Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: G4+ branch misprediction

Author: Eugene Nalimov

Date: 22:00:19 01/29/02

Go up one level in this thread


POWER4 is really very good processor, but you must read fine print before
trusting published POWER4 Spec2k results. Their wonderful submitted result is a
result from 8 CPUs system. Yes, only one CPU run the benchmarks, but it used
shared L3 caches of all the 8 CPUs – 128Mb of L3 cache total :-)

Eugene

On January 29, 2002 at 23:48:32, Tom Kerrigan wrote:

>On January 29, 2002 at 15:30:51, Dan Andersson wrote:
>
>>Sorry! I was less that clear in my wording. I eluded to the ratio between the
>>pipeline lengths of the different G# CPU's. And the ratio is definitelly larger
>>than one, moreso due to the previous very low stage count. I also realize that
>>it is only one of the factors slowing it down. Ramping a low power embedded
>>processor is close to madness, IMO. Go MIPS Apple, gawd'dagnit!
>
>I don't think MIPS would do Apple any good. The 500MHz R14000 is about as fast
>as a 700MHz Pentium 3. The GHz G4s are faster than that. If Apple wants to get
>serious about faster computers, they need to make a deal with IBM to get a
>personal computer version of the POWER4. That would be awesome.
>
>-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.