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
Why only the AMD section looks like it was copy pasted from source that had white background?
okayyyyyyyyyy
Or maybe they just misinterpreted his funny accent. I can't listen to it for any longer than 5 minutes, and since the man usually has 45+ minute monologues...
To be fair I think even Wccftech has a higher hit rate than him.
- my anus
We don't dislike him because he's an AMD fanboy.We dislike him because he's about as trustworthy as the current US president.
Repeating rumours or straight-up making them up is not news, it's FUD. And broken clocks are right twice a day.
You do know that he has called out AMD and their shitty Radeon products constantly and praises Nvidia products? Only Intel with his lacluster CPUs and very questionable tactics against smaller companies have been frowned upon by Adored, as it should be. And everybody in right mind should call out Intel about their tactics past 10year, that has led them to this 2 failed node releases (10nm and 7nm).
So get your facts straight and do your homework, before you start to call somebody a fanboy.
AdoredTV guy was good at analysis, blew it royally with NAVI leaks, which he was basing on some unknown dude sending him emails.
So, on one hand, nothing wrong with him, on the other, where do these leaks come from, another unknown dude sending him emails?
Greenboi might not know, chuckle, but you surely do know how hilariously far all that is from reality.
I don't like to pick on anyone for things they can't control, but that accent does no favors for the videos either...
This video is sheding much light to the issue and looking back at this old video, makes a lot clearer.
I am non English speaker and I actually like Scottish accent. Dunno, maybe just me, but that does not make me turn away. It's more painful to listen Chinese or Indian accent English, were it is truly hard to understand some pronunciations.
I think he has very good knowledge compared to most YouTubers, but where you venture into leaks and speculations it's hit and miss.
Also, CUDA dominates the existing GPU compute market. I wonder how much use one can make out of these.
Also, aren't leaks always supposed to be taken skeptically? And aren't all leaks, regardless of source, hit or miss? I just don't understand why they're holding AdoredTV personally accountable when he's just relaying some information, seems very personal.
Vega will crush 1080 Ti it's so powerful just wait
Navi 10 $249 and 150W against 2080! novideo is dead bois
Yeah see I'm firmly of the belief that if you can't cite your source you're full of shit, and if you can site your source you shouldn't be talking. People with inside info don't go blabbing it on Youtube. At this point if Adored gets anything right it's because whoever his "sources" are accidentally predicted something while they were coming up with more fake slides to dangle on a string in front of his face.
When I watch Jim's vidoes, I just think along, what if ... i don't trust that info and set myself up for a big letdown.
Hell, when I even read newspapers today, from reputable outlets, I don't 100% believe those, because nobody is right 100% and is trusted.
And now you go in believing some YT channel (and I dont mean only AdoredTV here) and expect the info you get from person who analyses the data, that has been leaked or given from unknown source half of the time. That is messed up! You just set yourself up...
Even if he got something wrong, it does not mean that info was wrong also. It may have been 6 months old info, where he made his analysis.
You only can predict 100% right all the time, when the product is actually out! Everything else is always up in the air and can change.
Unless You are Lisa Su's or Jensen Huang's family and have inside info from the source :p