SuperMicro on desktop for X86-64?

[JonMarsh]

It's amazing what information is starting to pop out of the woodworks about upcoming X86-64 systems (AMD)- especially from some of the most unlikely places.

Let's not even talk about the Sun and Dell rumors- let's go to something a little more obscure....



One of the MB brands which used to be my main favorite, especially because of their inlcusion of high performance SCSI and other cool features was Super Micro. My interest in them has waned in the last few years, what with their becoming a hard core intel house, and focussing relentlessly on the server market, including 2 and 4 way products. Hard to blame them, there's money to be made there, and better margins. And by hardcore Intel house, I mean that they have been quoted as saying "We'll never make an AMD based system".

Well, folks, the times, they are a changin. Looks like they're going to be using a new BRCM/Serverworks chipset with Opteron glue (think Hypertransport), and not only in servers, but in desktower personal Workstations. So, if you want to get the real deal (not the consumerized mass market ready in September Athlon64 with wimpy cache and only one Hypertransport link), start saving your pennies, quarters, and dollars. It won't be cheap, but then you can be the first person on your block with a two or four CPU cluster Opteron Desktop with AGP8X/Pro and optional SATA and SCSI.

Whooo, the thought of it fairly takes my breath away! Talk about being able to sneer at your buddy running a lowly 3GHz P4 with Hyperthreading... yowsa! :B

http://www.theinquirer.net/?article=8832

~Jon




Earth First!
_______________________________
We'll screw up the other planets later....

[Bing Fung]

Jon, I'm sure you have one with your name on it already

So do you think it best to hold of any 32 bit OS purchases and get something like the 64bit Windows XP, in anticipation of the Opterons and the glut of 64bit chips to follow?

It's funny to think that we will be sneering at a 3GHz P4 with Hyperthreading, but's I guess that's just the facts... I came across a 4 year old PC magazine the other day, it was hilarious to look at all the top line systems of Pentium II's at a breakneck speed of 450Mhz... The K7 AMD was not even out yet as K6's were all that were in AMD's line up.

One more thing, those PII 450Mhz with 128mb RAM were priced at around $3K us :rofl:




Bing

[JonMarsh]

AMD has injected a level of competition in the PC market since the introduction of the K7 Athlon which hadn't been seen for years, if ever. If not for them, and the pressure they put on Intel, we'd probably just now be getting to 1 GHz PIII CPU's.....

Now, seriously, for most plain folks, there is no immediate need for 64 bit desktop computing. If you have to do development for high end databases, or do demanding mechanical modeling, or large circuit electrical simultation (moi), then there could be some valid reasons other than future proofing. I bought my first 386 in order to be able to run a special version of SPICE for 32 bit on the PC, and that made a huge jump in capabilities- it would actually simulate PWM circuits, including a Class D amp I was working on, faster than my Sun workstation in the office. That was in '87. For a lot of applications, the jump from 16 to 32 bits was very significant in enabling capability; it proably won't be so significant for 32 to 64 bits, if only because most people don't really need a mainframe on their desktop! But with 64 bit processor and OS, and 2.5 GHz plus CPU speeds, believe me, that's what we'll have by any significant historical definition. I bet that 4 way Opteron box will even come up against the current DOD restrictions on exporting computers capable of designing and simulating nuclear weapons- performance bars which have been raised many times in the last ten years, because otherwise, our current desktops meet those capabilities as described legally ten years ago.

The Athlon64 will be 64 bits for the masses, at a very modest cost. Opteron desktops will be 64 bits for the hard core guys at very reasonable cost. I mean, we're talking about the potential for very serious problems for Intel, here, if they have "loyal" MB builders offereing 2 way and 4 way Opteron desktops, when the Itanium cost structure and Intels plans won't be supporting Itanium on the desktop in the next year or two. There's a HUGE cost differential in the cost to implement an Itanium with associated power pod, versus an Opteron, which is just basically a high power Athlon with a whole lot more pins....

And the leaked benchmarks to date show Opteron's VERY competitive in performance with Itanium, only being bested in floating point, but coming out on top in Integer benchmarks... this has gotta be keeping some folks at Intel up late at nights.... and they'd better bring out something more effective than their version of the "Republican Guard". Meanwhile, Otellini continues to fiddle away, denying any possibility of an X86 Yamhill...

Remember the old saying about history and those who forget it are doomed to repeat it? How many of you remember the Intel i432? What's that you say? It was Intel's first 32 bit processor: big, expensive chip, totally different instruction set from the 8086, totally incompatible. Sound familiar? Anyone remember the i860? Hot floating point, otherwise incompatible. A lot of add on boards were sold by Microway to engineering geeks for use as a numerical processor, with special compilers; never became any kind of "mainstream solution", though there were motherboards built to support them.

How many billions has Intel spent on Itanium so far? And how many processers have they sold? Last year I hear was a bit over 10k, a fairly significant ramp over the previous two years? How much money was essentially given away with each processor in the first and second generation?

If the Itanic was an introduction by an outside company, brought in as a replacement/alternative to Intel's Xeon offerings and IBM's PowerPC, would it have any traction in the marketplace at all?

My recomendation, is that for PC's for most folks, 32 bits will be OK for the rest of this year. The cost premium for Athlon64, once it's out, will be so small that for protection from future obsolesence, if you're a tekkie or gaming nut or in software development, you shouldn't buy anything short of an Athlon64 this year- otherwise, you may be replacing it in a year. PC's really should have a working life of three years or more for most folks; it's not healthy (financially) when that isn't the case. I've been rotating my main desktop almost yearly, but the systems get handled down to other applications, like my daughter, an HTPC, or to friends. So the cost (and waste) isn't what it might seem.

~Jon




Earth First!
_______________________________
We'll screw up the other planets later....

[JonMarsh]

And another Opteron workstation chipset is nearing release....

http://www.theinquirer.net/?article=8927

This sort of activity raises the possibility of a dual "consumer" path for X86-64- the original marketing designated model, based on the "consumer desktop" Athlon64, and higher performance systems (albeit at higher cost) based on the Opteron "server" CPU's.

Interesting, interesting.

~Jon




Earth First!
_______________________________
We'll screw up the other planets later....