Never really internalized before that 4K pages affect performance by limiting the total addressable memory by the TLB without a miss.
-
-
@cmuratori does it compensate for the less number of entries and the incresed cycles penalty?? http://www.7-cpu.com/cpu/IvyBridge.html … -
@cmuratori Anyway i'll personally ramp up even more then 2MiB, internal fragmentation is no longer an issues imho
End of conversation
New conversation -
-
-
@cmuratori I did a rasterizer working with tiles. Tried to avoid linearizing mem per tile. Used pad to make L123 ok. Vtune taught me abt TLB -
@anders_breakin@cmuratori it's so annoying that all the virtual addressing stuff isn't just completely free magic :( mental model fail! ;) - Show replies
New conversation -
-
-
@cmuratori Yup. KNF got significant perf wins from using 64k, 2M pages because of this. KNC grew # of 2M TLB entries even further. -
@tom_forsyth pity we'll never know the TLB entries for KLF.#AncientsOfMuMu - Show replies
New conversation -
-
-
@cmuratori Even for file paging, 2MB makes more sense than 4kB. Maybe 64kB would be OK. -
@jwatte@cmuratori OSes usually page at much higher granularities anyway. IIRC, Windows uses 64k. - Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.