nforce2 vs KT333 - KT333 Hands Down Champ

[Treeburst155]

I've been playing around quite a bit the past couple days with my computer and spare parts. What I did was put together a machine with an MSI KT3V (KT333) board. I did a fresh install of Win XP, drivers, etc.. I then ran Comanche4 bench several times, and 3DMark2001SE. Once I had that setup performing as well as possible, I swapped the KT333 for an nForce2 (Epox 8RDA3+) with all the same sound, video, nic card, memory and cpu etc.. The only thing changed was the mobo and the HD. I did another install for the nForce2 on a different, but identical hard drive.

Both Comanche 4 and 3DMark2001SE ran a good 10% slower on the nForce2. The KT333 could hit 11,700 on 3D Mark at default driver settings. The nForce2 could only do 10,400. On Comanche, the KT333 got over 46 FPS while the nForce2 could barely manage 42 FPS. The nForce2 overclocks much easier with no dividers; but heck, you have to overclock quite a bit just to get KT333 performance. Anyway, I was really surprised. I'm going to work on this nForce2 rig for awhile, and see if I can bring it up to KT333 performance. Hehe....any comments?

.

 

[gplracer]

Hmmm what video card are you running? I got somewhere around 13000 with my nforce2 running at 215x11=2365mhz @1.7v with a gf4 4200.

 

[boyRacer]

Originally posted by: gplracer
Hmmm what video card are you running? I got somewhere around 13000 with my nforce2 running at 215x11=2365mhz @1.7v with a gf4 4200.

:Q A GPL Racer?

 

[Treeburst155]

I'm running a Ti 4200. I ran the tests on both boards at 140 FSB. Everything was the same except the drivers for the boards. Maybe the nForce2 doesn't start performing until you get over 166 FSB or something. The install went well. Everything seems to be fine, it's just 10% slower than the KT333 with my XP 2400+.

 

[gplracer]

Yes! A GPL racer. I bought it the day it came out!

 

[boyRacer]

Originally posted by: gplracer
Yes! A GPL racer. I bought it the day it came out!

Nice. Pure racing sims are a dying breed.

 

[Sheriff]

PCI is locked on the NForce @ 33 and the 333 increases with the FSB. Run them both @ default speeds and then compare.

 

[Goose77]

Originally posted by: Treeburst155
I'm running a Ti 4200. I ran the tests on both boards at 140 FSB. Everything was the same except the drivers for the boards. Maybe the nForce2 doesn't start performing until you get over 166 FSB or something. The install went well. Everything seems to be fine, it's just 10% slower than the KT333 with my XP 2400+.

ur problem is that that the kt board's agp is being overclocked. the NV board has an agp lock, this explains the difference. try running both boards at the same bus speed of 133(both memory and cpu fsb).

 

[Treeburst155]

Aaaah yes, the AGP was out of spec with the KT333! I didn't think of that. Thanks! That and the new BIOS I just flashed, with a couple more memory tweaks in it, will probably take care of that 10%. My KT BIOS has a few obscure memory tweaks that work. I'm sure they have been a factor in that 10% difference. If I can get nForce2 on a par with the KT at 133FSB, the overclockability of the newer chip takes on real meaning. Thanks again!

 

[chocoruacal]

You have a problem somewhere. I use kt333 and nForce2 on a daily basis, I can assure you that the nForce2 boards eat the kt333 for dinner when it comes to 3D games. I've seen very little difference in anything else though.

Common parts would be...

XP2200
ti4200
512MB RAM
120gig 8mb cache

 

[Regs]

A kt333 with a newer bios should be on par with a nforce2, with the exception of Duel DDR witch means little for a AMD set up. Overclock a nforce2 to a 200 FSB, however, then the Kt333 will be eaten alive.


Which reminds me, I'm still running my PC3200 XMS on T2 2.5 Cas on my Kt333, Im still getting high 3d mark scores above 15,000. If I could flash to the new bios with Ram timing tweaks, I could hopefully brake the 16k mark.

 

[Treeburst155]

I now have the nForce 2 board performing ALMOST as good as the KT333 thanks to a BIOS update with a memory tweak called "auto ram precharge". All tests at the stock 133 FSB today. In CPU intensive tests, the nForce2 actually pulls ahead by a very slight margin. When it comes to graphics intensive things, especially 3DMark2001se, it's still running 8% behind the KT. I'm not used to buying new hardware and seeing 1000 points drop off 3DMark. It bothers me, even though I don't play graphics intensive games. I'm a flight simmer so the Comanche4 benchmark is more meaningful to me, and the nForce2 is doing well in that one.

 

[Regs]

Performance wise via 3dmark, they might come out the same. However Ive seen testimonials from more than a few that the IDE interface works faster on a nforce 2. Rumor or not, you still have to consider nforces capacity to overclock, onboard audio, onboard video, and other goodies. The real hamper for the kt333 is the lack of the support for a 200 FSB, and native support for low latency memory, sata, and raid.

 

[JBT]

I went from a MSI KT3 Ultra 2 to a ASUS A7N8X DLX and everything was much faster for me with the nforce 2 board. Though I have pretty much replaced all my components so I woulnd't be able to test which is faster I know I could never run above 166 fsb on the KT3 though.

 

[Treeburst155]

There are lots of good reasons to move to an nForce2 from KT333; but I'm 100% convinced, as of this afternoon, that within the FSB range of the KT333, the KT is a better performer than the nForce2 in graphics intensive games. FSB for FSB/Clock for Clock the KT333 is a better performer up until it hits its limit. If you aren't going to run your memory AND your CPU at 166+, you WILL suffer a performance drop in graphics intensive games. CPU intensive stuff is on a par with KT333.

This is a fact that will probably not be discovered unless you try to run a 133 FSB CPU with PC 2700 or less in your nForce2. If you slap your high end 200 FSB processor into an nForce2 with PC 3200 RAM and crank it up to 200 FSB, you will notice a performance gain; but not as much as you logically should. The increase in performance will NOT scale well with the increase in FSB/Clock compared to the KT333. Here's a hypothetical example of what I mean:

KT 333 at 133 FSB (RAM at FSB speed)
3DMark = 11,500

nForce2 at 133 FSB (RAM at FSBspeed)
3DMark = 10,200

KT 333 at 166 FSB (RAM same)
3DMark = 12,500

nForce2 at 166 FSB (RAM same)
3DMark = 11,800

KT333 topped out

nForce2 at 200 FSB (RAM same)
3DMark = 12,800

You have to take the nForce2 to its upper FSB range in order to see a significant improvement in graphics intensive games over a KT333. This means you need the latest CPUs and fastest RAM to go with your nForce2. As you lower your nForce2 FSB, you quickly degrade to KT333 performance. By the time you're moving down through 180 FSB, that KT333 is pulling ahead in performance.

 

[RanDum72]

You can't really say, based on your limited testing, that the KT333 is ALWAYS faster at the same clock than the Nforce2. Different motherboard brands, different BIOSes, different memory, etc... will have different results. You change the video card, it may be faster on one mobo than the other. Did you even run the nforce2 in dual-memory mode (two sticks of RAM)?

 

[InlineFive]

Originally posted by: RanDum72
You can't really say, based on your limited testing, that the KT333 is ALWAYS faster at the same clock than the Nforce2. Different motherboard brands, different BIOSes, different memory, etc... will have different results. You change the video card, it may be faster on one mobo than the other. Did you even run the nforce2 in dual-memory mode (two sticks of RAM)?

Aha! But this approach is unfair, we need someone to benchmark multiple KT333s and nForce2s at same speeds (cpu/mem/agp/pci/etc.) with Single-Channel memory. If the KT333 still prevails that means that it is a more refined engine, so to speak, and that the nForce2 needs extra horsepower to be able to outrun it.

 

[Treeburst155]

You're right. I can only say the nForce2 is slower with MY single set of peripherals. I can definitely state this however, because I'm essentially testing exactly identical systems but for the different mobos. I did not test the dual channel feature because I don't have two sticks of DDR. I've heard the performance gain with Athlon is minimal though. Here's Anand's article comparing KT333 and nForce2. Note how close the two boards are at 166 FSB. The article does not quite support my position, but it definitely lends some credibility to it. The nForce2 just ain't that much faster than a KT333. It's more overclockable; but not faster, clock for clock. You have to run it way up to really see any performance gain. The nForce2 was designed with high FSBs in mind I think. If you're running older RAM & CPU, it ain't worth it.

http://www.anandtech.com/chipsets/showdoc.html?i=1719&p=1

 

[CraigRT]

I went from a KT266A to an Nforce2, and the NForce2 is hands down faster.

the KT266A is not much slower than the KT333, either.

 

[Treeburst155]

It could be I have a rather slow nForce2 board (Epox8RDA3+), and a rather fast KT333 board (MSI KT3V).

I wish I could say I've seen an increase in gaming benchmarks with nForce2. I've certainly tried to make it so; but it just ain't happening in my case.

 

[PrinceXizor]

Originally posted by: PorBleemo

Originally posted by: RanDum72
You can't really say, based on your limited testing, that the KT333 is ALWAYS faster at the same clock than the Nforce2. Different motherboard brands, different BIOSes, different memory, etc... will have different results. You change the video card, it may be faster on one mobo than the other. Did you even run the nforce2 in dual-memory mode (two sticks of RAM)?

Aha! But this approach is unfair, we need someone to benchmark multiple KT333s and nForce2s at same speeds (cpu/mem/agp/pci/etc.) with Single-Channel memory. If the KT333 still prevails that means that it is a more refined engine, so to speak, and that the nForce2 needs extra horsepower to be able to outrun it.

So I suppose Intel's chips are suxor because they quad pump their data rate transfers? At the end of the day, does it run my stuff faster when I use it all. I'm not going to BUY an nforce2 chips to NOT run in dual data mode. I'm not going to buy a turbocharged car and then disable the turbocharger so my engine is "pure". That just doesn't make any sense.

P-X

 

[Treeburst155]

"So I suppose Intel's chips are suxor because they quad pump their data rate transfers? At the end of the day, does it run my stuff faster when I use it all. I'm not going to BUY an nforce2 chips to NOT run in dual data mode. I'm not going to buy a turbocharged car and then disable the turbocharger so my engine is "pure". That just doesn't make any sense."
____________________________________________________________________________________

Exactly. That's why an nForce2 is a waste of money if you don't put the high FSB RAM and CPU with it, unless you want the SATA and other little niceties. This is a fact I did not realize until I did my testing.

A better analogy: nForce2 is the engine, CPU is the turbocharger, RAM is the high performance exhaust, dual channel is the leather seats.

 

[RanDum72]

I did not test the dual channel feature because I don't have two sticks of DDR. I've heard the performance gain with Athlon is minimal though

Well, using dual-channel instead of single channel memory on my Asus A7N8X deluxe netted somewhere between 5-15% gain, depending on application. It makes a lot more difference in games than regular 'productive' stuff, and, to me, it's not minimal at all. This is with regular Crucial PC2100 memory and an AMD 2400+ XP. You don't need uber fast memory to make dual-channel fly, regular PC2700 (and preferably a CPU with a matching 333fsb) would do the trick. If you have no intention of using more than one memory stick, then the KT333 is a bargain. But in the long run, the nforce2 is the more logical choice. It upgrades to faster CPU's, supports higher FSB, has PCI/AGP lock so damage to your cards in nill, then when you add another memory stick, you not only increase memory but you get a performance bonus as well (which is enabling dual-channel mode).
Depending on model, you calso get a 'free' high performance sound card. Anybody starting out on Athlon XP-based system would be blind to ignore the overall benefits of the nforce2 over the KT333.

But if you already have a KT333 system and pretty happy with it, then there's no reason to upgrade.

 

[arcenite]

Originally posted by: Treeburst155
"So I suppose Intel's chips are suxor because they quad pump their data rate transfers? At the end of the day, does it run my stuff faster when I use it all. I'm not going to BUY an nforce2 chips to NOT run in dual data mode. I'm not going to buy a turbocharged car and then disable the turbocharger so my engine is "pure". That just doesn't make any sense."
____________________________________________________________________________________

Exactly. That's why an nForce2 is a waste of money if you don't put the high FSB RAM and CPU with it, unless you want the SATA and other little niceties. This is a fact I did not realize until I did my testing.

A better analogy: nForce2 is the engine, CPU is the turbocharger, RAM is the high performance exhaust, dual channel is the leather seats.

Well, with all due respect to your analogy... A car could run without a turbocharger, a computer could not run without a CPU. Leather seats do not improve car performance either, that I know of... We knew what you meant though ;-)

Bill

Edit: My 8rda+ is going strong @ 200 FSB (feel no need to try to clock higher... Mmm... 200 fsb goodness

 

[Treeburst155]

Yeah, the analogy was a little weak. The main point was the leather seats. Apparently, some think there is a noticeable gain with the dual DDR, even with AMD stuff. This interests me. I'll have to do some research.

As for whether it's smart to buy an nForce2 or not, I'll just say that I'm definitely keeping mine. As has been mentioned, it has upgrade potential. That alone is worth it. However, people should not make the mistake I did, and expect a quicker gaming machine with an nForce2 and their 133 FSB CPU and there PC2100 RAM. The chipset is not inherently faster. I think it's actually slower. It's running on a lower gear ratio than KT333. You have to rev it higher (FSB) to get the same speed. There's my analogy.