From a7664aee7d7a224b64b5ccd3c47d9debc0c2b9a5 Mon Sep 17 00:00:00 2001 From: biondizzle Date: Thu, 7 May 2026 14:26:05 +0000 Subject: [PATCH] Add BF16 upcast script and Blackwell DeepGEMM patch --- .../fp8_to_nvfp4_streaming.cpython-310.pyc | Bin 0 -> 14742 bytes 1 file changed, 0 insertions(+), 0 deletions(-) create mode 100644 __pycache__/fp8_to_nvfp4_streaming.cpython-310.pyc diff --git a/__pycache__/fp8_to_nvfp4_streaming.cpython-310.pyc b/__pycache__/fp8_to_nvfp4_streaming.cpython-310.pyc new file mode 100644 index 0000000000000000000000000000000000000000..d505c59d4dab9ebb57d37ecc02895ce61470861a GIT binary patch literal 14742 zcmaKTdu$v@dSAch)tnhVL{Su_m3k$u#F027k(BmyJgrt*iTgT<%jk7yo|MLLs)uZH zUaIaP#c_|l4)>C^6Xn*vyl*!do&&=dz;-S$Y#^~6;0{UtN?^o6Z1-P=4aBzzV)&34 zFm%3?e_vJ4aL8MEh_0@8SAF%>_hvnlF%|rVfBy&W$vcYjCsgSFB~iGB$NLRcQJAu% zFqLT)bx9R(ZAlYveMuK@W62Qj*iuZq<4bY8^-7|eTuQ1m)~KYa=8`G$vC2R-y_6Pt zt&&;Fpgms6RtJ{`MO~sYR2^O#7Wrglq&l)RqADv2OMO#aI^gJ?QD$}yvH_OsZAVdh3A2xV zQ$_9=JM>grIv&-H@75J9c6jM!HnDVq9ij1K-*b%o?ZArSvu!TvFUY<|DkhkTLILxyW+{sH2+mhQSSQODOTOuiuLB=W$Tas z?k`vi58hn7XqD=<4Tt*vGSkmo2AQTD6=f*s(ulS{4TsT#k*z*PM!1!1orp<2if->#9`kRls%1a;nP?qmuwEsW-1T>HO*A;w*R- zFCc(hW6l$YpZ7|2j^6Z9+Gh3b8~C{AvL^Olb)V3YZaVJDst;iK7W%HZH8eH2Q*s3g zU>OqqqBG1dZ{@{d7OX|5xNdFM`8v%~b*go~Wx3dHt&31%|9wZ!u!`uIt;wQQ2C!F~ zHAaBA{RVx%TK9Yb6MxmQDn-8Hpd76aNSn9rI9{__T&_5DoEG-pX!`QNa9$Eo2r)4g zcghZTY9(h5AGNHFVx{R=uSE!Cmm3$Yv(e19z=M57dqxb~T*V}E2AV~pAB~=~1U>>0 zfAYeV^__2DC|sh~v`A!rhVrwNT%_dH_LMavmc6{?J7RX4QtnQ!VXlYDQv!;#$LUw| zPQ|G@Ks<3*K69WwLqlh01WFgV_=?2!sR&(7^S+0hFSpXz@>lCMtK7J3`*pjvf&JKC zm&OztTh_zgSMqs}mwKcn|KMUCq$N)`uD=h%F?GpLw~siDd~e!4{RwG6Jgm#6`Cfy_ zMzJDSmd~#=n$`jVCi>m?9P7sYo7aIx8*a%#8{HYg51jIw!WGC-|+~il5txMKD!YSZ+AL3Zl8g*xUE6B4WW0-rYP%?blK zuo&?B3zVKgB~cCYQ{uV{IW-)J7Q@_-%stu%sixt4oa zAQ~|FupkjC_xh<;vT(MB%Yd$!DIy3cWD93kw}4tU*hdZJ6wOmzEy;eXCA;YRRBxhR z$;RBh1|mVS5RnxWGKI6$-}cNZ2 zEz}AZLbbt3BofK9Yv*~r?w_x1xXdl)tB&J?Or2+7k{+03e&eF#GUs1-3&PZu6{iNG z&9A;oE25=o?@uT^i%jdpBo0Inbnju|Av=cynB1HWHfNEMM{p`MYQ?G(#<}A+d2P>D zY1=g44Llyr=PSP2(E?>h-B#}^-%=NHdT1;IbHaheJM(wv?>v~d@63lWAbkng!v`^D zce6Z$VmRP>;5(9KEYidT(5%W2Qll9W8f+J+N9@sdYa1o%(*327fwUnPHH0MwN=MlN z+}AZP`z{2MKv}~%e4=fue@CVKL#?f~^*|N9bcio6V}!PY)1q`#xl1{k!9d%6l;m24kP-zt`9<9a@t3e%BW3*#|78r-{#c?GtCX{wO z(1Tc1j-i~8Yg>y4ai+7xQw_NUGbqrb|7g^aTN_UYpy)8|btw z4bQ={2Ea1Vum=c^8T)d7?3f_<_SZ@eni|8HkX}1OrDTEMS zuM!h&IF=CNJnN6&|3{X;)o_6(m~wjN@)JA-G3%#d;r#yxX0S|&J6UGQR> zhVc7SN?H>)inSUVAR?~Rtm&I_!D2V90fJEkt#7HJ&fMxkF2RprTo@Az3S(tRo_-js zHLFe~Of_rXSom*QiXg&(4+)Z3xD5Nci( zPc77z@m{Wlro05@W~H+KA{~f;3q~(EWVtar^JUEBSxAn?O*O3zX=(h8scCgg9al|# zNF7p-s-r0L96IHaD@esJ%KXvmQRdIPnZN92lxtB_vYQ#Y_NQ|2_%-l@7{ggOxkS|K zce~@h*UkL@ZsxDL8KXN>x;xY8HGURzMsoz$LiNcWibNN`cln8W1a$F10DN^#+fm!N z@N|9G3|v9&soK{3cn5vf5@^#wyKrR!t&`eOcC_9YBY*~Z6&LXhx;#ef7w^;|?paUh z*lSJ^s7#8f+ zKKWD!9CBu;`Smb{t5;eLV?yr;_2RM@#!HoAwPA+|n&|fRFNaymyz$=sg`3y!yfc3{ zj7ym%OncaDX%!q1HzqW}cYSxIS#Nr|7|-Jqp+PGSW0=%_5}L~dr6PA7FHB(Jw#%OE z$IwHN`)iD(H?= zggCwyXK@f(mGYg0pIl1;cR-MXM8_m{0b*+e$<6?=gKl4pO!?^`wXL?zz?9zuhGljX zDl-slBG>p!eN`6(d!ScKt=-ymFc75Kz*Xf9MSMAnFK2>GkSLQVkuENPR{gC=F@>Te zh3Q>7E;Wl4cf+ZHhhH3 zZwj=R>X6XD@h#u6CM)$#tL84#$IH-JeJMcG`XjkPETT8#^sBUs7pUEmPYeN%#ARwl z8hoVmdJy4X!S<}zUL%>En8v5i@#7O1@TZ;c{GT6v@$Fy#Jh1~-E@rz9`s z2000a+(Hr>8?Lj-XGOM%>&h=t(?v=yP%=#|8Nom<+s?)6Y`Ml?qK4y?(6!*7r(_06 zF3Vr1{548WP;wGU7#FPF!<{77Y13E2A(@ZZ2VJVcy)_W9oN^iL5%?0uML!lM=%V5# zhdMwV8mrK`!?--B{g->VYyRkIXTN~82cszoXuXKbx@B`d=X@?`x4b)!sb>9#-7k7Lu24Tr~s&?W# zM9E{V%^MK%VR?`@AUfcFv&ybvgvW=dB+GAXB0pOe_P=y!J5%~eULe<|7 zd-F=g^?Xucu6~!MrEiQYtx2lEer>r&sje5U04a(U?OuAZZ z6v6~_b(^_790YpULZ7#ZWO*eWo9X^2PpEJMk4HBfTt%@3A6p~*hUNRIrb78Zt>&WM z)`^$d*gevisEHH$*a;?Nm4$oX(!xXoGD^+&!ld&A2W;2ZrF4+w5Ag{oW9wKLe?<92 zsR>!S=7*UmBNqu(i-069epds-w!%SKO$tBhh#qNso>M7{iN%cjPd{mc!@It1|1NED z2uU3GKHmE?)z-@|I*>wl?4IjG@anr@+d({A@Z?6zoI-og&g>O01LnD(Mj%*jXMTtrv$X) z3xdKwiwj$GKGZK-vrPx8J*UbN)`D4{78vn0y(l=8O_L5Ki|X)M0Ys z4i(E(j29a)v9K^!f?~jV33WMxlNb!$Lg?p40|uGTRH);2xzhAjLv;2?ortT3n#CI{sp~i}7OaFfMPy9<(K)i%?gfEnQuR&=KVlWLOV$5~z?GLO?PIsQfqq zl(byD6YB<*B}gHH6VTNw@&M6C9AKP)%!e_3E=4ja=+SDZmtgr9?C%&F!gOS0gdL~l zNVpQHBq}y&XiNZ2!s#(JJ)m)rT+x3YrP7OxeGjc(8p)6b_M|2Os1dbw^o5|pkUiK* zprP|5@OU30!6pG375WG^34Cj-uJW@40u31Vioit;rUYOagW`^My5Znjz*>MryyKt5 zAfpn9v?EG~3<$BnI0Bms@Y47Yzz4Qw|DIA-8R*{!Q2o~u1X$vWL>Z_I%9mPrrsf60 zau@63Ye-rKVe7MWKa%d*4sCk`9raM=1!|P^AKuK7LsF`YyK;AO%a1Y-&AJ zHy{xu;GVESJ|w(_&Bwe##zw)f$Tac)h+j=8t(h0Xxm2lpP9Nv$4Ts81=M4G8smZIO zL%4+=-l({~BzUl7(Tx$eD~y5ES4o4Bcge5&MJVX?3bW<4(n;7!(TYV`uK6eN1zUvi>XfNakX{qZ{Ed1E)#+*(voaDG;JH?qY1l~9j0v?GE^mM zN!T{4mu#C8wd3zovPubUm#WG#! zN3-daXn}skOT@1B96lX-5|6m%zXVmo$Kf+j&~5c=nYJn{f56+e_9bXq259fLt}3Sp1GJ^PF!s{4e^y4#vr-OJ9{3P=Nk9>tfSTbV4rql!0ldQEE&w4rwe$zb5H5~Oi6uK5GoN;|Wn zw6oCCa#B#|9xocPu+2P60f){mWj9qY`X?pGdv-5(@fk^g2e z+~0~h1~A8Wf}y_J{vLo$0GI~ZU?><4Mp*jeIBt5T_-!248zh(ro165vl3UDbR3N65 z$#5oQmBN|_cj*hXu}Fi$#htSVC9V8b>9c~WCHeoZswboca>|fY0F95zy>2v+Bux^b zLFR94X(cnWyIP-x-)BBuNA%O;!tN<82=cY(eiG3n;w%&iBF;jQfa8G_ z2?p%sAkk8l$f?f~T^EZAxv<7O@vJi2oEatz$yUwF!IrXz;G~ zK#HzVdBao>&|xBi^s{(D0G5A*UO5nByuz$FdZ8&2%7i07e)wT#5BS4m7o6dcJSbux z#TwAs%MFDFxuima5IQu-?av83$O7^BHm$EB%EUOr?EIG(=kMIJ7q7o_2mT38q&$oX z6oVk<_>!3ZJk=&(;Hwl%j&Mih4)A|Sm1-G$gbY_=GyI>RPD)E!xfTw<`fSU~<3T_X z&K7~x9w&u3RI6~26g?+2;DKggn(*CW^6Kz^NL*(RhvGWcrm$HA(oejL(Z7G*;>(3aFD&_{!)Ll5Ol z($F;xtW1mk;?MZ6V)_!EJwL*}Yk0glBzrWp4D>U|-9kr;vG@lu=x7NCb}O=OK`ToN z?Mzupu{0=FHZ%p#ya5rb6IU)zMBkas!W$lOyGelr0EpNVQ!FR2MVxe~Z3h0+*H{P6Kid z2@$(3`;JJk>qR;nk_buXfz*9qccVT1CLEBrK}Z~XOq|5}3CLoN#Nma4qJU|b5bIDn zjN30nOCjv7Y3nh5h8QiiLb)L+jru`K>p(@Kza21XS@)XalQ0Pt=Lpz3shUuL^gfv! z^Ft=raX>m85e;$wdz{P;q(GLyMrAfen@(I0<)AHppq)RYA|AmMfw_?a{L&(a(}N6Fu( zVnp7pNwxeGucf=FoM*lnq!@?WHi)cli@Bce0FYknIwl17u~IgHCM zb)9I$$Dt{ZS9#wlh6jm-L{!|yb(f~IRe~^?>_$N%K{`Pmp(7gUA8H!%KTPWUyJ#%o z>HW?jM`xnr@zMZ0nJ!_fBy1GG6(&k~Cp88q6`T#m({35!FA+>;J~hOx)!2Y5a_a5_%Fbf{={_@N4 zgZGzDz#HBxW1WZif5fPhG@9=9hsxSGV63rI1R?)0_KyHaP`;Eu@gdL_BaW~$qoAE^ z6167Qn}hN);2-TA1@zMXOQ^w0DKwHzJtc8}1TL9W=NJ@+jJOWR1B8M?*)Tg6Ts7Gp zeCP4^#5nAn0w|+#orijRuszfs_Fsk~Fc1uF>rhKZ+6SWYuqcnV58@n7>?mJr0|O2) zrKx_T@`8U7iiCi25J7uetHn`gz=AC5+hf7OsP{a2k4C)@%D3n}+SmJ#e<~OY4sC1g zasL(M#atT7t*%2m)67<>*|k-pVt*-MqZ7?!~*&yx6|o{3_-> z0}ML>T*4c;IVns!v+UvrNk12y=$zdFR+3VTHJxOaf|ITqoM5khthG=1=RQ>E9<$GF zYhN2`zv54=<%3hfE9^4+JW+}tgQn)q_kSORz99JGZ2NR@x^sR<`3M{i)}7O97ho_w z6PzIm@sS46c)BwaoDR;=PCwEfFQCUPn_Jh=zaEt@uu)Ooq%!(n3{H1v>W;!{zkqos zgGo6r{{#PAFe%ImSbOJE@Ct5F?lV^1>0dRw!d_!n+3PDRDBv}AoxOp3edFU8yZN!^ zs{X5N{$nG^u{WD0;Q){#yt0iQ76PDN^j<&zG%d${*XhM@F`9s zOhi$D{6B$HwFdVEA#xI+Aol!cy!d~i9;tjjiiZ@EOsjsQUW=Rv-N5WV!$UW85*}MH zc=DxGp(gcQEz5{lE4%&|5SHvyjPF~E_XYnaUmaq85KinO5~l?n9X@6e(vYJCM1has z=M-UN`@Bhkh;d8;t&o3>mN|@A@LesMqIFxbQj-8Iwiqo94T1SN8gkqoi zmM=8lLP4l!`}DHbY0ILpBJM{f1!1|cCWVu|Yf{L~SUtL@gd zViXYv6OMF8?dwAW#C#;Z?8-_Wts23`4(sdRJcF&8HxcJdQKA$jMY1_RyL^t94_u<$}SDd1`% z#UBxukwHnwTdl!8MSG@|*#ns$ljKQ$n#JRh-T!AGJv}npWPgV4LI(CVHHxw9$?jMd zT$nygo4)W4M79q8OBfHMw2tc|g+5{L$fEX-NV=n)Lp4SKqWQdas9JRUEgs##%Fq;c z<7nP6S%q_g6@9qfFgjE>g)M|Ikub03b8=I)25>TEcSYFTS|cRg;)oGGpN|f>H7Y`y z^WiA&lR0Df&pn0(5=l(oV*&WByz$Od{OOruZRoO~WL12xWL>3qmp> zrC2Upds|hkm%AmLTmow*8(5f#3>G;fObe@)jXtY&7RF^@Ivj*z^^su?J6l1Jt`i=J zOkm=_C@3@+J8Y%Qo8&%+mCqdlU|V`&rp!rKY9%GT!Zl0ZBg?ypVn#_XOqUT?_ZGtG z<^LtL5|K)u`&^jB8etKE#UzXP6ySEvU!)=r({R7oXccwRL{2{G&=lRN#S^*|j5@qT zrW{z55Lozo)F&>SqaK>JfTwgJggu>qjv95)&Cnn^5vCEITHGkQ6$Vw7>lvxslgG(kSq8Or;fqL6vry@NhjOIa3*78H%`ns#wENclm*?;1&V?HMzJ`G0|E4bq@EAl7u~cc+ z3gd{Q23ZSDR77M8xF}U7Pv~IKyo8TkUXTQCBoNP+UbzP}?Z2VVCnCdcH#3W2(-h5! z2*_#VXQTXVsBKQ;ndR5SFd2_B4YOKheVW|g^xrnKqBtAI;K{)Zs>lB)eTa;up^lE> zP?`Q`q^skRf9iiwoACWejfLhpIa~bd}uVB|cc_PazVZx~t5tz*+CHe2v zpU6pQ5aDPMJ^3Dr6zYQT7d&KMi$8~94u*^@jB-QjS-edHCOhGnCFd;lrN21p;JAgR z2^|ypvDheV2}cnjhk3MuT0kmhfk^}roJhyiW8XL49~-6bm}KVDlj19)