Kedis ku xidhan hagaha · Adag Heerka

FP8 iyo Kediska Qaababka Saxan Hoose

Ku celceli xulashooyinka jibbaha FP8 iyo mantissa, cabbiridda, iyo shuruudaha qalabka, oo ay ku jiraan sababta qaabka tirada yar aanu u dammaanad qaadin wax soo saarka isku midka ah.

Wadooyinka haga ee la xidhiidhaFp8 Iyo Qaababka Saxnaanta Hoose
Su'aal 1 ee 8

Immisa xabbo ayuu nambarka FP8 adeegsadaa marka la barbar dhigo lambarka caadiga ah ee FP32?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Immisa xabbo ayuu nambarka FP8 adeegsadaa marka la barbar dhigo lambarka caadiga ah ee FP32?
    • 8 bits vs 16 bits (nus)
    • 8 bits vs 64 bits (halkii sideedaad)
    • 8 bits vs 32 bits (rubuc)
    • 8 bits vs 8 bits (isku mid)
  2. Maxay yihiin calaamadaha 'E4M3' iyo 'E5M2' waxay qeexayaan qaabka FP8?
    • Heerka qaladka iyo cabbirka moodeelka
    • Tirada jibbaarada ee ka soo horjeeda mantissa bits
    • Xawaaraha saacada iyo kanaalka xusuusta
    • Waayihii tababarka iyo lakabyada moodeelka
  3. Waa maxay sababta dhuumaha FP8 u adeegsadaan qodobbada is-qarinta tenorsada?
    • Si loo sireeyo miisaanka amniga
    • Si loo kordhiyo tirada lakabyada moodeelka
    • Si loogu beddelo qaabka luqad kale
    • Si loo ilaaliyo qiyamka gudaha xadka cidhiidhiga ah ee FP8 lagana fogaado qulqulka/qulqulka hoose
  4. Waa kuwee ganacsi dhimisteeda ugu weyn ee isticmaalka FP8?
    • Saxnaanta tirada oo la dhimay (khaladaadka wareega)
    • Isticmaalka xusuusta sare
    • Isku dhufashada matrix tartiib tartiib ah
    • Awood la'aanta in lagu shaqeeyo GPU kasta
  5. Jiilkee NVIDIA GPU markii ugu horreysay ku daray taageerada qalabeed ee u go'ay xisaabta matrix FP8?
    • Pascal (P100)
    • Hopper (H100)
    • Kepler (K80)
    • Maxwell (M40)
  6. Waa maxay isbeddelka qaababka saxda ah ee ka gudbaya FP8?
    • Ku soo noqoshada 64-bit ee saxda ah ee labanlaabka u-fiirsashada
    • Tirtiridda barta sabaynaysa gebi ahaanba qoraalka
    • Isticmaalka FP32 kaliya ee dhammaan miisaanka moodeelka
    • Xataa qaababka cabbir-yar ee-yar-yar sida MXFP4 iyo NVFP4
  7. A tensor’s values need more dynamic range. Which common FP8 encoding allocates more exponent bits?
    • E5M2
    • E4M3
    • Both allocate zero exponent bits
    • Neither encoding uses an exponent
  8. Why can an FP8 result differ from the same operation using FP32 inputs?
    • Eight-bit input storage is automatically exact for every real value
    • FP8 removes all mathematical operations
    • The lower-precision input representation can introduce rounding error
    • Higher-precision accumulation restores every lost input detail