Ross Technology? Let's talk! Message Board

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.

Technology Stocks : Ross Technology? Let's talk!

RTEC 28.50

-0.5%

Oct 25 5:00 PM EST

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: Peter Bernhardt who wrote (154)	11/27/1996 11:22:00 AM
From: Sean McGee	of 475

<<That said, I'd like to hear more about the potential threat to Solaris-based computing from newer generations of x86 processors (particularly from the engineers reading this board).>> If we look strictly at processor integer performance, the Pentium Pro does well when compared to an UltraSPARC. According to my numbers (from the CPU info center at Berkeley) the PPro 200Mhz is slightly better (<10%) than the UltraSPARC at 200Mhz (however the source for the PPro numbers in Intel itself and you never know if you can ever get their numbers on a real machine so let's just say it is a wash). The glaring difference is in the Floating Point performance. UltraSPARC has almost twice the SpecFP95 performance of the fastest PPro processor. That is the workstation market right there. If you are doing something which requires FP performance you are going to buy a workstation and not a PC. FP performance is required for Mechanical design, circuit simulation, mathematical analysis, etc. One big piece of the equation that is hidden in these details is that the FP performance is largely based on how quickly you can get the data into the processor for processing. Some of the Spec95 datasets are larger than 4MB which tend to blow out most caches so memory system performance plays a significant part in how high your FP95 number will be. PC systems tend to lack the data throughput of a workstation class machine, so if you are working with large data sets you are probably going to get better performance on a workstation. The other point I would make about PC's killing workstations is that you must be careful to compare apple to apple (no pun intended). When you deck out a PPro system with 64MB and a 21" monitor, graphics card, SCSI hard drive and network card it winds up costing about $5000. Sun is offering it's 142MHz Ultra I system for under $7000 now. 64MB is not exactly a large amount of memory either. Most serious workstation users want at least 256MB. As these things factor in the price difference between the PPro system and the workstation tends to diminish. If I'm going to give up half my FP performance and have to use Windows NT I'm going to want to save a whole lot more than $2000. :) The last point I will make is that from what I have heard, Windows NT is not yet a suitable replacement for Unix. I think that the lack of a serious PC operating system with a large software base is primarily what has held back the PC from being used more seriously as an engineering tool. That could change in the near future, it may not. There's an awful lot of momentum behind the SPARC architecture and the workstation market in general. I wouldn't write it off just yet. -Sean

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next