Silicon Investor (SI) -- The First Internet Community

STOCKTALK

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 -- Ignore unavailable to you. Want to Upgrade?

To: Petz who wrote (83313)	12/16/1999 7:37:00 PM
From: Tenchusatsu	Read Replies (1) \| Respond to of 1582527

Petz, <How much is a multitude of banks? Each open bank draws more current than the rest of the memory. Thats why i820 and i840 had to reduce them. I'm just saying that a multitude of banks might not be enough, you might need two or four multitudes to handle hundreds of tasks.> The study of DRAM performance characteristics can get rather complicated, and I don't feel very qualified to act like an expert on such things. So take what I say with a grain of salt. In servers, banks are closed a short time after they are accessed (a.k.a. auto-precharge). Page hits are rare in servers. Therefore, keeping a bank open has very little performance benefit (it could even hurt performance a little). Having a multitude of banks in servers, however, is a good thing because when you have heavy loads of traffic from multiple sources, there's less of a chance that a given bank will be overloaded with multiple accesses at once. If two different but concurrent accesses hit different rows in the same bank (or even the same row in the same bank), one will have to wait while the other is handled. That creates a lot of dead time in between accesses, which translates into wasted bandwidth. Having a large number of banks minimizes this possibility. <Its clear from the tomshardware.com review of the OR840 motherboard that streaming data applications can benefit significantly from RDRAM :-) but can they get the cost down to something reasonable?> RDRAM on the 840 is an excellent choice for workstations. The limitations that I mentioned before regarding RDRAM in servers don't really apply to workstations. The size limit isn't an issue, since workstations don't usually need more than one or even two GB of memory. And once the price of RDRAM comes down to a few notches above where DDR will be, the cost difference of 1-2 GB will not be that much. (Hopefully no more than a $500 difference per gigabyte.) I'm not as excited about 840 as a server chipset, but hey, I guess that's where RCC fits in. Tenchusatsu

To: Petz who wrote (83313)	12/17/1999 2:37:00 AM
From: Paul Engel	Read Replies (1) \| Respond to of 1582527

Petz - Re: "streaming data applications can benefit significantly from RDRAM :-) but can they get the cost down to something reasonable? " When will AMD get the cost of the 750 MHz AthWIPER DOWN to something reasonable ? Did it ever occur to you that the 750 MHz AthWIPER will - in due time - drop from RAMBUS range of $800 to around $200 ? It will happen - it will - just like RAMBUS prices will drop from $800 to $200 (per 128 MByte). So where is your OUTRAGE for AMD not dropping THEIR ATHWIPER prices right NOW to $200 - where hard working, honest Americans and Citizens of the WORLD can easily afford them? Why does AMD keep ripping off John Q Public and Joe SixPack? Huh ? And what was the IBM AThWIPER Model number you saw at Fry's last week ? I have asked you this 5 times now and your lack of response tells me we caught your little PUTZ in a big "FALSEHOOD"!!!! Paul