<table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Issue</th>
<td>
<a href=https://github.com/llvm/llvm-project/issues/119222>119222</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>
Increased spilling in x264 with rva22u64 after bidirectional scheduling/register pressure tracking
</td>
</tr>
<tr>
<th>Labels</th>
<td>
backend:RISC-V
</td>
</tr>
<tr>
<th>Assignees</th>
<td>
</td>
</tr>
<tr>
<th>Reporter</th>
<td>
lukel97
</td>
</tr>
</table>
<pre>
After #115445 we seem to have significantly more spills in this function from x264 when compiling for rva22u64 -O3 (without RVV): https://godbolt.org/z/fcKdrjqxG
Unfortunately this seems to be the cause of a 1.8% regression on 525.x264_r from LLVM 19.1.5 to today: https://lnt.lukelau.me/db_default/v4/nts/profile/13/56/54
I don't think this necessarily warrants a revert given that the patch significantly improved performance for other benchmarks, at least with RVV, but I thought it would be worth flagging anyway.
</pre>
<img width="1px" height="1px" alt="" src="http://email.email.llvm.org/o/eJxkk02PpDYQhn-N-1Ia1BTfBw6diYhG2SjSRpnrqoACvG1sYhf0dn595O5NIs1e4GKXnnrfxxSCni1zq4qfVPHziXZZnG_NfmXTVKfejff2Mgl7UJilaZHnBdwYAvMK4mChgyFO0JMeyIq5w-o8Q9i0MQG0BVl0gGm3g2hnYfJuhW9Y5nBb2MLg1k0bbWeYnAd_EOJe5vDyewYK65uWxe0Cn9_fFTYqu8AisgWVXRR2CrvZjb0zkjg_K-z-VthNw6-j__rXt1_U-aLOlz_t5LzsloTN_UkSwUMk7xlkYRhoDwxuAoI0qRUW4Hn2HEKkdRYKLJLI-8U_2T99ev8N0iZJkyJOETfS_UcyYyV5REh7srLCbuy_jDzRbkRhd-QKOytBYbd5N2kTT6SZwq4o4yd_0r_B6KzCSiK5vT75LQ8cAnlt7nAj78lKAALPB3uBWR8cIyd5LLeRDMuHevS6eXfwCBv7yfmV7MCP9J0s7KFnOywr-WtQ-AokYJiCQKzi2cMr9LvAG8Rq5kVAC9zcbsYY6M15WWAyNM-xU7L3G90TOI1tNjZZQydu0yrLqqKqEE9LO-bnqqGyTLkgzikr-6ac-iofyro-15iedItnzFM8N2mBeMak7suqqXIu6nyoax5VfuaVtEmMOdYowkmHsHObpg0ingz1bMLDbcSehivbUWWXz29_vL68K8RovG_j3Zd-n4PKz0YHCf9PEy2G2zc7eKbA49PruJu23zWOwfwnLj1eSq9H7fkhPBkIw8LjHi8p7DzPOsQzW3Rs9wziabhqO592b9oPfmtZ9j4Z3BqVMse_v5fNu688RJUe20aRvi98tPhPAAAA__-o3kRb">