815 tested on tomshardware

[The PIT]

The first test I've seen on this chipset if you believe Tom (No further comment) the Bx is still king.
In the meantime I think I've found one of w2k bugs, the dialup networking. This keeps messing up my username and password. Sigh at least I didn't have to re-install Norton antivirus this time.

[paulcs]

After years of providing us with faster and cheaper components, Intel and VIA have obviously decided that what we want and need are slower and more expensive parts. But look at all the fine features we'll be getting:

The Audio Modem Riser! I've yet to see anything you can actually stick in the AMR slot, but it's nice to know it's there, replacing a potentially useful PCI or ISA slot. Remember, these devices are software-based, and therefore use more CPU resources while delivering less in terms of performance. The correct use of AMR devices should guarantee a slower machine, and we *want* to go slower.

Your Choice of an Integrated i753 or S3 Display Adapter! What better way to get your computer to run noticably slower than these two solutions? Neither Intel or S3/Diamond couldn sell the damn things on the retail market, because we were previously mislead to believe faster was better. Couple Intel's leadership in the graphics performance arena with VIA's and S3/Diamond's long traditions of cutting-edge driver support and we should be back to 286 levels of performance in no time.

Slower is better!

(I'm sure Dr. Moore is spinning in his grave, and he's still alive! )

Paul
paulcs@flashcom.net

[RoGuE]

LOL

[Himself]

Hilarious.

[paulcs]

SCSI is a different animal, I think. It's a high end solution, used primarily for servers and by peformance buffs or users with devices that require it. It's not like a huge company is trying to impose the SCSI standard on anyone. In fact, the opposite seems to be true. It's always been expensive, and the learning curve seems pretty steep for your average upgrader.

There are plenty of arguements for and against SCSI in a desktop system. But looks who is having the arguement. Computer wonks. It's kind of an elitist (and I mean that in the best sense of the word) standard. Few efforts are being made to mainstream it, and if it wasn't for scanners and CD burners, it would be less common in desktops than it already is.

On the other hand, some people believe Intel is ramming Rambus down the throat of an unwilling public. Intel doesn't want Rambus in high end systems only. They want Rambus in *every* system, which will, in effect, create yet another monopoly.

Fortunately or unfortunately, depending on your perspective, Intel and Rambus have dropped the ball repeatedly, and the Cape Cod recall can be interpreted yet another example of Intel tripping over its own big, feat.

Paul
paulcs@flashcom.net

[The PIT]

Brilliant Paulcs. You forgotten to mention the random crashes with via designed to keep the driver programmers working longer hours. Plus the Intel memory manager that causes lockups to keep the helpdesks and techies busy.
Now in future for "instant on" you'll need the via xt266 and intel 8200 (With mutliple crashing memory hubs)and either an Amd 9000 Marthon or Piv Goldmine 8999 processor. Qdr sdram or Bumbus 2050 memory.

[Joxx]

paulcs,

SCSI is a high end solution mainly for servers and workstations, Rambus is NOT a high end solution for servers and workstations. Intels upcoming high end chipsets for servers and workstations does not support rambus mainly because it offers bad performance and it is too expensive.

For high end solutions intel has choosen DDR-SDRAM...

[This message has been edited by Joxx (edited 15 June 2000).]

[paulcs]

That was my point.

Paul
paulcs@flashcom.net

[Rags]

Concerning RDRAM and servers, the reason for not going with Rambus is simple. The more modules you add to the board, the higher the latency becomes. 1, 2, and 3 Rambus modules compared to a DDR SDRAM modules in a server situation, the RAMBUS will outperform overall, this is also true for having larger sized RDRAM modules. But add 4+ modules, and SDRAM starts to perform SIGNIFICANTLY better. We all know that servers usually have a whole slew of memory, thus RDRAM is NOT a good choice.

Rags



------------------
Partnership for an idiot free America

[paulcs]

Let's get our priorities straight here. Will it make Quake go faster?

Paul
paulcs@flashcom.net

[This message has been edited by paulcs (edited 16 June 2000).]

[The PIT]

Nope but it'll make Intels bank balance better.

[xortam]

RAMBUS stock holders are pretty darn happy today too.