Re: AVX-512 for LAME: code to try, if you have the hardware
Tobias Damisch via Lame-dev <[email protected]> Sun, 26 Jul 2026 16:47:26 +0200
| Newsgroups | gmane.comp.audio.mp3.lame |
|---|---|
| Message-ID | <[email protected]> |
SSBqdXN0IG5vdGljZWQgdGhhdCBJIHVzZWQgYSAyNGJpdCBtb25vIHRlc3RmaWxlIGFzIGlucHV0 LCB3aGljaCBpcwpjb21wbGV0ZWx5IG5vdCB3aGF0IHlvdSBhc2tlZCBmb3IgaW4gdGhlIGluc3Ry dWN0aW9ucywgc28gSSByZXJhbiB0aGUKdGVzdHMgd2l0aCBhY3R1YWwgMTZiaXQgU3RlcmVvIDQ0 LjFLSHogbXVzaWMuCgpTcG9pbGVyIGFsZXJ0OiBJdCBzZWVtaW5nbHkgZG9lc24ndCBjaGFuZ2Ug dGhlIG91dGNvbWUgb2YgdGhlIHRlc3RzLgoKT3V0cHV0IGlzIHN0aWxsIGlkZW50aWNhbCwgYW5k IHRoZSB0aW1pbmdzIGFyZToKCm5vYXNtIGF2eDUxMiAtVjIKcmVhbCAgICAwbTQsNjE1cwp1c2Vy ICAgIDBtNCw1NjhzCnN5cyAgICAgMG0wLDA0OXMKCnJlYWwgICAgMG00LDU3OXMKdXNlciAgICAw bTQsNTIxcwpzeXMgICAgIDBtMCwwNjBzCgpyZWFsICAgIDBtNCw1OTlzCnVzZXIgICAgMG00LDU1 MXMKc3lzICAgICAwbTAsMDQ5cwoKCkp1c3QgLVYyOgoKcmVhbCAgICAwbTQsNTgzcwp1c2VyICAg IDBtNCw1NTFzCnN5cyAgICAgMG0wLDAzNHMKCnJlYWwgICAgMG00LDU5MHMKdXNlciAgICAwbTQs NTM4cwpzeXMgICAgIDBtMCwwNTNzCgpyZWFsICAgIDBtNCw1NzlzCnVzZXIgICAgMG00LDUzOHMK c3lzICAgICAwbTAsMDQzcwoKCm5vYXNtIGF2eDUxMiAtYiAzMjA6CgpyZWFsICAgIDBtNiw1MzFz CnVzZXIgICAgMG02LDQ5MnMKc3lzICAgICAwbTAsMDQwcwoKcmVhbCAgICAwbTYsNTE4cwp1c2Vy ICAgIDBtNiw0NzNzCnN5cyAgICAgMG0wLDA0NnMKCnJlYWwgICAgMG02LDUwMnMKdXNlciAgICAw bTYsNDY4cwpzeXMgICAgIDBtMCwwMzZzCgoKSnVzdCAtYiAzMjA6CgpyZWFsICAgIDBtNiw1ODJz CnVzZXIgICAgMG02LDU0NnMKc3lzICAgICAwbTAsMDM4cwoKcmVhbCAgICAwbTYsNTgycwp1c2Vy ICAgIDBtNiw1NTBzCnN5cyAgICAgMG0wLDAzM3MKCnJlYWwgICAgMG02LDU5NnMKdXNlciAgICAw bTYsNTQycwpzeXMgICAgIDBtMCwwNTVzCgotLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0t LS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCgoKU3RhZ2UgMjoKLS0tLS0tLS0tLS0t LS0KCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KICAgIGxhbWUgNC4x OiB0ZXN0L3VuaXQvdGVzdC1zdWl0ZS5sb2cKPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PQoKIyBUT1RBTDogMTMKIyBQQVNTOiAgMTIKIyBTS0lQOiAgMAojIFhGQUlMOiAw CiMgRkFJTDogIDEKIyBYUEFTUzogMAojIEVSUk9SOiAwCgouLiBjb250ZW50czo6IDpkZXB0aDog MgoKRkFJTDogdGVzdF9jYWxjX3NmYl9ub2lzZV92ZWN0b3IKPT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT0KCls9PT09PT09PT09XSB0ZXN0czogUnVubmluZyA1IHRlc3QocykuClsgUlVO ICAgICAgXSB0ZXN0X2xlbmd0aHMKWyAgICAgICBPSyBdIHRlc3RfbGVuZ3RocwpbIFJVTiAgICAg IF0gdGVzdF9pbmRleF9ib3VuZGFyaWVzClsgICAgICAgT0sgXSB0ZXN0X2luZGV4X2JvdW5kYXJp ZXMKWyBSVU4gICAgICBdIHRlc3RfYWxsX3plcm8KWyAgICAgICBPSyBdIHRlc3RfYWxsX3plcm8K WyBSVU4gICAgICBdIHRlc3RfcmVmZXJlbmNlX2Nhbl9kaXNhZ3JlZQpbICAgICAgIE9LIF0gdGVz dF9yZWZlcmVuY2VfY2FuX2Rpc2FncmVlClsgUlVOICAgICAgXSB0ZXN0X21hdGNoZXNfc3NlMl9l eGFjdGx5ClsgIEVSUk9SICAgXSAtLS0gYSA9PSBiClsgICBMSU5FICAgXSAtLS0gdGVzdF9jYWxj X3NmYl9ub2lzZV92ZWN0b3IuYzoyNTM6IGVycm9yOiBGYWlsdXJlIQpbICBGQUlMRUQgIF0gdGVz dF9tYXRjaGVzX3NzZTJfZXhhY3RseQpbPT09PT09PT09PV0gdGVzdHM6IDUgdGVzdChzKSBydW4u ClsgIFBBU1NFRCAgXSA0IHRlc3QocykuClsgIEZBSUxFRCAgXSB0ZXN0czogMSB0ZXN0KHMpLCBs aXN0ZWQgYmVsb3c6ClsgIEZBSUxFRCAgXSB0ZXN0X21hdGNoZXNfc3NlMl9leGFjdGx5CgogIDEg RkFJTEVEIFRFU1QoUykKRkFJTCB0ZXN0X2NhbGNfc2ZiX25vaXNlX3ZlY3RvciAoZXhpdCBzdGF0 dXM6IDEpCgoKQ2hlZXJzLAogICAgICAgICAgVG9iaWFzCgoKCkFtIDI2LjA3LjI2IHVtIDE1OjI0 IHNjaHJpZWIgQWxleGFuZGVyIExlaWRpbmdlcjoKPiBBbSAyMDI2LTA3LTI2IDEyOjQ0LCBzY2hy aWViIFRvYmlhcyBEYW1pc2NoIHZpYSBMYW1lLWRldjoKPj4gSSBkaWQgYXMgaW5zdHJ1Y3RlZCwg bm93IEFWWC01MTIgaXMgaW4gdmVjdG9yIHJvdXRpbmVzLAo+PiB0aGUgMWMgdGVzdCBzaG93ZWQg SURFTlRJQ0FMLgo+Pgo+PiBDUFU6IEFNRCBSeXplbiA5IDc5NTBYM0QgKDMyKSBAIDUuNzU5R0h6 Cj4+PiBnY2MgLS12ZXJzaW9uCj4+IGdjYyAoVWJ1bnR1IDEzLjMuMC02dWJ1bnR1Mn4yNC4wNC4x KSAxMy4zLjAKPj4gT1M6IExpbnV4IE1pbnQgMjIuMyB4ODZfNjQKPj4KPj4+IC4vZnJvbnRlbmQv bGFtZSAtLXZlcmJvc2UgZXhhbXBsZS53YXYKPj4gTEFNRSA0LjEgKGFscGhhIDAsIEp1bCAyNiAy MDI2IDExOjQzOjA4KSA2NGJpdHMgKGh0dHBzOi8vIAo+PiBsYW1lLnNvdXJjZWZvcmdlLmlvKQo+ PiB3YXJuaW5nOiBhbHBoYSB2ZXJzaW9ucyBzaG91bGQgYmUgdXNlZCBmb3IgdGVzdGluZyBvbmx5 Cj4+IENQVSBmZWF0dXJlczogTU1YLCBTU0UsIFNTRTIsIEFWWDIsIEFWWC01MTIKPj4gdmVjdG9y IHJvdXRpbmVzOiBBVlgtNTEyCj4+Cj4+Cj4+IFN0YWdlIDFkIChJIGRpZCBydW4gYSB3YXJtdXAg cm91bmQgYmVmb3JlIHRpbWUgbWVhc3VyZW1lbnRzKToKPiAKPiBHcmVhdCEgVGhhbmtzIQo+IFVu Zm9ydHVuYXRlbHkgdGhlIEFWWC01MTIgY29kZSBpcyBzbG93ZXIgb24geW91ciBoYXJkd2FyZSB0 aGFuIHRoZSBBVlgyIAo+IGNvZGUuIFplbiA1IHNlZW1zIHRvIHNoYXJlIHRoZSBzYW1lIDI1NmJp dCBkYXRhcGF0aCBhcyBaZW4gNCBkb2VzLCBzbyAKPiBBVlgtNTEyIGlzIG5vdCBiZW5lZmljaWFs IGZvciBjdXJyZW50IEFNRDY0IGhhcmR3YXJlLgo+IAo+IEluIGNhc2Ugc29tZW9uZSBoYXMgYW4g SW50ZWwgQ1BVIHdpdGggQVZYLTUxMiwgaXQgd291bGQgYmUgaW50ZXJlc3RpbmcgCj4gdG8gc2Vl IHRoZSBudW1iZXJzIHRoZXJlLCBpdCBzZWVtcyB0byBoYXZlIGEgNTEyYml0IGRhdGFwYXRoLgo+ IAo+PiBTdGFnZSAyOgo+PiAtLS0tLS0tLS0tLS0KPj4gY29uZmlndXJlIGFuZCBtYWtlIGNvbXBs ZXRlZCwgYnV0IG1ha2UgY2hlY2sgcHJvZHVjZWQ6Cj4+Cj4+IFBBU1M6IHRlc3Rfc2FuaXR5Cj4+ IFBBU1M6IHRlc3RfZ2V0X2F1ZGlvX2FpZmYKPj4gUEFTUzogdGVzdF9wYXJzZV9wYXRobGVuCj4+ IFBBU1M6IHRlc3RfbWVyZ2VfYXJndgo+PiBQQVNTOiB0ZXN0X2lkM3RhZwo+PiBQQVNTOiB0ZXN0 X3BhcnNlX2lkMwo+PiBQQVNTOiB0ZXN0X2VuY29kZV9ub25maW5pdGUKPj4gUEFTUzogdGVzdF9l bmNvZGVfbW9ub19pbnRlcmxlYXZlZAo+PiBQQVNTOiB0ZXN0X3RvdGFsZnJhbWVzCj4+IFBBU1M6 IHRlc3Rfc2V0X2dldAo+PiBQQVNTOiB0ZXN0X2Nob29zZV90YWJsZV92ZWN0b3IKPj4gUEFTUzog dGVzdF9xdWFudGl6ZV9saW5lc192ZWN0b3IKPj4gRkFJTDogdGVzdF9jYWxjX3NmYl9ub2lzZV92 ZWN0b3IKPj4gPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PQo+PiBUZXN0c3VpdGUgc3VtbWFyeSBmb3IgbGFt ZSA0LjEKPj4gPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PQo+PiAjIFRPVEFMOiAxMwo+PiAjIFBBU1M6wqAg MTIKPj4gIyBTS0lQOsKgIDAKPj4gIyBYRkFJTDogMAo+PiAjIEZBSUw6wqAgMQo+PiAjIFhQQVNT OiAwCj4+ICMgRVJST1I6IDAKPj4gPT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PQo+PiBTZWUgdGVzdC91bml0 L3Rlc3Qtc3VpdGUubG9nCj4+IFBsZWFzZSByZXBvcnQgdG8gbGFtZS1kZXZAbGlzdHMuc291cmNl Zm9yZ2UubmV0Cj4+ID09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KPj4KPj4gSSBhc3N1bWUgdGhlIHBlcmZv cm1hbmNlIGNvbXBhcmlzb25zIGZvciBzdGFnZSAyIGFyZSBvYnNvbGV0ZSBpZgo+PiAibWFrZSBj aGVjayIgc2hvd3MgZmFpbHMuLi4/Cj4gCj4gUGxlYXNlIHByb3ZpZGUgdGVzdC91bml0L3Rlc3Qt c3VpdGUubG9nIHRvIHVuZGVyc3RhbmQgd2hhdCBleGFjdGx5IAo+IGZhaWxlZC4gVGhpcyBpcyBt b3JlIG91dCBvZiBjdXJpb3NpdHksIGl0IHdpbGwgbm90IGNoYW5nZSB0aGUgb3V0Y29tZS4KPiAK PiBUaGFua3MsCj4gQWxleGFuZGVyLgo+IAoKCgpfX19fX19fX19fX19fX19fX19fX19fX19fX19f X19fX19fX19fX19fX19fX19fXwpMYW1lLWRldiBtYWlsaW5nIGxpc3QKTGFtZS1kZXZAbGlzdHMu c291cmNlZm9yZ2UubmV0Cmh0dHBzOi8vbGlzdHMuc291cmNlZm9yZ2UubmV0L2xpc3RzL2xpc3Rp bmZvL2xhbWUtZGV2Cg==