TazmnianDv@... wrote: >What? Only thousands of Plaintiffs? >Intel, PC Makers Sued Over P4 Performance >http://www.pcworld.com/news/article/0,aid,104075,tk,dn081602X,00.asp ;-) Hmm -- looks like I missed out on a piece of that action. When the Willamette P4 came out nearly 2 years ago at 1.4 and 1.5 GHz and with 256kB L2 cache it was expensive and actully a poorer performer on the existing code than the then fastest PIIIs -- (X87 FPU particularly) . The RDRAM memory that Intel paired the P4 with was way more expensive (eg 6x) than PC 133 SDRAM at that time. The verdict of the PC hardware sites was -- don't buy one at this stage. Athlon's T'birds were way better for the money -- there were some heating and chipset/compatibility issues which scared a few DAW people but lots of fast cheap DAWs ran on Athlons. With the arrival of 0.13 micron Northwood P4 with 512KB of cache, much higher clock speeds (2.8 Ghz coming out in the next few days) faster memory bus (533 Mhz) and fast and now cheaper memory (RDRAM and DDR) this situation no longer applys. The SSE2 units on P4 (which are analogous to the Altivec units on G4s) are now being exploited by a lot more software (eg Logic/Cubase/Nuendo and soon PSP plugs ). When all/most of your favourite software has been revised to use the new instructions, the performance advantages of the P4 architecture over the PIII architecture become evident. Sure -- the PIII is still faster clock for clock but this architecture will not scale to the speeds of P4 and AthlonXP/Hammer because it lacks the memory bandwidth and the deep pipelines. Intel is actually making a kind of PIII/P4 hybrid called Banias for low power applications like laptops and low power servers. Banias is more or less a PIII core with the much faster P4 memory bus grafted on and it is due to hit the fan early next year starting at 1.6 GHz or so. Of course when Intel brought out the P4 they didn't say -- Just wait a couple of years people it's gonna be great. Nevertheless P4 is currently the best game in town and has been for a while now. AMD's Hammer may change that and it will also support the P4's SSE2 instructions. Another thing to consider -- Intel are by all reports going to enable multi-threading on a single CPU in new chipsets due this year. Apparently Northwood CPUs support this capability already but current chipsets don't. The OS sees a "hyperthreading" CPU as 2 CPUs and so an OS version capable of supporting more CPUs is necessary under Windows NT/2000/XP -- perhaps Intel and M$ will work something out here. It will be interesting to see what impact (if any) this capability will have on music apps like Cubase/Nuendo. The end Logic development on the Windows platform will almost certainly mean Logic never supports multi-processing or multi-threading on Windows. Regards, Murray
Message
Re: [L-OT] Intel, PC Makers Sued Over P4 Performance Claims
2002-08-20 by Murray McDowall
Attachments
- No local attachments were found for this message.