Friday, July 31st 2020
AMD Radeon MI100 "Arcturus" Alleged Specification Listed, the GPU Could be Coming in December
AMD has been preparing to launch its MI100 accelerator and fight NVIDIA's A100 Ampere GPU in machine learning and AI horizon, and generally compute-intensive workloads. According to some news sources over at AdoredTV, the GPU alleged specifications were listed, along with some slides about the GPU which should be presented at the launch. So to start, this is what we have on the new Radeon MI100 "Arcturus" GPU based on CDNA architecture. The alleged specifications mention that the GPU will feature 120 Compute Units (CUs), meaning that if the GPU keeps the 64-core per CU configuration, we are looking at 7680 cores powered by CDNA architecture.
The leaked slide mentions that the GPU can put out as much as 42 TeraFLOPs of FP32, single-precision compute. This makes it more than twice as fast compared to NVIDIA's A100 GPU at FP32 workloads. To achieve that, the card would need to have all of its 7680 cores running at 2.75 GHz, which would be a bit high number. On the same slide, the GPU is claimed to have 9.5 TeraFLOPs of FP64 dual-precision performance, while the FP16 power is going to be around 150 TeraFLOPs. For comparison, the A100 GPU from NVIDIA features 9.7 TeraFLOPS of FP64, 19.5 TeraFLOPS of FP32, and 312 (or 634 with sparsity enabled) TeraFLOPs of FP16 compute. AMD GPU is allegedly only more powerful for FP32 workloads, where it outperforms the NVIDIA card by 2.4 times. And if that is really the case, AMD has found its niche in the HPC sector, and it plans to dominate there. According to AdoredTV sources, the GPU could be coming in December of this year.
Source:
AdoredTV via VideoCardz
The leaked slide mentions that the GPU can put out as much as 42 TeraFLOPs of FP32, single-precision compute. This makes it more than twice as fast compared to NVIDIA's A100 GPU at FP32 workloads. To achieve that, the card would need to have all of its 7680 cores running at 2.75 GHz, which would be a bit high number. On the same slide, the GPU is claimed to have 9.5 TeraFLOPs of FP64 dual-precision performance, while the FP16 power is going to be around 150 TeraFLOPs. For comparison, the A100 GPU from NVIDIA features 9.7 TeraFLOPS of FP64, 19.5 TeraFLOPS of FP32, and 312 (or 634 with sparsity enabled) TeraFLOPs of FP16 compute. AMD GPU is allegedly only more powerful for FP32 workloads, where it outperforms the NVIDIA card by 2.4 times. And if that is really the case, AMD has found its niche in the HPC sector, and it plans to dominate there. According to AdoredTV sources, the GPU could be coming in December of this year.




34 Comments on AMD Radeon MI100 "Arcturus" Alleged Specification Listed, the GPU Could be Coming in December
To each his own... I think its a complete waste of time and so far havent been proven wrong. This sort if reporting has zero impact on anything, its just clickbait.
He called out first that Vega64 is much weaker than RTX2080Ti.
He was first one to release info about RTX naming scheme for Nvidia and the silicon names.
There were some Zen2 stuff also that he nailed, can't remember right now.
Jim actually analyses the info he gets, Wccftech and similar pages just publish the info they get without deep diving into it.
What have You been doing lately except calling others out?
I remember when everyone was all hyped up about AMDs Arcturus architecture, but then I believe it was Adoredtv that mentioned Arcturus was one product rather than an architecture. And now it is starting to surface that the top CDNA chip is codenamed Arcturus.
The 2nd bold, should maintain that every website contains some form of regurgitated rumors and wetnaps to wipe up drool.
That is really all this is. He doesn't analyse shit, he makes educated guesses and has a few sources. The vast majority just can't be arsed to even look into these kind of rumors because like me, they honestly don't give a damn because product only counts when its on shelves and specs are combined with the other necessary metrics such as price point, TDPs, actual release time and availability, etc. Those are the minutia that make or break a product. Not the idea of it having 3500 or 4000 shaders. Remember the HBM hype? Everyone and his dog had to have it. Today we're looking at GDDR6. What's the point of all of it?
This whole tech rumor mill has never been my thing, all Adored does for me is take it to another level with an annoying accent and overall tone of voice, and a brutally long monologue of boredom and graphs. That is all, really. So sure, if you want to step into the rumor arena and get thrown from left to right all the time with what you're supposed to believe, Adored and a range of other 'Tubers are your first stop.
I'm not that guy. Sure, some speculating based on semi-confirmed/credible info, but I don't attribute much value to it in any way shape or form, and neither should anyone else, because in the end, that is all AdoredTV does: speculate. Hopefully its true but again, what's that going to help you with? The only value this sort of info has is to start yet another completely misguided hype train forum topic only to get disappointed or completely overhyped when product hits shelves. I've never really seen the merit of that, I'm not that kind of slave to commerce or herd mentality... To each his own...
I appraise technology and releases differently altogether, to be honest. I'm way past chasing the cutting edge and diving on every hype that gets thrown out there, because once you've seen a few generations... you've really seen them all. Adored knows this... now his subscribers still need to have that penny drop. He's got a good business model and a good following of gullible fools.
I know this is not going to land as a popular message. I hope you can get past the emotional first response because yes this is a straight up judgment on all his followers including you. Some things need to be said, apparently. Its not personal though - nothing against you or any other individual. Its something to consider, usually we only realize things like this have been happening to us in hindsight.What I've been doing? Being a dad, having a life beyond the internet, you know, the usual things. TPU takes a half hour out of it as well, a half hour ten times more enjoyable and useful than ever watching 30 minutes of AdoredTV (or other Tuber) nonsense ;) I don't think its a relevant question though, I'm not looking for subs.
It doesn't even register to me that he's often right. I simply don't care because I don't consider him a credible or trustworthy source and likely never will. You only need to be wrong a few times to land in that position with me, btw.