luyu_wu

7 ms·

31.
by luyu_wu 2y ago
Here is a commercial product using the older 'Nanhu' architecture by XiangShan. Doesn't seem to be released yet, but interesting nonetheless! https://milkv.io/ja/ruyibook
32.
by luyu_wu 2y ago
I posted a link to this project a few weeks ago as well! It's really interesting to see an academic project like this one. For people interested, there's bi-weekly blogs (some in English) linked here: https://docs.xiang
33.
XiangShan: An Open-Source High-Performance RISC-V Processor (2021) [pdf](raw.githubusercontent.com)
3 points by luyu_wu 2y ago|1 comments
34.
AMD Disables Zen 4's Loop Buffer(chipsandcheese.com)
338 points by luyu_wu 2y ago|141 comments
35.
Pushing AMD's Infinity Fabric to Its Limits(old.chipsandcheese.com)
19 points by luyu_wu 2y ago|1 comments
36.
by luyu_wu 2y ago
Took me a while to figure out that the actual pricing was $200 USD for a mainboard (you have to untick value-added services). Definitely not the best pricing, but also not completely bad considering you get a 64GB microSD and case alongsid
37.
by luyu_wu 2y ago
There've been a lot of studies and meta studies on correlation between cellphone usage and academic performance. IIRC the Pearson correlation coefficient for that has always been hilariously low. Something like -0.1 to -0.2. The argume
38.
Mediatek Dimensity 9400 Benchmarking [Geekerwan] [video](youtube.com)
1 points by luyu_wu 2y ago|0 comments
39.
by luyu_wu 2y ago
Just watched this video, it's really amazing what Wenting does! Have been a big fan of his for a long time. DSLR cameras are one of those places where there just doesn't seem to be much DIY activity, so I'm especially excite
40.
Lion Cove: Intel's P-Core Roars(chipsandcheese.com)
126 points by luyu_wu 2y ago|76 comments
41.
by luyu_wu 2y ago
The section on speculative execution is interesting. "This approach allows each forward pass to generate multiple tokens without compromising performance, thereby significantly reducing memory access consumption, and enabling several o
42.
Hot Chips 2024: Irrational Recap(irrationalanalysis.substack.com)
2 points by luyu_wu 2y ago|0 comments
43.
by luyu_wu 2y ago
Because you won't be gaming at all on a Mac? Surely this is satire.
44.
by luyu_wu 2y ago
Yes, that's correct as long as you don't worry about transient boosts like PL1/2!
45.
by luyu_wu 2y ago
Of course not, my point was just that the 37W wasn't an NPU power figure, but a package power.
46.
by luyu_wu 2y ago
Where are you getting that the NPU draws up to 37W? I believe that is full package power, which Intel states is roughly the combined CPU+GPU+NPU TOPS of 120.
47.
by luyu_wu 2y ago
Thanks for giving some context! I should've done it yesterday, but was a bit busy when the article popped up.
48.
Qualcomm Oryon CPU Core Design(chipsandcheese.com)
103 points by luyu_wu 2y ago|6 comments
49.
by luyu_wu 2y ago
Spent an hour or so playing around with various Python interpreters/compilers and I have to say I was a bit disappointed by Codon. It defonetly has potential, but the compiled binary ended up being slower than CPython? In any case Numb
50.
by luyu_wu 2y ago
APUs have their own small niche in the DIY market! From what I understand, some people build their own computer without a dGPU first, then later purchase a GPU and prefer to be able to change the CPU at that point as well? Hopefully this ra
51.
by luyu_wu 2y ago
Very valid concerns! AMD's current die-to-die interconnects have some pretty abysmal energy/bit. Really hope they can pull off something similar to Intel's EMIB maybe?
52.
AMD's Radeon 890M: Strix Point's Bigger iGPU(chipsandcheese.com)
158 points by luyu_wu 2y ago|69 comments
53.
by luyu_wu 2y ago
I'm one of thosr who have, Microsoft office webapps love to misbehave with Linux useragents! I know Linux friends who spoof useragents for privacy reasons as well!
54.
by luyu_wu 2y ago
An interesting part of Zen 5 architecture was related to lowering the absurd instruction passthrough time/power if I remember correctly! This definetly used to be the case on the older Intel implementations of large SIMD instructions (
55.
by luyu_wu 2y ago
Slightly meta comment, but it's amusing the number went from 400 to 700 to 1000 pieces of debris through the article!
56.
by luyu_wu 2y ago
Maybe try the T key for theater mode!
57.
by luyu_wu 2y ago
This is a hardware deepdive by a couple of uni students and enthusiasts... Some people are interested in things that aren't as shallow as fluctuating performance leads !
58.
by luyu_wu 2y ago
I'm not sure if you're familiar with CPU history, but this is roughly true. Intel's catchup to multicore offerings was trippy and severely lagged behind AMD. I think it's often forgotten that CPU leadership has fluctuate
59.
by luyu_wu 2y ago
I'm not sure emulating CUDA would be legal, you can look at ZLUDA as an example. It was originally funded by AMD, but got cut for what I presume would be legal reasons. ZLUDA does work amazingly well though from my experience!
60.
by luyu_wu 2y ago
The obvious issue with both your points is that NVidia's competitors did do as such. AMD has had workable Linux drivers for many years now and there were numerous alternatives to CUDA pushed.

More ›