SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Advanced Micro Devices
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Elmer who wrote (49529)2/16/1999 7:29:00 AM
From: wily  Read Replies (1) of 1573127
 
Elmer,

Here is an even more extreme example of the lopsided performance gain possible with dual cpu's. It is part of a review of the Tyan Tiger motherboard on this site:

sysopt.com

How much of a performance increase does dual CPUs give? Well, I spent some time trying to find a test that could translate back to Win95 performance and that most people could understand. Since any testing must be efficiently multithreaded to clearly show the potential gains, normal games would not suffice. Added the fact that I was running in NT and game scores tend to seriously lag behind on an NT platform, I needed a different benchmark. Solution, RC5 block rate. If you are not familiar with RC5, click here. Basically, the RC5 contest is about distributed processing. People download clients that each work on a section of the encrypted data's potential keys. It has full support for dual CPUs and is mostly processor intensive. Both tests were run on PII 266 MHz systems running NT 4.0 with service pack 3. 


                                     RC5 Block Rate  
Single PII   
                   266MHz            740 kkeys/sec 

Dual PII  
                  266 MHz  
                           CPU #1   730 kkeys / sec 
                  266Mhz  
                           CPU #2  1070 kkeys / sec 
                       Dual Total  1800 kkeys / sec



This is about a 240% increase in performance over the single processor system. Why is the second processor 146% faster then the first? It all has to do with how NT handles threads and tasking for the OS. The 300 kkeys/sec loss is a result in OS overhead that is lost in every single CPU system.

The author points out that the second cpu isn't involved in O/S tasks and that this explains the added performance of the second cpu. This benchmark may be more pronounced in showing the duals' advantage (in multi-threaded apps) since it is designed to be cpu intensive (I believe it is designed to use your cpu 100% 24/7).

AnandTech reports excellent overclocking results on the new batch of Malay Celerons:

anandtech.com

Tom's Hardware Guide speculates that the PPGA celerons may soon be bus-locked:

www5.tomshardware.com

Regarding RTIII and multi-threading (if anybody's interested), a friend sent me this:

The application may not purposely create worker threads to do work concurrently. {...}the application uses a separate thread for each and every window that you open in RT3. This is pretty standard to have a separate thread for each UI (User Interface - window or dialog) component to keep the UI very responsive to the user even if a lot of work is going on.

So, as you can see I have every incentive to plug away at this project. I'm still in the process of choosing and gathering components.

wily

Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  

HomeMailHotSubjectMarksPeopleMarksKeepersSettings
Terms Of UseContact UsCopyright/IP PolicyPrivacy PolicyAbout UsFAQAdvertise on SI
© 2025 Knight Sac Media.  Data provided by Twelve Data, Alpha Vantage, and CityFALCON News