Author: Tom Kerrigan
Date: 11:33:25 06/10/03
Go up one level in this thread
On June 10, 2003 at 13:13:28, Robert Hyatt wrote: >On June 10, 2003 at 12:56:58, Tom Kerrigan wrote: > >>On June 10, 2003 at 11:05:53, Robert Hyatt wrote: >> >>>On June 10, 2003 at 02:37:56, Tom Kerrigan wrote: >>> >>>>On June 09, 2003 at 22:31:48, Robert Hyatt wrote: >>>> >>>>>On June 08, 2003 at 17:29:30, Tom Kerrigan wrote: >>>>> >>>>>>On June 08, 2003 at 08:25:17, Peter Berger wrote: >>>>>> >>>>>>>On June 08, 2003 at 07:43:51, Michael P. Nance Sr. wrote: >>>>>>> >>>>>>>>Tell Me how You think that a P/C with ONLY 650 mhz and ONLY 512 OF Ram is even >>>>>>>>worth considering? Wouldn't You think that a Computer like that is >>>>>>>>obsolete?>>>>Mike >>>>>>> >>>>>>>Those are not PCs so you can't simply compair the MHz numbers if you want to >>>>>>>compair speed/performance. They are not obsolete, but I agree you wouldn't want >>>>>>>to buy one (only) for computerchess at all :). >>>>>> >>>>>>You're right, an UltraSPARC IIi MHz is worth less than a Pentium 3, Pentium 4, >>>>>>or Athlon MHz. :) >>>>>> >>>>>>-Tom >>>>> >>>>> >>>>>I'm not sure that is totally true. But the problem is they don't make those >>>>>3+ gigahertz processors. They are so far behind they will never catch up. And >>>>>I really don't believe they intend to try. >>>> >>>>No, it's true. According to SPEC 2k submissions, the US-IIi is the slowest >>>>processor you can buy (per GHz) except for the US-IIe. >>>> >>>>http://www.aceshardware.com/SPECmine/index.jsp?b=0&s=2&v=4&if=0&r1f=2&r2f=0&m1f=0&m2f=0&o=0&o=1 >>>> >>>>Even the Pentium 4 gets slightly more SPECints/GHz, the difference being that >>>>the P4 runs at 3GHz and the IIi runs at 650MHz. Whoops... too bad for Sun. >>>> >>>>-Tom >>> >>> >>>SPECINT is not the perfect test, however. The sparc _can_ do 64 bit operations, >>>which means it gets more per instruction than a PIV, for applications that need >>>64 bits. IE 64 bit adds, etc. >> >>Well, sure, the US is a completely 64-bit chip. It may do more per instruction >>but it's still in-order so it doesn't necessarily do more per clock. Also, >>imagine a P4 running at 650MHz... no more memory bottleneck, so it'll perform >>WAY better per GHz. >> >>-Tom > > >Its "in order" but it is still super-scalar, which means optimizing turns it >into a more than one instruction per cycle processor. For certain cycles, but it can't average more than one instruction per clock. The P3 averages ~1.2 uop retires/clock on SPECint2k, so I imagine the P4 averages <= 1.0, and the P4 still does better than the US-IIi per clock, even with the bigger memory bottleneck. SPARCs have always been an embarrassment. You know your RISC architecture sucks when it can't keep up with a 486. Here are TSCP MIPS/MHz: 68020: 0.2 MicroSPARC: 0.658 SPARC: 0.665 486: 0.678 MicroSPARC II: 0.816 UltraSPARC II: 0.88 Pentium: 0.904 Athlon: 1.117 Pentium 4: 1.134 Pentium II(I): 1.15 -Tom
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.