>>>If it is a K6 and the motherboard has a Super Seven socket (and BIOS and chipset that supports the K6-2), the easiest upgrade would be to slap in a K6-2. Keep in mind that you will need PC100 RAM if you go with the K6-2 300 or up.
>>>
>>>200 MHz...can't be a K6-2 as far as I know.
>>>
>>
>>The description is vague enough that it could be; it doesn't state that it's Super7, and if it isn't, well, let's just leave it as a huge mistake.
>>
>
>Well, sure, if AMD ever made a 200 MHz K6-2 processor. I don't think they did. Are you saying they did? Trivial argument, but I think I am right.
>
>>>Also, regarding where/what to buy: I have a home-based business (part-time) and I can put together better systems for a better price and with better service (at least to home users; to business' I warn them that I am not available 9-5) than any of the big companies. And I generally have better warranties (2 or 3 years minimum on everything, some stuff longer). If you decide to go with a non-brand name, just make sure you are getting a good mainboard and good components. I prefer Asus mainboards.
>>
>>I don't have a home-based, part-time business. I build my own systems because I consider myself to be a hobbyist, and I'm probably more knowledgable about the aspects of current hardware than you migth suspect.
>
>I don't suspect anything...in fact, I don't really care. I will just comment on what you say here.
>
>>My preference runs to the SuperMicro boards; I use their P6DGS and P6DGU both at home and at the office for small workgroup servers. My clients deal with hundreds, or in at least one case, thousands of systems, and a small business selling the motherboard du jour isn't going to be able to repeatedly crank out the systems they need in the course of a year, much less provide the support. From a TCO standpoint, I've found that name-brand, business-targeted systems make more sense for a small business, because of the range of support options and vendors offered.
>>
>
>Are you suggesting that Asus motherboards are a "motherboard du jour?" They have long been considered the best motherboard out there by many experts. And among the best by all the other experts.
Yes, I consider the Asus to be a board du jour, in the same way I consider other 'major' vendor's boards (companies like SuperMicro, Tyan, Abit) to be; they are treated as interchangable, the vendors aren't good about keeping a given revision of a board available, and I have to wade through several layers to get to an engineer if I really need one.
I really consider Super 7 to be the dead horse of the year. Super 7 is a single processor architecture; with the likely direction of Wintel operating systems heading for NT, I want an architecture that can scale to >1 CPU. The boards that I buy for myself are dual-processor boards, on the theory that even if I can't buy a CPU enhancement that will give me 1.5x the real throughput of a single processor now, I can certainly get that level of performance from a second CPU and an SMP kernal under NT.
As for Asus being 'the best', we differ in opinion. I like the P2B, but I'd rather have the GX chipset, which both Tyan and SuperMicro have been shipping for several months in both Slot 1 and Slot 2 flavors. The Abit B6's soft configuration features make it an attractive, stable BX platform, with, IMO, a better BIOS than either Asus or SM ship today, and for the overclocking crowd (I'm not a fan of overclocking) may be the single best buy available.
>Sure, a small business could not support thousands of systems, so a company with thousands of computers possible should choose a different route. But small computer shops can easily support a few hundred machines. And the good shops provide better service than the name-brand vendors.
>
During normal business hours, maybe; there's no comparison for the home and small business user who need basic phone support after hours. The large vendors can afford to staff a help line 24/7. In the case of the server marketplace, the big guys have an advantage, too; most of them can keep spares for their systems within 4 hours of a customer site. I probably can roll my own low-end server around a P6DGS or DGU (or the S2DGU if the customer wants to go Slot 2) and an ARO1130CA RAIDPort and get better bang for the buck (that's the configuration I run at home, a P6DGS with dual PII/400s, 256MB of ECC memory, two 2GB CP2107s as a mirrored pair for the boot/system partition, and a 4 drive array of Quantum Atlas IIIs in a striped set with parity, controlled by the AR01130CA), but it's unlikely that I can afford to keep all the necessary spares to keep that server up with <4 hours downtime in mission-critical environments. If nothing else, I have to troubleshoot the system on site before I can get the system back up and running. HP, Compaq, Micron and Dell can roll in a complete, new system to meet service requirements and troubleshoot the hardware off-site at their leisure. You can be up and running as quickly as the last complete backup can be put on the new box.
If I have a need for a strong service staff, a major vendor that has a lot of depth, in terms of spare parts and staff, offers me a better chance of keeping things up and running.
>For a lot of companies, a small shop provides a lot more than a large manufacturer can provide. Can you call Dell and say "What ISP is a good one in town?"? Will Dell come over and set-up their network? A good computer shop has people with a wide scope of knowledge and can provide a complete solution.
>
Actually, yes, Dell can refer you to an ISP - they own a couple. They will come and set up your network for you for a price. And they have greater depth of knowledge on their staff going for them. I can guarentee that if I have a service contract with HP on an LH3, and I call in with serious trouble, they can connect me with an MCSE or CNE at almost any time of night or day, regardless of vacation schedules, personal emergency or illness. And they can get me spares on virtually any hardware that I bought through them.
>>The advantages of volume make, in the long run, the small vendor a really bad idea for a business that doesn't have unusual hardware requirements. That's an opinion formed from being on several sides of the issue, including having done custom hardware integration and design, IT management for a small-sized, highly automated business, and consulting for some fairly large organizations.
>
>Volume of what? Hardware? If that is what you meant, are you only talking about the point you already made (and that I agree with -> that a large 1000+ computer organization should go with brand-name)? But lets give an example that I am familiar with: A large corporation has set a company policy to buy only a few certain models of
machines. They have in-house computer technicians, and felt that with standardization, the techs would be more efficient. Well, two machines with the same model number and specs can have two different motherboards, hard drives, CD-ROM drives, video cards, BIOSs, etc. etc. So they haven't standardized at all. In fact, IMHO (and many of the knowledgeable employees), they now have a bunch of over-priced and under-powered computers with only a one year warranty. The machines I sell (same as some other small computer shops), for instance, are backed by a 3 Yr P&L warranty (some parts longer) by a company that has been around since 1983 and is
>doing quite well financially (ie. they should be around in 3 years).
>
A three year life on hardware is not very realistic. This year's high-end 450MHz PI/BX chipset/AGP video/UDMA drive workstation will be below the entry-level curve within two years; BIOS and chipset advances will make a new piece of equipment, with a new warranty, a better buy in terms of new software and standards. I may upgrade my own stuff (as I said, I consider myself to be a hobbyist, and spend far too much time playing with sand) but I wouldn't buy a piece of equipment today on the basis of future upgrades - I made that mistake with Socket 8. How many of the people who bought Pentium Pros actually bought the OverDrive? I didn't; it was less expensive to buy a new BX or GX chipset board and a 400MHz P II than it was to replace the Pro with the Pro Overdrive, which gave you the equivalent of perhaps a P II/333, without the advantages in the newer chipset and BIOS.
>BTW, I've done custom hardware integration. I do IT management for a small-sized, highly automated business. I haven't done consulting for large organizations. But I have for a number of small companies.
>
>>
>>That's why I recommended the Celeron 300A and 333 as the baseline for entry-level systems. It's easy to find stable, inexpensive toasters with an 18 month to two year life span; IMO, the Super7 marketplace is a waste of time, especially with the introduction of Socket370 and some higher-end Celeron processors, even though they're limited to a 66MHz FSB. Buying below the curve; systems like the lower-end Compaqs, Aptivas and even HPO Pavillions is counterproductive. As mentioned, I can get an HP model that I like with plenty of processing power for under $800. Why jerk around trying to mix and match components when I can get a toaster off the sheld that does what I want, on a reproducible basis. It's the same logic as the T-34 tank in WWII.
>>
>
>I am not familiar with tanks.
>
>I would bet that HP jerks around with mix and match components (see example above). Try to get Dell to guarantee a particular brand of hard drive. Or Compac. Or ... Only difference is that when you buy a brand-name machine, often it is 6-month old technology because that is how long it take them to get the product to market.
>
>Look, I can give exact examples of better machines at better prices with better warranties.
>
>I am not saying "A small shop is ALWAYS better." I am saying "A small shop is SOMETIMES better."
>
And I'm not saying "A big vendor is ALWAYS better." I am saying that "A small shop is rarely better, unless unique hardware or software requirements exist that override the advantages offered by a large vendor."
>Does HP recommend a BX board and PC100 memory so the system can be upgraded to a PII 450 in a year for a hundred bucks or so? I do. Value Added.
>
Several major vendors, including HP, have offered processor upgrades for their systems, ranging from daughterboards that went (in theory anyway) from the P54C to the P55 to Slot 1. No one in their right mind did it, because by the time the cost of the upgrade became economically attractive, it would only bring the machine up to what was the entry-level box at the time of the upgrade. If Katmai/Pentium III/whatever does take off, and PC98 becomes a viable standard, and vendors change standard RAM types agains (a distinct possibility, since RDRAM, RAMBus and the like will be needed to get better throughput to memory as FSB speeds increase; we're already close to the limits of SDRAM/SyncDRAM implementations today with PC100. I've got a bunch of money in my own UW SCSI stuff, including UW2; FireWire and FCHAN are going to probably move us away from the parallel copper SCSI implementations used in high-end systems today, even with the flexibility gained by the LVD UW2 cabling scheme.
Eventually the buying public will realize that piecemeal upgrading isn't going to keep them near the baseline of new, entry-level systems. Processor upgrades stopped being attractive, significant performance enhancements which cost less than subsystem replacement with the 486 - the OverDrive debacle, and the brief attempt by companies like AMD to produce clock-quadrupled processors to keep up with the low-end Pentiums, without the advantages of the new architecture represented the real end of single-component upgrades. In most cases, when a new processor technology became available after the DX4/100, the motherboard needed to change to take fullest advantage of the enhancements..
