Radeon 128MB vs. GeForce 4 . . . close contest?

[XMan]

I have been trying to decide between these two variants of cards for my next system, and I was thinking the GeForce performance was a good deal higher. But then I read John Carmack's comments about the Radeon in regards to Doom 3, and began to wonder. So I pawed through some of Anand's reviews and grabbed the following numbers.

One caveat - the 128MB Radeon numbers are from the Radeon All-In-Wonder; I'm not sure if the clock speeds are the same on the two variants.

Since I can't do any fancy graphs, I'm ordering them from highest score to lowest, winner in bold.

Quake III Arena - demo four 1024x768x32

Ti4600 - 218.4
Ti4400 - 216.0
Radeon 128MB - 213.5

The Ti4600 wins here, but not by much . . . considering price and image quality, I gotta go with the ATI.

Comanche 4 - 1024x768x32

Ti4600 - 41.6
Ti4400 - 40.8
Radeon 128MB - 32.6

Not so good here . . . I haven't played Comanche 4, is it a good indicator of what games coming down the pike will be like?

I'm curious about something. In this review, the Radeon 8500 - presumably 64MB - scores 58.7 on the 1024 Unreal Performance Test 2002. In this review, the 128MB AIW version nets only 35.4. Is the clock speed different on the two? EDIT - According to the review, the Radeon AIW 128MB is clocked at 275/275, just like the standard retail Radeon. So where does the discrepancy come in?

I'm still undecided. Keeping in the budget I've set myself, if I get the Radeon 128MB, I can have an additional 512MB of RAM, whereas with the GeForce, I'd have to cut the memory or get a slower CPU.

 

[richleader]

Not so good here . . . I haven't played Comanche 4, is it a good indicator of what games coming down the pike will be like?

I doubt it. Flight and racing sims have traditionally been bad barometers to compare to other genres--and ATI has always gotten spanked in that department, which 3dfx usually won back in the "old" days, with Nvidia on its heels.

In future games, it is possible that the 8500 might take the lead, but who knows--I think the real pros of each are:

ATI-great 2d / TV features if you spring for it

NVidia, FSAA. Supersampling just isn't going to cut it on the 8500 in future games. With the G4, 1024x768 with quincunx *might* be possible in Unreal 2003 (the demo test is more intensive than regular scenes and they keep optimizing), I doubt that the 8500 could pull off anything equivalent, though it might keep up without FSAA.

 

[kami]

The reason the card scores differently in those Unreal tests is because they are different builds of the benchmark. The newer ones stresses the card a lot more. I have been researching for a while and have decided on a 128MB Ti4200. It can overclock to Ti4400 speeds guaranteed, and there's a good chance it can do Ti4600 speeds. Also it is pretty cheap. Newegg has 128MB version at $211 and a 64MB version at $160

 

[XMan]

Now this is interesting, found a review of the AIW 128MB on Hard OCP. They used the new 6058 ATI Radeon drivers.

Serious Sam 2 - Extreme Quality 1600/1280/1024

Radeon 128MB - 43/59.4/82.2
Ti4600 - 38.8/56.3/86.6

Jedi Knight 2 Maximum Settings 1600/1280/1024

Radeon 128MB - 77.6/94.6/95.4
Ti4600 - 82.3/97.8/99.1

Basically a dead heat . . . hmm.

 

[XMan]

It can overclock to Ti4400 speeds guaranteed

Guaranteed, eh?

I have learned the hard way that nothing is guaranteed. I think the last "guaranteed" overclock that actually worked out for me was my Celeron 300A. I had a PII 333 when everyone else was getting 500+ MHZ out of them, and I only got 416. I had a PIII 450, when everybody else was getting 600+, and I could only get 558 stable. My Athlon 500 was great - it did 750MHz stable, and was on the bleeding edge at 800MHz. But my AXIA T-bird 1.2 was a flop, I can't even get 1.33 out of it.

So I don't bank on guarantees.

 

[kami]

Well, I'd say a 90% chance you get Ti4400 speeds.

Oh well, even a stock Ti4200 owned the Radeon 8500 in most tests I saw, so it pretty much ended there for me

 

[XMan]

Oh well, even a stock Ti4200 owned the Radeon 8500 in most tests I saw, so it pretty much ended there for me

I dunno, it looks like the newest drivers kicked it up to just under Ti4600 performance in most situations.

I'm only going to get a 19" monitor, so I can't see myself playing any games at any resolution higher than 1280x1024, anyway.

 

[Willoughbyva]

You probably already know about rage3d, but thought I would just mention it. www.rage3d.com has some good info about ati cards and there is a forum there.

Could you wait until the next series of cards is released in the fall? The ones out now are bound to drop in price. Plus the option of the newer cards with newer features. They'll be pricey though. I'll probably wait until next spring to replace my card, that is if I am still gaming.

On another note I had an gf2mx400 and went and got an ati aiw 7500. There was a noticable difference in the 2d quality. The 7500 has much better 2d than the gf2 I had.

Either way good luck.

Will

 

[HowDoesItWork]

Why would your monitor size influence what res you run games in? Am I smoking crack again? I understand why you wouldn't want your desktop at 1600x1200 on a 19 inch monitor, the icons get all small and stuff. But in games when I adjust the res the object stays the same size regardless of resolution, it just gets more or less detailed? It is late, I'm tired, let me know if I have a screw loose.

You have to watch benchmarks, if you pull them from different places or even different articles at the same place, the test bed might be (likely) different. The benchmarks I have seen give the 4600 a pretty hefty lead in most games/apps. Some games seem to stress the CPU more, ex. Star Wars and hence may not be very good benchmarks of the graphics card.

I have not seen the new drivers, the last rev (v7.65?) though really improved the Radeon in some benchmarks, and totally hosed it in others. Unreal2003 benchmarks went way up (but not to 4600 levels) but then Serious Sam marks were less than half what a 4600 was running. I don't know if this is a positive sign 'Hey once they get the drivers hammered out this card will rock!' or a negative 'AIT still can't get drivers right!' We shall see, the more competition the better though.

 

[richleader]

Why would your monitor size influence what res you run games in? Am I smoking crack again? I understand why you wouldn't want your desktop at 1600x1200 on a 19 inch monitor, the icons get all small and stuff. But in games when I adjust the res the object stays the same size regardless of resolution, it just gets more or less detailed? It is late, I'm tired, let me know if I have a screw loose.

Two reasons, both tied to refresh rate which decreases as resolution goes up:

1. Flicker is introduced. An image at 120 HZ at 1024x768 is much more stable than an image at 72 HZ at 1600x1200.

2. Refresh rate is tied to FPS, even to some degree when vsync is disabled as tearing starts to occur if there aren't enough screen updates. IF you're hitting 60-70 hz with your monitor at a given resolution, it can impact your frame rate to some degree.

For those two reasons, many choose to run at lower resolution with FSAA rather than jumping straight to 1600x1200.

For more info, you can check out FAQs on my website here in the glossary section:

http://www.richleader.com/bargainbinreview.htm

 

[HowDoesItWork]

Oh yeah, forgot about refresh rate. That has to do with the quality of your monitor, not the size though. In my experience anything above 60-70 hz is flicker free.

Maybe I was reading too much into it, it just seemed that the size of the monitor and max gaming res were being linked.

Edit: ""Is this a good game for $10?" is remarkably easier to answer, though especially painful when the answer is "sadly, no.""

Hehe, and I thought I was the only one who bought Dominion Wars!