News Custom SFF Radeon R9 Nano Rig Packs 10 TFlops Performance

AMD’s Radeon R9 Nano is an impressive SFF (Small Form Factor) graphics card that packs a lot of punch and shows what’s possible with the HBM architecture. DGLee managed to get a sample of the Radeon R9 Nano graphics card and made a custom SFF Mini-ITX rig with it that packs a total of 10 TFlops of performance under its hood.

Would you like to know more ? <click>

 

[Phuncz]

No ofcourse not, most reviewers don't spend much time with higher thermal limits and boost clock profiles when comparing driver versions. But it did get noticed. It was also the fabled driver that was going to give over 70% performance increases, only to end up giving 11% in two specific games, that were basically not properly optimized earlier.

Don't get it the wrong way, I don't care who buys what, but AMD doesn't deserve the amount of trash-talk it gets with Nvidia's track record. But aren't saints, that's a fact.

But let's get back on-topic, since the reviews are out on the Nano card: https://smallformfactor.net/forum/threads/now-this-could-be-big-r9-nano.100/page-5#post-3273

 

[iFreilicht]

Maybe link to the R9 Nano thread, so we can all discuss the reviews there. Don't want to split the discussion up.

 

[Vittra]

Cherry-picking is hard to prove without multiple sources because it could just very well be a good chip at random. But The GPU manufacturers know how good a chip is before it gets soldered, which they can sell at a premium, reserve for higher binned SKUs or for "marketing purposes".

But I've seen nastier practices than that, like Nvidia's driver released just after AMD's Mantle driver, passively OC'ing certain GPUs by adjusting their thermal limit.

To clarify, I meant the review unit's baseclock (which was OC'd beyond reference base clock of course) was higher than what the retail models were specced for - not the boost clock which is subject to specific chip quality.