Recent content by del42sa

it is relevant, because he was trying to show you, that Radeon Cores 2.0 = VLIW 4 graphic ;)

what features ?

JF-AMD is not working anymore for AMD

There´s plenty of things to fix in Bulldozer µarch. Not only slow caches, write-throught policy, but also weak front-end and fetch and FPU. Agner Fog did great work to summarize all weak points in BD.µarch. : Agner Fog BD arch. other, but very important is this...

it´s quite funny, while GloFo abandoning SOI for 28nm, TSMC seems go opposite route ;) GloFo is movin from Gate First to the Gate Last, TSMC goes for Gate First instead. Mad world :D ibm-and-globalfoundries-go-gate-last-for-20nm FD SOI TSMC TSMC 28nm FDSOI

Bulldozer bottlenecks by Agner: http://pctforum.tyden.cz/viewtopic.php?p=8289810#p8289810

Sure we can, or maybe AMD engineers are on crack, given that poor BD result ;)

AMD should change cache write policies first ! Bulldozer L1 write speed is way slower then any other processor! How can it happen? Remember that Bulldozer L1 cache use a write-through design, while Phenom II L1 cache is a write-back one. Write-through caches imply that each write is immediately...

I am not sure, if anyone post this before: http://semiaccurate.com/forums/showthread.php?t=5993

I have question about cache: Why L1 cache uses write-through instead of write-back approaches, when write-through cache means worse results ? (not to mention, that L1 Data Cache shrunk to only 16kB) What impact to performance have shared write coalescing cache between L1 and L2 cache in terms...