The 290X by its very nature throttles, which is why we also have the "uber" results in there. Those aren't throttling, and are consistent from run-to-run.
  • TheinsanegamerN - Saturday, July 4, 2015 - link

    So you guys are using a stock 290x? that would make sense then. Third party cooled models work much better and more consistently, and many have overclocks.
  • chizow - Sunday, July 5, 2015 - link

    Yes this goes to AT's long-standing policy to use the reference cooler and stock clocks provided by the IHVs, going back to a dust-up over AnandTech using non-reference cooled and overclocked EVGA cards. AMD fanboys got super butthurt over it and now they reap the policy that they sowed.

    The solution for AMD is to design a better cooler or a chip that is adequately cooled by their reference coolers; it looks like they got the memo. Expect air cooled Fury and all the 300 series Rebrandeons to not run into these problems as all will be custom/air-cooled from the outset.
  • FMinus - Tuesday, July 7, 2015 - link

    It's not really realistic tho, I can't get a reference AMD 280, 280x, 290, 290x anywhere in Europe except maybe used, and I couldn't get those said reference cards a month after launch of the said cards. Same pretty much goes for nvidia, but a lot of high-end nvidia cards still use the blower design for some reason, so I can at least get a reference look alike there, not so with AMD cards.

    Which again makes the results a bit skewed if the performance of said cards is really that much dependent on better cooling solutions, since better cooling solutions is all people can buy.

Log in

Don't have an account? Sign up now