Thinking about buying GeForce4 Ti4200 but..

[Guest]

Hi.

I currently have a voodoo 3 3000 an its have problems running the newer games even at its lowest settings. I was thinking about upgrading to a GeForce4 Ti4200 64m but I only have a duron 700 and was wondering if the 700duron would run it without any problems. I'm not looking for the best just something that will let me play the newer games fairly well with good frame rates and with good picture quality.

 

[williamc]

You will get performance increase, but not a big one. If you have no plans to upgrade the rest of your pc soon i highly suggest just trying a low end GF4 MX 440. Anything more isnt going to be justified on that computer. My personal advise is upgrade the whole pc...if you dont have the whole amount now, save up. 700Mhz is just too far outdated for a gamer.

"WtF? Package just said silver stuff, why won't the HSF come off?" - Failed A+ 6 times

 

[Guest]

what would be the min processor speed i could get away with to get good performance? 1ghz duron?

 

[bront]

Go with an athlon XP over a duron. The price isn't that much different, and the performance is much better.

You will get some great performance from a GF4 TI4200 on your current system, just not compairable to performance on a high end CPU. If gaming is all you do, get the video card update, and update your computer later. If you do anything else, upgrade the CPU first (Athlon XP, PC2100, DDR MB min) and then go for the video card upgrade. Heck, for a decient price, you could go with the Nforce board, get a small game boost now, and then get a better card later.

The Windows Gods demand money to appease the BSOD! - Rev. Bill Gates

 

[phsstpok]

I think you will get a big enough performance boost to justify the expense but judge for yourself. Look at CPU scaling test in <A HREF="http://www.anandtech.com/video/showdoc.html?i=1608" target="_new">Anandtech's sub-$200 video card article</A>. They do CPU testing at speeds from 800 to 1733 mhz, with an Athlon XP not a Duron but the info should help you decide.

Two thoughts, though.

1. You need a beefy power supply for these high powered video cards.

2. The Ti4200 scales well. If you upgrade the rest of your system later the video card won't hold you back.

<b>I have so many cookies I now have a FAT problem!</b>

 

[Guest]

thanks for the advice =)

think i'll go with the Ti4200 then upgrade the CPU later if need be

 

[DDR64]

hmm, bear in mind, that Athlon's larger L2 cache over Duron's does make a considerable difference in frame rates. Just something to consider when you're looking at CPU scaling test. You'll indeed get higer score in benchmarks, but don't be surprised when you get a choppy frame rate even with G4Ti4200. But you'll be able to run the games at highest setting(using lower setting won't give you faster frame rate anyway in your system with G4). I say at least 1.4-1.5GHz CPU should be paired with highend video card.

 

[phsstpok]

that Athlon's larger L2 cache over Duron's does make a considerable difference in frame rates

Where is your evidence of this claim? If there is a difference I would not use the word "considerable".

Here's some evidence to contradict your claim. Taking some examples from Madonion's online browser. These are the best scores for the CPU/videocard combinations.

<font color=green>Duron 700 (or less) + Geforce4 Ti4400 = 4928 (only one person with this combo so maybe it's not the best possible).
Athlon Tbird 700 + Geforce4 Ti4400 = 5855.

Duron 800 + Geforce4 Ti4400 = 5983.
Athlon Tbird 800 + Geforce4 Ti4400 = 6083.

Duron 1000 + Geforce4 Ti4400 = 7297.
Athlon 1000 + Geforce4 Ti4400 = 8207.

Duron 1200 + Geforce4 Ti4400 = 9254 (maybe this is a mistake. Second best is 8546).
Athlon Tbird 1200 + Geforce4 Ti4400 = 8790.
Athlon XP/MP/4 1200 + Geforce4 Ti4400 = 8440.

Duron 1300 + Geforce4 Ti4400 = 9560.
Athlon (Tbird) 1333 mhz + Geforce4 Ti4400 = 9709.
Athlon XP/MP/4 1333 mhz + Geforce4 Ti4400 = 9115.</font color=green>

Hmmmm, doesn't seem like much of difference between the large cache Athlons and the small cache Durons, nothing conclusive at least.

Lets throw in the highest Athlon Tbird and XP scores (with Geforce4 Ti4400) vs the highest Duron score (same video card), CPU speed is not limited.

<font color=red>Duron 1398mhz + Geforce4 Ti4400 = 9536
Athlon Tbird 1528mhz + Geforce4 Ti400 = 11,694
Athlon XP/MP/4 1954mhz + Geforce4 Ti4400 = 12,941</font color=red>

Duron is showing slower here. Wait a minute, the Athlon XP has a 40% superior clockspeed and it scores 36% better than the Duron. The Athlon Tbird has a 9% superior clockspeed and it scores 22% better than the Duron. Nothing conclusive here, either.

I realize 3DMark2001 is a synthetic benchmark but it was the easiest way to gather some numbers on Duron, Athlon Tbird, and Athlon XP and Geforce4 Ti4400 performance. Still, I just don't see big differences, not consistant ones.

Feel free to check results yourself.


<b>I have so many cookies I now have a FAT problem!</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 04/24/02 09:03 PM.</EM></FONT></P>

 

[chuck232]

I realize 3DMark2001 is a synthetic benchmark but it was the easiest way to gather some numbers on Duron, Athlon Tbird, and Athlon XP and Geforce4 Ti4400 performance. Still, I just don't see big differences, not consistant ones.

Most if not all of the mentioned scores are by people who have overclocked to the extreme. That is the only way that they could get these amazingly high scores. You'd probably see more conclusive evidence if these were stock scores.

 

[phsstpok]

Are you trying to claim that only Duron owners overclock their video cards? Or are you trying to claim that Duron owners padded the results in all cases? I'm afraid that's not the case. ALL of the video cards are overclocked. Think about it. If the processor cache mattered then Durons and Athlons would not get similar results.

Cache is not the most important factor. At a given clockspeed it is floating point performance that matters more. Durons and Athlons have the same floating point units and thus similar results.

<b>I have so many cookies I now have a FAT problem!</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 04/24/02 09:29 PM.</EM></FONT></P>

 

[DDR64]

When I said "considerable" I meant the difference would to as much that you can't state Athlon and Duron at same clock speed will give you same frames per second. That the difference won't be negligible. Thus, I would equate Duron800MHz to Athlon700MHz or lower in terms of frame rate.

http://www.tomshardware.com/cpu/02q1/020121/duron1300-05.html

wouldn't you say this is considerable difference?

http://www.anandtech.com/showdoc.html?i=1336&p=8
http://www.anandtech.com/showdoc.html?i=1336&p=9
http://www.anandtech.com/showdoc.html?i=1336&p=10

even over clocked to 900, Duron performs equal or lower than Athlon800.

Up to some point, lager cache size can make up for slower clock speed when comparing processors of same architecture, that's why the current AMD roadmap shows the Barton(future Athlon core) shows double the L2 cache size(512KB from 256KB) instead of going 166FSB since the architecture of the Athlon would benefit more from larger cache size than higer FSB as confirmed by anandtech.com

Some games are higly CPU dependent and one game can be more L2 cache dependent than others. Well, SSE and 3Dnow of modern CPUs are designed for Floating point calculation for 3D, but that was for when GPU's didn't have T&L unit. The main role of CPU in 3D is to send vertices of polygon to the 3D chip. How fast CPU can deliver that geometry data to the GPU is the key point. In this regard, larger L2 cache size is obiviously beneficial.

using 3Dmark2001 is somewhat unreliable because the effect of mobo(chipsets) and memory users have. Chipsets make a quite a difference in frame rates. If you want to find out the effect of L2 cache size of Athlon and Duron, you have to benchmark it with same system. Using 3Dmark score can be misleading since there is variation in different chipsets/memory used and as well as degree of overclocking.

And yes, 3Dmark2001 is synthetic. It doesn't even use sound so it does not take the effect of CPU utilization by sound chips(or onbard sound) into account, for example.

<P ID="edit"><FONT SIZE=-1><EM>Edited by DDR64 on 04/24/02 10:35 PM.</EM></FONT></P>

 

[phsstpok]

Chipsets make a quite a difference in frame rates.

Bus speeds make more difference. Those Durons, in the tests to which you linked, are run with a 100mhz FSB (200 DDR) and the Athlons are run with 133mhz FSB (266 DDR). It's bus speed difference you are seeing there not cache size.

The best Madionion results are done with systems overclocked to the hilt, video card, memory, bus speed. I my opinion that puts them on a much more equal footing than the test you linked to.

If you have the time, you can wade through results, yourself. You can find similar bus speeds, and identical chipsets. Those are both listed. You will have to take many samples though because the level of video card overclock is not available. Choose any video card as long as it doesn't limit the results, like old cards that don't have enough performance and run out of steam.


<b>I have so many cookies I now have a FAT problem!</b>

 

[DDR64]

http://www6.tomshardware.com/mainboard/02q2/020422/chipset-11.html

chipsets do make a difference.

yeah, yeah, I agree that at same clock speed, higher FSB give you more performance. But, even with different FSB, if clock frequency is the same(meaing that difference in FSB do not result in different clock speed), wouldn't you say performance difference is going to come mainly from different L2 cache size? I never seen CPU benchmark that compares same speed Athlon and Duron and discuss effect of FSB rather than L2 cache size. They all talk about L2 cache size.

I mean, look here
http://www.anandtech.com/cpu/showdoc.html?i=1595&p=12

% difference between 133FSB/DDR266 vs. 166FSB/DDR333 in Athlon(both same clock speed) is usually 1%.

This is why AMD decided to double the size of future Athlon(Barton)'s L2 cache to 512KB as their roadmap shows instead of going for 166FSB.<P ID="edit"><FONT SIZE=-1><EM>Edited by DDR64 on 04/24/02 11:00 PM.</EM></FONT></P>

 

[phsstpok]

I never said chipset don't make a difference. I said, "Bus speeds make more difference". You gave me a link where the testing of Durons was at a different bus speeds than the Athlons then later you went back and edited the post and adding an old useless benchmark. Nobody cares if a 2-year-old benchmark done with a Geforce2 at 640x480x16 is 128 fps or 121 fps. Nobody plays at this resolution.

You've been sidetracked. The important question is whether Aeolus can get any benefit from a Geforce4 Ti4200 with his current system or not. Currently he has a Voodoo3 which with a Duron 700 might score maybe 1500 points in 3DMark2001 (sorry to bring up this benchmark again) and that's 16-bit color (because the Voodoo3 can't do 3D acceleration in 32-bit color). I'm saying with Geforce4 Ti4200, he can turn on 32-bit color, turn up the level of detail pretty high, probably bump up the resolution, turn on some new features, and just plain get much more enjoyment out of his games. Aeolus CAN upgrade his video card and get more performance. A fast CPU is not required to do this. Sure, if he wants even more performance then he will have upgrade the CPU but he can do that later.

<b>I have so many cookies I now have a FAT problem!</b>

 

[phsstpok]

If you want to say something new please do not "back edit" your posts just insert something or comment your edits. Other people have to read them.

Now replying to the words you are now presenting. Yes cache matters but given the same architechture Duron/Athlon CPU speed matter most then between cache and bus speed I can't answer that. If one CPU has a much too small cache then yes it will have inferior performance. However defining "too small" becomes the key for optimization. Both the Duron and the Athlon have the same size L1 cache. It's 128K. This cache provide the processors with most of their memory performance, not the the L2 cache doesn't matter. I have no idea what is an optimal size for an L2 cache, hopefully the AMD engineers do. The larger the cache the greater the chance for a hit but this comes at the risk of more time wasted when there is not a cache hit. If the cache is too big you loose performance. That's the tradeoff.

This is why AMD decided to double the size of future Athlon(Barton)'s L2 cache to 512KB as their roadmap shows instead of going for 166FSB

I don't know that this is true, that the engineers chose one over the other. I'm sure more cache will increase performance but I would think that if you have more cache that increasing memory bandwidth couldn't hurt so why not do both.

<b>I have so many cookies I now have a FAT problem!</b>

 

[DDR64]

umm, actually, the reason benchmark is done at 640x480x16 is to eliminate the possibility of either fill rate or memory bandwidth restriction. That way, difference is entirely dependent on CPU performance when you're benchmarking same video card with various processors.

Yes, yes. He'll be able to enjoy higer resolution, color bit depth, and texture resolution. What I'm saying is he may get disappointed that even with Ti4200, he may be getting choppy performance because of CPU. You really have to experience this effect that slow CPU can have in a system with high end card. For example, with a popular game like RTCW, he runs checkpoint timedemo benchmark and it will give him result that are less than 50% of what Ti4200 is capable of and that fps won't increase even he tries to lower resolution and color depth. Say he tries to play online multi-player game of popular games like MOHAA, RTCW, or may be Global Operations. If he joins a server with lots of players, the end result can be described as hitting a brick wall. Single digit frame rate. With Duron700 this is inevitable. I've seen this happening with 1GHz Duron. So, he may feel that he wasted his money buying Ti4200 and would have been better off getting faster platform with low-end card. Still, Ti4200 would be good investment only if he plans to upgrade platform soon.

 

[chuck232]

Are you trying to claim that only Duron owners overclock their video cards? Or are you trying to claim that Duron owners padded the results in all cases?

I meant that these scores aren't conclusive since the level of overclock is unknown, so you really can't compare the scores.

 

[phsstpok]

You can compare, just not based on the handful of scores I listed. That was just for an example. If you look at enough scores you will see a pattern. If a Duron a 1200 has one score that nearly matches the best Athlon 1200 you may wonder but it doesn't prove anything. However, if you look at page after page of scores that are similar then you have to think there is something to it. If you see (just for example) 100 similar scores on both sides, Athlon and Duron, you can't claim that 100 Duron users were more succesful at overclocking their video cards making it possible to match the Athlon scores. You have to think Duron systems are simply able to facilitate equally high scores.

My conclusion is based on my rudimentary understanding of probabilities. An analogy, a coin tossed five times comes up "heads" five straight times. Well that could be just a chance occurance. A coin tossed 100 times and they are all "heads". Well, that is simply a unfair coin.

<b>I have so many cookies I now have a FAT problem!</b>

 

[DDR64]

What are you trying to prove, man?
You listed 3Dmark score for Duron and Athlon at same speed with the same video card which shows people with Athlon mostly have the higer score.
So you want to say Duron nearly match the score of Athlon at same speed with the same video card. SO WHAT?
System with Duron will always be slower than Athlon of same speed other components being equal.
You're trying to do statistical analysis where there are too many unknown variations. Like, the mobo(chipsets), system memory configuration, degree of overclocking.
This renders your comparsion completely useless.
You've been looking at 3Dmark scores too much. You need to get away from it for a while.

If you want to prove Duron can perform nearly as good as Athlon at same speed, you do benchmark with identical components which you can find all over the web. You don't look at the 3Dmark scores. If one was able to produce nearly identical score with his Duron machine in comparison to another who has Athlon of same speed, that just means if you put Athlon on that Duron machine, the score would improve further.

 

[phsstpok]

What are you trying to prove, man?
You listed 3Dmark score for Duron and Athlon at same speed with the same video card which shows people with Athlon mostly have the higer score.

Don't know where you are looking. My little sample had the Duron ahead of both the Tbird and XP once and the Duron was ahead of the XP two more times but this was not what I was trying to prove. What I was trying to prove is Athlons are not outscoring Durons.

The claim was that Athlons always outperform Durons because they have a larger cache. I was trying to prove that this is not always true and that there is more to video performance than that simplification.

Does no one see the signifcance that Athlon systems ARE NOT outperforming Duron systems of the same clock speed?

<b>I have so many cookies I now have a FAT problem!</b>

 

[DDR64]

...

So do the both system has exactly same components?
Noooooooo...
Then that comparison is completely useless.

Well, it is significant that Duron can outperform Athlon system at same speed, but that just means the Duron system has better components. With 2 systems with identical components, Athlon should always outperform Duron at same speed in this hypothetical situation. That difference coming from lager L2 cache since clock speed is equal.
I don't know why this is hard to understand and someone would try to contradict it by comparing two systems where components are not equal.

Nah... I know what you were trying to say. you wanted to emphasize the fact that Duron systems can outperform Athlon systems at same clock speed. But, why are you using this fact to denounce effect of L2 cache size? How can you asses effect of L2 cache size when systems you're comparing have different components? That just prove's the Duron system has better mobo(chipsets), memory, and overclocking. And this was enuf to overcome Duron's disadvantage of samller L2 cache.

You understand what i'm saying, don't you?

 

[phsstpok]

do the both system has exactly same components?

I'm not talking about two systems I'm talking about dozens.

I was trying to say (and we got way sidetracked) that clock for clock the performance difference between an Athlon and Duron is small enough that you can still use the CPU scaling results to make predictions for a Duron's performance with a Geforce4 Ti4200. So you just adjust downward a little bit for a Duron. That is what Aeolus wanted to know. Will he gain performance going from a Voodoo3 to a Geforce4 Ti4200. I say, yes.

<b>I have so many cookies I now have a FAT problem!</b><P ID="edit"><FONT SIZE=-1><EM>Edited by phsstpok on 04/26/02 04:26 AM.</EM></FONT></P>

 

[DDR64]

"So you just adjust downward a little bit for a Duron"

It doesn't seem like this is what you were trying to say. You were trying to say the difference was negligible and not worth considering. If, it were, you supplied wrong data and one mighty twisted logic to prove this point.

"you can still use the CPU scaling results to make predictions for a Duron's performance with a Geforce4 Ti4200"

well, ok. But isn't it another way of saying what I was claiming? That you can't exactly assume that Duron would produce same frame rate as Athlon at same clock frequency? That you have to assume Duron would equal to Athlon at lower clock frequency? Just look at my previous postings.

You just denounced my claims and post 3Dmark scores and said Duron can perfom equal or better than Athlon. I do not get impression that the above statements in quote is what you were trying to prove.

You're contradicting yourself and not making sense.

I'm getting tired and beginning to lose respect. You're headstrong in a weird way that makes me believe you're just a runny nose kid with...

man...this is getting tired. Saying same things over and over again.

 

[phsstpok]

"So you just adjust downward a little bit for a Duron"

It doesn't seem like this is what you were trying to say. You were trying to say the difference was negligible and not worth considering. If, it were, you supplied wrong data and one mighty twisted logic to prove this point.

"you can still use the CPU scaling results to make predictions for a Duron's performance with a Geforce4 Ti4200"

well, ok. But isn't it another way of saying what I was claiming? That you can't exactly assume that Duron would produce same frame rate as Athlon at same clock frequency? That you have to assume Duron would equal to Athlon at lower clock frequency? Just look at my previous postings.

You just denounced my claims and post 3Dmark scores and said Duron can perfom equal or better than Athlon. I do not get impression that the above statements in quote is what you were trying to prove.

You're contradicting yourself and not making sense.

I'm getting tired and beginning to lose respect. You're headstrong in a weird way that makes me believe you're just a runny nose kid with...

man...this is getting tired. Saying same things over and over again.

Let me see. Here was my very first assertion to Aeolus.

"Look at CPU scaling test in Anandtech's sub-$200 video card article. They do CPU testing at speeds from 800 to 1733 mhz, with an Athlon XP not a Duron but the info should help you decide."

then you seemed to be contradicting my claim with

"hmm, bear in mind, that Athlon's larger L2 cache over Duron's does make a considerable difference in frame rates."

Then I tried to prove that there isn't that much difference using 3DMark benchmark results.

Next you gave me counter benchmarks however Durons and Athlons were clearly running different bus speeds.

After I replied to that you edited your post and changed the links to different benchmarks.

Following that you gave me some very telling information about online gaming which I did not know. To which I started to reply because I wanted more information but I was tired. I forgot to get back to it the next day because you let into me again after I responded to Chuck232's querry (his not yours by the way).

So which one of us is not sticking to the course.

<b>I have so many cookies I now have a FAT problem!</b>

 

[DDR64]

oh my god, you're just a kid aren't you?
Man, this is just too childish.
Forget this. I'm wasting my time here.
Grow up, man...