Athlon XP 2800+, AMD Hit 333!

Author : Wayne Date : 1st October 2002

Pov-Ray :

The Persistence of Vision Raytracer is a high-quality, totally free tool for creating stunning three-dimensional graphics. It is available in official versions for Windows, Mac OS/Mac OS X and i86 Linux. The source code is available for those wanting to do their own ports.

To give some idea of how this result shapes up, an Athlon 1.4 running on an Abit KG7 RAID Motherboard took 469 seconds to complete rendering while a P4 1.7GHz on an i845 board took 609 seconds. The 2800+ on the nForce2 turned in a stunning 288seconds!

Chess2 render time - 800x600 AA0.3

 

VillageMark :

Traditionally used as a DirectX graphics card benchmark VillageMark uses decent sized textures and lots of overdraw as it takes you on a flight around this imaginary virtual village. Both scores are impressive but in particular a score of 75FPS at 1600x1200 is a real eye opener.

 

 

Quake 3 Arena :

Running on the nForce2 Q3A really wasn't posing any great challenge to this system so I decided to whack the quality sliders all the way over. 66 plus frames per second at 1600x1200 with 8xAA and 16x anistropic filtering enables is just plain scary and shows how potent the 2800+ is when combined with nForce2 and the Radeon 9700 Pro.

 

Jedi Knight :

Despite its Quake 3 roots Jedi Knight is a vastly more demanding piece of code that once again showcases the nForce and 2800+ tag team.

Overclocking :

Being preproduction and silicon and due to it having to wend its way back to AD again I was fairly limited in what I was able to do on the overclocking front. I certainly wasn't in a position to deface the package with our usual conductive pain graffiti so it was a case of turning up the FHB until it choked. I didn't spend very long on this but did manage a fairly stable 170MHz (2295MHz) without a whole lot of fuss.

 

Conclusion :

It's been quite a couple of days here. Quite apart from the fact that I had to work through the night to get this and the nForce2 reviews completed (DHL is now an official swear word round here) I have had the privilege of handling the fastest single processor AMD platform on the planet. By combining the XP 2800+ plus with NVIDIA's nForce2, throwing in a couple of sticks of Corsair XMS3200 and piping it all through the Radeon 9700 Pro you have one awesome machine that's ready to power its way through anything you can throw at it right now. The fact that I have to return all of this bleeding edge kit borders on mental cruelty.......I wonder if I'd have a case with the court of human rights?

AMD have finally taken the plunge and done what we've all been asking of them for some time, they've bowed to the pressure from an anxious public who wanted to see them keep in touch and released a solid, high performance processor with a stable 333MHz FSB. So why not do it sooner? Well I'm guessing it's all down to marketing and product placement. Enthusiast have been running Athlons at 333MHz plus for some time so I don't believe for one second it was a technical issue that caused the delay. More likely it was the desire to keep prices low and not do anything that might take the shine off either Barton or Hammer.

The biggest negative I can find here is the fact that this is yet another phantom launch. You're not likely to be able to get your hands on either the 2700+ or the 2800+ or even the nForce2 come to think of it, until at least November/December and even then the 2800+ will be directed primarily towards the performance orientated OEMs. AMD say numbers should increase next year (2003) but they don't have plans to make it a major line and personally I think the 2800+ will prove to be rarer than sheep's eggs. What ever happened to the good old days when you read about something them went out and bought it a week or two later? I see absolutely no point in AMD publicizing this processor if they're not prepared to produce it in worthwhile numbers and I really hope they rethink the plan, the community wants and needs the fastest dies that AMD can pile on store shelves if they're not to jump ship.

Despite the fact that the 333MHz bus isn't bringing the kind of massive performance boosts we all thought we'd see it does offer an advantage and it does mean we can now run memory and CPU synchronously which has to be a good thing in the long run. 333MHz may not be a huge leap in muscle but it is a significant leap in stamina, what I mean by that is that although raw numbers don't see massive increases it's now a lot more consistent when the going gets tough. I still think something has been "relaxed" a little to increase stability, probably latency or pre-fetch aggression but I'd have to ask AMD about that, and will. The other potential issue was related to stability when used on the KT333CE driven 8K3A+. Initially the system refused to run using aggressive memory timings on a few but remarkably removing and se-seating the memory cured this?? This is all the more curious because the system would boot and run perfectly happily with memory set to "normal" in the BIOS but not "Turbo". I can only assume there was a slightly iffy contact somewhere and would also therefor guess that any kind of memory flaw (or using cheap memory even) will cause problems. Otherwise the system was quite happy even though the latest Epox BIOS doesn't correctly identify the 2800+.

I didn't mention thermal performance in the review because AMD tell me this is pre-production silicon and is not accurately reporting temperatures from its internal diode. I can say for fairly certain though that it runs cooler than the infamous first revision 2200+ and I don't see heat being a big problem with sufficient cooling. The HSF that AMD supplied coped perfectly and this was a very basic skived aluminium affair with a low profile fan on top.

All in all this is a very hot product that will become even hotter by the nature of its rarity. I did question my morality at a few stages, should I be publicly talking up a product that you guys can't even get your hands on? I mean I've just launched a 5GHz CPU with a 1066MHz FSB but unfortunately my only 10 samples went to IBM to power their latest number cruncher so you'll have to take my word for it. It's all getting a bit frustrating for the people who keep these companies in white shirts and Jaguars, (that's you by the way). It's just like the early days of the PC when standards were left to fight it out between themselves and the only way to be certain if product "A" would work with product "B" was to take it home and try it. The race for speed has left compatibility behind at the first bend and much as we all want the fastest rig money can buy I suggest we should learn to be patient and leave these guys to sort out their mess before we get too excited about anything that's launched.

Despite all that negative guff, which I must say is aimed primarily at VIA, JEDEC and anyone who publicly launches a product you've no hope of buying within the month, I still have to proclaim that I'm impressed by the 2800+ and I'm theoretically equally impressed with nForce2. Take a Radeon9700 (or presumably an NV30 shortly) with its cold boot issues and lack of AGP 8X support from just about all motherboards, plug it in to an nForce2 which you can't yet buy and which currently has a semi-crippled BIOS and throw in an Athlon 2800+ which is unlikely to be seen in the wild this side of Christmas and you have yourself one of the fastest systems that money can't buy. I think they should all take a week off then go back to their offices and start over. We don't want to read about these things, we want to use them. </rant>

<<< Back to Review Index >>>

Home