Optimizing ClickHouse for Intel's ultra-high core count processors
1–10 of 53 posts
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#2This is my favorite type of HN post, and definitely going to be a classic in the genre for me.
> Memory optimization on ultra-high core count systems differs a lot from single-threaded memory management. Memory allocators themselves become contention points, memory bandwidth is divided across more cores, and allocation patterns that work fine on small systems can create cascading performance problems at scale. It is crucial to be mindful of how much memory is allocated and how memory is used.
In bioinformatics, one of the most popular alignment algorithms is roughly bottlenecked on random RAM access (the FM-index on the BWT of the genome), so I always wonder how these algorithms are going to perform on these beasts. It's been a decade since I spent any time optimizing large system performance for it though. NUMA was already challenging enough! I wonder how many memory channels these new chips have access to.
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#3This post looks like excellent low-level optimisation writing just in the first sections, and (I know this is kinda petty, but...) my heart absolutely sings at their use of my preferred C++ coding convention where & (ref) neither belongs to the type nor the variable name!
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#4288 cores is an absurd number of cores.
Do these things have AVX512? It looks like some of the Sierra Forest chips do have AVX512 with 2xFMA…
That’s pretty wide. Wonder if they should put that thing on a card and sell it as a GPU (a totally original idea that has never been tried, sure…).
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#5288 cores is an absurd number of cores. Do these things have AVX512? It looks like some of the Sierra Forest chips do have AVX512 with 2xFMA… That’s pretty wide. Wonder if they should put that thing on a card and sell it as a GPU (a totally original idea that has never been tried, sure…).
Sadly, no! On the bright side, they support new AVX2 VNNI extensions, that help with low precision integer dot products for Vector Search!
SimSIMD (inside USearch (inside ClickHouse)) already has those SIMD kernels, but I don’t yet have the hardware to benchmark :(
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#6This is my favorite type of HN post, and definitely going to be a classic in the genre for me. > Memory optimization on ultra-high core count systems differs a lot from single-threaded memory management. Memory allocators themselves become contention points, memory bandwidth is divided across more cores, and allocation patterns that work fine on small systems can create cascading performance problems at scale. It is…
My expectation, they will perform great! I’m now mostly benchmarking on 192 core Intel, AMD, and Arm instances on AWS, and in some workloads they come surprisingly close to GPUs even on GPU-friendly workloads, once you get the SIMD and NUMA pinning parts right.
For BioInformatics specifically, I’ve just finished benchmarking Intel SPR 16-core UMA slices against Nvidia H100, and will try to extend them soon: https://github.com/ashvardanian/StringWa.rs
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#7Great work!
I like duckdb, but clickhouse seems more focused on large scale performance.
I just thought that the article is written from the point of view of a single person, but has multiple authors, which is a bit weird. Did I misunderstood something?
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#8Great work! I like duckdb, but clickhouse seems more focused on large scale performance. I just thought that the article is written from the point of view of a single person, but has multiple authors, which is a bit weird. Did I misunderstood something?
Not sure what happened here, but it's not uncommon for a post to have one primary author and then multiple reviewers/supporters also credited
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#9288 cores is an absurd number of cores. Do these things have AVX512? It looks like some of the Sierra Forest chips do have AVX512 with 2xFMA… That’s pretty wide. Wonder if they should put that thing on a card and sell it as a GPU (a totally original idea that has never been tried, sure…).
Sadly, no! On the bright side, they support new AVX2 VNNI extensions, that help with low precision integer dot products for Vector Search! SimSIMD (inside USearch (inside ClickHouse)) already has those SIMD kernels, but I don’t yet have the hardware to benchmark :(
Something that could help is to use llvm-mca or similar to get an idea of the potential speedup.
Re: Optimizing ClickHouse for Intel's ultra-high core count processors
#10Earlier quoted context omitted.
Sadly, no! On the bright side, they support new AVX2 VNNI extensions, that help with low precision integer dot products for Vector Search! SimSIMD (inside USearch (inside ClickHouse)) already has those SIMD kernels, but I don’t yet have the hardware to benchmark :(
Something that could help is to use llvm-mca or similar to get an idea of the potential speedup.
A basic block simulator like llvm-mca is unlikely to give useful information here, as memory access is going to play a significant part in the overall performance.