This looks right to me. The UltraSPARC-IV chip contained 2*UltraSPARC-III "chips" bolted together to make a dual-core (not dual-thread) CPU.
The chip numbering will be based on the position of the chip within the system. If I were to guess,I'd say this is a dual-board SFx800 system, whose numbering is 0,1,2,3 for System Board 0 (SB0) and 8,9,10,11 for System Board 2 (SB2). These numberings are effectively determined by the location of the chip within the system for core 0. Given the number of cpus and numbering of them, I'd guess this is a SF3800, but that's just a guess :-) Core 1 of each chip was catered for by adding 512 to the cpu number of core 0 from the same chip. Thus "cpus" 0 and 512 were chip 0 (512 - 512 == 0), 1 and 512 were chip 1 (513 - 512 == 1), etc.... As far as I can see, psrinfo and kstat agree in this respect: CHIP 0 from psrinfo: The physical processor has 2 virtual processors (0 512) <==== 0 & 512 UltraSPARC-IV (portid 0 impl 0x18 ver 0x23 clock 1200 MHz) CHIP 0 from kstat: chip_id: 0, core_id: 0 chip_id: 0, core_id: 512 CHIP 1 from psrinfo: The physical processor has 2 virtual processors (1 513) <==== 1 & 513 UltraSPARC-IV (portid 1 impl 0x18 ver 0x23 clock 1200 MHz) CHIP 1 from kstat: chip_id: 1, core_id: 1 chip_id: 1, core_id: 513 ...etc... This is how the numbering works for US-IV systems. If you have any further questions, I would suggest opening a support call to address them. Hope this helps, Brian JoeBilish wrote: > Hi All. > > I'm not sure if the is the right mail list to post this. > > I have a Solaris 10 box with 8 UltraSPARC iV 2 way threads. > > And kstat and psrinfo provide differnet information about core. > > Seems the "core_id" get from kstat is incorrect. Is this a bug of kstat? > > Thanks for your help in advance. > > # psrinfo -pv > The physical processor has 2 virtual processors (0 512) > UltraSPARC-IV (portid 0 impl 0x18 ver 0x23 clock 1200 MHz) > The physical processor has 2 virtual processors (1 513) > UltraSPARC-IV (portid 1 impl 0x18 ver 0x23 clock 1200 MHz) > The physical processor has 2 virtual processors (2 514) > UltraSPARC-IV (portid 2 impl 0x18 ver 0x23 clock 1200 MHz) > The physical processor has 2 virtual processors (3 515) > UltraSPARC-IV (portid 3 impl 0x18 ver 0x23 clock 1200 MHz) > The physical processor has 2 virtual processors (8 520) > UltraSPARC-IV (portid 8 impl 0x18 ver 0x23 clock 1200 MHz) > The physical processor has 2 v irtual processors (9 521) > UltraSPARC-IV (portid 9 impl 0x18 ver 0x23 clock 1200 MHz) > The physical processor has 2 virtual processors (10 522) > UltraSPARC-IV (portid 10 impl 0x18 ver 0x23 clock 1200 MHz) > The physical processor has 2 virtual processors (11 523) > UltraSPARC-IV (portid 11 impl 0x18 ver 0x23 clock 1200 MHz) > > chip_id and core_id get from kstat: > > chip_id: 0, core_id: 0 > chip_id: 1, core_id: 1 > chip_id: 2, core_id: 2 > chip_id: 3, core_id: 3 > chip_id: 8, core_id: 8 > chip_id: 9, core_id: 9 > chip_id: 10, core_id: 10 > chip_id: 11, core_id: 11 > chip_id: 0, core_id: 512 > chip_id: 1, core_id: 513 > chip_id: 2, core_id: 514 > chip_id: 3, core_id: 515 > chip_id: 8, core_id: 520 > chip_id: 9, core_id: 521 > chip_id: 10, core_id: 522 > chip_id: 11, core_id: 523 > > > ------------------------------------------------------------------------ > > _______________________________________________ > opensolaris-discuss mailing list > opensolaris-discuss@opensolaris.org -- Brian Ruthven Solaris Network RPE (Sustaining) Oracle UK
_______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
