{"id":122838,"date":"2025-04-02T12:55:26","date_gmt":"2025-04-02T10:55:26","guid":{"rendered":"https:\/\/prohoster.info\/?p=122838"},"modified":"2025-04-02T12:55:26","modified_gmt":"2025-04-02T10:55:26","slug":"novaya-statya-radeon-rx-9070-xt-protiv-geforce-rtx-5070-ti","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sk\/blog\/news\/novaya-statya-radeon-rx-9070-xt-protiv-geforce-rtx-5070-ti","title":{"rendered":"Nov\u00fd \u010dl\u00e1nok: Radeon RX 9070 XT vs GeForce RTX 5070 Ti","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Charakteristiky GPU rodiny Blackwell a testy GeForce RTX 5080<\/a><\/noindex> nazna\u010dili, \u017ee v\u0161etky modely 50. s\u00e9rie, s v\u00fdnimkou GeForce RTX 5090, s\u00fa len miernym upgradeom zodpovedaj\u00facich modelov predch\u00e1dzaj\u00facej gener\u00e1cie. Situ\u00e1ciu zhor\u0161il nedostatok nov\u00fdch grafick\u00fdch kariet a n\u00e1sledne astronomick\u00e9 ceny.<\/p>\n

Preva\u017ene v\u00e4\u010d\u0161ina kupuj\u00facich zariaden\u00ed najvy\u0161\u0161ej kateg\u00f3rie, ktor\u00fdch monopolne ovl\u00e1da NVIDIA, vyvol\u00e1va len teoretick\u00fd z\u00e1ujem. Av\u0161ak na jednej \u00farovni ni\u017e\u0161ie technologick\u00fd stagn\u00e1cia \u201ezelenej\u201c grafiky vytvorila situ\u00e1ciu, v ktorej by AMD mohla jedn\u00fdm skokom dobehn\u00fa\u0165 zaost\u00e1vanie v tak\u00fdch problematick\u00fdch aspektoch architekt\u00fary RDNA, ako s\u00fa sledovanie l\u00fa\u010dov a upscale-ing. Napriek tomu predpove\u010f, \u017ee AMD obnov\u00ed paritu s konkurentom (aj ke\u010f bez n\u00e1rokov na vlajkov\u00fa lo\u010f) sme po\u010duli u\u017e d\u00e1vno, a zaka\u017ed\u00fdm znie \u010doraz menej presved\u010divo, preto ozn\u00e1menie grafick\u00fdch kariet Radeon RX 9000 nepril\u00e1kalo ve\u013ek\u00fa pozornos\u0165 na pozad\u00ed b\u00farlivej \u010dinnosti NVIDIA. A predsa AMD vykonalo rozsiahlu pr\u00e1cu na chyb\u00e1ch a ak nie\u017ee odstr\u00e1nilo, tak aspo\u0148 kompenzovalo nedostatky minul\u00fdch GPU.<\/p>\n

Preh\u013ead Radeon RX 9070 XT predstavujeme v neobvyklom form\u00e1te porovnania pre 3DNews, preto\u017ee \u201e\u010derven\u00e1\u201c novinka a jej hlavn\u00fd konkurent \u2014 GeForce RTX 5070 Ti \u2014 sa dostali do na\u0161ich r\u00fak s\u00fa\u010dasne, a jedno zariadenie nemo\u017eno posudzova\u0165 bez druh\u00e9ho, ve\u010f v Rusku sa pred\u00e1vaj\u00fa za pribli\u017ene rovnak\u00e9 peniaze.<\/p>\n\n\n\n
\"Nov\u00fd <\/td>\n <\/td>\n \"Nov\u00fd <\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

<\/center> \t\t\t\t <\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex> Grafick\u00fd procesor Navi 48 a architekt\u00fara RDNA 4<\/strong><\/h2>\n

O tom, \u017ee NVIDIA sa stala monopolistom na trhu vysoko v\u00fdkonn\u00fdch GPU a spr\u00e1va sa tomu zodpovedaj\u00facim sp\u00f4sobom, nehovoril len lenivec. Rarer sa venuje pozornosti tomu faktu, \u017ee v\u0161etky spolo\u010dnosti navrhuj\u00face ve\u013ek\u00e9 procesory s\u00fa n\u00faten\u00e9 vyu\u017e\u00edva\u0165 slu\u017eby jedn\u00e9ho dod\u00e1vate\u013ea \u2014 taiwanskej TSMC. Dopyt po kapacit\u00e1ch TSMC sa ur\u010dite stal jednou z pr\u00ed\u010din, pre\u010do AMD zastavila v\u00fdvoj ve\u013ek\u00fdch spotrebite\u013esk\u00fdch GPU, a vo v\u0161eobecnosti v rade \u201e\u010derven\u00fdch\u201c grafick\u00fdch procesorov novej gener\u00e1cie zatia\u013e existuje len jeden \u010dip \u2014 Navi 48.<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
V\u00fdrobca<\/th>\nAMD<\/th>\n<\/tr>\n
N\u00e1zov<\/td>\nNavi 32<\/td>\nNavi 31<\/td>\nNavi 48<\/td>\n<\/tr>\n
Kde sa pou\u017e\u00edva<\/td>\n\n

Radeon RX 7700 XT;<\/p>\n

Radeon RX 7800 XT<\/p>\n<\/td>\n

\n

Radeon RX 7900 GRE;<\/p>\n

Radeon RX 7900 XT;<\/p>\n

Radeon RX 7900 XTX<\/p>\n<\/td>\n

\n

Radeon RX 9070;<\/p>\n

Radeon RX 9070 XTX<\/p>\n<\/td>\n<\/tr>\n

Architekt\u00fara<\/td>\nRDNA 3<\/td>\nRDNA 3<\/td>\nRDNA 4<\/td>\n<\/tr>\n
Technologick\u00fd proces, nm<\/td>\nTSMC N5\/N6<\/td>\nTSMC N5\/N6<\/td>\nTSMC N4P<\/td>\n<\/tr>\n
Po\u010det tranzistorov, mld<\/td>\n28,1<\/td>\n57,7<\/td>\n53,9<\/td>\n<\/tr>\n
Plocha \u010dipu, mm2<\/td>\n346<\/td>\n522<\/td>\n357<\/td>\n<\/tr>\n
Po\u010det CU\/WGP\/SE<\/td>\n<\/tr>\n
Compute Units (CU)<\/td>\n60<\/td>\n96<\/td>\n64<\/td>\n<\/tr>\n
Workgroup Processors (WGP)<\/td>\n30<\/td>\n48<\/td>\n32<\/td>\n<\/tr>\n
Shader Arrays (SA)<\/td>\nNie<\/td>\nNie<\/td>\nNie<\/td>\n<\/tr>\n
Shader Engines (SE)<\/td>\n4?<\/td>\n6<\/td>\n4<\/td>\n<\/tr>\n
Konfigur\u00e1cia Compute Unit<\/td>\n<\/tr>\n
Vektorov\u00e9 ALU (FP32)<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n<\/tr>\n
Vektorov\u00e9 ALU (FP32\/INT32)<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n<\/tr>\n
Skal\u00e1rn\u00ed ALU<\/td>\n2<\/td>\n2<\/td>\n2<\/td>\n<\/tr>\n
ALU speci\u00e1ln\u00edho ur\u010den\u00ed (SFU)<\/td>\n2 \u00d7 8<\/td>\n2 \u00d7 8<\/td>\n2 \u00d7 8<\/td>\n<\/tr>\n
Bloky trasov\u00e1n\u00ed paprsk\u016f (Ray Accelerators)<\/td>\n1<\/td>\n1<\/td>\n1<\/td>\n<\/tr>\n
Bloky texturov\u00e1n\u00ed (TMU)<\/td>\n4<\/td>\n4<\/td>\n4<\/td>\n<\/tr>\n
Vektorov\u00e9\/skal\u00e1rn\u00ed registry, KByte<\/td>\n384\/10<\/td>\n384\/10<\/td>\n384\/16<\/td>\n<\/tr>\n
Objem cache L0, KByte<\/td>\n32<\/td>\n32<\/td>\n32<\/td>\n<\/tr>\n
Konfigurace Workgroup Processor (WGP)<\/td>\n<\/tr>\n
Objem Shared Memory, KByte<\/td>\n128<\/td>\n128<\/td>\n128<\/td>\n<\/tr>\n
Objem cache instrukc\u00ed, KByte<\/td>\n32<\/td>\n32<\/td>\n32<\/td>\n<\/tr>\n
Objem skal\u00e1rn\u00ed cache, KByte<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n<\/tr>\n
V\u00fdpo\u010detn\u00ed bloky GPU<\/td>\n<\/tr>\n
Vektorov\u00e9 ALU (FP32)<\/td>\n3 840 (7 680)<\/td>\n6 144 (12 288)<\/td>\n4 096 (8 192)<\/td>\n<\/tr>\n
Bloky texturov\u00e1n\u00ed (TMU)<\/td>\n240<\/td>\n384<\/td>\n256<\/td>\n<\/tr>\n
Bloky operac\u00ed rastrov\u00e1n\u00ed (ROP)<\/td>\n96<\/td>\n192<\/td>\n128<\/td>\n<\/tr>\n
Bloky trasov\u00e1n\u00ed paprsk\u016f (Ray Accelerators)<\/td>\n60<\/td>\n96<\/td>\n64<\/td>\n<\/tr>\n
Konfigurace pam\u011bti<\/td>\n<\/tr>\n
Objem cache L1, KByte<\/td>\n2 048<\/td>\n3 072<\/td>\n2 048<\/td>\n<\/tr>\n
Objem cache L2, MByte<\/td>\n4<\/td>\n6<\/td>\n8<\/td>\n<\/tr>\n
Objem Infinity Cache, MByte<\/td>\n64<\/td>\n96<\/td>\n64<\/td>\n<\/tr>\n
\u0160\u00ed\u0159ka sb\u011brnice VRAM, bit<\/td>\n256<\/td>\n384<\/td>\n256<\/td>\n<\/tr>\n
Typ \u010dip\u016f VRAM<\/td>\nGDDR6<\/td>\nGDDR6<\/td>\nGDDR6<\/td>\n<\/tr>\n
<\/td>\n<\/tr>\n
Rozhran\u00ed PCI Express<\/td>\n4.0 x16<\/td>\n4.0 x16<\/td>\n5.0 x16<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Mezi star\u00fdmi \u010dipy se nejv\u00edce podob\u00e1 modelu druh\u00e9 vrstvy, Navi 32, podle hlavn\u00edch charakteristik, ale obsahuje t\u00e9m\u011b\u0159 stejn\u00e9 mno\u017estv\u00ed tranzistor\u016f jako Navi 31 (53,9 a 57,7 miliard odpov\u00eddaj\u00edc\u00edm zp\u016fsobem). Navi 48 je vyrobena technologi\u00ed TSMC N4P, kter\u00e1 pat\u0159\u00ed do stejn\u00e9ho fotolitografick\u00e9ho uzlu 5 nm, jen\u017e byl osvojeno p\u0159ed dv\u011bma lety. Nicm\u00e9n\u011b Navi 48 zab\u00edr\u00e1 pouze 68 % plochy Navi 31 d\u00edky tomu, \u017ee AMD zanechalo \u010dipletov\u00e9ho designu, proto\u017ee MCD-\u010diplet, kter\u00e9 obsahuj\u00ed t\u0159et\u00ed \u00farove\u0148 cache a pam\u011b\u0165ov\u00e9 kontrol\u00e9ry, jsou vyr\u00e1b\u011bny podle hrub\u0161\u00ed technologie N6.<\/p>\n

\"Nov\u00fd <\/p>\n

Architekt\u00fara nov\u00e9ho grafick\u00e9ho procesora RDNA 4 sa \u0161truktur\u00e1lne nel\u00ed\u0161i od predch\u00e1dzaj\u00facich iter\u00e1ci\u00ed. Navi 48 obsahuje 64 Compute Units, alebo 4 096 FP32-kompatibiln\u00fdch shaderov\u00fdch ALU, \u010do je celkom typick\u00e1 konfigur\u00e1cia pre GPU druhej \u00farovne \u2013 ako aj 256-bitov\u00e1 zbernica grafickej pam\u00e4te. Av\u0161ak AMD v hodnoten\u00ed v\u00fdkonu postav\u00ed Navi 48 proti vlajkovej lodi Navi 31, ktor\u00e1 m\u00e1 6 144 re\u00e1lne \u010d\u00edseln\u00fdch shaderov\u00fdch ALU \u0161tandardnej presnosti. Rozdiel by mali dostato\u010dne kompenzova\u0165 viacer\u00e9 vylep\u0161enia logiky a na ich pochopenie odpor\u00fa\u010dame o\u017eivi\u0165 si pam\u00e4\u0165 o architekt\u00fare RDNA 3, ktor\u00fa sme podrobne rozobrali v recenzii. Radeon RX 7900 XTX<\/a><\/noindex>.<\/p>\n

V porovnan\u00ed so \u201ezelenou\u201c architekt\u00farou Blackwell<\/a><\/noindex> RDNA 4 p\u0159edstavuje v\u011bt\u0161\u00ed a hlub\u0161\u00ed vylep\u0161en\u00ed p\u0159edchoz\u00edch n\u00e1vrh\u016f. Nicm\u00e9n\u011b popis v\u0161ech novinek neuv\u011b\u0159iteln\u011b vych\u00e1z\u00ed do kr\u00e1tk\u00e9ho seznamu, proto\u017ee NVIDIA tradi\u010dn\u011b v\u011bnuje zv\u00fd\u0161enou pozornost slo\u017eit\u00fdm softwarov\u00fdm nebo poloprogamov\u00fdm \u0159e\u0161en\u00edm, kter\u00e1 vy\u017eaduj\u00ed rozs\u00e1hl\u00e9 koment\u00e1\u0159e, zat\u00edmco inovace AMD obvykle za\u010d\u00ednaj\u00ed a kon\u010d\u00ed v hardwaru.<\/p>\n

\"Nov\u00fd <\/p>\n

Fundament\u00e1lne zmeny sa udiali v zariaden\u00ed Compute Unit, nedelite\u013enom v\u00fdpo\u010dtovom bloku, ktor\u00fd je anal\u00f3gom SM u NVIDIA a jadier CPU. Tak\u017ee pl\u00e1nova\u010d in\u0161trukci\u00ed teraz podporuje tzv. rozdelen\u00e9 bari\u00e9ry, ktor\u00e9 umo\u017e\u0148uj\u00fa odovzd\u00e1va\u0165 in\u0161trukcie na vykonanie po\u010das ich platnosti. Takisto bola optimalizovan\u00e1 predv\u00fdber in\u0161trukci\u00ed z cache a skalarni ALU, ktor\u00e9 sa pou\u017e\u00edvaj\u00fa na \u00fasporn\u00e9 vykon\u00e1vanie oper\u00e1ci\u00ed podmienkov\u00e9ho skoku a vetvenia, zdedili schopnos\u0165 pracova\u0165 s re\u00e1lnymi \u010d\u00edseln\u00fdmi d\u00e1tami od mobilnej architekt\u00fary RDNA 3.5. Z\u00e1rove\u0148 sa zv\u00fd\u0161il aj objem skalarni registri: z 10 na 16 KB.<\/p>\n

\"Nov\u00fd <\/p>\n

Jedn\u00fdm z najd\u00f4le\u017eitej\u0161\u00edch inov\u00e1ci\u00ed RDNA 4 je zdvojen\u00fd v\u00fdkon vykon\u00e1vania maticov\u00fdch in\u0161trukci\u00ed WMMA, ktor\u00e9 sa pou\u017e\u00edvaj\u00fa v \u00faloh\u00e1ch strojov\u00e9ho u\u010denia a spracovania \u00fadajov pomocou neur\u00f3nov\u00fdch siet\u00ed. Okrem toho, s\u00fabor in\u0161trukci\u00ed RDNA 4 obsahuje pr\u00edkazy na v\u00fdpo\u010dty nad \u0161truktur\u00e1lne sparsn\u00fdmi maticami a 8-bitov\u00fdmi \u00fadajmi s pl\u00e1vaj\u00facou desatinnou \u010diarkou (FP8 a BF8).<\/p>\n

\"Nov\u00fd <\/p>\n

Napriek tomu priepustnos\u0165 1 024 FLOP za cyklus znamen\u00e1, \u017ee \u010dipy Navi dosiahli len \u00farove\u0148 predminulej \u201ezelenej\u201c architekt\u00fary \u2013 Ampere. Okrem toho, ak \u010dipy NVIDIA a Intel pou\u017e\u00edvaj\u00fa na maticov\u00e9 oper\u00e1cie samostatn\u00e9 pole ALU, v architekt\u00fare RDNA 4 t\u00e1to pr\u00e1ca na\u010falej spo\u010d\u00edva na shaderov\u00fdch SIMD.<\/p>\n

Nezab\u00fadajme, \u017ee v re\u00e1lnych aplik\u00e1ci\u00e1ch grafick\u00e9 procesory zalo\u017een\u00e9 na RDNA 3 nikdy nedosahuj\u00fa pln\u00e9ho za\u0165a\u017eenia. Uveden\u00e9 teraflopy pri FP32 v\u00fdpo\u010dtoch v\u0161eobecn\u00e9ho \u00fa\u010delu s\u00fa dosiahnute\u013en\u00e9 iba za podmienky, \u017ee sa pou\u017e\u00edva form\u00e1t VOPD (Vector Operation Dual), ktor\u00fd umo\u017e\u0148uje zoskupi\u0165 dve nez\u00e1visl\u00e9 vektorov\u00e9 in\u0161trukcie. Ten v\u0161ak prin\u00e1\u0161a mno\u017estvo obmedzen\u00ed, tak\u017ee polovica 32-bitov\u00fdch ALU v Compute Unite \u010dasto z\u016fst\u00e1va nevyu\u017eit\u00e1. Pri zmie\u0161anom za\u0165a\u017een\u00ed (napr\u00edklad pri hre s upscalingom) maticov\u00e9 oper\u00e1cie, samozrejme, odober\u00fa \u010das\u0165 v\u00fdpo\u010dtov\u00fdch zdrojov od shaderov, ale tieto straty nebud\u00fa tak\u00e9 ve\u013ek\u00e9, ako sa na prv\u00fd poh\u013ead zd\u00e1.<\/p>\n\n\n\n\n\n\n
<\/th>\nCompute Unit (AMD RDNA 3)<\/th>\nCompute Unit (AMD RDNA 4)<\/th>\nStreaming Multiprocessor (NVIDIA Ada Lovelace)<\/th>\nStreaming Multiprocessor (NVIDIA Blackwell)<\/th>\nXe-core (Intel Xe-HPG)<\/th>\nXe-core (Intel Xe2)<\/th>\n<\/tr>\n
V\u00fdkonn\u00e9 bloky<\/td>\n\n

2 × SIMD32 (FP32\/INT32);<\/p>\n

2 × SIMD32 (FP32);<\/p>\n

2 × SIMD2 (FP64);<\/p>\n

2 × SIMD8 (SFU);<\/p>\n

2 × skal\u00e1rnych ALU<\/p>\n<\/td>\n

\n

2 × SIMD32 (FP32\/INT32);<\/p>\n

2 × SIMD32 (FP32);<\/p>\n

2 × SIMD2 (FP64);<\/p>\n

2 × SIMD8 (SFU);<\/p>\n

2 × skal\u00e1rnych ALU<\/p>\n<\/td>\n

\n

4 × SIMD16 (FP32\/INT32);<\/p>\n

4 × SIMD16 (FP32);<\/p>\n

2 × SISD? (FP64);<\/p>\n

4 × SIMD4 (SFU);<\/p>\n

4 × skal\u00e1rnych ALU;<\/p>\n

4 × tenzorov\u00fdch jadier<\/p>\n<\/td>\n

\n

8 × SIMD16 (FP32\/INT32);<\/p>\n

2 × SISD? (FP64);<\/p>\n

4 × SIMD4 (SFU);<\/p>\n

4 × skal\u00e1rnych ALU;<\/p>\n

4 × tenzorov\u00fdch jadier<\/p>\n<\/td>\n

\n

16 × SIMD8 (FP32);<\/p>\n

16 × SIMD8 (INT32);<\/p>\n

16 × SISD (FP64);<\/p>\n

16 × SIMD2 (SFU);<\/p>\n

16 × XMX<\/p>\n<\/td>\n

\n

8 × SIMD16 (FP32);<\/p>\n

8 × SIMD16 (INT32);<\/p>\n

8 × SIMD2 (FP64);<\/p>\n

8 × SIMD4 (SFU);<\/p>\n

8 × XMX<\/p>\n<\/td>\n<\/tr>\n

Oper\u00e1cie na linke SIMD za takt<\/td>\n\n

128 × FP32;<\/p>\n

64 × INT32;<\/p>\n

256 × FP16;<\/p>\n

4 × FP64;<\/p>\n

16 × transcendent\u00e1lne funkcie<\/p>\n<\/td>\n

\n

128 × FP32;<\/p>\n

64 × INT32;<\/p>\n

256 × FP16;<\/p>\n

4 × FP64;<\/p>\n

16 × transcendent\u00e1lne funkcie<\/p>\n<\/td>\n

\n

128 × FP32;<\/p>\n

64 × INT32;<\/p>\n

128 × FP16;<\/p>\n

2 × FP64;<\/p>\n

16 × transcendent\u00e1lne funkcie<\/p>\n<\/td>\n

\n

128 × FP32;<\/p>\n

128 × INT32;<\/p>\n

128 × FP16;<\/p>\n

2 × FP64;<\/p>\n

16 × transcendent\u00e1lne funkcie<\/p>\n<\/td>\n

\n

128 × FP32;<\/p>\n

128 × INT32;<\/p>\n

256 × FP16;<\/p>\n

16 × FP64;<\/p>\n

32 × transcendent\u00e1lne funkcie<\/p>\n<\/td>\n

\n

128 × FP32;<\/p>\n

128 × INT32;<\/p>\n

256 × FP16;<\/p>\n

16 × FP64;<\/p>\n

32 × transcendent\u00e1lne funkcie<\/p>\n<\/td>\n<\/tr>\n

Matrica oper\u00e1ci\u00ed, FLOP za takt (FP16)<\/td>\n512<\/td>\n1 024<\/td>\n2 048<\/td>\n2 048<\/td>\n2 048<\/td>\n2 048<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Zv\u00fd\u0161enie sn\u00edmok pomocou neur\u00f3nov\u00fdch siet\u00ed \u2013 u\u017e nie je teoretick\u00e1 situ\u00e1cia pre akceler\u00e1tory Radeon. Nov\u00e1, \u0161tvrt\u00e1 verzia FSR kone\u010dne vych\u00e1dza z strojov\u00e9ho u\u010denia, vr\u00e1tane generovania sn\u00edmok, pri\u010dom pou\u017eit\u00e9 modely predstavuj\u00fa hybridn\u00fa verziu konvolu\u010dn\u00fdch siet\u00ed a transformerov. \u017dia\u013e, na rozdiel od predch\u00e1dzaj\u00facich verzi\u00ed upscalera, ktor\u00e9 fungovali prakticky na akomko\u013evek hardv\u00e9ri, FSR 4 podporuj\u00fa iba grafick\u00e9 procesory architekt\u00fary RDNA 4 (v s\u00favislosti s roz\u0161\u00edren\u00edm mno\u017einy in\u0161trukci\u00ed WMMA pre d\u00e1ta FP8). Univerz\u00e1lny re\u017eim s jednoduch\u00fdmi v\u00fdpo\u010dtov\u00fdmi jadrami, ako je u Intelu XeSS, nie je k dispoz\u00edcii.<\/p>\n

\"Nov\u00fd <\/p>\n

\u010eal\u0161\u00edm smerom reformy architekt\u00fary RDNA je sledovanie l\u00fa\u010dov. R\u00fdchlos\u0165 testovania interakci\u00ed l\u00fa\u010da s boxom BVH a s trojuholn\u00edkom sa v prepo\u010dte na jeden Ray Accelerator (a teda aj jeden Compute Unit) zdvojn\u00e1sobila: z 4 a 1 na takt na 8 a 2. Napriek tomu RT-bloky \u010dipov Intel Battlemage pracuj\u00fa r\u00fdchlos\u0165ou 18 a 2, zatia\u013e \u010do grafick\u00e9 procesory Blackwell dok\u00e1\u017eu po\u010das taktu z\u00edska\u0165 \u0161tyri interakcie l\u00fa\u010da s trojuholn\u00edkom a nezverejnen\u00fd po\u010det interakci\u00ed l\u00fa\u010da s boxom BVH. U konkurencie sa testovania dvoch typov (proti boxu BVH a trojuholn\u00edku) vykon\u00e1vaj\u00fa s\u00fa\u010dasne, zatia\u013e \u010do u AMD to st\u00e1le nie je mo\u017en\u00e9. Do istej miery je paralelizmus mo\u017en\u00fd, preto\u017ee ka\u017ed\u00fd Ray Accelerator teraz obsahuje dvoch nez\u00e1visl\u00fdch \u201etestova\u010dov\u201c, ale nie na plnej r\u00fdchlosti.<\/p>\n

\"Nov\u00fd <\/p>\n

Okrem toho, Ray Accelerator z\u00edskal schopnos\u0165 spracovania l\u00fa\u010dov mimo poradia. L\u00fa\u010de, ktor\u00e9 nevy\u017eaduj\u00fa pr\u00edstup k pomal\u00fdm \u00farovniam z\u00e1sobn\u00edka pam\u00e4te, sa spracov\u00e1vaj\u00fa najsk\u00f4r, a v\u00fdsledky testovania sa uchov\u00e1vaj\u00fa tak, aby ich shaderov\u00e1 program dostala v o\u010dak\u00e1vanom porad\u00ed.<\/p>\n

Logika ray tracingu teraz pou\u017e\u00edva hustej\u0161iu \u0161trukt\u00faru BVH8 namiesto BVH4 (osem namiesto \u0161tyroch uzlov na ka\u017ed\u00fa vetvu hierarchie objemov), \u010do zvy\u0161uje r\u00fdchlos\u0165 prechodu a z\u00e1rove\u0148 zni\u017euje tlak na grafick\u00fa pam\u00e4\u0165. Ve\u013ekos\u0165 BVH v pam\u00e4ti sa zmen\u0161ila aj v\u010faka nov\u00e9mu algoritmu kompresie primit\u00edv. Nakoniec, AMD pon\u00faka odviaza\u0165 orient\u00e1ciu boxov BVH od absol\u00fatnych s\u00faradn\u00edc sc\u00e9ny. Rot\u00e1cia boxov v s\u00falade s geometriou objektov umo\u017e\u0148uje zn\u00ed\u017ei\u0165 po\u010det krokov v h\u013eadan\u00ed.<\/p>\n

\"Nov\u00fd <\/p>\n

Je smutn\u00e9, \u017ee napriek v\u0161etk\u00fdm uveden\u00fdm zmen\u00e1m, RDNA 4 st\u00e1le vykon\u00e1va spracovanie BVH na shaderov\u00fdch ALU, a nie prostredn\u00edctvom \u0161pecializovanej logiky. Pod\u013ea klasifik\u00e1cie prostriedkov hardv\u00e9rovej ray tracingu, navrhnutej spolo\u010dnos\u0165ou Imagination Technologies, rie\u0161enia AMD uviazli na druhom leveli, grafick\u00e9 procesory Intel s\u00fa na tre\u0165om, zatia\u013e \u010do hardv\u00e9r NVIDIA u\u017e prech\u00e1dza z levelu 3 na level 4 v\u010faka funkcii SER \u2014 dynamickej skupinovej organiz\u00e1cii pr\u00edkazov\u00fdch tokov s cie\u013eom zv\u00fd\u0161i\u0165 koherenciu pr\u00edstupov do pam\u00e4te a t\u00fdm aj zn\u00ed\u017ei\u0165 latenciu.<\/p>\n

\"Nov\u00fd <\/p>\n

Anal\u00f3g SER v architekt\u00fare RDNA 4 tie\u017e neexistuje, ale objavila sa funkcia Out of Order Memory, ktor\u00e1 in\u00fdm sp\u00f4sobom pom\u00e1ha vyrie\u0161i\u0165 t\u00fa ist\u00fa \u00falohu. Predch\u00e1dzaj\u00face verzie RDNA, pracuj\u00face s viacer\u00fdmi skupinami in\u0161trukci\u00ed (wavefrontami v terminol\u00f3gii AMD), vracaj\u00fa d\u00e1ta z pam\u00e4te striktne v tom istom porad\u00ed, v akom boli vykonan\u00e9 po\u017eiadavky. Ke\u010f s\u00fa d\u00e1ta po\u017eadovan\u00e9 shaderom v r\u00fdchlostnom cache, vr\u00e1tenie m\u00f4\u017ee by\u0165 napriek tomu oneskoren\u00e9, ak predt\u00fdm svoju po\u017eiadavku urobil in\u00fd shader, a jeho d\u00e1ta s\u00fa v relat\u00edvne pomalej VRAM. Pravdou je, \u017ee toto obmedzenie sa zd\u00e1 by\u0165 \u00faplne nezmyseln\u00e9 v kontexte viacvl\u00e1knov\u00e9ho spracovania, ale nepochybne sa objavilo s cie\u013eom zjednodu\u0161i\u0165 organiz\u00e1ciu GPU a do\u010dasne (\u010d\u00edtaj: vo rasterizovan\u00fdch hr\u00e1ch) ost\u00e1valo rozumn\u00fdm kompromisom. RDNA 4 umo\u017e\u0148uje skupin\u00e1m in\u0161trukci\u00ed vraca\u0165 d\u00e1ta nez\u00e1visle od poradia po\u017eiadaviek.<\/p>\n

\"Nov\u00fd <\/p>\n

Poruchovan\u00fd pam\u00e4\u0165ov\u00fd syst\u00e9m \u2014 len jedna z mnoh\u00fdch zmien v architekt\u00fare RDNA 4. Pridelenie registrov Compute Unit'a shaderov\u00fdm programom prebiehalo predt\u00fdm staticky, s oh\u013eadom na maxim\u00e1lnu spotrebu po\u010das \u017eivotnosti shaderu, \u010do znamenalo, \u017ee ur\u010dit\u00e1 \u010das\u0165 s\u00faboru registrov zost\u00e1vala vlastne pr\u00e1zdna. RDNA 4 umo\u017e\u0148uje shaderom dynamicky po\u017eadova\u0165 a uvo\u013e\u0148ova\u0165 registre. T\u00fdm sa objem s\u00faboru registrov vyu\u017e\u00edva efekt\u00edvnej\u0161ie.<\/p>\n

Zauj\u00edmav\u00e9 je, \u017ee Shader Engine \u2014 ve\u013ek\u00e1 \u0161trukt\u00fara GPU, sp\u00e1jaj\u00faca s\u00fabor CU, rasteriz\u00e1tor, blok primit\u00edv a bloky rasteriza\u010dn\u00fdch oper\u00e1ci\u00ed \u2014 v RDNA 4 pri\u0161iel o vlastn\u00fa prv\u00fa \u00farove\u0148 cache. Zato ka\u017ed\u00fd segment cache L2 obsahuje dvakr\u00e1t to\u013eko bankov. T\u00fdm sa zv\u00fd\u0161il nielen jeho objem, ale aj priepustnos\u0165, a to v oboch smeroch \u2014 k Shader Engine a k Infinity Cache (in\u00fdmi slovami, L3). Objem samotnej Infinity Cache zostal rovnak\u00fd ako v Navi 32 \u2014 64 MB.<\/p>\n

Navi 48 disponuje 16 linkami zbernice PCI Express piatej gener\u00e1cie, ale samozrejme v hern\u00fdch \u00faloh\u00e1ch nem\u00f4\u017ee plne vyu\u017ei\u0165 tak r\u00fdchly rozhranie.<\/p>\n

\"Nov\u00fd <\/p>\n

Nakoniec, AMD hl\u00e1si, \u017ee hardwarov\u00fd video kodek v novej gener\u00e1cii GPU zais\u0165uje vy\u0161\u0161iu kvalitu k\u00f3dovania pri popul\u00e1rnych streamovac\u00edch n\u00edzkych bitrate, ale nie viac. To znamen\u00e1, \u017ee podpora farebnej subdemonstrovania YUV 4:2:2, ktor\u00fa maj\u00fa GPU NVIDIA a Intel, sa tak st\u00e1le neobjavila.<\/p>\n

Pok Pokia\u013e ide o prostriedky na zobrazenie obrazu, kontrol\u00e9r displeja Navi 48 je kompatibiln\u00fd s najnov\u0161\u00edmi verziami rozhran\u00ed DisplayPort a HDMI: 2.1a a 2.1b.<\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex>Technick\u00e9 \u0161pecifik\u00e1cie, ceny<\/strong><\/h2>\n

Radeon RX 9070 XT je vybaven\u00fd plne funk\u010dn\u00fdm \u010dipom Navi 48 s 64 Compute Unit'ami, \u010do znamen\u00e1 4 096 re\u00e1lnych ALU \u0161tandardnej presnosti. N\u00e1js\u0165 form\u00e1lneho predchodcu novinky medzi \u201e\u010derven\u00fdmi\u201c akceler\u00e1tormi minul\u00e9ho gener\u00e1cie nie je jednoduch\u00e9, preto\u017ee sa zmenil aj modelov\u00fd rad, ako aj princ\u00edp \u010d\u00edslovania GPU (\u010dipy Navi nikdy predt\u00fdm nedosiahli \u010d\u00edslo osem). Pod\u013ea vzorca v\u00fdpo\u010dtov\u00fdch blokov a charakterist\u00edk pam\u00e4\u0165ov\u00e9ho syst\u00e9mu je Radeon RX 9070 XT podobn\u00fd Radeon RX 7800 XT, ale AMD ho pozicionuje ako n\u00e1hradu za Radeon RX 7900 XTX.<\/p>\n

Hern\u00e9 hodiny a Boost Clock (cie\u013eov\u00e1 frekvencia v hr\u00e1ch a limit frekvencie) grafickej karty v\u00fdrazne vzr\u00e1stli v porovnan\u00ed s parametrami vlajkovej lode z minul\u00e9ho roka a dosahuj\u00fa 2,4 a 2,97 GHz. Teoretick\u00fd v\u00fdkon Radeon RX 7900 XTX na z\u00e1klade maxim\u00e1lnej frekvencie je o 26 % vy\u0161\u0161\u00ed, ale AMD m\u00e1 v \u00famysle kompenzova\u0165 tento rozdiel prostredn\u00edctvom vylep\u0161en\u00ed architekt\u00fary RDNA 4.<\/p>\n

Radeon RX 9070 XT dostal 16 GB pam\u00e4te GDDR6 s priepustnos\u0165ou 20 Gbit\/s na kontakte 256-bitovej zbernice. Pod\u013ea spr\u00e1v AMD zva\u017eovalo zv\u00fd\u0161i\u0165 objem VRAM na 32 GB, ale nakoniec t\u00fato konfigur\u00e1ciu pova\u017eovalo za nadbyto\u010dn\u00fa pre spotrebite\u013esk\u00e9 rie\u0161enia. Je to \u0161koda, preto\u017ee po\u017eiadavky hier neust\u00e1le rast\u00fa a pr\u00e1ve objem grafickej pam\u00e4te bol d\u00f4le\u017eitou konkuren\u010dnou v\u00fdhodou mnoh\u00fdch predch\u00e1dzaj\u00facich zariaden\u00ed AMD.<\/p>\n

Referen\u010dn\u00e1 spotreba energie Radeon RX 9070 XT je 304 W a odpor\u00fa\u010dan\u00e1 cena \u2014 599 USD (o 100 USD viac ako pri Radeon RX 7800 XT).<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
V\u00fdrobca<\/th>\nAMD<\/th>\nNVIDIA<\/th>\n<\/tr>\n
Model<\/td>\nRadeon RX 7900 XT<\/td>\nRadeon RX 7900 XTX<\/td>\nRadeon RX 9070<\/td>\nRadeon RX 9070 XT<\/td>\nGeForce RTX 4070 Ti SUPER<\/td>\nGeForce RTX 4080 SUPER<\/td>\nGeForce RTX 5070 Ti<\/td>\nGeForce RTX 5080<\/td>\n<\/tr>\n
Grafick\u00fd procesor<\/td>\n<\/tr>\n
N\u00e1zov<\/td>\nNavi 31 XT<\/td>\nNavi 31 XTX<\/td>\nNavi 48 XT<\/td>\nNavi 48 XTX<\/td>\nAD103<\/td>\nAD103<\/td>\nGB203<\/td>\nGB203<\/td>\n<\/tr>\n
Architekt\u00fara<\/td>\nRDNA 3<\/td>\nRDNA 3<\/td>\nRDNA 4<\/td>\nRDNA 4<\/td>\nAda Lovelace<\/td>\nAda Lovelace<\/td>\nBlackwell<\/td>\nBlackwell<\/td>\n<\/tr>\n
Technologick\u00fd proces<\/td>\nTSMC N5\/N6<\/td>\nTSMC N5\/N6<\/td>\nTSMC N4C<\/td>\nTSMC N4C<\/td>\nTSMC 4N<\/td>\nTSMC 4N<\/td>\nTSMC 4NP<\/td>\nTSMC 4NP<\/td>\n<\/tr>\n
Po\u010det tranzistorov, mld<\/td>\n57,7<\/td>\n57,7<\/td>\n53,9<\/td>\n53,9<\/td>\n45,9<\/td>\n45,9<\/td>\n45,6<\/td>\n45,6<\/td>\n<\/tr>\n
Frekvencia (Base Clock\/Game Clock\/Boost Clock), MHz<\/td>\n1 500 \/2 025\/2 394<\/td>\n1 855 \/2 269\/2 499<\/td>\n1 330 \/2 070\/2 520<\/td>\n1 660 \/2 400\/2 970<\/td>\n2 310\/2 610<\/td>\n2 205\/2 550<\/td>\n2 295\/2 452<\/td>\n2 295\/2 617<\/td>\n<\/tr>\n
Shader ALU (FP32)<\/td>\n5 376 (10 752)<\/td>\n6 144 (12 228)<\/td>\n3 584 (7 168)<\/td>\n4 096 (8 192)<\/td>\n8 448<\/td>\n10 240<\/td>\n8 960<\/td>\n10 752<\/td>\n<\/tr>\n
Bloky texturov\u00e1n\u00ed (TMU)<\/td>\n336<\/td>\n384<\/td>\n224<\/td>\n256<\/td>\n264<\/td>\n320<\/td>\n280<\/td>\n336<\/td>\n<\/tr>\n
Bloky operac\u00ed rastrov\u00e1n\u00ed (ROP)<\/td>\n192<\/td>\n192<\/td>\n128<\/td>\n128<\/td>\n112<\/td>\n112<\/td>\n96<\/td>\n112<\/td>\n<\/tr>\n
Tensorov\u00e9 jadr\u00e1<\/td>\nNie<\/td>\nNie<\/td>\nNie<\/td>\nNie<\/td>\n264<\/td>\n320<\/td>\n280<\/td>\n336<\/td>\n<\/tr>\n
RT-jadr\u00e1<\/td>\n84<\/td>\n96<\/td>\n112<\/td>\n128<\/td>\n66<\/td>\n80<\/td>\n70<\/td>\n84<\/td>\n<\/tr>\n
Objem cache poslednej \u00farovne, MB<\/td>\n80<\/td>\n96<\/td>\n64<\/td>\n64<\/td>\n48<\/td>\n64<\/td>\n64<\/td>\n64<\/td>\n<\/tr>\n
Opera\u010dn\u00e1 pam\u00e4\u0165<\/td>\n<\/tr>\n
\u0160\u00edrka zbernice, bit<\/td>\n320<\/td>\n384<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n<\/tr>\n
Typ \u010dipov<\/td>\nGDDR6 SGRAM<\/td>\nGDDR6 SGRAM<\/td>\nGDDR6 SGRAM<\/td>\nGDDR6 SGRAM<\/td>\nGDDR6X SGRAM<\/td>\nGDDR6X SGRAM<\/td>\nGDDR7 SGRAM<\/td>\nGDDR7 SGRAM<\/td>\n<\/tr>\n
Priepustnos\u0165 na kontakte, Gbit\/s<\/td>\n20<\/td>\n20<\/td>\n20<\/td>\n20<\/td>\n21<\/td>\n23<\/td>\n28<\/td>\n30<\/td>\n<\/tr>\n
Celkov\u00e1 priepustnos\u0165, GB\/s<\/td>\n800<\/td>\n960<\/td>\n640<\/td>\n640<\/td>\n672<\/td>\n736<\/td>\n896<\/td>\n960<\/td>\n<\/tr>\n
Objem, GB<\/td>\n20<\/td>\n24<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n<\/tr>\n
V\u00fdkon<\/td>\n<\/tr>\n
\u0160pi\u010dkov\u00fd v\u00fdkon FP32, TFLOPS<\/td>\n25,7 (51,5)<\/td>\n30,7 (61,1)<\/td>\n18,1 (36,1)<\/td>\n24,3 (48,7)<\/td>\n44,1<\/td>\n52,2<\/td>\n43,9<\/td>\n56,3<\/td>\n<\/tr>\n
V\u00fdkon FP64\/FP32<\/td>\n1\/32 (1\/64)<\/td>\n1\/32 (1\/64)<\/td>\n1\/32 (1\/64)<\/td>\n1\/32 (1\/64)<\/td>\n1\/64<\/td>\n1\/64<\/td>\n1\/64<\/td>\n1\/64<\/td>\n<\/tr>\n
V\u00fdkon FP16\/FP32<\/td>\n2\/1<\/td>\n2\/1<\/td>\n2\/1<\/td>\n2\/1<\/td>\n1\/1<\/td>\n1\/1<\/td>\n1\/1<\/td>\n1\/1<\/td>\n<\/tr>\n
In\u00e9<\/td>\n<\/tr>\n
Zbernica PCI Express<\/td>\nPCI Express 4.0 x16<\/td>\nPCI Express 4.0 x16<\/td>\nPCI Express 5.0 x16<\/td>\nPCI Express 5.0 x16<\/td>\nPCI Express 4.0 x16<\/td>\nPCI Express 4.0 x16<\/td>\n5.0 x16<\/td>\n5.0 x16<\/td>\n<\/tr>\n
Rozhrania v\u00fdstupu obrazu<\/td>\nDisplayPort 2.1, HDMI 2.1a<\/td>\nDisplayPort 2.1, HDMI 2.1a<\/td>\nDisplayPort 2.1a, HDMI 2.1b<\/td>\nDisplayPort 2.1a, HDMI 2.1b<\/td>\nDisplayPort 1.4a, HDMI 2.1<\/td>\nDisplayPort 1.4a, HDMI 2.1<\/td>\nDisplayPort 2.1b, HDMI 2.1b<\/td>\nDisplayPort 2.1b, HDMI 2.1b<\/td>\n<\/tr>\n
TDP\/TBP, W<\/td>\n315<\/td>\n335<\/td>\n220<\/td>\n304<\/td>\n285<\/td>\n320<\/td>\n300<\/td>\n360<\/td>\n<\/tr>\n
Maloobchodn\u00e1 cena (USA), $<\/td>\n899 (odpor. k d\u00e1tumu uvedenia)<\/td>\n999 (odpor. k d\u00e1tumu uvedenia)<\/td>\n549 (odpor. k d\u00e1tumu uvedenia)<\/td>\n599 (odpor. k d\u00e1tumu uvedenia)<\/td>\n799 (odpor. k d\u00e1tumu uvedenia)<\/td>\n999 (odpor. k d\u00e1tumu uvedenia)<\/td>\n749 (odpor. k d\u00e1tumu uvedenia)<\/td>\n999 (odpor. k d\u00e1tumu uvedenia)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Na druhej strane, GeForce RTX 5070 Ti je zalo\u017een\u00e1 na tom istom grafickom procesore GB203, ktor\u00fd sa pou\u017e\u00edva v GeForce RTX 5080, ale ni\u017e\u0161\u00ed model pri\u0161iel o 14 streamov\u00fdch multiprocesorov, teda 1 792 FP32-kompatibiln\u00fdch shaderov\u00fdch ALU, a hern\u00e1 taktov\u00e1 frekvencia bola zn\u00ed\u017een\u00e1. V d\u00f4sledku toho je medzi GeForce RTX 5080 a RTX 5070 Ti rozdiel 27 % v teoretickom v\u00fdkone, pri\u010dom samotn\u00e9 GeForce RTX 5070 Ti v tomto parametri nevybo\u010duje od RTX 4070 Ti SUPER. Tak ako v pr\u00edpade Radeon RX 9070 XT, nedostatok hrubej v\u00fdpo\u010dtovej sily m\u00e1 kompenzova\u0165 nov\u00e1 architekt\u00fara GPU, predov\u0161etk\u00fdm funkcia generovania viacer\u00fdch sn\u00edmok pomocou DLSS (a\u017e do troch generovan\u00fdch sn\u00edmok medzi dvoma \u201eskuto\u010dn\u00fdmi\u201c). <\/p>\n

GeForce RTX 5070 Ti je navr\u017een pro spot\u0159ebu 300 W a s doporu\u010denou cenou $749 je o $50 levn\u011bj\u0161\u00ed ne\u017e GeForce RTX 4070 Ti SUPER.<\/p>\n

Rozdiel v $150 odpor\u00fa\u010danej hodnoty medzi Radeon RX 9070 XT a GeForce RTX 5070 Ti nie je vidite\u013en\u00fd pri aktu\u00e1lnych slovensk\u00fdch cen\u00e1ch: najdostupnej\u0161ie partnersk\u00e9 modely sme na\u0161li v DNS za 99 999 a 104 999 \u20ac. Pr\u00e1ve ich budeme testova\u0165 v benchmarkoch. Radeon RX 9070 XT predstavuje grafick\u00e1 karta GIGABYTE GAMING OC s tov\u00e1rensk\u00fdm pre\u0165a\u017een\u00edm GPU a\u017e do 2,52\/3,06 GHz, zatia\u013e \u010do GeForce RTX 5070 Ti \u2013 akceler\u00e1tor Palit GamingPro s referen\u010dn\u00fdmi \u0161pecifik\u00e1ciami.<\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex>GIGABYTE Radeon RX 9070 XT GAMING OC: konstrukce<\/strong><\/h2>\n

Radeon RX 9070 XT vo variante GIGABYTE m\u00e1 pomerne skromn\u00e9 rozmery pre grafick\u00fa kartu s energetickou spotrebou minim\u00e1lne (a, ako uk\u00e1\u017eu testy, ove\u013ea viac) 300 W: 288 mm na d\u013a\u017eku a tri sloty na \u0161\u00edrku. Jedin\u00fdm prvkom LED osvetlenia je kr\u00e1tky p\u00e1sik na boku. N\u00e1lepka s logom v\u00fdrobcu je pohybliv\u00e1: ak ju presunieme na LED p\u00e1s, logo za\u010dne svieti\u0165.<\/p>\n

\"Nov\u00fd <\/p>\n

Chladic\u00ed syst\u00e9m zaji\u0161\u0165uj\u00ed t\u0159i ventil\u00e1tory o pr\u016fm\u011bru 87 mm. P\u0159es malou sekci chladi\u010de vzduch proch\u00e1z\u00ed skrze n\u011bj d\u00edky v\u00fd\u0159ez\u016fm v backplate.<\/p>\n

\"Nov\u00fd <\/p>\n

Grafick\u00fd procesor a \u010dipy grafickej pam\u00e4te odovzd\u00e1vaj\u00fa teplo tepeln\u00e9mu potrubiu, av\u0161ak prv\u00fd \u2013 priamo, zatia\u013e \u010do posledn\u00e9 \u2013 cez U-tvarovan\u00fd kovov\u00fd podlo\u017eku. Na li\u0161t\u00e1ch chladi\u010da s\u00fa pripevnen\u00e9 samostatn\u00e9 dosky, ktor\u00e9 sl\u00fa\u017eia ako tepeln\u00e9 odn\u00edma\u010de komponentov VRM. Tepeln\u00fdch r\u00farok je celkom sedem, ich priemer je 6 mm. V\u0161imnime si, \u017ee GIGABYTE uprednostnila kvapaln\u00e9 tepeln\u00e9 podlo\u017eky pred be\u017en\u00fdmi.<\/p>\n

\"Nov\u00fd <\/p>\n

Backplate se pod\u00edl\u00ed na odv\u00e1d\u011bn\u00ed tepla z ti\u0161t\u011bn\u00e9ho obvodu pomoc\u00ed n\u011bkolika teplovodiv\u00fdch podlo\u017eek (zde u\u017e standardn\u00edch, nikoli tekut\u00fdch).<\/p>\n

\"Nov\u00fd <\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex>GIGABYTE Radeon RX 9070 XT GAMING OC: ti\u0161t\u011bn\u00fd obvod<\/strong><\/h2>\n

Niektor\u00e9 verzie Radeon RX 9070 XT z\u00edskavaj\u00fa energiu cez konektor 12V-2\u00d76, no GIGABYTE uprednostnila tri univerz\u00e1lne osemkontaktov\u00e9 konektory. S oh\u013eadom na vysok\u00fa spotrebu energie grafickej karty a nepretr\u017eit\u00e9 s\u0165a\u017enosti na nov\u00fd konektor, je to spo\u013eahlivej\u0161ie, aj ke\u010f nie tak pohodln\u00e9 a estetick\u00e9 rie\u0161enie. Z 17 f\u00e1zy nap\u00e1jacieho syst\u00e9mu s\u00fa 14, riaden\u00e9 PWM ovl\u00e1da\u010dom Monolithic Power Systems MP2868A, vyhraden\u00e9 pre grafick\u00fd procesor, \u010fal\u0161ie tri \u2013 pre grafick\u00fa pam\u00e4\u0165. Oba VRM s\u00fa vybaven\u00e9 nap\u00e1jac\u00edmi stup\u0148ami MP87993 s predpokladan\u00fdm menovit\u00fdm pr\u00fadom 90A.<\/p>\n

VRAM je osazena \u010dipy SK hynix, v jejich\u017e ozna\u010den\u00ed H56G42AS8DX-014 je zak\u00f3dov\u00e1na propustnost 20 Gbit\/s.<\/p>\n

Okrem k\u00f3pie BIOS-u, ktor\u00e1 je akt\u00edvna z v\u00fdroby, m\u00e1 grafick\u00e1 karta z\u00e1lo\u017en\u00fd \"tich\u00fd\" firmware.<\/p>\n

\"Nov\u00fd \"Nov\u00fd <\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex>Palit GeForce RTX 5070 Ti GamingPro: konstrukce<\/strong><\/h2>\n

Palit GeForce RTX 5070 Ti GamingPro m\u00e1 d\u013a\u017eku 331,9 mm, ale iba tri roz\u0161iruj\u00face sloty. Pod logom GamingPro a ved\u013ea segmentu predn\u00e9ho panela sa nach\u00e1dzaj\u00fa RGB LED di\u00f3dy. Pr\u00e1cu LED je mo\u017en\u00e9 synchronizova\u0165 s z\u00e1kladnou doskou cez \u0161tandardn\u00fd ARGB konektor, ktor\u00fd sa nach\u00e1dza pri vstupe nap\u00e1jania 12V-2\u00d76.<\/p>\n

\"Nov\u00fd <\/p>\n

Bo\u010dn\u00ed plochy sk\u0159\u00edn\u011b tvo\u0159\u00ed lit\u00e9 hlin\u00edkov\u00e9 r\u00e1my s ventila\u010dn\u00edmi otvory na dlouh\u00fdch stran\u00e1ch. V ochrann\u00e9m plechu na zadn\u00ed stran\u011b PCB je tradi\u010dn\u00ed m\u0159\u00ed\u017eka pro pr\u016fchod vzduchu chladi\u010dem. \u010celn\u00ed lopatky ventil\u00e1tor\u016f maj\u00ed pr\u016fm\u011br 94 mm.<\/p>\n

\"Nov\u00fd <\/p>\n

Na spodnej \u010dasti chladi\u010da sa nach\u00e1dza v\u00fdparn\u00e1 komora dostato\u010dnej plochy, aby pokryla GPU \u010dip a okolo neho umiestnen\u00e9 mikrosch\u00e9my pam\u00e4te, pri\u010dom pre komponenty VRM s\u00fa zabezpe\u010den\u00e9 samostatn\u00e9 plechov\u00e9 chladi\u010de. Lamely chladi\u010da s\u00fa navle\u010den\u00e9 na \u00f4smych tepelne trubic\u00edch s priemerom 6 mm.<\/p>\n

\"Nov\u00fd <\/p>\n

Kovov\u00fd backplate pln\u00ed len ochrann\u00fa a dekorat\u00edvnu funkciu, preto\u017ee medzi n\u00edm a PCB nie je \u017eiadna tepeln\u00e1 podlo\u017eka.<\/p>\n

\"Nov\u00fd <\/p>\n

Grafick\u00e1 karta je dod\u00e1van\u00e1 s adapt\u00e9rom z dvoch osemkol\u00edkov\u00fdch nap\u00e1jac\u00edch konektorov na 12V-2\u00d76 a malou tkanou podlo\u017ekou pod my\u0161.<\/p>\n

\"Nov\u00fd <\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex>Palit GeForce RTX 5070 Ti GamingPro: plo\u0161n\u00fd spoj<\/strong><\/h2>\n

Na rozdiel od grafickej karty Palit GameRock<\/a><\/noindex>, ktor\u00fa sme presk\u00famali v ned\u00e1vnej recenzii GeForce RTX 5080, je \u010fal\u0161\u00edm modelom s\u00e9rie, ktor\u00fd je postaven\u00fd na priestrannej PCB, ale jej nap\u00e1jac\u00ed syst\u00e9m je jednoduch\u0161\u00ed. Grafick\u00e9mu procesoru patr\u00ed 15-f\u00e1zov\u00fd VRM riaden\u00fd PWM regul\u00e1torom Monolithic Power Systems MP29816. Nap\u00e1janie pam\u00e4\u0165ov\u00fdch \u010dipov GDDR7 je trojf\u00e1zov\u00e9 s kontrol\u00e9rom MP2988. Obe regul\u00e1tory s\u00fa vybaven\u00e9 tranzistorov\u00fdmi zostavami MP87993 s predpokladan\u00fdm menovit\u00fdm pr\u00fadom 90 A.<\/p>\n

Ozna\u010denie \u010dipov pam\u00e4te Samsung K4VAF325ZC-SC28 nazna\u010duje priepustnos\u0165 28 Gbit\/s.<\/p>\n

Rovnako ako Radeon RX 9070 XT od GIGABYTE, aj akceler\u00e1tor Palit m\u00e1 dve varianty BIOSu \u2013 \u201ev\u00fdkonn\u00fd\u201c a \u201etich\u00fd\u201c.<\/p>\n

\"Nov\u00fd \"Nov\u00fd <\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex>Testovacia zostava, met\u00f3da testovania<\/strong><\/h2>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Testovac\u00ed st\u00e1nok<\/th>\n<\/tr>\n
CPU<\/td>\nAMD Ryzen 9 7950X3D (PBO +150 MHz, CU -20)<\/td>\n<\/tr>\n
Z\u00e1kladn\u00e1 doska<\/td>\nASUS ROG Crosshair X670E Hero<\/td>\n<\/tr>\n
Opera\u010dn\u00e1 pam\u00e4\u0165<\/td>\nG.Skill Trident Z5 Neo RGB (F5-6000J3040G32GX2-TZ5NR), 2 \u00d7 32 GB (6200 MT\/s, CL30)<\/td>\n<\/tr>\n
Pevn\u00fd disk<\/td>\nSolidigm P44 Pro, 2 TB<\/td>\n<\/tr>\n
Nap\u00e1jac\u00ed zdroj<\/td>\nCorsair AX1600i, 1600 W<\/td>\n<\/tr>\n
Chladi\u010d CPU<\/td>\nZak\u00e1zkov\u00e1 AIO (EK-Quantum Velocity\u00b2 DDC 4.2 PWM D-RGB + EK-Quantum Surface X280M)<\/td>\n<\/tr>\n
Kryt<\/td>\nOtvoren\u00fd stojan<\/td>\n<\/tr>\n
Opera\u010dn\u00fd syst\u00e9m<\/td>\nWindows 11 Pro<\/td>\n<\/tr>\n
Softv\u00e9r pre GPU AMD<\/td>\n<\/tr>\n
V\u0161etky grafick\u00e9 karty<\/td>\nAMD Software Adrenalin Edition 25.2.1\/25.3.1<\/td>\n<\/tr>\n
Softv\u00e9r pre GPU NVIDIA<\/td>\n<\/tr>\n
V\u0161etky grafick\u00e9 karty<\/td>\nNVIDIA GeForce Game Ready Driver 532.60\/532.70<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Hry bez sledovania l\u00fa\u010dov<\/th>\n<\/tr>\n
Hra<\/td>\nAPI<\/td>\nMet\u00f3da testovania<\/td>\nNastavenia grafiky<\/td>\n<\/tr>\n
Alan Wake 2<\/td>\nDirectX 12<\/td>\nOCAT, lokalita Bright Falls<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
Black Myth: Wukong<\/td>\nDirectX 12<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
Cyberpunk 2077<\/td>\nDirectX 12<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
F1 24<\/td>\nDirectX 12<\/td>\nIntegrovan\u00fd benchmark, tra\u0165 Monaco (d\u00e1\u017e\u010f)<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
Hogwarts Legacy<\/td>\nDirectX 12<\/td>\nOCAT, jazda na voz\u00edku v Path to Hogwarts<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
Horizon Zero Dawn Remastered<\/td>\nDirectX 12<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
Metro Exodus.<\/td>\nDirectX 12<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky; Shading Rate: 100%<\/td>\n<\/tr>\n
Red Dead Redemption 2<\/td>\nVulkan<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
Returnal<\/td>\nDirectX 12<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n
Total War: WARHAMMER III<\/td>\nDirectX 11<\/td>\nIntegrovan\u00fd benchmark (Mirrors of Madness Benchmark)<\/td>\nMax. kvalita grafiky<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Hry so sledovan\u00edm l\u00fa\u010dov<\/th>\n<\/tr>\n
Hra<\/td>\nAPI<\/td>\nMet\u00f3da testovania<\/td>\nNastavenia grafiky<\/td>\n\u0160k\u00e1lovanie sn\u00edmok<\/td>\n<\/tr>\n
AMD<\/td>\nIntel<\/td>\nNVIDIA<\/td>\n<\/tr>\n
Alan Wake 2<\/td>\nDirectX 12<\/td>\nOCAT, lokalita Bright Falls<\/td>\nMax. kvalita grafiky, vysok\u00e1 kvalita sledovania l\u00fa\u010dov<\/td>\nFSR Balanced<\/td>\nFSR Balanced<\/td>\nDLSS Balanced + Ray Reconstruction (+ Frame Generation)<\/td>\n<\/tr>\n
Black Myth: Wukong<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky a sledovania l\u00fa\u010dov (Path Tracing)<\/td>\nFSR Balanced (+ Frame Generation)<\/td>\nXeSS Balanced\/FSR Balanced + Frame Generation<\/td>\nDLSS Balanced (+ Frame Generation)<\/td>\n<\/tr>\n
Cyberpunk 2077<\/td>\nIntegrovan\u00fd benchmark (OCAT pre generovanie sn\u00edmok)<\/td>\nMax. kvalita grafiky a sledovania l\u00fa\u010dov (Path Tracing)<\/td>\nFSR Balanced (+ Frame Generation)<\/td>\nXeSS Balanced\/FSR Balanced + Frame Generation<\/td>\nDLSS Balanced (Transformer Model) + Ray Reconstruction (+ Frame Generation)<\/td>\n<\/tr>\n
F1 24<\/td>\nIntegrovan\u00fd benchmark, tra\u0165 Monaco (d\u00e1\u017e\u010f)<\/td>\nMax. kvalita grafiky a sledovania l\u00fa\u010dov<\/td>\nFSR Balanced (+ Frame Generation)<\/td>\nXeSS Balanced\/FSR Balanced + Frame Generation<\/td>\nDLSS Balanced (+ Frame Generation)<\/td>\n<\/tr>\n
Hogwarts Legacy<\/td>\nOCAT, jazda na voz\u00edku v Path to Hogwarts<\/td>\nMax. kvalita grafiky a sledovania l\u00fa\u010dov<\/td>\nFSR Balanced<\/td>\nXeSS Balanced<\/td>\nDLSS Balanced + Ray Reconstruction (+ Frame Generation)<\/td>\n<\/tr>\n
Indiana Jones a Ve\u013ek\u00fd Kruh<\/td>\nOCAT, lokalita Sukhothai<\/td>\nMax. kvalita grafiky, vysok\u00e1 kvalita sledovania l\u00fa\u010dov (Path Tracing); Ray Traced Shadows: Iba slne\u010dn\u00e9 svetlo<\/td>\nFSR Balanced (+ Frame Generation)<\/td>\nXeSS Balanced\/FSR Balanced + Frame Generation<\/td>\nDLSS Balanced (Transformer Model) + Ray Reconstruction (+ Frame Generation)<\/td>\n<\/tr>\n
Metro Exodus Enchanced Edition<\/td>\nIntegrovan\u00fd benchmark<\/td>\nMax. kvalita grafiky a sledovania l\u00fa\u010dov<\/td>\nN\/D<\/td>\nN\/D<\/td>\nDLSS Vyv\u00e1\u017een\u00e9<\/td>\n<\/tr>\n
Returnal<\/td>\nIntegrovan\u00fd benchmark (OCAT pre generovanie sn\u00edmok)<\/td>\nMax. kvalita grafiky a sledovania l\u00fa\u010dov<\/td>\nFSR Balanced (+ Frame Generation)<\/td>\nXeSS Balanced\/FSR Balanced + Frame Generation<\/td>\nDLSS Balanced (+ Frame Generation)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

V v\u00e4\u010d\u0161ine hier sa priemer a minimum (uv\u00e1dzame 1. percentil rozdelenia) sn\u00edmok za sekundu vypo\u010d\u00edtavaj\u00fa z po\u013ea \u010dasov renderovania jednotliv\u00fdch sn\u00edmok alebo okam\u017eit\u00e9ho sn\u00edmkovania, z\u00edskan\u00e9ho pomocou vstavan\u00e9ho benchmarku. V\u00fdnimkami s\u00fa hry bez vstavan\u00e9ho benchmarku a testy s vyu\u017eit\u00edm generovania sn\u00edmok: v t\u00fdchto pr\u00edpadoch na zachyt\u00e1vanie medzi sn\u00edmkov\u00fdmi intervalmi pou\u017e\u00edvame program OCAT.<\/p>\n\n\n\n\n\n\n\n\n
Pracovn\u00e9 aplik\u00e1cie<\/th>\n<\/tr>\n
Aplik\u00e1cia<\/td>\nBenchmark<\/td>\nNastavenia<\/td>\n<\/tr>\n
Adobe Premiere Pro 25.x<\/td>\nPugetBench for Premiere Pro 1.1.1 (zlo\u017eenie testov pod\u013ea odkazu)<\/a><\/noindex><\/td>\n\u0160tandard (4K)<\/td>\n<\/tr>\n
Blender 4.x<\/td>\nDemo Agent 327 Barbershop z webu Blender<\/td>\nRenderer Cycles<\/td>\n<\/tr>\n
Blackmagic Design DaVinci Resolve Studio 19.x<\/td>\nPugetBench for DaVinci Resolve 1.1.1 (zlo\u017eenie testov pod\u013ea odkazu)<\/a><\/noindex><\/td>\n\u0160tandard (4K); H.264\/HEVC K\u00f3dovanie re\u017eim: Automaticky<\/td>\n<\/tr>\n
CAD aplik\u00e1cie<\/td>\nSPECviewperf 2020 v3.1<\/td>\nRozl\u00ed\u0161enie obrazovky: 3840 \u00d7 2160<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Dek\u00f3dovanie videa (ffmpeg 5.x)<\/th>\n<\/tr>\n
Form\u00e1t<\/td>\nRozl\u00ed\u0161enie<\/td>\nParametre k\u00f3dovania<\/td>\nAPI<\/td>\n<\/tr>\n
H.264 (YUV 4:2:0, 8 bit\/kan\u00e1l)<\/td>\n1920 \u00d7 1080<\/td>\nHigh Profile, L4.1<\/td>\nD3D11VA<\/td>\n<\/tr>\n
3840 \u00d7 2160<\/td>\nHigh Profile, L5.1<\/td>\n<\/tr>\n
HEVC (YUV 4:2:0, 8 bit\/kan\u00e1l)<\/td>\n1920 \u00d7 1080<\/td>\nMain Profile, L4.0<\/td>\n<\/tr>\n
3840 \u00d7 2160<\/td>\nMain Profile, L5.0<\/td>\n<\/tr>\n
7680 \u00d7 4320<\/td>\nMain Profile, L6.0<\/td>\n<\/tr>\n
VP9 (YUV 4:2:0, 8 bit\/kan\u00e1l)<\/td>\n1920 \u00d7 1080<\/td>\nN\/D<\/td>\n<\/tr>\n
3840 \u00d7 2160<\/td>\n<\/tr>\n
7680 \u00d7 4320<\/td>\n<\/tr>\n
AV1 (YUV 4:2:0, 8 bit\/kan\u00e1l)<\/td>\n1920 \u00d7 1080<\/td>\nMain Profile, L4.0<\/td>\n<\/tr>\n
3840 \u00d7 2160<\/td>\nMain Profile, L5.0<\/td>\n<\/tr>\n
7680 \u00d7 4320<\/td>\nMain Profile, L6.0<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
K\u00f3dovanie videa (ffmpeg 5.x)<\/th>\n<\/tr>\n
Form\u00e1t<\/td>\nRozl\u00ed\u0161enie<\/td>\nParametre k\u00f3dovania<\/td>\nAPI<\/td>\n<\/tr>\n
AMD<\/td>\nIntel<\/td>\nNVIDIA<\/td>\nAMD<\/td>\nIntel<\/td>\nNVIDIA<\/td>\n<\/tr>\n
H.264 (YUV 4:2:0, 8 bit\/kan\u00e1l)<\/td>\n1920 \u00d7 1080<\/td>\n-c:v h264_amf -quality speed -coder cabac -refs 1 -b:v 3M<\/td>\n-c:v h264_qsv -preset veryfast -profile:v main -level 4.1 -b:v 3M<\/td>\n-c:v h264_nvenc -preset fast -coder cabac -refs 1 -b:v 3M<\/td>\nAMF<\/td>\noneVPL<\/td>\nNVENC<\/td>\n<\/tr>\n
3840 \u00d7 2160<\/td>\n-c:v h264_amf -quality speed -coder cabac -refs 1 -b:v 7.5M<\/td>\n-c:v h264_qsv -preset veryfast -profile:v main -level 5.1 -b:v 7.5M<\/td>\n-c:v h264_nvenc -preset fast -coder cabac -refs 1 -b:v 7.5M<\/td>\n<\/tr>\n
HEVC (YUV 4:2:0, 8 bit\/kan\u00e1l)<\/td>\n1920 \u00d7 1080<\/td>\n-c:v hevc_amf -quality speed -b:v 3M<\/td>\n-c:v hevc_qsv -preset veryfast -tier main -b:v 3M<\/td>\n-c:v hevc_nvenc -preset fast -b:v 3M<\/td>\n<\/tr>\n
3840 \u00d7 2160<\/td>\n-c:v hevc_amf -quality speed -b:v 7.5M<\/td>\n-c:v hevc_qsv -preset veryfast -tier main -b:v 7.5M<\/td>\n-c:v hevc_nvenc -preset fast -b:v 7.5M<\/td>\n<\/tr>\n
7680 \u00d7 4320<\/td>\n-c:v hevc_amf -quality speed -b:v 20M<\/td>\n-c:v hevc_qsv -preset veryfast -tier main -b:v 20M<\/td>\n-c:v hevc_nvenc -preset fast -b:v 20M<\/td>\n<\/tr>\n
AV1 (YUV 4:2:0, 8 bit\/kan\u00e1l)<\/td>\n1920 \u00d7 1080<\/td>\n-c:v hevc_amf -quality speed -b:v 3M<\/td>\n-c:v av1_qsv -preset veryfast -profile main -b:v 3M<\/td>\n-c:v hevc_nvenc -preset fast -b:v 3M<\/td>\n<\/tr>\n
3840 \u00d7 2160<\/td>\n-c:v hevc_amf -quality speed -b:v 7.5M<\/td>\n-c:v av1_qsv -preset veryfast -profile main -b:v 7.5M<\/td>\n-c:v hevc_nvenc -preset fast -b:v 7.5M<\/td>\n<\/tr>\n
7680 \u00d7 4320<\/td>\n-c:v hevc_amf -quality speed -b:v 20M<\/td>\n-c:v av1_qsv -preset veryfast -profile main -b:v 20M<\/td>\n-c:v hevc_nvenc -preset fast -b:v 20M<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

V\u00fdkon grafick\u00fdch kariet sa zaznamen\u00e1va samostatne od CPU a \u010fal\u0161\u00edch komponentov PC pomocou zariadenia NVIDIA PCAT. Ako z\u00e1\u0165a\u017e pre testy v\u00fdkonu a hladiny hluku sa pou\u017e\u00edva hra Cyberpunk 2077 pri rozl\u00ed\u0161en\u00ed 3840 \u00d7 2160 a maxim\u00e1lnych parametroch kvality grafiky (bez ray tracingu), rovnako ako stresov\u00fd test FurMark s najagres\u00edvnej\u0161\u00edmi nastaveniami (rozl\u00ed\u0161enie 3840 \u00d7 2160, MSAA 8x). Merania v\u0161etk\u00fdch parametrov sa vykon\u00e1vaj\u00fa po zahrian\u00ed grafickej karty, ke\u010f sa teplota GPU a taktovacie frekvencie stabilizuj\u00fa.<\/p>\n

\u2b07<\/a><\/noindex>#<\/a><\/noindex>\u00da\u010dastn\u00edci testovania<\/strong><\/h2>\n

V testovan\u00ed v\u00fdkonnosti sa z\u00fa\u010dastnili nasleduj\u00face grafick\u00e9 karty:<\/p>\n