Volari Duo V8 Benchmarks&impressions

agreed

 

I didn't say that, I said " why would you cheat if you do not meet the requirements to participate in the race"

 

Well, the geforce 4 mx 440 was faster than even the TI 500 I believe, and considering the speed a geforce 3 will run in a dx 8 or 9 game with shaders, well, it didn't really lose much by dropping them.(and for some reason, my ti 200 couldn't even do them right, there were tons of display errors, pretty pathetic when even nvidia's tech demos don't run right)

BTW, maybe the V8 Duo will see some major performance increases. As I can recall, the voodoo 5, original radeon, and geforce 2 all had enourmous performance increases later on compared to their original drivers.

 

That I'd love to see established first heh....

 

Actually, a MX440 is slower than a GeForce 3 Ti200 in everything that is not limited by the T&L unit -> everything but not the 3dmark01 T&L test.

 

Wierd, I could have sworn I saw a Geforce 4 MX beating the TI 200 by as much as 40% when the cards first came out.

Well, after looking at some benchmarks, the Geforce 4 MX 460 does beat the Geforce 3 TI 200 in many cases, though the 440 doesn't.

So would it go something like this?
400- between a geforce 2 mx and a geforce 2
420- geforce 2
440- slightly below a geforce 3 ti 200
460- normal geforce 3 to ti 500 in most cases, sometimes less than a ti 200
Radeon 8500- somewhere inbetween a geforce 3 and a geforce 4 ti(I guess similar speeds to a geforce 4 ti, except in t&l related stuff?)

 

From these benchmarks it looks like the gf4mx(all series) get crushed by the gf3 in any GPU demanding games. http://www20.tomshardware.com/graphic/20021218/vgacharts-02.html#aquanox And rightly so as the memory bandwidth is a lot lower. The only place where the gf4mx can keep up to the gf3 is in Jedi Knight which is very CPU limited.

 

I had the same results the last time I used these graphics boards for a review. http://cf01.tgv.net/~tt-hardware/article.php?sid=2980 (it's french but eh numbers are numbers )

 

http://www.digit-life.com/articles/gf4mx/

Well, here the 460 wins in quite a few 3dmark2001 tests.
Both the 460 and 440 outperform the TI 200 at high resolutions in quake 3.(and usually at low resolutions as well)
In return to castle wolfenstein, even the 420 sometimes beats the ti 200.
Would have been nice if the tests used some other games as well, but the 460 does tend to do exceptionally well in the 3dmark2001 game tests.

 

Hmmmm, wonder why that is....

 

Do CPU bound applications ring a bell?

The GF4MX is a value offering and that it has always been and as such it performed, period. Mainstream and high end sollutions of the time used to show their real strenghts especially with IQ improving features or partially dx8.1 applications, which in reality isn't that much different today with accelerators either.

*ahem*:

http://www.3dcenter.de/artikel/r9500+9700+9800_vs_gffx5200+5600+5800/zj_bench_rtcw_a.php

Just look at a Ti4600 compared to a 9800PRO.

http://www.3dcenter.de/artikel/r9500+9700+9800_vs_gffx5200+5600+5800/zj_bench_rtcw_d.php

Ooops suddenly the R350 turns out twice as fast in 1280*960. Where are the surprises?

 

3dmark2001 may be cpu limited(though a P4 2100 is probably twice as fast as what was available at the time 3dmark 2001 was made and the 460 MX tends to do best at 1600x1200, which should be fill rate limited, but not at 800x600 which should be cpu limited), but the geforce 4 460 mx is still winning by a substantial amount, so either there is less cpu overhead, it is a more powerful card, or it is doing less work.

 

Is 3D Mark 2003 a CPU bound application (this is an honest question) ?

 

Eh....might be, but the graphics accelerator seems to effect performance far more than the cpu, whereas 3dmark2001 should scale quite well with faster cpus.

 

No the CPU has almost no impact at performance at all except in GT1 (the "DX7" test).

 

BARELY if at all from my experiences. Bouncing a card from a PIII800@928 to a Celeron1.4@1.68 to a Barton 2500+@2.1 I saw almost no difference using the same cards.

It's my biggest "not like" about 3dm2k3, I still think 3dm2k1se gives a better performance picture.

 

Without a graphic board limitation, every game test shows more than 300 FPS with a Barton 3000+. 3Dmark03 is definitely not CPU bound. The Game Test 1 can of course be CPU bound with some older CPU.

 

3DMark03 Performance Factors @ aceshardware compares several processors using different Radeons in 3DMark and a couple of games.

 

I would have liked to see them try a few more games in their tests....
Halo.
Tomb Raider: Angel of Darkness
Splinter Cell
and what's another game that's shader intensive?(actually, I don't even know if splinter cell uses shaders)
I'd say Doom 3 if it was out, since it seems to most closely resemble battle of proxycon.

 

I don't know what to make of the above. In any case watch this for a second:

GF4MX460@300MHz = 2*2*300 = 1200MTexels/s
GF3Ti200@175MHz = 4*2*175MHz = 1400MTexels/s
GF3Ti500@250MHz = 4*2*250MHz = 2000MTexels/s

The digit-life link you provided above, verifies the fillrate ballpark between those cards. For the record the specific review does ONLY use a GF3Ti200 and NOT a GF3Ti500. The GF2Ti@250MHz that they use in that review belongs to the GF2 line of processors.

The significant difference between the two is that the GF2 line is capable of only dual multitexturing, while the GF3 can do up to 4 textures per pass.

Geometry throughput (as in the Poly throughput tests in 3dmark2001) are going to inevitably higher on a GF2Ti compared to a GF3Ti200, since clockspeed plays a major role with HW T&L.

Coming to the overall 3dmark2001 scores of the review in question, the GF2Ti's and GF4MXs truly do not run all tests (those envolving EMBM and Pixel Shaders) while GF3 family accelerators do.

What follows in the review are Q3a and RtCW benchmarks, where the latter also uses the q3a engine. The particular engine will force past dual texturing to multipass and that's one of the reasons why a GF3Ti200 cannot cope that well against a MX460 with almost twice as high clockspeed.

A clearer picture here:

http://www.firingsquad.com/hardware/geforce4/page7.asp

http://www.firingsquad.com/hardware/geforce4/page8.asp

In fact read through the whole review. If you find a GF3Ti500 to be slower than a GF4MX460 let me know.