Seriously is there really a difference

[Dark1]

Before this thread becomes a Amd vs. Intel debate, I am only looking for the opinions of those that work on both platforms at comparable speeds. I am about to build a computer and I was wondering could you really see (how noticeable is it) the difference between Amd (2000+or 2100+) and Intel (Northwood 2.2 or 2.26 533fsb). I?fm not planning to overclock and I?fm not planning to upgrade for at least a year. The programs that I using right now and I?fm sure that I?fll be using programs just like them a year from
Now are Dreamweaver 4, Flash 5 (soon I?fll be using MX on both), Adobe Photoshop 7,Premier 6,divx and Return to Castle Wolfenstien (I?fll be playing any fps for ID or Epic)
Other then those apps 30% of the time I?fm going to be surfing the web. If you guys can help me with my decision it would be greatly appreciated. The specs for the rest of the system are as follows
AMD
MOBO: SOYO Dragon Ultra KT333
INTEL
MOBO: ABIT IT7 MAX

The rest of the parts will be the same either way I go.
CHIEFTEC Server) Model DX-01WD-420W; Turbo Link Power supply
Samsung Original PC-2700 512mb CAS2.5
1 60 gig 7200 rpm Maxtor drive
2 20 gig 7200 rpm Maxtor drives
Geforce 4TI 4200(might wait on this one to see how well the R300 performs)
Lite-On 16x DVD Model LTD-163 Black
Lite On 32x12x40 CDRW Model LTR-32123S BLACK Color
MITSUMI D359M3 BLACK FDD 1.44MB 3.5INCH
TextText

 

[busmaster11]

You shouldn't notice a difference in performance between the two, really. The P4's SSE2 will become a noticeable boost in performance once more proggies are developed to take advantage of them, but then, we've been saying that for years.

Everything you've chosen is good, except for two things... I'd recommend an nForce 415-d based motherboard coupled with standard pc2100 ddr. You will notice no benefits of the KT333 - especially when combined with the latencies of an asynchronous system:memory bus speeds. It's a dead end, and VIA makes some very suspect chipsets.

 

[Dark1]

Thanks, I?fll take that into consideration

 

[Siddhartha]

I do not think you would notice a big real world difference between the two systems. I have not had any problems with the VIA based motherboards I used in the last two years or so. I could be lucky but I doubt it.

 

[Pabster]

Busmaster11 wrote:

"It's a dead end, and VIA makes some very suspect chipsets."

To say the least.

Yes, there seriously is a difference. The motherboard selection in the Socket478 arena is considerably better than what is available for Socket A. And Pentium 4 Northwood, for a number of reasons, is a better choice than AthlonXP. (Thermal management being high on the list.)

If you're looking for a super-speedy, rock-solid machine, I recommend the Northwood 1.6A coupled with ASUS P4B266(-C). Intel's i845-D is a rock solid chipset. Yes, it is a few percentage points slower in synthetic benchies than other P4 chipsets, but, IMHO, the tradeoff is worth it.

Avoid VIA like the plague.

 

[JungleJoe]

The real difference you can see is the price difference between Athlon XP 2100 and P4 2.2 Ghz...

 

[Dreadogg]

Go AMD/Nforce all the way dont be silly!

 

[Scipionix]

I'm convinced it's personal preference, unless you have to have that 2.53.

 

[Mem]

VIA makes some very suspect chipsets.

There`re quite a few KT333 and KT266a boards out that are very good,I`m using the Epox 8KHA+(VIA KT266A) board which is rock solid,don`t listen to Pabster and his disciples he`s well known for being anti-VIA period,ask for feedback from members here that have KT333 or KT266a boards for honest opinions.

I`ve yet to have a problem with my board period.

 

[MadRat]

<<Busmaster11 wrote:

"It's a dead end, and VIA makes some very suspect chipsets."

Pabster replied:

"To say the least.">>

Both of you are truly misinformed. VIA has done well in the chipset arena, but you harp on one PCI bug that has everything to do with the cardmakers inflexible support of PCI, not specifically VIA's implementation. VIA is not the only southbridge chipset maker to leave that option out of their PCI design. Your logic basically is: "If there is one bug then that company should never make chipsets again!" Using your logic we should can avoid all past and future Intel, ALi, SiS, and nForce chipsets.

<<Pabster wrote:

"Yes, there seriously is a difference. The motherboard selection in the Socket478 arena is considerably better than what is available for Socket A. And Pentium 4 Northwood, for a number of reasons, is a better choice than AthlonXP. (Thermal management being high on the list.)

If you're looking for a super-speedy, rock-solid machine, I recommend the Northwood 1.6A coupled with ASUS P4B266(-C). Intel's i845-D is a rock solid chipset. Yes, it is a few percentage points slower in synthetic benchies than other P4 chipsets, but, IMHO, the tradeoff is worth it.">>

I would recommend Pentium4 right now for two reasons, neither of which has to do with Pabster's recommendations. 1) The motherboard will support processors in the future that have higher speed grades than present Socket-A design. There is no garauntee that .13 T-breds will work in existing Socket-A boards. 2) The Pentium4 runs significantly cooler if you are using the Northwood design. The overall system temperature will be a few degrees lower, but the cost to run the system over several years will be enough difference in my opinion to warrant the P4. Cost of ownership, especially running costs, is an important consideration in my opinion, because my systems run 24/7.

<<Pabster wrote:

"Avoid VIA like the plague.">>

Go ahead. Some of us will use VIA in the future and be perfectly happy with doing so.

 

[Pabster]

MadRat wrote:

"Both of you are truly misinformed. VIA has done well in the chipset arena, but you harp on one PCI bug that has everything to do with the cardmakers inflexible support of PCI, not specifically VIA's implementation. VIA is not the only southbridge chipset maker to leave that option out of their PCI design. Your logic basically is: "If there is one bug then that company should never make chipsets again!" Using your logic we should can avoid all past and future Intel, ALi, SiS, and nForce chipsets."

Let's see. I'm misinformed? Let's do a quick Google check:

Query1: "VIA chipset issues" produces ~46,700 hits.
Query2: "SiS chipset issues" produces ~13,400 hits.
Query3: "Intel chipset issues" produces ~69,900 hits.
Query4: "ALi chipset issues" produces ~6,420 hits.
Query5: "nForce chipset issues" produces ~2,790 hits.

Now, do a quick sampling of the first couple pages of hits in each bank. Intel's hit total is high due to the sheer number of chipsets and (more so) the fact that, unlike the others, they actually admit to errata and publish it. Also, the figure includes issues relating to software (INFs, IAA, et al.)

It doesn't take a rocket scientist or a brain surgeon to comprehend this stuff. I'll reiterate. Just how many VIA chipset problems/issues are written about just here @ AT on a daily basis? And no, it isn't just "old" VIA chipsets. I'm seeing quite a proliferation of issues with the "latest and greatest" KT333 boards as well. Care to do a quick search and look at the facts, or will you continue to ignore the obvious and bash me for "VIA hating"? I don't hate VIA. I do refuse to purchase their shoddy core logic products, and I actively discourage others to do so. But I don't hate them; I just want them to admit to the friggin' problems and, more importantly, FIX them. There's no excuse for the same errata appearing in chipset after chipset, revision after revision. Period.

And, please, don't start with the "it was just one 'little' PCI bus issue..." We both know that wasn't (and is not) the case. VIA's chipset errata spawns a plethora of areas well above and beyond the PCI bus.

"I would recommend Pentium4 right now for two reasons, neither of which has to do with Pabster's recommendations. 1) The motherboard will support processors in the future that have higher speed grades than present Socket-A design. There is no garauntee that .13 T-breds will work in existing Socket-A boards. 2) The Pentium4 runs significantly cooler if you are using the Northwood design. The overall system temperature will be a few degrees lower, but the cost to run the system over several years will be enough difference in my opinion to warrant the P4. Cost of ownership, especially running costs, is an important consideration in my opinion, because my systems run 24/7."

To each their own Certainly, the thermal management area is where Pentium 4 (Willamette or Northwood) holds its greatest advantage over the competition. Northwood's .13u process, lower TDP, and Intel's amazingly quiet retail HS/F all combine for a very pleasant experience. I can't speak about electricity consumption, except to say that when I swapped my PVR rig from an XP 1800+ to a 1.6A (running 2.4 with ease) the UPS load (with all other components identical) dropped from 67% to 51%, under full load conditions. Running 24/7, it could certainly add up. But probably not enough for you to purchase that new NV30 or anything

"Go ahead. Some of us will use VIA in the future and be perfectly happy with doing so."

Like I said ... to each their own. I can't force anyone to do anything, and I wouldn't want to. Buy whatever you like and I hope it works well for you. But I'll continue to offer my experiences and provide as much information as possible to help prospective purchasers make informed decisions. I can't justify spending my hard earned money on flawed core logic. But that's just me. Do whatever you like

 

[Dark1]

thanks guys

 

[MrGrim]

Pabster I think you are putting way too much effort into these discussions. State your opinion, share your knowledge and experienceand and then let them decide if you are right or wrong. I used to be just like you but now I have a new moto:

"Somebody has to buy the crap stuff so that the decent stuff stays cheap".

 

[tweakmm]

Does anybody else find it funny that pabster doesn't have an Intel box but has 5 AMD boxes?

 

[Dark1]

Not to stray to far from the subject, but what do you guys think of the rumors that the T-breds will be released Monday at PR ratings of 2400+ and 2500+
and that the new Athlon (Clawhammer) will be released at the end of August. I'm sure the prices of the Athlon and P4 series will come down considerably in the next couple of weeks. If so I?fm upping my choices to Athlon (2400+) and P4 (2.4 @533 fsb). As for the Hammer I?fll wait a year for better mobo bundles to come out (i.e. onboard raid possibly SATA raid, IEEE 1394b at 800 mbps, and Dolby Digital 5.1 sound or whatever new standard from Dolby) and to make sure that all the bugs if any have been fixed (hey this new Architecture I?fm sure not every legacy component or app will play nicely with it). By that time (late 2003 or early 2004) Microsoft will make their intentions known about BLACKCOMB (Windows 2004/5, Windows XP 2 or whatever the heck they?fll call it) in regards to native x86-64 support for home versions of windows.
:|

 

[Pabster]

tweakmm wrote:

"Does anybody else find it funny that pabster doesn't have an Intel box but has 5 AMD boxes?"

You're actually the first to notice my rig profile, evidently. I haven't updated it for some time. The only AMD rig I still own is my MP3 jukebox machine (XP1800+, K7S5A). The other 4 are P4 Northwood machines. I just haven't bothered to update the rig list.

 

[Hendrik]

You're actually the first to notice my rig profile, evidently. I haven't updated it for some time. The only AMD rig I still own is my MP3 jukebox machine (XP1800+, K7S5A). The other 4 are P4 Northwood machines. I just haven't bothered to update the rig list.

Pabster has joined the Dark Side!

(The CPU in his jukebox machine must be one of the most under-utilized pieces of electronics on the planet!)

 

[Dark1]

yes I'm bumping this again

 

[busmaster11]

Originally posted by: MadRat
<<Busmaster11 wrote:

"It's a dead end, and VIA makes some very suspect chipsets."

Pabster replied:

"To say the least.">>

Both of you are truly misinformed. VIA has done well in the chipset arena, but you harp on one PCI bug that has everything to do with the cardmakers inflexible support of PCI, not specifically VIA's implementation. VIA is not the only southbridge chipset maker to leave that option out of their PCI design. Your logic basically is: "If there is one bug then that company should never make chipsets again!" Using your logic we should can avoid all past and future Intel, ALi, SiS, and nForce chipsets.

VIA has done well due to their relationship with all the major oems, AMD, and until their P4 chipset design, Intel. Historically they've been among the first to come out with the newest designs, and this allows them the deepest market penetration. People give them credit for this, despite the fact that their intial attempts are pretty much the hardware equivalent of crippleware.

Remember the original PC133 Slot/SocketA chipset? Yup. No official 133 FSB support for Athlons as promised- only memory bus.

Remember the original KT266? Slow arse POS... Couple tiny mods and the KT266a actually competes. Instead of realizing that VIA had seriously underperformed in their initial attempt, people like you will remember them as innovators, and remember only how they "leapfrogged" the competition on their second attempt, the KT266a. They should be known for underachieving in their initial attempt.

VIA has proven they cannot be trusted. IMO they have even broupt down the reputations of Asus and Abit until they wised up and offered competition from nVidia.

 

[MadRat]

Busmaster11-

You have some seriously selective memory.

<<VIA has done well due to their relationship with all the major oems, AMD, and until their P4 chipset design, Intel. Historically they've been among the first to come out with the newest designs, and this allows them the deepest market penetration. People give them credit for this, despite the fact that their intial attempts are pretty much the hardware equivalent of crippleware.>>

VIA's market is in the low to mid range, not in the top end. They do not use near as complex of designs as the original equipment makers, therefore they can eek out a meager living. Your idea of crippleware is looked upon as "more affordable" to others. I have no problem with SiS, ALi, and VIA earning a living with stripped down chipsets.

<<Remember the original PC133 Slot/SocketA chipset? Yup. No official 133 FSB support for Athlons as promised- only memory bus.>>

Because there were no 133fsb Athlon CPUs when VIA released the first Athlon chipset, duh. VIA's northbridge was pushing 120MHz whereas AMD's was only pushing 110MHz, so obviously the VIA was the better overclocker.

<<Remember the original KT266? Slow arse POS... Couple tiny mods and the KT266a actually competes.>>

The KT266A was a total redesign whereas the KT266 was an early design of a DDR-compatible chipset. The KT266 and KT266A share model numbers but not much else.

<<VIA has proven they cannot be trusted. IMO they have even broupt down the reputations of Asus and Abit until they wised up and offered competition from nVidia.>>

You message came out pretty garbled. Perhaps you can restate what you meant. Back up your claims, too, instead of posting garbage ike "VIA has proven they cannot be trusted." Seems to be that nForce has some bugs in its design that cannot be overcome with firmware upgrades, which really makes your statement look like total horsesh!t.

 

[Rainsford]

The only difference I see is the cost. If you are worried about VIA chipsets (and I'll admit I am just a little bit, even though I've never had a problem with them), get an nForce board, the 415-D based boards are good.

Some of what Pabster said is true, there is a greater variety of P4 boards out there, many of which are rock solid and very stable. But an Athlon XP 2000+ will beat a P4 1.6a (as long as you don't overclock, which you say you aren't) in almost every aspect of computing. Also, I see thermal management as a null issue because as long as you mount your heatsink correctly you shouldn't have any problems.

A P4 system does have some advantages, but I see cost as a bigger issue than any of those Pabster brings up. JMHO.

 

[busmaster11]

Originally posted by: MadRat


VIA's market is in the low to mid range, not in the top end. They do not use near as complex of designs as the original equipment makers, therefore they can eek out a meager living. Your idea of crippleware is looked upon as "more affordable" to others. I have no problem with SiS, ALi, and VIA earning a living with stripped down chipsets.

Stripped down? exactly what core features exist on Intel/AMD chipsets that VIA doesn't claim? What did Iron Gate do that the KX133 didn't? How exactly was irongate more "complex"?

<<Remember the original PC133 Slot/SocketA chipset? Yup. No official 133 FSB support for Athlons as promised- only memory bus.>>

Because there were no 133fsb Athlon CPUs when VIA released the first Athlon chipset, duh. VIA's northbridge was pushing 120MHz whereas AMD's was only pushing 110MHz, so obviously the VIA was the better overclocker.

The point was everyone was expecting explicit support.

<<Remember the original KT266? Slow arse POS... Couple tiny mods and the KT266a actually competes.>>

The KT266A was a total redesign whereas the KT266 was an early design of a DDR-compatible chipset. The KT266 and KT266A share model numbers but not much else.

LOL... Completely contrary to anything I've read. Can you or anyone back this up?

<<VIA has proven they cannot be trusted. IMO they have even broupt down the reputations of Asus and Abit until they wised up and offered competition from nVidia.>>

You message came out pretty garbled. Perhaps you can restate what you meant. Back up your claims, too, instead of posting garbage ike "VIA has proven they cannot be trusted." Seems to be that nForce has some bugs in its design that cannot be overcome with firmware upgrades, which really makes your statement look like total horsesh!t.

There's been exactly one complaint (about PCI IDE read performance being lower than write performance for some odd reason) thats been officially acknowledged that the 415 didn't resolve AFAIK. I couldn't even begin to list the ones about VIA's 266a.

As for my comment - well I've owned about five boards with VIA based Athlon chipsets, each of them from Asus, Abit or MSI. All but one had serious issues, from performance to that funky IDE/LIVE! issue to other IDE controller related issues to random lockups related to the 4in1 drivers. It's also ridiculous to have to put up with phantom "SCSI" devices that device manager detects when none exists. All my Asus/Abit/MSI boards that weren't VIA performed relatively... Win9x issues notwithstanding.

The KT333 is a pretty transparent scheme to squeeze every last penny out of the newbie hardware enthusiast, promising memory performance 20% greater, but as benchmarks prove, the latency issues pretty much even it all out. Then there's the fact it requires expensive PC2700 DDR, which may or may not be a dead end but AMD certainly isn't releasing Athlon support for it...

I feel sorry for those people buying KT333 based boards expecting great performance. It may edge out the nforce by a negligible margin but in no way is the risk of instability and the added cost worth it.

Madrat if you wanna keep boasting your own POS just because you made the mistake of buying it, thats fine. (Don't feel too bad, I made that mistake five times) But I consider my message a public service.

 

[MadRat]

<<Stripped down? exactly what core features exist on Intel/AMD chipsets that VIA doesn't claim? What did Iron Gate do that the KX133 didn't? How exactly was irongate more "complex"?>>

For one thing the VIA chipsets have half to one-fourth the buffering within their chipsets, which tends to be the single biggest difference. The southbridges were cut down even more, especially when you consider the whole "VIA/SB Live! bug" was from a stripped down PCI design. The backbone of communication between north-south bridges tends to be thinner on VIA designs, too. The pinouts of the chipsets tend to be fewer on VIA, SiS, and ALi chipsets; fewer pins means simpler layout for motherboard makers. The simple truth is that these chipset makers have room to make money on the margins of cost between their own chipset design and that of AMD/Intel.

<<The point was everyone was expecting explicit support.>>

No, actually they were not. The 133fsb Thunderbirds did not even exist at the time, so there was never a credible reason to believe they would support 133fsb. AMD had to release the specifications for the new bus before VIA would have been able to do a 133fsb design.

<<((The KT266A was a total redesign whereas the KT266 was an early design of a DDR-compatible chipset. The KT266 and KT266A share model numbers but not much else.))
LOL... Completely contrary to anything I've read. Can you or anyone back this up?>>

You haven't looked very far then. You can save us all time by posting links to where you got the idea that these were the SAME design only rebadged. The onus is on your original claim that they were identical. VIA claims to have improved the memory controller and tightened their design specifications, obviously in an attempt to outperform the SiS735 chipset. Funny how the KT266A's release seems to have made the SiS735 obsolete.

Quote from Anand: "The VT8366A (the original KT266 North Bridge was the VT8366) has two major enhancements: an improved memory controller and deeper internal buffers." link

<<There's been exactly one complaint (about PCI IDE read performance being lower than write performance for some odd reason) thats been officially acknowledged that the 415 didn't resolve AFAIK. I couldn't even begin to list the ones about VIA's 266a.>>

You have it backwards, WRITE performance in the nForce chipset is crippled. Not just crippled, but restricted to ATA33 speeds of 20MB/sec.

<<As for my comment - well I've owned about five boards with VIA based Athlon chipsets, each of them from Asus, Abit or MSI. All but one had serious issues, from performance to that funky IDE/LIVE! issue to other IDE controller related issues to random lockups related to the 4in1 drivers. It's also ridiculous to have to put up with phantom "SCSI" devices that device manager detects when none exists. All my Asus/Abit/MSI boards that weren't VIA performed relatively... Win9x issues notwithstanding.>>

1. Asus, Abit, and MSI have had their troubles on different motherboards, just like every other manufacturer.
2. 4in1 drivers have resolved all of the other issues
3. phantom "SCSI" devices are the add-on ATA100 controllers
4. You ended the thought with a hanging modifier and no conclusion. "Relatively" to what?

<<The KT333 is a pretty transparent scheme to squeeze every last penny out of the newbie hardware enthusiast, promising memory performance 20% greater, but as benchmarks prove, the latency issues pretty much even it all out. Then there's the fact it requires expensive PC2700 DDR, which may or may not be a dead end but AMD certainly isn't releasing Athlon support for it...>>

NVidia's claims on their own memory performance of nForce chipsets are more farcical. Here is VIA's exact claim: "The VIA Apollo KT333 is the first VIA chipset to feature DDR333 memory offering 25% more memory bandwidth to the CPU." link

<<I feel sorry for those people buying KT333 based boards expecting great performance. It may edge out the nforce by a negligible margin but in no way is the risk of instability and the added cost worth it.>>

The reason people choose a KT333 is likely because of its ability to officially support DDR333. Even if its the same KT266A design with a new badge its important to the customer to know that it does support DDR333. At the same time VIA releases a VT8233A south bridge wih ATA133 support.

<<Madrat if you wanna keep boasting your own POS just because you made the mistake of buying it, thats fine. (Don't feel too bad, I made that mistake five times) But I consider my message a public service.>>

What exact POS have I boasted about? Oh, yeah, I never boasted. This claim is just more FUD from you.

 

[busmaster11]

Originally posted by: MadRat
<<The point was everyone was expecting explicit support.>>

No, actually they were not. The 133fsb Thunderbirds did not even exist at the time, so there was never a credible reason to believe they would support 133fsb. AMD had to release the specifications for the new bus before VIA would have been able to do a 133fsb design.

The expectation that AMD would goto a 133FSB was fully there at the time. Otherwise, and the end result was, it was only marginally better than irongate, which was a disappointment because it came later with more promises. The KX133a was everything people expected from the KX133. Funny how we can say the same about the KT266/266a.

<<((The KT266A was a total redesign whereas the KT266 was an early design of a DDR-compatible chipset. The KT266 and KT266A share model numbers but not much else.))
LOL... Completely contrary to anything I've read. Can you or anyone back this up?>>

You haven't looked very far then. You can save us all time by posting links to where you got the idea that these were the SAME design only rebadged. The onus is on your original claim that they were identical. VIA claims to have improved the memory controller and tightened their design specifications, obviously in an attempt to outperform the SiS735 chipset. Funny how the KT266A's release seems to have made the SiS735 obsolete.

Quote from Anand: "The VT8366A (the original KT266 North Bridge was the VT8366) has two major enhancements: an improved memory controller and deeper internal buffers." link

Let's finish the quote. Funny what you decided to leave out...

The end result is the VIA KT266A which from an architectural standpoint is nothing more than the original KT266 with an updated North Bridge. ...VIA is being very tight lipped about the exact improvements surrounding the new memory controller however we have some educated guesses...

I'm not an engineer. But I know what you call major enhancements can mean one of two things that aren't necessarily dependent on each other.
1. Major performance increase <--- This we know is true
2. Major REVISIONS in architecture <--- This is definitely argueable at best

As for the SIS735 - I'm not surprised. I don't doubt that because the KT266a is as much better than the 735 as the 735 is from AMD's and VIA's own KT266... And once the low end KT266a chipsets were able to be priced within spitting distance of the 735, people didn't need it anymore. VIA is still a bigger name than SIS.

<<There's been exactly one complaint (about PCI IDE read performance being lower than write performance for some odd reason) thats been officially acknowledged that the 415 didn't resolve AFAIK. I couldn't even begin to list the ones about VIA's 266a.>>

You have it backwards, WRITE performance in the nForce chipset is crippled. Not just crippled, but restricted to ATA33 speeds of 20MB/sec.

So do you know of any other ones?

<<As for my comment - well I've owned about five boards with VIA based Athlon chipsets, each of them from Asus, Abit or MSI. All but one had serious issues, from performance to that funky IDE/LIVE! issue to other IDE controller related issues to random lockups related to the 4in1 drivers. It's also ridiculous to have to put up with phantom "SCSI" devices that device manager detects when none exists. All my Asus/Abit/MSI boards that weren't VIA performed relatively... Win9x issues notwithstanding.>>

1. Asus, Abit, and MSI have had their troubles on different motherboards, just like every other manufacturer.
2. 4in1 drivers have resolved all of the other issues
3. phantom "SCSI" devices are the add-on ATA100 controllers
4. You ended the thought with a hanging modifier and no conclusion. "Relatively" to what?

1. No doubt. I'm just merely pointing out a pattern from my own experience. Take it as you may.
2. VIA + CL is still a combo many people refuse to venture into. Whose fault it is I won't judge. But in the time it took VIA to fix their 4in1's, mobos could and have gone obsolete... Whereas my nforce worked flawlessly out of the box. Interesting concept. Most people couldn't say the same about VIA boards till recently, if ever.
3. Not to mention the IDE devices on those controllers...
4. Typo there. The other mobos were relatively stable.

<<The KT333 is a pretty transparent scheme to squeeze every last penny out of the newbie hardware enthusiast, promising memory performance 20% greater, but as benchmarks prove, the latency issues pretty much even it all out. Then there's the fact it requires expensive PC2700 DDR, which may or may not be a dead end but AMD certainly isn't releasing Athlon support for it...>>

NVidia's claims on their own memory performance of nForce chipsets are more farcical. Here is VIA's exact claim: "The VIA Apollo KT333 is the first VIA chipset to feature DDR333 memory offering 25% more memory bandwidth to the CPU." link

Your link's not working. Do you deny any of my claims? Do you suggest that we actually see much of that 25% boost in memory bandwidth in benchmarks? Are you saying the added memory cost is actually worth whatever performance boost you get?

<<I feel sorry for those people buying KT333 based boards expecting great performance. It may edge out the nforce by a negligible margin but in no way is the risk of instability and the added cost worth it.>>

The reason people choose a KT333 is likely because of its ability to officially support DDR333. Even if its the same KT266A design with a new badge its important to the customer to know that it does support DDR333. At the same time VIA releases a VT8233A south bridge wih ATA133 support.

Two negligible things. Whats so special about DDR333? For today's cpus, we've seen the benchmarks, and their not impressive. For tomorrow's, well, you tell me who or what will support them. And ATA133? Well you've just managed to max out your PCI bus and more. And if you don't, who needs it?

You can't blame VIA for releasing the KT333 - its an evolutionary step that people should look into. Admittedly it does support a host of other features like USB 2.0 and AGP 8x... But overall it's just painful to watch when people here purchase it expecting hugh performance increases over their current DDR setups.

 

[bluemax]

So between this heated debate, what chipsets do you both RECOMMEND for being fast, featured, stable and reliable?
SIS 7xx?
Intel 8xx?