AMD Explains Why 4GB of HBM Is Not an Issue for Fiji GPUs

[bp9801]

There have been some concerns lately about AMD's upcoming Fiji GPU and its use of HBM memory. Namely the amount of it, as the card could use 4GB of HBM, which is the same amount the current high end cards from both AMD and NVIDIA have. If Fiji is as powerful as the rumors suggest, wouldn't that 4GB HBM be a limitation and a possible hindrance to pushing the card as far as possible for graphical fidelity? Well, AMD CTO Joe Macri recently explained why that is nothing to worry about, as according to Macri:

You're not limited in this world to any number of stacks, but from a capacity point of view, this generation-one HBM, each DRAM is a two-gigabit DRAM, so yeah, if you have four stacks you're limited to four gigabytes. You could build things with more stacks, you could build things with less stacks. Capacity of the frame buffer is just one of our concerns. There are many things you can do to utilise that capacity better. So if you have four stacks you're limited to four [gigabytes], but we don't really view that as a performance limitation from an AMD perspective.

If you actually look at frame buffers and how efficient they are and how efficient the drivers are at managing capacities across the resolutions, you'll find that there's a lot that can be done. We do not see 4GB as a limitation that would cause performance bottlenecks. We just need to do a better job managing the capacities. We were getting free capacity, because with [GDDR5] in order to get more bandwidth we needed to make the memory system wider, so the capacities were increasing. As engineers, we always focus on where the bottleneck is. If you're getting capacity, you don't put as much effort into better utilising that capacity. 4GB is more than sufficient. We've had to go do a little bit of investment in order to better utilise the frame buffer, but we're not really seeing a frame buffer capacity [problem]. You'll be blown away by how much [capacity] is wasted.

Macri did not say exactly how much HBM would be on the Fiji GPU, especially since HBM supports a 1024-bit bus and over 100GB/s of bandwidth per stack. That is an insanely huge number on both accounts, because four stacks would mean a 512GB/s bandwidth. HBM is also much smaller than GDDR5, so where a1GB of GDDR5 takes up 672mm², a 1GB stack of HBM occupies only 35mm². According to AMD, it allows for a PCB 50% smaller than the one with the R9 290X, which bodes exceptionally well for a dual-GPU card or small form factor cards.

So, a Fiji GPU equipped with 4GB of HBM1 potentially sounds like it will be extremely competitive with what NVIDIA has to offer. Of course, first we actually need that Fiji GPU to see for sure, which Macri says should arrive sometime in the next two months. Maybe we'll see it during Computex and then get our hands on it near the end of June. But we'll just have to wait and see.

Source: Ars Technica via WCCFtech

Back to original news post