Certhas

21 ms·

361.
by Certhas 2y ago
I don't know about Lumi specifically, but top tier scientific supercomputers should typically have 80+% utilisation rate: https://doku.lrz.de/usage-statistics-for-supermuc-ng-1148309... Smaller machines will tend lower
362.
by Certhas 2y ago
Also, from what I understand after watching some of the video demonstration in the Q&A, I could just have another EFI entry point towards the nmbl configuration with a grub like menu, and get an exact replica of the grub experience. Hav
363.
by Certhas 2y ago
Nowhere did you say that you think trying to build an interface that is as close as possible to cuda is the problem. You still haven't said why it's a problem. You only said "look at the implementation" and said that if
364.
by Certhas 2y ago
Again, your argument is ad hominem ("you don't understand") and not naming an actual problem but declaring that it's obvious to anyone who wants to look. The AMD and the CUDA backend in pytorch are actually derived from
365.
by Certhas 2y ago
I have said nothing of the sort anywhere in this thread. All I said is that AMD is viable for some large customers. You chose to continue to misinterpret that after several clarifications. Then you throw out vague statements, and claim that
366.
by Certhas 2y ago
Frontier will run for more than 2 years before being replaced, but there are many more national lab type organizations than FAANG companies. If you decide to classify anything below FAANG scale as "not large", well... And anyway,
367.
by Certhas 2y ago
Frontier has 38k MI250. Google and Amazon apparently bought 50k H100 in 2023. (Behind Meta and Microsoft at 150k) I believe 38k definitely qualifies as large. Maybe you misread me. I didn't mean to write that this was the default choic
368.
by Certhas 2y ago
You have no facts and only cliche lines for reasons either. What does software penetrative or power laws have to do with anything? I don't know what relevance term sheets should have here, but AMD success at the HPC top end is pretty o
369.
by Certhas 2y ago
I don't see it. Large customers are choosing AMD over nVidia. NVidia has a massive headstart, but there are diminishing returns to their software work. If AMD performance per dollar is 10% better at the hardware level, and their softwa
370.
by Certhas 2y ago
Have you ever organized anything of size? Financing is not the bottleneck. Organizational capacity might well be, though. As an organization, AMDs survival depended not on competing with nVidia but on competing with Intel. Now they are esta
371.
by Certhas 2y ago
When you ask your funding agency for an HPC upgrade or a new machine, the first thing they will want from you are utilisation numbers of current infrastructure. The second thing they will ask is why you don't just apply for time on a
372.
by Certhas 2y ago
In Europe at least supercomputer are organised in tiers. Tier 0 are the highest grade, tier 3 are small local university clusters like the one you describe. Tier 2 or Tier 1 machines and upward usually require you to apply for time. They ar
373.
by Certhas 2y ago
This is a common sentiment, no doubt also driven by the wish thay AMD would cater to us. But I see no evidence that the strategy is wrong or failing. AMD is already powering a massive and rapidly growing share of Top 500 HPC: https:/&
374.
by Certhas 2y ago
Very well put.
375.
by Certhas 2y ago
The article also explains that the natural way to look at this property is in terms of matrix exponentials/logarithm. The space of rotation matrices is not a linear space. It's obvious that interpolation is meaningless there. The
376.
by Certhas 2y ago
I don't get the love of quaternions in this context. How are quaternions more intuitive than matrices? A matrix acts on a vector. Rotations act on vectors. What could be more natural than to look at rotations as matrices? The matrix ex
377.
by Certhas 2y ago
> EU GDP per capita 2022 is the same as US GDP per capita 2017. Can't edit anymore, but: That were nominal numbers, and thus useless. See here: https://news.ycombinator.com/item?id=40673552
378.
by Certhas 2y ago
Can't edit anymore, but this little factoid: > More importantly, even GDP per Capita wise: > > https://data.oecd.org/gdp/gross-domestic-product-gdp.htm > > EU per Capita GDP in 2022 is the same as
379.
by Certhas 2y ago
The 2017 number was a mistake by me, I wanted to look at inflation adjusted/PPP, and that was nominal. You can clearly see the massive effect of the Russian invasion of Ukraine and the resulting inflation in the data. If we were lookin
380.
by Certhas 2y ago
Sources please? Are these nominal dollars? World bank data in PPP dollars is reported as 64,600 vs 45,900 here: https://ourworldindata.org/grapher/gdp-per-capita-worldbank?... Germany is at 53,900 there, but a good chu
381.
by Certhas 2y ago
These numbers are PPP, so inflation adjusted.
382.
by Certhas 2y ago
EU GDP per capita 2022 is the same as US GDP per capita 2017. Unless you want to say that the US was much poorer in 2017 than it was in 2022 that's a fairly ridiculous statement. Also, the highest productivity places in the EU have muc
383.
by Certhas 2y ago
.... while all of the content is hosted on Linux servers, and you're listening to music through a Swedish app. That is running on silicone made in Taiwan. Using equipment that can currently only be manufactured in Belgium and Germany.
384.
by Certhas 2y ago
Nvidia's valuation is in part driven by the fact that their chips potentially disrupt the vehicle for ad revenue that is search. For all his insanity, the one thing I respect Musk for, is that he actually started successful companies t
385.
by Certhas 2y ago
As someone with kids... What if there is? The presence of dried Semen on a seat on which my kids sit will have zero negative impact on them. At all. Is it Yuck? Yes of course. But it also seems extremely unlikely. And it's a lot less Y
386.
by Certhas 2y ago
I think you were being absolutely precise, but I want to give a tiny bit of constructive feedback anyway: In my experience, to not be misunderstood it is more important to understand the state of mind/frame of reference of your audienc
387.
by Certhas 2y ago
The vast majority of what we know would be due to public funding. Not private donors.
388.
by Certhas 2y ago
I am German. German culture does not have the Anglo Saxon obsession with swear words to begin with. The US/British obsession with keeping dirty words from children's ears is a never-ending source of amusement for me. I think you m
389.
by Certhas 2y ago
I mean... I _do_ think that language sensitivities are overblown these days. E.g. the majority of women participants did not "agree or strongly agree" that NIPS should become NeurIPS, yet it did. GIMP and Git are words that most n
390.
by Certhas 2y ago
I used the term "as established as google". In my mind this certainly meant the market has to be established. As long as an industry is brand new and rapidly developing, things are obviously different. Many early market leaders di

More ›