Plateforme Level Extreme
Abonnement
Profil corporatif
Produits & Services
Support
Légal
English
Abysmal table open times on one workstation
Message
Information générale
Forum:
Visual FoxPro
Catégorie:
Autre
Divers
Thread ID:
00480266
Message ID:
00480638
Vues:
10
>>>>I don't know if this will help, but check out www.emulators.com/Pentium4.htm. There are all kinds of problems apparently. I've heard that in some instances, through put slows down to as little as 200 Mhz. From what I can see, the chip is a dog and should be registered with the AKC. Right now, my plan is to buy a new Athlon machine.
>>>
>>>Wow, that's a pretty revealing article re: P4s. It doesn't help me to determine if the processor is the problem, but it makes me feel good that the last two machines I have built for myself have been Athlons.
>>
>>The reason I mentioned this was the various caches (or rather lack thereof) on the P4. Frankly, I'm clueless here, but I'm wondering if VFP's internal caching tied to these changes might be part of the problem. About the only way to find out is to try a different box (read PIII) similarly configured. Of course, setting that up might not be practical.
>
>Only indirectly. The L1 and L2 caches internal to the processor are independent of the memory address space visible to the CPU. What does affect them is the size of the current task's working set - the range of memory actively in use for code and data, and VFP often has a large working set. If the working set is bigger than the L2 cache, then frequent updates of the L2 cache from main memory will occur. If the main working segment of the working set is larger than the L1 cache, then L1 has to update from L2 (which may need to update from main RAM) and this causes stalls in the execution stream, which are what makes the P4 under-perform in many cases. The L2 cache size issue is what differentiated the original Celeron from the Celeron 'A'; the Athlon and Xeon both invest lots of real estate in on-board cache, which helps their performance tremendously (they do get hot, quickly, though.)


>Until we see code specifically optimized for the P4, I'd avoid it.

I was hoping you'd jump in here. So basically, it is likely that that level of performance you might expect to see from the higher clock speed won't be present. Is that correct?
George

Ubi caritas et amor, deus ibi est
Précédent
Suivant
Répondre
Fil
Voir

Click here to load this message in the networking platform