{"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\/sq\/blog\/news\/novaya-statya-radeon-rx-9070-xt-protiv-geforce-rtx-5070-ti","title":{"rendered":"Artikulli i ri: Radeon RX 9070 XT kund\u00ebr GeForce RTX 5070 Ti","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Karakteristikat e GPU familjes Blackwell dhe testet GeForce RTX 5080<\/a><\/noindex> tregon se t\u00eb gjitha pozitat e seris\u00eb 50, p\u00ebrve\u00e7 GeForce RTX 5090, jan\u00eb pak m\u00eb shum\u00eb se nj\u00eb p\u00ebrmir\u00ebsim i but\u00eb i modeleve p\u00ebrkat\u00ebse t\u00eb gjenerat\u00ebs s\u00eb kaluar. Situat\u00ebn e p\u00ebrkeq\u00ebsoi mungesa e kartave t\u00eb reja grafike dhe, si pasoj\u00eb, \u00e7mimet e papara.<\/p>\n

Pothuajse t\u00eb gjith\u00eb bler\u00ebsit e pajisjeve t\u00eb nivelit t\u00eb lart\u00eb, t\u00eb cilat i kontrollon monopolizi NVIDIA, kan\u00eb vet\u00ebm nj\u00eb interes teorik. Por tani, n\u00eb nj\u00eb shkall\u00eb m\u00eb t\u00eb ul\u00ebt, stagnimi teknologjik i grafik\u00ebs 'jeshile' ka krijuar nj\u00eb situat\u00eb ku AMD do t\u00eb mund t\u00eb rikuperonte menj\u00ebher\u00eb vonesat n\u00eb aspekte problematike p\u00ebr arkitektur\u00ebn RDNA, si\u00e7 jan\u00eb rrezet dhe upscaling. Megjithat\u00eb, profecia q\u00eb AMD do t\u00eb rikrijoj\u00eb barazimin me konkurrenc\u00ebn (n\u00ebse jo me pretendime n\u00eb sektorin kryesor) e d\u00ebgjojm\u00eb prej koh\u00ebsh, dhe \u00e7do her\u00eb ajo ting\u00ebllon gjithnj\u00eb e m\u00eb pak bind\u00ebse, prandaj njoftimi i kartave grafike Radeon RX 9000 nuk t\u00ebrhoqi shum\u00eb v\u00ebmendje n\u00eb sfondin e aktiviteteve t\u00eb gjalla t\u00eb NVIDIA. Megjithat\u00eb, AMD ka kryer nj\u00eb pun\u00eb t\u00eb madhe p\u00ebr t\u00eb p\u00ebrmir\u00ebsuar gabimet e saj dhe n\u00ebse nuk i ka eliminuar plot\u00ebsisht, t\u00eb pakt\u00ebn e ka kompensuar disavantazhin e GPU-ve t\u00eb kaluar.<\/p>\n

Ne paraqesim nj\u00eb rishikim t\u00eb Radeon RX 9070 XT n\u00eb nj\u00eb format t\u00eb pazakont\u00eb p\u00ebr 3DNews t\u00eb krahasimit, pasi noviteti 'i kuq' dhe konkurrenti i tij kryesor \u2014 GeForce RTX 5070 Ti \u2014 arrit\u00ebn n\u00eb duar tona n\u00eb t\u00eb nj\u00ebjt\u00ebn koh\u00eb, dhe nj\u00eb pajisje nuk mund t\u00eb shqyrtohet e ndar\u00eb nga tjetra, pasi n\u00eb Rusi ato shiten p\u00ebr rreth t\u00eb nj\u00ebjtin \u00e7mim.<\/p>\n\n\n\n
\"Artikulli <\/td>\n <\/td>\n \"Artikulli <\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

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

\u279c<\/a><\/noindex>#<\/a><\/noindex> Grafik procesori Navi 48 dhe arkitektura RDNA 4<\/strong><\/h2>\n

Q\u00eb NVIDIA \u00ebsht\u00eb b\u00ebr\u00eb monopol n\u00eb tregun e GPU-ve me performanc\u00eb t\u00eb lart\u00eb dhe sillet p\u00ebrkat\u00ebsisht, nuk ka th\u00ebn\u00eb vet\u00ebm ai q\u00eb \u00ebsht\u00eb len\u00ebs. M\u00eb rrall\u00eb i kushtohet v\u00ebmendje faktit se t\u00eb gjitha kompanit\u00eb q\u00eb projektuan procesor\u00eb t\u00eb m\u00ebdhenj, p\u00ebr prodhimin e tyre jan\u00eb t\u00eb detyruara t\u00eb p\u00ebrdorin sh\u00ebrbimet e nj\u00eb kontraktori \u2014 TSMC n\u00eb Tajvan. K\u00ebrkesa p\u00ebr kapacitetet e TSMC me siguri ka qen\u00eb nj\u00eb nga arsyet pse AMD ndaloi zhvillimin e GPU-ve t\u00eb m\u00ebdhenj p\u00ebr konsumator\u00ebt, dhe n\u00eb t\u00eb v\u00ebrtet\u00eb n\u00eb gam\u00ebn e procesor\u00ebve grafik\u00eb 't\u00eb kuq' t\u00eb brezit t\u00eb ri aktualisht ka vet\u00ebm nj\u00eb kristal \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
Prodhuesi<\/th>\nAMD<\/th>\n<\/tr>\n
Emri<\/td>\nNavi 32<\/td>\nNavi 31<\/td>\nNavi 48<\/td>\n<\/tr>\n
Ku p\u00ebrdoret<\/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

Arkitektura<\/td>\nRDNA 3<\/td>\nRDNA 3<\/td>\nRDNA 4<\/td>\n<\/tr>\n
Teknologjia, nm<\/td>\nTSMC N5\/N6<\/td>\nTSMC N5\/N6<\/td>\nTSMC N4P<\/td>\n<\/tr>\n
Numri i transistor\u00ebve, miliard<\/td>\n28,1<\/td>\n57,7<\/td>\n53,9<\/td>\n<\/tr>\n
Sip\u00ebrfaqa e \u00e7ipit, mm\u00b2<\/td>\n346<\/td>\n522<\/td>\n357<\/td>\n<\/tr>\n
Numri CU\/WGP\/SE<\/td>\n<\/tr>\n
Nj\u00ebsit\u00eb e Llogaritjeve (CU)<\/td>\n60<\/td>\n96<\/td>\n64<\/td>\n<\/tr>\n
Procesor\u00ebt e Grumbullit t\u00eb Pun\u00ebs (WGP)<\/td>\n30<\/td>\n48<\/td>\n32<\/td>\n<\/tr>\n
Ndarjet e Shader\u00ebve (SA)<\/td>\nJo<\/td>\nJo<\/td>\nJo<\/td>\n<\/tr>\n
Motor\u00ebt e Shader\u00ebve (SE)<\/td>\n4?<\/td>\n6<\/td>\n4<\/td>\n<\/tr>\n
Konfigurimi i Nj\u00ebsis\u00eb s\u00eb Llogaritjeve<\/td>\n<\/tr>\n
ALU vektoriale (FP32)<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n<\/tr>\n
ALU vektoriale (FP32\/INT32)<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n2 \u00d7 32<\/td>\n<\/tr>\n
ALU skalar\u00eb<\/td>\n2<\/td>\n2<\/td>\n2<\/td>\n<\/tr>\n
ALU t\u00eb destinuara p\u00ebr q\u00ebllime t\u00eb ve\u00e7anta (SFU)<\/td>\n2 \u00d7 8<\/td>\n2 \u00d7 8<\/td>\n2 \u00d7 8<\/td>\n<\/tr>\n
Blloqet e gjurmimit t\u00eb rretheve (Ray Accelerators)<\/td>\n1<\/td>\n1<\/td>\n1<\/td>\n<\/tr>\n
Blloqet e vendosjes s\u00eb teksturave (TMU)<\/td>\n4<\/td>\n4<\/td>\n4<\/td>\n<\/tr>\n
Regjistrat vektorial\u00eb\/skalar\u00eb, KBytes<\/td>\n384\/10<\/td>\n384\/10<\/td>\n384\/16<\/td>\n<\/tr>\n
Hap\u00ebsira e memorjes L0, KBytes<\/td>\n32<\/td>\n32<\/td>\n32<\/td>\n<\/tr>\n
Konfigurimi i Procesorit t\u00eb Grupit t\u00eb Pun\u00ebs (WGP)<\/td>\n<\/tr>\n
Hap\u00ebsira e Memories s\u00eb Ndashur, KBytes<\/td>\n128<\/td>\n128<\/td>\n128<\/td>\n<\/tr>\n
Hap\u00ebsira e caches t\u00eb instruksioneve, KBytes<\/td>\n32<\/td>\n32<\/td>\n32<\/td>\n<\/tr>\n
Hap\u00ebsira e caches skalar, KBytes<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n<\/tr>\n
Blloqet e llogaritjeve t\u00eb GPU<\/td>\n<\/tr>\n
ALU vektoriale (FP32)<\/td>\n3 840 (7 680)<\/td>\n6 144 (12 288)<\/td>\n4 096 (8 192)<\/td>\n<\/tr>\n
Blloqet e vendosjes s\u00eb teksturave (TMU)<\/td>\n240<\/td>\n384<\/td>\n256<\/td>\n<\/tr>\n
Blloqet e operacioneve t\u00eb rasterizimit (ROP)<\/td>\n96<\/td>\n192<\/td>\n128<\/td>\n<\/tr>\n
Blloqet e gjurmimit t\u00eb rretheve (Ray Accelerators)<\/td>\n60<\/td>\n96<\/td>\n64<\/td>\n<\/tr>\n
Konfigurimi i memories<\/td>\n<\/tr>\n
Hap\u00ebsira e caches L1, KBytes<\/td>\n2 048<\/td>\n3 072<\/td>\n2 048<\/td>\n<\/tr>\n
Hap\u00ebsira e caches L2, MBytes<\/td>\n4<\/td>\n6<\/td>\n8<\/td>\n<\/tr>\n
Hap\u00ebsira e Infinity Cache, MBytes<\/td>\n64<\/td>\n96<\/td>\n64<\/td>\n<\/tr>\n
Gjer\u00ebsia e autobusit VRAM, bit<\/td>\n256<\/td>\n384<\/td>\n256<\/td>\n<\/tr>\n
Lloji i \u00e7ip\u00ebve VRAM<\/td>\nGDDR6<\/td>\nGDDR6<\/td>\nGDDR6<\/td>\n<\/tr>\n
<\/td>\n<\/tr>\n
Interfeisi PCI Express<\/td>\n4.0 x16<\/td>\n4.0 x16<\/td>\n5.0 x16<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Midis \u00e7ip\u00ebve t\u00eb vjet\u00ebr, ai m\u00eb shum\u00eb i ngjan modelit t\u00eb nivelit t\u00eb dyt\u00eb, Navi 32, n\u00eb karakteristikat kryesore, por ka pothuajse kaq shum\u00eb tranzistor\u00eb sa Navi 31 (53.9 dhe 57.7 miliard\u00eb p\u00ebrkat\u00ebsisht). Navi 48 po krijohet p\u00ebrmes teknologjis\u00eb TSMC N4P, e cila i p\u00ebrket t\u00eb nj\u00ebjtit nod fotolitografik 5 nm, e cila u thith dy vjet m\u00eb par\u00eb. Megjithat\u00eb, Navi 48 z\u00eb vet\u00ebm 68% t\u00eb sip\u00ebrfaqes s\u00eb Navi 31 fal\u00eb faktit se AMD i \u00ebsht\u00eb shmangur dizajnit t\u00eb \u00e7iplet\u00ebve, pasi \u00e7iplet\u00ebt MCD, t\u00eb cil\u00ebt p\u00ebrmbajn\u00eb caches t\u00eb nivelit t\u00eb tret\u00eb dhe kontrollet e memories, prodhohen p\u00ebrmes nj\u00eb procesi m\u00eb t\u00eb trash\u00eb N6.<\/p>\n

\"Artikulli <\/p>\n

Arkitektura e procesorit t\u00eb ri grafik, RDNA 4, strukturalisht nuk ndryshon nga iteracionet e m\u00ebparshme. Navi 48 p\u00ebrmban 64 Nj\u00ebsi Computimi, ose 4 096 ALU shenjuesh me pajtim FP32, q\u00eb \u00ebsht\u00eb nj\u00eb formul\u00eb mjaft tipike p\u00ebr GPU-t\u00eb e nivelit t\u00eb dyt\u00eb - ashtu si shina e memories video me 256 bit. Megjithat\u00eb, AMD n\u00eb vler\u00ebsimin e performances i p\u00ebrball\u00eb Navi 48 me flakmanin Navi 31, i cili ka 6 144 ALU shenjuesh me sakt\u00ebsi standarde. Diferenca duhet t\u00eb kompensohet mjaft nga p\u00ebrmir\u00ebsimet e shumta t\u00eb logjik\u00ebs, dhe p\u00ebr t'i kuptuar ato, k\u00ebshillohet t\u00eb rifreshoni memorien p\u00ebr arkitektur\u00ebn RDNA 3, t\u00eb cil\u00ebn e kemi p\u00ebrshkruar n\u00eb detaje n\u00eb shqyrtim. Radeon RX 7900 XTX<\/a><\/noindex>.<\/p>\n

N\u00eb krahasim me arkitektur\u00ebn \"e gjelb\u00ebr\" Blackwell<\/a><\/noindex> RDNA 4 paraqet nj\u00eb p\u00ebrmir\u00ebsim m\u00eb t\u00eb shtrir\u00eb dhe m\u00eb t\u00eb thell\u00eb t\u00eb punimeve t\u00eb m\u00ebparshme. Megjithat\u00eb, p\u00ebrshkrimi i t\u00eb gjitha novacioneve n\u00eb m\u00ebnyr\u00eb t\u00eb habitshme formohet n\u00eb nj\u00eb list\u00eb t\u00eb shkurt\u00ebr p\u00ebr shkak se NVIDIA tradicionalisht i kushton v\u00ebmendje t\u00eb shtuar zgjidhjeve komplekse t\u00eb softuer\u00ebve ose t\u00eb pjes\u00ebmarr\u00ebsve t\u00eb zakonsh\u00ebm, q\u00eb k\u00ebrkojn\u00eb komente t\u00eb gjera, kur inovacionet e AMD zakonisht fillojn\u00eb dhe p\u00ebrfundojn\u00eb n\u00eb harduer.<\/p>\n

\"Artikulli <\/p>\n

Ndryshimet themelore ndodh\u00ebn n\u00eb nd\u00ebrtimin e Nj\u00ebsis\u00eb Computimi - bllokut t\u00eb pandar\u00eb t\u00eb llogaritjes, i cili \u00ebsht\u00eb ekuivalent me SM-n\u00eb te NVIDIA dhe b\u00ebrthamave t\u00eb CPU. K\u00ebshtu, planifikuesi i instrukcioneve tani mb\u00ebshtet barrierat e ndara, q\u00eb lejojn\u00eb dh\u00ebnien e instrukcioneve p\u00ebr ekzekutim gjat\u00eb periudh\u00ebs s\u00eb veprimit. Gjithashtu, \u00ebsht\u00eb optimizuar parap\u00ebrzgjedhja e instrukcioneve nga cache, dhe ALU-t\u00eb skalar\u00eb, t\u00eb cilat p\u00ebrdoren p\u00ebr ekzekutimin t\u00eb kursyer t\u00eb operacioneve t\u00eb kalimit kushtor dhe deg\u00ebzimit, trash\u00ebguan mund\u00ebsin\u00eb e pun\u00ebs me t\u00eb dh\u00ebnat me numra t\u00eb v\u00ebrtet\u00eb nga arkitektura mobile RDNA 3.5. Po ashtu \u00ebsht\u00eb rritur edhe shuma e regjistrave skalar\u00eb: nga 10 n\u00eb 16 KB.<\/p>\n

\"Artikulli <\/p>\n

Nj\u00eb nga risit\u00eb m\u00eb t\u00eb r\u00ebnd\u00ebsishme t\u00eb RDNA 4 \u00ebsht\u00eb dyfishimi i ritmit t\u00eb ekzekutimit t\u00eb instrukcioneve matri\u00e7ike WMMA, t\u00eb cilat p\u00ebrdoren n\u00eb detyra t\u00eb m\u00ebsimit t\u00eb makinerive dhe p\u00ebrpunimit t\u00eb t\u00eb dh\u00ebnave me ndihm\u00ebn e rrjeteve nervore. P\u00ebr m\u00eb tep\u00ebr, grupi i instrukcioneve RDNA 4 p\u00ebrmban komanda p\u00ebr llogaritjet mbi matrica strukturore t\u00eb hapura dhe t\u00eb dh\u00ebna me 8-bit\u00eb me presje flotante (FP8 dhe BF8).<\/p>\n

\"Artikulli <\/p>\n

Megjithat\u00eb, kapaciteti i 1 024 FLOP p\u00ebr cik\u00ebl do t\u00eb thot\u00eb se \u00e7ipat Navi kan\u00eb arritur vet\u00ebm n\u00eb nivelin e arkitektur\u00ebs \"e gjelb\u00ebr\" t\u00eb dy viteve m\u00eb par\u00eb - Ampere. P\u00ebrve\u00e7 k\u00ebsaj, n\u00ebse \u00e7ipat NVIDIA dhe Intel p\u00ebrdorin matrica t\u00eb ve\u00e7anta ALU p\u00ebr operacionet matricore, n\u00eb arkitektur\u00ebn RDNA 4, puna e till\u00eb ende bie mbi shenjuesit SIMD.<\/p>\n

Megjithat\u00eb, mos harrojm\u00eb se n\u00eb aplikacione reale, procesor\u00ebt grafik\u00eb t\u00eb bazuar n\u00eb RDNA 3 kurr\u00eb nuk arrijn\u00eb nj\u00eb ngarkes\u00eb t\u00eb plot\u00eb. Tera-flop\u00ebt e deklaruar n\u00eb llogaritjet e FP32 p\u00ebr q\u00ebllime t\u00eb p\u00ebrgjithshme arrihen vet\u00ebm n\u00eb kushte t\u00eb p\u00ebrdorimit t\u00eb formatit VOPD (Vector Operation Dual), i cili lejon t\u00eb paketohen s\u00eb bashku dy instrukcione vektoriale t\u00eb pavarura. Dhe vet\u00eb ai imponon shum\u00eb kufizime, prandaj gjysma e ALU-ve 32-bit brenda Nj\u00ebsis\u00eb Ngrit\u00ebse shpesh q\u00ebndron n\u00eb pritje. N\u00eb nj\u00eb ngarkes\u00eb t\u00eb p\u00ebrzier (si\u00e7 \u00ebsht\u00eb nj\u00eb loj\u00eb me upscaling), operacionet matriksore do t\u00eb marrin nga shenjuesit nj\u00eb pjes\u00eb t\u00eb burimeve t\u00eb llogaritjeve, por k\u00ebto humbje nuk do t\u00eb jen\u00eb aq t\u00eb m\u00ebdha sa duket n\u00eb shikimin e par\u00eb.<\/p>\n\n\n\n\n\n\n
<\/th>\nUniteti i Llogaritjes (AMD RDNA 3)<\/th>\nUniteti i Llogaritjes (AMD RDNA 4)<\/th>\nMikroprocesori i Shkarkimit (NVIDIA Ada Lovelace)<\/th>\nMikroprocesori i Shkarkimit (NVIDIA Blackwell)<\/th>\nXe-core (Intel Xe-HPG)<\/th>\nXe-core (Intel Xe2)<\/th>\n<\/tr>\n
Blloqet Ekzekutive<\/td>\n\n

2 \u00d7 SIMD32 (FP32\/INT32);<\/p>\n

2 \u00d7 SIMD32 (FP32);<\/p>\n

2 \u00d7 SIMD2 (FP64);<\/p>\n

2 \u00d7 SIMD8 (SFU);<\/p>\n

2 \u00d7 ALU scalar<\/p>\n<\/td>\n

\n

2 \u00d7 SIMD32 (FP32\/INT32);<\/p>\n

2 \u00d7 SIMD32 (FP32);<\/p>\n

2 \u00d7 SIMD2 (FP64);<\/p>\n

2 \u00d7 SIMD8 (SFU);<\/p>\n

2 \u00d7 ALU scalar<\/p>\n<\/td>\n

\n

4 \u00d7 SIMD16 (FP32\/INT32);<\/p>\n

4 \u00d7 SIMD16 (FP32);<\/p>\n

2 \u00d7 SISD? (FP64);<\/p>\n

4 \u00d7 SIMD4 (SFU);<\/p>\n

4 \u00d7 ALU scalar;<\/p>\n

4 \u00d7 b\u00ebrthama tensor;<\/p>\n<\/td>\n

\n

8 \u00d7 SIMD16 (FP32\/INT32);<\/p>\n

2 \u00d7 SISD? (FP64);<\/p>\n

4 \u00d7 SIMD4 (SFU);<\/p>\n

4 \u00d7 ALU scalar;<\/p>\n

4 \u00d7 b\u00ebrthama tensor;<\/p>\n<\/td>\n

\n

16 \u00d7 SIMD8 (FP32);<\/p>\n

16 \u00d7 SIMD8 (INT32);<\/p>\n

16 \u00d7 SISD (FP64);<\/p>\n

16 \u00d7 SIMD2 (SFU);<\/p>\n

16 \u00d7 XMX<\/p>\n<\/td>\n

\n

8 \u00d7 SIMD16 (FP32);<\/p>\n

8 \u00d7 SIMD16 (INT32);<\/p>\n

8 \u00d7 SIMD2 (FP64);<\/p>\n

8 \u00d7 SIMD4 (SFU);<\/p>\n

8 \u00d7 XMX<\/p>\n<\/td>\n<\/tr>\n

Operacione n\u00eb vij\u00ebn SIMD p\u00ebr taktim<\/td>\n\n

128 \u00d7 FP32;<\/p>\n

64 \u00d7 INT32;<\/p>\n

256 \u00d7 FP16;<\/p>\n

4 \u00d7 FP64;<\/p>\n

16 \u00d7 funksione transcendente;<\/p>\n<\/td>\n

\n

128 \u00d7 FP32;<\/p>\n

64 \u00d7 INT32;<\/p>\n

256 \u00d7 FP16;<\/p>\n

4 \u00d7 FP64;<\/p>\n

16 \u00d7 funksione transcendente;<\/p>\n<\/td>\n

\n

128 \u00d7 FP32;<\/p>\n

64 \u00d7 INT32;<\/p>\n

128 \u00d7 FP16;<\/p>\n

2 \u00d7 FP64;<\/p>\n

16 \u00d7 funksione transcendente;<\/p>\n<\/td>\n

\n

128 \u00d7 FP32;<\/p>\n

128 \u00d7 INT32;<\/p>\n

128 \u00d7 FP16;<\/p>\n

2 \u00d7 FP64;<\/p>\n

16 \u00d7 funksione transcendente;<\/p>\n<\/td>\n

\n

128 \u00d7 FP32;<\/p>\n

128 \u00d7 INT32;<\/p>\n

256 \u00d7 FP16;<\/p>\n

16 \u00d7 FP64;<\/p>\n

32 \u00d7 funksione transcendente;<\/p>\n<\/td>\n

\n

128 \u00d7 FP32;<\/p>\n

128 \u00d7 INT32;<\/p>\n

256 \u00d7 FP16;<\/p>\n

16 \u00d7 FP64;<\/p>\n

32 \u00d7 funksione transcendente;<\/p>\n<\/td>\n<\/tr>\n

Operacione matricore, FLOP p\u00ebr taktim (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

Masa e frame-ve p\u00ebrmes nj\u00eb rrjeti nervor \u2014 nuk \u00ebsht\u00eb m\u00eb nj\u00eb situat\u00eb teorike p\u00ebr akcelerator\u00ebt Radeon. Versioni i kat\u00ebrt i FSR s\u00eb fundmi mb\u00ebshtetet n\u00eb m\u00ebsimin e makinerive, p\u00ebrfshir\u00eb edhe p\u00ebr gjenerimin e frame-ve, dhe modelet e p\u00ebrdorura p\u00ebrb\u00ebjn\u00eb nj\u00eb variant hibrid t\u00eb rrjeteve konvencionale dhe transformuesve. Fatkeq\u00ebsisht, ndryshe nga versionet e m\u00ebparshme t\u00eb upscaler, t\u00eb cilat punonin praktikisht n\u00eb \u00e7do harduer, FSR 4 mb\u00ebshtet vet\u00ebm procesor\u00ebt grafik\u00eb nga arkitektura RDNA 4 (p\u00ebr shkak t\u00eb zgjerimit t\u00eb setit t\u00eb instruktimeve WMMA p\u00ebr t\u00eb dh\u00ebnat FP8). Nj\u00eb mod i universale me b\u00ebrthama llogarit\u00ebse t\u00eb thjeshta, si n\u00eb XeSS t\u00eb Intel, nuk \u00ebsht\u00eb i parashikuar.<\/p>\n

\"Artikulli <\/p>\n

Nj\u00eb drejtim tjet\u00ebr i reform\u00ebs s\u00eb arkitektur\u00ebs RDNA ishte gjurmimi i rrezatimit. Shpejt\u00ebsia e testimit t\u00eb nd\u00ebrprerjeve t\u00eb rrezit me kutin\u00eb BVH dhe me trek\u00ebnd\u00ebshin u dyfishua p\u00ebr \u00e7do Ray Accelerator (dmth, edhe p\u00ebr \u00e7do Nj\u00ebsin\u00eb Llogarit\u00ebse): nga 4 dhe 1 p\u00ebr cik\u00ebl n\u00eb 8 dhe 2, p\u00ebrkat\u00ebsisht. Megjithat\u00eb, blloku RT i \u00e7ipave Intel Battlemage funksionon me nj\u00eb rit\u00ebm prej 18 dhe 2, nd\u00ebrsa n\u00eb procesor\u00ebt grafik\u00eb Blackwell trajtohen menj\u00ebher\u00eb kat\u00ebr nd\u00ebrprerje rrezatimi me trek\u00ebnd\u00ebshin n\u00eb \u00e7do cik\u00ebl dhe nj\u00eb num\u00ebr t\u00eb paspublikuar nd\u00ebrlidhjesh me kutin\u00eb BVH. P\u00ebr m\u00eb tep\u00ebr, tek konkurent\u00ebt testimet e dy llojeve (ndaj kutis\u00eb BVH dhe trek\u00ebnd\u00ebshit) kryhen nj\u00ebkoh\u00ebsisht, nd\u00ebrsa tek AMD - akoma jo. N\u00eb nj\u00eb far\u00eb mase, paralelizmi \u00ebsht\u00eb i mundur, pasi \u00e7do Ray Accelerator tani p\u00ebrmban dy \"testues\" t\u00eb pavarur, por jo n\u00eb shpejt\u00ebsi t\u00eb plot\u00eb.<\/p>\n

\"Artikulli <\/p>\n

P\u00ebr m\u00eb tep\u00ebr, Ray Accelerator ka marr\u00eb mund\u00ebsin\u00eb p\u00ebr t\u00eb procesuar rrezet jasht\u00eb radh\u00ebs. Rrezet q\u00eb nuk k\u00ebrkojn\u00eb akses n\u00eb nivelet e ngadalta t\u00eb grumbullimit t\u00eb memories trajtohen s\u00eb pari, dhe rezultatet e testimit ruhen, p\u00ebr t'i dh\u00ebn\u00eb programit t\u00eb shader-it ato n\u00eb rendin e pritur.<\/p>\n

Logjika e ritransimit tani p\u00ebrdor nj\u00eb struktur\u00eb m\u00eb t\u00eb dendur BVH8 n\u00eb vend t\u00eb BVH4 (tet\u00eb n\u00eb vend t\u00eb kat\u00ebr nyjave n\u00eb \u00e7do deg\u00eb t\u00eb hierarkis\u00eb s\u00eb volumit), e cila rrit shpejt\u00ebsin\u00eb e kalimit dhe nj\u00ebherazi zvog\u00eblon presionin n\u00eb memorie grafike. Madh\u00ebsia e BVH n\u00eb memorie \u00ebsht\u00eb reduktuar gjithashtu p\u00ebr shkak t\u00eb nj\u00eb algoritmi t\u00eb ri kompresimi t\u00eb primitiveve. S\u00eb fundi, AMD ofron mund\u00ebsin\u00eb p\u00ebr t\u00eb shk\u00ebputur orientimin e kutive BVH nga koordinatat absolute t\u00eb sken\u00ebs. Rrotullimi i kutive n\u00eb p\u00ebrputhje me gjeometrin\u00eb e objekteve lejon t\u00eb zvog\u00eblohet numri i hapave n\u00eb k\u00ebrkim.<\/p>\n

\"Artikulli <\/p>\n

\u00cbsht\u00eb e dhimbshme q\u00eb, ndon\u00ebse t\u00eb gjitha ndryshimet e renditura, RDNA 4 ende realizon kalimin e BVH n\u00eb sh\u00ebrbimet ALU t\u00eb shaderit, dhe jo p\u00ebrmes logjik\u00ebs s\u00eb specializuar. Sipas klasifikimit t\u00eb mjeteve p\u00ebr rrethin drit\u00ebs q\u00eb propozohet nga Imagination Technologies, zgjidhjet e AMD jan\u00eb ngjitur n\u00eb nivelin e dyt\u00eb, ndryshe nga procesor\u00ebt grafik\u00eb t\u00eb Intel q\u00eb jan\u00eb n\u00eb nivelin e tret\u00eb, nd\u00ebrsa pajisjet NVIDIA tashm\u00eb po kalojn\u00eb nga niveli 3 n\u00eb nivelin 4 fal\u00eb funksionit SER - grumbullimit dinamik t\u00eb rrjedhave t\u00eb instruksioneve me q\u00ebllim p\u00ebr t\u00eb rritur koherenc\u00ebn e aksesit n\u00eb memorje dhe, si pasoj\u00eb, p\u00ebr t\u00eb reduktuar latenc\u00ebn.<\/p>\n

\"Artikulli <\/p>\n

Nuk ka nj\u00eb analog t\u00eb SER n\u00eb arkitektur\u00ebn RDNA 4, por ka nj\u00eb funksion t\u00eb ri, Memoria Jasht\u00eb Rendit, e cila ndihmon p\u00ebr t\u00eb zgjidhur po at\u00eb problem n\u00eb nj\u00eb m\u00ebnyr\u00eb tjet\u00ebr. Versionet e m\u00ebparshme t\u00eb RDNA, duke punuar me disa grupe instruksionesh (\u00ebndrrat n\u00eb terminologjin\u00eb e AMD), kthenin t\u00eb dh\u00ebnat nga memoria rigorozisht n\u00eb t\u00eb nj\u00ebjtin rend si\u00e7 ishin b\u00ebr\u00eb k\u00ebrkesat. Kur t\u00eb dh\u00ebnat e k\u00ebrkuara nga shaderi ndodhen n\u00eb cache t\u00eb shpejt\u00eb, kthimi mund t\u00eb vonohet, n\u00ebse m\u00eb par\u00eb nj\u00eb shader tjet\u00ebr ka b\u00ebr\u00eb k\u00ebrkes\u00ebn e tij dhe t\u00eb dh\u00ebnat e tij ndodhen n\u00eb VRAM relativisht t\u00eb ngadalshme. N\u00eb t\u00eb v\u00ebrtet\u00eb, kjo kufizim duket krejt\u00ebsisht e pap\u00ebrshtatshme n\u00eb kontekstin e llogaritjeve shum\u00ebthesaret, por sigurisht \u00ebsht\u00eb shfaqur p\u00ebr t\u00eb simplifikuar organizimin e GPU dhe deri n\u00eb koh\u00eb (lexo: n\u00eb loj\u00ebrat e rasterizuara) ka mbetur nj\u00eb kompromis i arsyesh\u00ebm. RDNA 4 lejon grupe t\u00eb instruksionesh q\u00eb t\u00eb kthejn\u00eb t\u00eb dh\u00ebnat pavar\u00ebsisht nga rendi i k\u00ebrkesave.<\/p>\n

\"Artikulli <\/p>\n

Memoria e jasht\u00ebzakonshme \u2013 \u00ebsht\u00eb vet\u00ebm nj\u00eb nga ndryshimet e stack-ut t\u00eb memories n\u00eb arkitektur\u00ebn RDNA 4. Cil\u00ebsimi i regjistrave p\u00ebr programet e shader-ave ndodhte m\u00eb par\u00eb n\u00eb m\u00ebnyr\u00eb statike, duke u bazuar n\u00eb konsumin maksimal gjat\u00eb jet\u00ebs s\u00eb shader-it, \u00e7ka do t\u00eb thot\u00eb se nj\u00eb pjes\u00eb e caktuar e skedarit t\u00eb regjistrave n\u00eb fakt mbetej e zbraz\u00ebt. RDNA 4 lejon q\u00eb shader-art t\u00eb k\u00ebrkojn\u00eb dhe t\u00eb lirojn\u00eb regjistrat n\u00eb m\u00ebnyr\u00eb dinamike. K\u00ebshtu, volumi i skedarit t\u00eb regjistrave p\u00ebrdoret m\u00eb ekonomisht.<\/p>\n

E \u00e7uditshme, Shader Engine \u2013 nj\u00eb struktur\u00eb e madhe e GPU-es q\u00eb kombinon nj\u00eb grup CU, rasterizuesin, bllokun e primitiveve dhe blloqet e operacioneve t\u00eb rasterizimit \u2013 n\u00eb RDNA 4 nuk ka m\u00eb cache-n e vet t\u00eb nivelit t\u00eb par\u00eb. Megjithat\u00eb, \u00e7do segment i cache L2 p\u00ebrmban dy her\u00eb m\u00eb shum\u00eb banka. K\u00ebshtu, jo vet\u00ebm q\u00eb \u00ebsht\u00eb rritur volumi i tij, por edhe kapaciteti i tij i kalimit, p\u00ebrkat\u00ebsisht n\u00eb t\u00eb dyja drejtimet \u2013 drejt Shader Engine dhe drejt Infinity Cache (n\u00eb fjal\u00eb t\u00eb tjera, L3). Volumi i vet\u00eb Infinity Cache ka mbetur i nj\u00ebjt\u00eb si n\u00eb Navi 32, \u2013 64 MB.<\/p>\n

Navi 48 ka 16 linja PCI Express t\u00eb gjenerat\u00ebs s\u00eb pest\u00eb, por n\u00eb detyrat loj\u00ebra, natyrisht, nuk mund t\u00eb z\u00ebr\u00eb plot\u00ebsisht nj\u00eb nd\u00ebrfaqe aq t\u00eb shpejt\u00eb.<\/p>\n

\"Artikulli <\/p>\n

S finally, AMD njofton se kodeku i ri harduerik i videos q\u00eb p\u00ebrfshihet n\u00eb GPU-n\u00eb e gjenerat\u00ebs s\u00eb re ofron cil\u00ebsi m\u00eb t\u00eb lart\u00eb kodimi me bitrates t\u00eb ul\u00ebta, t\u00eb njohura p\u00ebr transmetimin, por jo m\u00eb shum\u00eb se kaq. K\u00ebshtu, mb\u00ebshtetje p\u00ebr sub-sampling-ngjyr\u00eb YUV 4:2:2, t\u00eb cil\u00ebn e kan\u00eb procesor\u00ebt grafik\u00eb NVIDIA dhe Intel, nuk \u00ebsht\u00eb prezantuar.<\/p>\n

Sa i p\u00ebrket mjeteve t\u00eb daljes s\u00eb imazhit, kontrolluesi i ekranit Navi 48 \u00ebsht\u00eb n\u00eb p\u00ebrputhje me versionet m\u00eb t\u00eb reja t\u00eb nd\u00ebrfaqeve DisplayPort dhe HDMI: 2.1a dhe 2.1b p\u00ebrkat\u00ebsisht.<\/p>\n

\u279c<\/a><\/noindex>#<\/a><\/noindex>Specifikat teknike, \u00e7mimet<\/strong><\/h2>\n

Radeon RX 9070 XT vjen me nj\u00eb crystal plot\u00ebsisht funksional Navi 48 me 64 Compute Unit\u00eb, \u00e7ka do t\u00eb thot\u00eb 4,096 ALU t\u00eb sakt\u00ebsis\u00eb standarde t\u00eb llogaritjeve. T\u00eb gjesh nj\u00eb pararend\u00ebs formal t\u00eb novitetit mes acelerator\u00ebve \"t\u00eb kuq\" t\u00eb gjenerat\u00ebs s\u00eb kaluar \u00ebsht\u00eb e v\u00ebshtir\u00eb, sepse \u00ebsht\u00eb ndryshuar si rrjeti modelor ashtu edhe parimi i num\u00ebrimit t\u00eb GPU-ve (chip-\u00ebt Navi kurr\u00eb m\u00eb par\u00eb nuk kan\u00eb arritur n\u00eb numrin tet\u00eb). Duke u bazuar n\u00eb formul\u00ebn e blloqeve t\u00eb llogaritjeve dhe karakteristikat e stack-ut t\u00eb memories, Radeon RX 9070 XT \u00ebsht\u00eb e ngjashme me Radeon RX 7800 XT, por AMD e pozicionon at\u00eb si z\u00ebvend\u00ebsim p\u00ebr Radeon RX 7900 XTX.<\/p>\n

Game Clock dhe Boost Clock (frekuenca celulare n\u00eb loj\u00eb dhe kufiri i frekuenc\u00ebs celulare) t\u00eb procesorit grafik kan\u00eb p\u00ebsuar nj\u00eb rritje t\u00eb qen\u00ebsishme krahasuar me karakteristikat e modelit kryesor t\u00eb gjenerat\u00ebs s\u00eb kaluar dhe arrijn\u00eb 2.4 dhe 2.97 GHz, p\u00ebrkat\u00ebsisht. Performanca teorike e Radeon RX 7900 XTX bazuar n\u00eb maksimumin e frekuenc\u00ebs celulare \u00ebsht\u00eb 26% m\u00eb e lart\u00eb, por AMD ka n\u00eb plan ta kompensoj\u00eb k\u00ebt\u00eb diferenc\u00eb p\u00ebrmes p\u00ebrmir\u00ebsimeve t\u00eb arkitektur\u00ebs RDNA 4.<\/p>\n

Radeon RX 9070 XT ka marr\u00eb 16 GB memorie GDDR6 me nj\u00eb band\u00eb t\u00eb gjer\u00eb prej 20 Gbit\/s mbi nj\u00eb autobus 256-bit. Sipas thashethemeve, AMD kishte shqyrtuar mund\u00ebsin\u00eb e rritjes s\u00eb kapacitetit VRAM n\u00eb 32 GB, por p\u00ebrfundimisht e konsideroi k\u00ebt\u00eb konfigurim t\u00eb tep\u00ebrt p\u00ebr zgjidhjet p\u00ebr konsumator\u00ebt. E pik\u00ebllueshme, p\u00ebrderisa k\u00ebrkesat e loj\u00ebrave vazhdojn\u00eb t\u00eb rriten, nd\u00ebrsa kapaciteti i memorjes grafike \u00ebsht\u00eb nj\u00eb avantazh konkurues i r\u00ebnd\u00ebsish\u00ebm p\u00ebr shum\u00eb prej pajisjeve t\u00eb m\u00ebparshme t\u00eb AMD.<\/p>\n

Konsumimi referues i energjis\u00eb p\u00ebr Radeon RX 9070 XT \u00ebsht\u00eb 304 W, nd\u00ebrsa \u00e7mimi i rekomanduar \u00ebsht\u00eb $599 (me $100 m\u00eb shum\u00eb se 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
Prodhuesi<\/th>\nAMD<\/th>\nNVIDIA<\/th>\n<\/tr>\n
Modeli<\/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
Procesori grafike<\/td>\n<\/tr>\n
Emri<\/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
Arkitektura<\/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
Procesi teknologjik<\/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
Numri i transistor\u00ebve, miliard<\/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
Frekuenca celulare (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
Blloqet e vendosjes s\u00eb teksturave (TMU)<\/td>\n336<\/td>\n384<\/td>\n224<\/td>\n256<\/td>\n264<\/td>\n320<\/td>\n280<\/td>\n336<\/td>\n<\/tr>\n
Blloqet e operacioneve t\u00eb rasterizimit (ROP)<\/td>\n192<\/td>\n192<\/td>\n128<\/td>\n128<\/td>\n112<\/td>\n112<\/td>\n96<\/td>\n112<\/td>\n<\/tr>\n
Kern t\u00eb tensor\u00ebve<\/td>\nJo<\/td>\nJo<\/td>\nJo<\/td>\nJo<\/td>\n264<\/td>\n320<\/td>\n280<\/td>\n336<\/td>\n<\/tr>\n
Kern RT<\/td>\n84<\/td>\n96<\/td>\n112<\/td>\n128<\/td>\n66<\/td>\n80<\/td>\n70<\/td>\n84<\/td>\n<\/tr>\n
Kapaciteti i cache-it t\u00eb nivelit t\u00eb fundit, MB<\/td>\n80<\/td>\n96<\/td>\n64<\/td>\n64<\/td>\n48<\/td>\n64<\/td>\n64<\/td>\n64<\/td>\n<\/tr>\n
Memoria operuese<\/td>\n<\/tr>\n
Gjer\u00ebsia e autobusit, bit<\/td>\n320<\/td>\n384<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n256<\/td>\n<\/tr>\n
Lloji i \u00e7ipave<\/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
Bandwidth p\u00ebr kontakt, Gbit\/s<\/td>\n20<\/td>\n20<\/td>\n20<\/td>\n20<\/td>\n21<\/td>\n23<\/td>\n28<\/td>\n30<\/td>\n<\/tr>\n
Kapaciteti total, GB\/s<\/td>\n800<\/td>\n960<\/td>\n640<\/td>\n640<\/td>\n672<\/td>\n736<\/td>\n896<\/td>\n960<\/td>\n<\/tr>\n
Kapaciteti, GB<\/td>\n20<\/td>\n24<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n16<\/td>\n<\/tr>\n
Performanca<\/td>\n<\/tr>\n
Performanca maksimale 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
Performanca 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
Performanca 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
T\u00eb tjera<\/td>\n<\/tr>\n
Autobusi 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
Interface p\u00ebr daljen e imazhit<\/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
\u00c7mimi me pakic\u00eb (Shtetet e Bashkuara), $<\/td>\n899 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n999 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n549 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n599 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n799 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n999 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n749 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n999 (rekomanduar n\u00eb dat\u00ebn e daljes)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Nga ana tjet\u00ebr, GeForce RTX 5070 Ti \u00ebsht\u00eb e bazuar n\u00eb t\u00eb nj\u00ebjtin procesor grafik GB203, i cili p\u00ebrdoret n\u00eb GeForce RTX 5080, por modeli m\u00eb i ri ka humbur 14 shum\u00ebpro\u00e7esor\u00eb t\u00eb rrjedh\u00ebs, dometh\u00ebn\u00eb 1,792 ALU shader t\u00eb p\u00ebrshtatsh\u00ebm me FP32, dhe frekuenca e loj\u00ebs \u00ebsht\u00eb reduktuar. Si rezultat, GeForce RTX 5080 ndan nj\u00eb 27% t\u00eb performanc\u00ebs teorike nga RTX 5070 Ti, dhe vet\u00eb GeForce RTX 5070 Ti n\u00eb k\u00ebt\u00eb aspekt nuk ndryshon nga RTX 4070 Ti SUPER. Si n\u00eb rastin e Radeon RX 9070 XT, mungesa e fuqis\u00eb bruto llogarit\u00ebse pritet t\u00eb kompensohet nga arkitektura e re GPU, dhe n\u00eb radh\u00eb t\u00eb par\u00eb \u2013 funksioni i gjenerimit t\u00eb kadrave t\u00eb shumt\u00eb p\u00ebrmes DLSS (deri n\u00eb tre kadrave t\u00eb gjeneruar n\u00eb intervalin midis dy \"t\u00eb v\u00ebrtetave\"). <\/p>\n

GeForce RTX 5070 Ti \u00ebsht\u00eb e dizajnuar p\u00ebr nj\u00eb konsum energjie prej 300 W dhe me nj\u00eb \u00e7mim t\u00eb rekomanduar prej $749, \u00ebsht\u00eb $50 m\u00eb e lir\u00eb se GeForce RTX 4070 Ti SUPER.<\/p>\n

Dallimi prej $150 i vler\u00ebs s\u00eb rekomanduar valutore midis Radeon RX 9070 XT dhe GeForce RTX 5070 Ti nuk \u00ebsht\u00eb e dukshme p\u00ebr \u00e7mimet aktuale n\u00eb Rusi: modelet m\u00eb t\u00eb p\u00ebrballueshme partnere i gjet\u00ebm n\u00eb DNS p\u00ebr 99,999 dhe 104,999 rubla p\u00ebrkat\u00ebsisht. K\u00ebto do t\u00eb jen\u00eb ato q\u00eb do t'i testojm\u00eb n\u00eb benchmark-et. Radeon RX 9070 XT \u00ebsht\u00eb karta grafike GIGABYTE GAMING OC me nj\u00eb ngarkes\u00eb fabrike GPU deri n\u00eb 2.52\/3.06 GHz, nd\u00ebrsa GeForce RTX 5070 Ti \u2014 akseleratori Palit GamingPro me specifikacione referuese.<\/p>\n

\u279c<\/a><\/noindex>#<\/a><\/noindex>GIGABYTE Radeon RX 9070 XT GAMING OC: dizajni<\/strong><\/h2>\n

Radeon RX 9070 XT n\u00eb versionin GIGABYTE ka p\u00ebrmasa mjaft modeste p\u00ebr nj\u00eb kart\u00eb grafike me konsum energjie jo m\u00eb pak (dhe, si\u00e7 do t\u00eb tregojn\u00eb testet, shum\u00eb m\u00eb tep\u00ebr) se 300 W: 288 mm n\u00eb gjat\u00ebsi dhe tre vende n\u00eb gjer\u00ebsi. Elementi i vet\u00ebm i ndri\u00e7imit LED \u00ebsht\u00eb nj\u00eb band\u00eb e shkurt\u00ebr n\u00eb an\u00eb. Shenj\u00eb me logon e prodhuesit \u2014 e l\u00ebvizshme: n\u00ebse e zhvendosni n\u00eb band\u00ebn LED, logo fillon t\u00eb ndri\u00e7oj\u00eb.<\/p>\n

\"Artikulli <\/p>\n

Sistemi i ftohjes p\u00ebrmban tre ventilator\u00eb me diamet\u00ebr rrotullimi prej 87 mm. Ajri kalon p\u00ebrmes nj\u00eb seksioni t\u00eb vog\u00ebl t\u00eb radiatorit fal\u00eb nj\u00eb prerjeje n\u00eb bacak.<\/p>\n

\"Artikulli <\/p>\n

Procesori grafik dhe \u00e7ipat e memories grafike l\u00ebshojn\u00eb nxeht\u00ebsi te kamera e avullit, por i pari \u2014 drejtp\u00ebrdrejt, nd\u00ebrsa t\u00eb fundit \u2014 p\u00ebrmes nj\u00eb ndihm\u00ebs metalik n\u00eb form\u00eb 'P'. Pjes\u00ebt q\u00eb i bashkohen lamelave t\u00eb radiatorit jan\u00eb pllaketa t\u00eb ve\u00e7anta q\u00eb sh\u00ebrbejn\u00eb si mbledh\u00ebs t\u00eb nxeht\u00ebsis\u00eb s\u00eb komponent\u00ebve VRM. Ka gjithsej shtat\u00eb tub ngritjeje, diametri i t\u00eb cil\u00ebve \u00ebsht\u00eb 6 mm. T\u00eb kemi parasysh se GIGABYTE preferoi p\u00ebrdorimin e padhjeve t\u00eb l\u00ebngshme mbi ato t\u00eb zakonshmet. Ndoshta kjo \u00ebsht\u00eb arsyeja pse termocircuitet VRAM nxehen gjat\u00eb ngarkes\u00ebs (shihni seksionin empiric t\u00eb shqyrtimit).<\/p>\n

\"Artikulli <\/p>\n

Batekpleti merr pjes\u00eb n\u00eb shp\u00ebrndarjen e nxeht\u00ebsis\u00eb nga tabela e printerit p\u00ebrmes disa termo-nd\u00ebrmjet\u00ebsve (k\u00ebtu tani jan\u00eb t\u00eb zakonsh\u00ebm, e jo t\u00eb l\u00ebngsh\u00ebm).<\/p>\n

\"Artikulli <\/p>\n

\u279c<\/a><\/noindex>#<\/a><\/noindex>GIGABYTE Radeon RX 9070 XT GAMING OC: tabela e printerit<\/strong><\/h2>\n

Disa variante t\u00eb Radeon RX 9070 XT marrin energji p\u00ebrmes nj\u00eb kone 12V-2\u00d76, por GIGABYTE preferoi tre kone t\u00eb p\u00ebrgjithshme me tet\u00eb kontakte. N\u00eb drit\u00ebn e konsumit t\u00eb lart\u00eb t\u00eb energjis\u00eb s\u00eb kart\u00ebs grafike dhe ankesave t\u00eb vazhdueshme p\u00ebr kone t\u00eb re, kjo \u00ebsht\u00eb nj\u00eb zgjidhje m\u00eb e besueshme, megjith\u00ebse jo aq e p\u00ebrshtatshme dhe estetike. Nga 17 faza t\u00eb sistemit t\u00eb furnizimit, 14, t\u00eb menaxhuara nga kontrollori PWM Monolithic Power Systems MP2868A, i atribuohen procesorit grafik, nd\u00ebrsa tre t\u00eb tjera \u2014 memories grafike. T\u00eb dy VRM jan\u00eb t\u00eb pajisur me kaskad\u00eb energjie MP87993 me nj\u00eb vler\u00eb t\u00eb mundshme t\u00eb rrym\u00ebs prej 90A.<\/p>\n

VRAM \u00ebsht\u00eb e p\u00ebrb\u00ebr\u00eb nga \u00e7ipet SK hynix, n\u00eb shenjimin e tyre H56G42AS8DX-014 \u00ebsht\u00eb koduar kapaciteti i band\u00ebs 20 Gbit\/s.<\/p>\n

P\u00ebrve\u00e7 kopjes s\u00eb BIOS-it, e cila \u00ebsht\u00eb aktive si parazgjedhje, karta grafike ka nj\u00eb firmware rezerv\u00eb 't\u00eb qet\u00eb'.<\/p>\n

\"Artikulli \"Artikulli <\/p>\n

\u279c<\/a><\/noindex>#<\/a><\/noindex>Palit GeForce RTX 5070 Ti GamingPro: struktura<\/strong><\/h2>\n

Palit GeForce RTX 5070 Ti GamingPro ka nj\u00eb gjat\u00ebsi prej 331.9 mm, por ka vet\u00ebm tri vende zgjerimi. N\u00ebn logon GamingPro dhe segmentin p\u00ebrkat\u00ebs t\u00eb panelit frontal ndodhen LED RGB. Funksionimi i LED-ve mund t\u00eb sinkronizohet me pllak\u00ebn am\u00eb p\u00ebrmes nj\u00eb kone standarde ARGB, e cila ndodhet pran\u00eb hyrjes s\u00eb energjis\u00eb 12V-2\u00d76.<\/p>\n

\"Artikulli <\/p>\n

Sip\u00ebrfaqet an\u00ebsore t\u00eb korpusit p\u00ebrb\u00ebhen nga nj\u00eb korniz\u00eb alumini e b\u00ebr\u00eb me derdhje me hap\u00ebsira ventilimi n\u00eb an\u00ebt e gjata. N\u00eb pllak\u00ebn mbrojt\u00ebse n\u00eb an\u00ebn e prapme t\u00eb PCB-s\u00eb ka nj\u00eb rrjet\u00eb t\u00eb zakonshme p\u00ebr ajrosjen p\u00ebrmes radiatorit. D\u00ebrrasat e ventilator\u00ebve kan\u00eb nj\u00eb diamet\u00ebr prej 94 mm.<\/p>\n

\"Artikulli <\/p>\n

N\u00eb baz\u00ebn e ftoh\u00ebsit ndodhet nj\u00eb kamer\u00eb avulluese me nj\u00eb sip\u00ebrfaqe t\u00eb mjaftueshme p\u00ebr t\u00eb mbuluar kristalin e GPU-s\u00eb dhe \u00e7ipat p\u00ebrreth t\u00eb memories grafike, nd\u00ebrsa p\u00ebr komponent\u00ebt e VRM \u00ebsht\u00eb parashikuar nj\u00eb radiatori t\u00eb ve\u00e7ant\u00eb. Lamela e radiatorit \u00ebsht\u00eb vendosur mbi tet\u00eb tubo termike me diamet\u00ebr 6 mm.<\/p>\n

\"Artikulli <\/p>\n

Panfalti metalik ka vet\u00ebm nj\u00eb funksion mbrojt\u00ebs dhe dekorativ, pasi midis tij dhe PCB-s\u00eb nuk ka asnj\u00eb p\u00ebr\u00e7ues termik.<\/p>\n

\"Artikulli <\/p>\n

Kart\u00ebs grafike i shoq\u00ebrohet me nj\u00eb adaptor nga dy konektor\u00eb t\u00eb energjis\u00eb me tet\u00eb kontakte n\u00eb 12V-2\u00d76 dhe nj\u00eb qilim t\u00eb vog\u00ebl p\u00ebr mouse.<\/p>\n

\"Artikulli <\/p>\n

\u279c<\/a><\/noindex>#<\/a><\/noindex>Palit GeForce RTX 5070 Ti GamingPro: plati formuese<\/strong><\/h2>\n

N\u00eb krahasim me kart\u00ebn grafike Palit GameRock<\/a><\/noindex>, e cila u shqyrtua n\u00eb nj\u00eb rishikim t\u00eb fundit t\u00eb GeForce RTX 5080, modeli tjet\u00ebr m\u00eb i lart\u00eb i seris\u00eb \u00ebsht\u00eb nd\u00ebrtuar n\u00eb nj\u00eb PCB t\u00eb gjer\u00eb, por sistemi i furnizimit \u00ebsht\u00eb m\u00eb i thjesht\u00eb. Procesorin grafik i takon nj\u00eb VRM 15-faz\u00ebsh n\u00ebn menaxhimin e kontrollorit PWM Monolithic Power Systems MP29816. Furnizimi i \u00e7ipave t\u00eb memories GDDR7 \u00ebsht\u00eb tri-fazor, mbi baz\u00ebn e kontrollorit MP2988. T\u00eb dy rregullator\u00ebt jan\u00eb t\u00eb p\u00ebrfunduar me grupe tranzistor\u00ebsh MP87993 me nj\u00eb nominal t\u00eb dyshimt\u00eb t\u00eb rrym\u00ebs prej 90 A.<\/p>\n

Shqiponja e \u00e7ipave t\u00eb memories grafike Samsung K4VAF325ZC-SC28 tregon nj\u00eb band\u00eb kalimi prej 28 Gbit\/s.<\/p>\n

Ashtu si Radeon RX 9070 XT n\u00eb versionin GIGABYTE, akseleratori Palit ka dy variante BIOS - \"me performanc\u00eb\" dhe \"t\u00eb qet\u00eb\".<\/p>\n

\"Artikulli \"Artikulli <\/p>\n

\u279c<\/a><\/noindex>#<\/a><\/noindex>Standi i testimit, metodologia e testimit<\/strong><\/h2>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Standi i testimit<\/th>\n<\/tr>\n
CPU<\/td>\nAMD Ryzen 9 7950X3D (PBO +150 MHz, CU -20)<\/td>\n<\/tr>\n
Motherboard<\/td>\nASUS ROG Crosshair X670E Hero<\/td>\n<\/tr>\n
Memoria operuese<\/td>\nG.Skill Trident Z5 Neo RGB (F5-6000J3040G32GX2-TZ5NR), 2 \u00d7 32 GB (6200 MT\/s, CL30)<\/td>\n<\/tr>\n
ROM<\/td>\nSolidigm P44 Pro, 2 TB<\/td>\n<\/tr>\n
Furnizimi i energjis\u00eb<\/td>\nCorsair AX1600i, 1600 W<\/td>\n<\/tr>\n
Sistemi i ftohjes s\u00eb CPU<\/td>\nSistem i personalizuar t\u00eb ftohjes (EK-Quantum Velocity\u00b2 DDC 4.2 PWM D-RGB + EK-Quantum Surface X280M)<\/td>\n<\/tr>\n
Kasa<\/td>\nStand i hapur<\/td>\n<\/tr>\n
Sistemi operativ<\/td>\nWindows 11 Pro<\/td>\n<\/tr>\n
Software p\u00ebr GPU AMD<\/td>\n<\/tr>\n
T\u00eb gjitha kartat grafike<\/td>\nAMD Software Adrenalin Edition 25.2.1\/25.3.1<\/td>\n<\/tr>\n
Software p\u00ebr GPU NVIDIA<\/td>\n<\/tr>\n
T\u00eb gjitha kartat grafike<\/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
Loj\u00ebra pa gjurmim rreze<\/th>\n<\/tr>\n
Loj\u00eb<\/td>\nAPI<\/td>\nMetoda e testimit<\/td>\nCil\u00ebsimet grafike<\/td>\n<\/tr>\n
Alan Wake 2<\/td>\nDirectX 12<\/td>\nOCAT, lokacioni Bright Falls<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
Black Myth: Wukong<\/td>\nDirectX 12<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
Cyberpunk 2077<\/td>\nDirectX 12<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
F1 24<\/td>\nDirectX 12<\/td>\nBenkmark i integruar, pista Monaco (shi)<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
Hogwarts Legacy<\/td>\nDirectX 12<\/td>\nOCAT, udh\u00ebtim me karroc\u00ebn n\u00eb Path to Hogwarts<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
Horizon Zero Dawn Remastered<\/td>\nDirectX 12<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
Metro Exodus<\/td>\nDirectX 12<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike; Shading Rate: 100%<\/td>\n<\/tr>\n
Red Dead Redemption 2<\/td>\nVulkan<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
Returnal<\/td>\nDirectX 12<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n
Total War: WARHAMMER III<\/td>\nDirectX 11<\/td>\nBenkmark i integruar (Mirrors of Madness Benchmark)<\/td>\nMax. cil\u00ebsia grafike<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Loj\u00ebra me gjurmim rreze<\/th>\n<\/tr>\n
Loj\u00eb<\/td>\nAPI<\/td>\nMetoda e testimit<\/td>\nCil\u00ebsimet grafike<\/td>\nSkalimi i kornizave<\/td>\n<\/tr>\n
AMD<\/td>\nIntel<\/td>\nNVIDIA<\/td>\n<\/tr>\n
Alan Wake 2<\/td>\nDirectX 12<\/td>\nOCAT, lokacioni Bright Falls<\/td>\nMax. cil\u00ebsia grafike, cil\u00ebsi e lart\u00eb e gjurmimit rreze<\/td>\nFSR Balance<\/td>\nFSR Balance<\/td>\nDLSS Balance + Ray Reconstruction (+ Generimi i Kornizave)<\/td>\n<\/tr>\n
Black Myth: Wukong<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike dhe gjurmimit rreze (Path Tracing)<\/td>\nFSR Balance (+ Generimi i Kornizave)<\/td>\nXeSS Balance\/FSR Balance + Generimi i Kornizave<\/td>\nDLSS Balance (+ Generimi i Kornizave)<\/td>\n<\/tr>\n
Cyberpunk 2077<\/td>\nBenkmark i integruar (OCAT p\u00ebr gjenerimin e kornizave)<\/td>\nMax. cil\u00ebsia grafike dhe gjurmimit rreze (Path Tracing)<\/td>\nFSR Balance (+ Generimi i Kornizave)<\/td>\nXeSS Balance\/FSR Balance + Generimi i Kornizave<\/td>\nDLSS Balance (Transformer Model) + Ray Reconstruction (+ Generimi i Kornizave)<\/td>\n<\/tr>\n
F1 24<\/td>\nBenkmark i integruar, pista Monaco (shi)<\/td>\nMax. cil\u00ebsia grafike dhe gjurmimit rreze<\/td>\nFSR Balance (+ Generimi i Kornizave)<\/td>\nXeSS Balance\/FSR Balance + Generimi i Kornizave<\/td>\nDLSS Balance (+ Generimi i Kornizave)<\/td>\n<\/tr>\n
Hogwarts Legacy<\/td>\nOCAT, udh\u00ebtim me karroc\u00ebn n\u00eb Path to Hogwarts<\/td>\nMax. cil\u00ebsia grafike dhe gjurmimit rreze<\/td>\nFSR Balance<\/td>\nXeSS Balanced<\/td>\nDLSS Balance + Ray Reconstruction (+ Generimi i Kornizave)<\/td>\n<\/tr>\n
Indiana Jones and the Great Circle<\/td>\nOCAT, lokacioni Sukhothai<\/td>\nMax. cil\u00ebsia e grafik\u00ebs, cil\u00ebsi e lart\u00eb e gjurmimit t\u00eb rrezeve (Path Tracing); Hijet e Gjurm\u00ebs s\u00eb Rrezeve: Vet\u00ebm Drita e Diellit<\/td>\nFSR Balance (+ Generimi i Kornizave)<\/td>\nXeSS Balance\/FSR Balance + Generimi i Kornizave<\/td>\nDLSS Balance (Transformer Model) + Ray Reconstruction (+ Generimi i Kornizave)<\/td>\n<\/tr>\n
Metro Exodus Enchanced Edition<\/td>\nBenkmark i integruar<\/td>\nMax. cil\u00ebsia grafike dhe gjurmimit rreze<\/td>\nN\/A<\/td>\nN\/A<\/td>\nDLSS Balanced<\/td>\n<\/tr>\n
Returnal<\/td>\nBenkmark i integruar (OCAT p\u00ebr gjenerimin e kornizave)<\/td>\nMax. cil\u00ebsia grafike dhe gjurmimit rreze<\/td>\nFSR Balance (+ Generimi i Kornizave)<\/td>\nXeSS Balance\/FSR Balance + Generimi i Kornizave<\/td>\nDLSS Balance (+ Generimi i Kornizave)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

N\u00eb shumic\u00ebn e loj\u00ebrave, treguesit e mes\u00ebm dhe minimal (ne tregojm\u00eb percentilin e par\u00eb t\u00eb shp\u00ebrndarjes) t\u00eb frekuencave t\u00eb kornizave del nga grumbulli i koh\u00ebs s\u00eb renderimit t\u00eb kornizave individuale ose frekuenc\u00ebs s\u00eb menj\u00ebhershme t\u00eb korniz\u00ebs, e marr\u00eb me an\u00eb t\u00eb benchmark-ut t\u00eb integruar. P\u00ebrjashtim b\u00ebjn\u00eb loj\u00ebrat q\u00eb nuk kan\u00eb benchmark t\u00eb integruar dhe testet q\u00eb aplikojn\u00eb gjenerimin e kornizave: n\u00eb k\u00ebto raste, p\u00ebr kapjen e intervaleve nd\u00ebrkornizore p\u00ebrdorim programin OCAT.<\/p>\n\n\n\n\n\n\n\n\n
Aplikacionet e pun\u00ebs<\/th>\n<\/tr>\n
Aplikacioni<\/td>\nBenchmark<\/td>\nCil\u00ebsimet<\/td>\n<\/tr>\n
Adobe Premiere Pro 25.x<\/td>\nPugetBench for Premiere Pro 1.1.1 (pjesa testuese n\u00eb lidhje)<\/a><\/noindex><\/td>\nStandard (4K)<\/td>\n<\/tr>\n
Blender 4.x<\/td>\nDemo Agent 327 Barbershop nga faqja Blender<\/td>\nRenderuesi Cycles<\/td>\n<\/tr>\n
Blackmagic Design DaVinci Resolve Studio 19.x<\/td>\nPugetBench for DaVinci Resolve 1.1.1 (pjesa testuese n\u00eb lidhje)<\/a><\/noindex><\/td>\nStandard (4K); Modaliteti i Kodimit H.264\/HEVC: Auto<\/td>\n<\/tr>\n
Aplikacione CAD<\/td>\nSPECviewperf 2020 v3.1<\/td>\nZgjidhja e ekranit: 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
Dekodimi i videos (ffmpeg 5.x)<\/th>\n<\/tr>\n
Formati<\/td>\nRezolucioni<\/td>\nParametrat e kodimit<\/td>\nAPI<\/td>\n<\/tr>\n
H.264 (YUV 4:2:0, 8 bit\/kanal)<\/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\/kanal)<\/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\/kanal)<\/td>\n1920 \u00d7 1080<\/td>\nN\/A<\/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\/kanal)<\/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
Kodimi i videos (ffmpeg 5.x)<\/th>\n<\/tr>\n
Formati<\/td>\nRezolucioni<\/td>\nParametrat e kodimit<\/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\/kanal)<\/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\/kanal)<\/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\/kanal)<\/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

P\u00ebrmasat e energjis\u00eb s\u00eb kartave grafike regjistrohen ve\u00e7mas nga CPU dhe komponent\u00ebt e tjer\u00eb t\u00eb PC-s\u00eb p\u00ebrmes pajisjes NVIDIA PCAT. Si ngarkes\u00eb p\u00ebr testet e energjis\u00eb dhe nivelit t\u00eb zhurm\u00ebs p\u00ebrdoret loja Cyberpunk 2077 me zgjidhjen 3840 \u00d7 2160 dhe parametrat maksimale t\u00eb cil\u00ebsis\u00eb grafike (pa ndri\u00e7im prej rrezeve), si dhe testi i stresit FurMark me cil\u00ebsime m\u00eb agresive (zgjidhja 3840 \u00d7 2160, MSAA 8x). Masat e t\u00eb gjitha parametrave zhvillohen pas ngrohjes s\u00eb kart\u00ebs grafike, kur temperatura e GPU-s\u00eb dhe frekuencat e taktimeve stabilizohen.<\/p>\n

\u279c<\/a><\/noindex>#<\/a><\/noindex>Pjes\u00ebmarr\u00ebsit e testimit<\/strong><\/h2>\n

Testimi i performanc\u00ebs p\u00ebrfshiu kartat grafike si m\u00eb posht\u00eb:<\/p>\n