camel-cdr

15 ms·

181.
by camel-cdr 1y ago
Cool idea. For the sake of being pedantic: > The whole bowl has fewer calories than two slices of garlic bread Really? Idk what the sizes are, but I looked up the weight of one of those micro wave popcorn packages, which usually fill the
182.
by camel-cdr 1y ago
absl::flat_hash_map performs quite well on the slight more "realistic" benchmarks that were added to ankerls map_benchmark after the article came out (GameOfLife and knucleotide). See this slide: https://www.youtube.com
183.
by camel-cdr 1y ago
Reminder: The Cortex-A76 cores in the Pi5 are 9.9 SPECint2006/GHz, only the UR-CP1000 is above that, but that doesn't support RVV. The SpacemiT X100 is slightly below that, but has 256-bit RVV and RVA23 support.
184.
by camel-cdr 1y ago
Here is my version of this concept, I tried to keep it as simple as possible: https://github.com/camel-cdr/cauldron/blob/main/cauldron/str...
185.
by camel-cdr 1y ago
Disallowing /[^\t]\t/ and /^ / is a good start.
186.
by camel-cdr 1y ago
Yeah, I wouldn't use this type of thing for these problems, this was a huge mess. But I think code designed for autovec is resonable as a scalar base implementation for a larger set of problems than people think. I've seen the pro
187.
by camel-cdr 1y ago
So I think I managed to do all of the above with varying degrees of success: https://godbolt.org/z/WY99vxs76 * parse_fract: I got that one down to 23 instructions on icx, although gcc took 81 and clang 91 Since both of
188.
by camel-cdr 1y ago
I think you can coax most of them into autovectorizing via openmp and a bit of restructuring. Can you give 1-3 examples of these type of problems where you think autovectorization is lacking? I want to give it a try.
189.
by camel-cdr 1y ago
Is it a lot different from autovec with #pragma omp simd? I played around with ISPC a bit and it didn't seem much different.
190.
by camel-cdr 1y ago
reposting a comment of mine from a few weeks ago: > All digitized books ever written/encoded compress to a few TB. I tied to estimate how much data this actually is in raw text form: # annas archive stats papers = 10571489
191.
by camel-cdr 1y ago
I wonder how much of the advantages from kagi are due to their yandex backend. For example, I recently tried to search for a text string from ao3 and google, bing, brave, qwant, ... all return no results, while yandex and by extension kagi
192.
by camel-cdr 1y ago
So sadlt no "V" support, but RVA22 with the 1.0 hypervisor extension and UEFI support. Four issue 2GHz OoO core. vs 1.8GHz SiFive P550: https://browser.geekbench.com/v5/cpu/compare/23667112?baseli...
193.
by camel-cdr 1y ago
Usually lexing isn't part of the performance equation compared to all other parts of the compiler, but SIMD can be used to speedup the number parsing.
194.
Apple Trace Cache(realworldtech.com)
2 points by camel-cdr 1y ago|0 comments
195.
by camel-cdr 1y ago
https://github.com/dzaima/intrinsics-viewer is like Intels Guide, but also for Arm, RISC-V and wasm. RISC-V and wasm are hosted here: https://dzaima.github.io/intrinsics-viewer/ You need to downlo
196.
by camel-cdr 1y ago
RTX5090 has 21,760 "CUDA cores", which surprisingly close. Although "CUDA core" is stupid marketing term anyway.
197.
by camel-cdr 1y ago
A RISC-V board with NVIDIA graphics is more likely: https://mp.weixin.qq.com/s/KiV13GqXGMZfZjopY0Xxpg NVIDIA Keynote from the upcoming RISC-V Summit China: "Enabling RISC-V application processors in NVIDIA compute
198.
by camel-cdr 1y ago
Here is an example using google highway: https://godbolt.org/z/Y8vsonTb8 See how the code has only been written once, but multiple versions of the same functions where generated targeting different hardware features (e
199.
by camel-cdr 1y ago
My personal gripe with Rust's std::simd in its current form is that it makes writing portable SIMD hard while making non-portable SIMD easy. [0] > the implementation of sin4f and sin8f are line by line equal, with the exception of t
200.
by camel-cdr 1y ago
> It is far better for software to run with a couple of percent slowdown on old cpus only than not run at all. The difference on geekbench is 2x and not every benchmark has been optimized for RVV yet: https://browser.geekbench
201.
by camel-cdr 1y ago
The link is in the reddit comments. It's Dead Dove though.
202.
by camel-cdr 1y ago
> There's no SLA for Harry Styles porn But what about my good night's sleep? How can I go to bed without reading about my favorite blorbos?
203.
by camel-cdr 1y ago
I wasn't sure what the percentage of porn is, so I counted the number of works for each maturity rating: 4,247,583: Teen And Up Audiences 4,173,082: General Audiences 2,816,083: Explicit 2,271,446: Mature 1,676,06
204.
by camel-cdr 1y ago
None of them are competitive with Zen5 on a per core basis, if you compare the published SPEC results. Veyron V2 has comparable perf per GHz to Zen4/5, but at a lower clock frequency (N4: 3.25, N3: 3.85): https://www.ventana
205.
by camel-cdr 1y ago
> All digitized books ever written/encoded compress to a few TB. I tied to estimate how much data this actually is: # annas archive stats papers = 105714890 books = 52670695 # word count estimates avrg_wor
206.
by camel-cdr 1y ago
@dang The website shows this comment as written 50 minutes ago, but I wrote it over a day ago.
207.
by camel-cdr 1y ago
> Are you saying the model used to simulate many different cpu models is the same, which makes comparing CPUs harder? Or are you saying the model is not accurate? Both, but mostly the former. You can view the scheduling models used for a
208.
by camel-cdr 1y ago
One thing to keep in mind with llvm-mca is that not all processors use their own scheduling model and different scheduling models are more or less accurate. E.g. Cortex-A72 uses the Cortex-A57 model, as does Cortex-A76, even Cortex-A78. The
209.
by camel-cdr 1y ago
Somebody uploaded recordings to youtube: https://www.youtube.com/channel/UCyHfAf6-wDsMKKGVf-7a3cg
210.
Sixth Championship of Branch Prediction (CBP2025)(ericrotenberg.wordpress.ncsu.edu)
2 points by camel-cdr 1y ago|1 comments

More ›