TalkChess.com

I mostly trust the compiler. Instruction scheduling, unrolling loops, avoiding branches if possible, auto-vectorization (to some extent, Microsoft compiler still lags behind quite a bit) - all this is what you get out of the box from modern compilers. SIMD intrinsics is one of the areas where it's s...

Roland Chastain wrote: ↑

Tue Dec 24, 2019 9:08 am

Now it's a mystery for me how the C code can work.

Well in C, the 0 prefix denotes octal, so 7*8+7 = 63
(does it mean that Eschecs will get an opening book soon?

phhnguyen wrote: ↑

Sun Dec 22, 2019 7:58 am

- The encoding (Little Endian) of integers from book data is not in the right order of typical OSs => you need to convert

You mean big endian

Fixed bug of application crashing when a piece was dropped out of the window. Hmm, I got another access violation when I mistakenly dropped a piece at the border of the chessboard while capturing (in a completely winning position against Moustique :) steps to reproduce: grab a piece and drop it at ...

The idea that GCC doesnt have a say is wrong. The goal of the compiler should be to produce the fastest binary given the information it has. If the information spawns from the "native" switch, then it should be doing more than just asking what instruction sets are supported, and the fact that it do...

dragontamer5788 wrote: ↑

Tue Dec 10, 2019 3:05 pm

GCC still needs the #pragma omp simd to see that this loop is vectorizable for some reason

why? the code you posted is auto-vectorized by gcc regardless of any pragma, clang even unrolls the loop as well as msc

int todo[MAX]; int n=0; for(i=0; i<MAX; i++) { todo[n] = i; n += (Condition(i) != 0); } for(i=0; i<n; i++) Action(todo[i]); I use something similar but with different motivation. I often extract elements/pointers based on a condition into a (thread)local array for further processing. This is benefi...

Comment: For some reason, I couldn't grok your code in the first pass. My brain understands the following more: for(i=0; i<MAX; i++) { todo[n] = i; n += (Condition(i) ? 1 : 0); } Condition(i) ? 1 : 0 will probably compile into a cmov instruction, so you won't have any branches either. In fact, you ...

Hello! Eschecs 5.0.1 is available . Thanks, I missed the release it seems. Playing against Moustique is ideal for me, it lets me build up an advantage and then I blunder the win away to ultimately blunder the draw and lose. I'm a very lousy player :) (Alouette seems a bit weak for me) 100px pieces ...

I've found the problem! In init_table there is this code: for (int t = 0; t < num; t++) { data = (uint8_t *)(((uintptr_t)data + 0x3f) & ~0x3f); ei[t].precomp->data = data; data += size[t][0][2]; if (split) { data = (uint8_t *)(((uintptr_t)data + 0x3f) & ~0x3f); ei[num + t].precomp->data = data; dat...