Sign in with
Sign up | Sign in
Your question

Toms VGA card review, comments?

Last response: in Graphics & Displays
Share
April 18, 2002 12:30:43 PM

This is the official discussion thread, discuss!

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 18, 2002 12:40:35 PM

Good work nVidia! Apparently Tom's Hardware used the new reference drivers from nVidia. Surprisingly, they were able to squeeze out another ~5% out of the GF3 and GF4 cards. Well, ATI's next official drivers (the 6058, which is currently unofficially release) also squeezes out 5% and in some cases 10% out of the R8500 more over the 6043 drivers. In any case, nVidia and ATI seem to now be playing the driver game. Hmm, I wonder... Is it just me or has nVidia released this driver to take the lead in the sub-$200 section with Ti4200? In any case, I'm not complaining. Round Google: Winner: nVidia. :wink:

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 18, 2002 12:48:32 PM

I think it´s interesting to see how old cards measure up to the new ones.


<font color=blue>Be yourself, do whatever <font color=red>you</font color=red> like</font color=blue> (Not at my expense though)
Related resources
Anonymous
a b U Graphics card
April 18, 2002 12:55:38 PM

I was just wodnering where the Ati 7200 fits into this group of cards, I thought it was equivalent to one of the cards that are up there but I am not sure, I lost track when Ati renamed their line of cards.
April 18, 2002 1:24:03 PM

Damn, I just read the article, and it looks like honestly ati got owned, by the ti500, LET ALONE, the gf4.


The benchmarks where the radeon8500 won, (with the exception of 3dmark2001) they won BARELY, but in the benchmarks where the 8500 lost, they lost big time.(the first one comes to mind).


The gf4 just smokes the radeon.

This cements my feeling that the gf3 ti500 is a better overall performer than the 8500, and that the gf4 is just sweeeeet.



:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 18, 2002 1:29:14 PM

i have an ati radeon 64 mb drr vivo and it shows up a 7200 in 3dmark and the display porpertiessay ati radeon/ sti radeon 7200

what is better then a 7000 rpm, a 8000 rpm delta. to cause more noise to kill your ears :smile:
April 18, 2002 1:33:13 PM

Quote:

This cements my feeling that the gf3 ti500 is a better overall performer than the 8500, and that the gf4 is just sweeeeet.


Hmm, it depends on the two drivers you compare, but I say they're neck and neck. The Ti500 is slightly faster with the new drivers but we haven't seen the OpenGL games where the R8500 truly shines with the 6043 drivers. Most of the improvements in the 6043 were focused on OpenGL performance not DirectX. The 6058 improves DirectX slightly. ATI is always two official driver revisions behind leaks.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 18, 2002 1:37:55 PM

Quote:
Hmm, it depends on the two drivers you compare, but I say they're neck and neck. The Ti500 is slightly faster with the new drivers but we haven't seen the OpenGL games where the R8500 truly shines with the 6043 drivers. Most of the improvements in the 6043 were focused on OpenGL performance not DirectX. The 6058 improves DirectX slightly. ATI is always two official driver revisions behind leaks.


You really cant compare leaked drivers to official ones IMO, but regardless, the open gl games the radeon won BARELY.(did anyone else think it was weird the origional gf3 beat the gf3 ti500 in jediknight 2? There is NO reason why that should happen, ever)

The direct x games the radeon got owned, and direct x is a large part of modern games, so you have a radeon which can keep up in ogl and gets punked in d3d, dosent make me want to go and buy it personally.


:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 18, 2002 1:38:07 PM

As you can see, ATI lost to the Ti500 in Aquanox and Max Payne but won in Q3, Jedi Knight II and 3dMark2001. I'd hardly say that the Ti500 owns the R8500. Besides, Jedi Knight II and Q3 were the only OpenGl games they tested. That's an incomplete comparison if you ask me.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 18, 2002 1:43:06 PM

Quote:
As you can see, ATI lost to the Ti500 in Aquanox and Max Payne but won in Q3, Jedi Knight II and 3dMark2001. I'd hardly say that the Ti500 owns the R8500. Besides, Jedi Knight II and Q3 were the only OpenGl games they tested. That's an incomplete comparison if you ask me.


Maybe incomplete, but lets analyze those figures shall we.



the radeon lost to the gf3 ti 500 in aquanox by ~10%
The radeon lost to the gf3 ti 500 in max payne by ~10-20%
The radeon beat the gf3 ti 500 in quake 3 by ~3-4%
The radeon beat the gf3 ti 500 in jedi knight by <1%
The radeon beat the gf3 ti 500 in 3dmark2001(pointless) by ~5%


as you can see, when the 8500 wins, it is by far less than when it loses, which shows me the ti500 is a stronger overall player, you can debate all you want, but as far as this review is concerned the writing is on the wall man.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 18, 2002 1:48:58 PM

Any reason why they didn't include the Radeon 8500 128MB card in the benchmarks? (Not the LE edition) From what I've heard it has a higher clock speed and should perform a little better than both the 8500 64MB and the 8500LE 128MB. Just curious.
April 18, 2002 1:50:06 PM

Quote:

as you can see, when the 8500 wins, it is by far less than when it loses, which shows me the ti500 is a stronger overall player, you can debate all you want, but as far as this review is concerned the writing is on the wall man.

I don't have to debate, the 6043 drivers hardly touch the DX core. I want some OpenGL tests. How about RTCW, SS:SE, etc?

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 18, 2002 1:55:00 PM

Quote:
Any reason why they didn't include the Radeon 8500 128MB card in the benchmarks? (Not the LE edition) From what I've heard it has a higher clock speed and should perform a little better than both the 8500 64MB and the 8500LE 128MB. Just curious


because were comparing 64megs of memory to 64 megs of memory for starters.



:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 18, 2002 1:56:38 PM

Quote:
don't have to debate, the 6043 drivers hardly touch the DX core. I want some OpenGL tests. How about RTCW, SS:SE, etc?


What good is a videocard which in its best environment can just match the gf3 but gets its ass handed to it in d3d?

Even if the 8500 came out ~5% ahead of the gf3 in ogl, it would still lose overall due to the 10-20% loss in d3d, both systems are used extensivly, its not like there are no d3d games and you can just ignore d3d as if it didnt exist.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 18, 2002 3:15:27 PM

I'm guessing that JKII makes better use of the 128MB of memory available in some of the cards (it was a GF3 Ti200 with 128MB). It looks like all of the 128MB cards get a boost in JKII.
April 18, 2002 3:20:31 PM

JKII seems bottlenecked by the processor.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 18, 2002 3:33:23 PM

Quote:
I'm guessing that JKII makes better use of the 128MB of memory available in some of the cards (it was a GF3 Ti200 with 128MB). It looks like all of the 128MB cards get a boost in JKII

i agree...even the slower gf3 ti200 128mb beat the ti500 and the slower R8500LE 128mb beat the bread and butter R8500...
all i know is that ATI is still playing the catch-up game w/ nVidia but it looks like the gap is getting smaller...except that ATI is by no means any where near the gf4 levels yet (maybe close to the ti4200)...

<b><font color=red>ATI</font color=red>'s drivers are like a broken faucet, they both keep on leaking...</b> :cool:
April 18, 2002 3:47:59 PM

"because were comparing 64megs of memory to 64 megs of memory for starters."

doesn't make sense since they even compared the 32meg Geforce 2 MX with the 64 meg version.

They could have included the 128 meg Radeon 8500, but my guess is that ATI never sent Tomshardware one, so they can't run a test on the card. As I've heard, if they don't get what they're testing for free... they don't test it.
April 18, 2002 4:27:07 PM

This remind me I have to get a new graphic card soon.
April 18, 2002 4:35:51 PM

I'm already waiting for a newer review with the RV250 (it's either a GF4 Ti4400 to TI4600 or a GF4MX killer, I've heard both rumours), and the 6058 or newer drivers under the R8500.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 18, 2002 4:46:39 PM

From the looks of things, I'm going to stick with my Geforce 3 for now. Even moving to a 4600 I dont see the point for me. 30+ fps lead in some games but I dont need 130fps over 100fps, and I play all my games in 1024x768x32 due to monitor rr restrictions.

In all I was very pleased with nvidias performance, Jedi Knight especially was good considering my original geforce3 came out on top of the ti500.

I think the facts show nvidia can get more out of less hardware than ati.
No one disputes that atis hardware isnt superior because it is.. but I concern myself with driver support when I buy products.

Athlon XP 1700+,KT266A,Geforce3, Audigy.. 'nuff said.
Anonymous
a b U Graphics card
April 18, 2002 6:19:35 PM

i would like to congradulate tomshardware on putting out another wonderfully useless article. i mean those graphs were soo unexpected! especially by the end! wow another triangle of tapering scores! good thing it covered so many different resolutions and detail levels, ohh wait it didnt! nevermind the complete lack of practical use for this limited stupid information. god damn why not try something useful.

( O.o)
(o.O )
April 18, 2002 7:00:59 PM

They have an interesting GF3 TI500 that seems to school the old GF3 by quite a bit in every test except Jedi Knight II (Where it actualy looses to the older GF3, and is very close to the TI200), and 3DMark2001 where it seems to be in line with the TI200 and GF3 in performance regards (Almost linier).

Why does that seem so inconsistant? Especialy with the Jedi Knight II test? Just seems strange to me.

Also, they only did 1024x768x32 in the tests. The 8500, and all Radeon cards for that matter, tend to do better in higher resolutions, as they don't loose as much performance. If you're getting 120+ FPS in 1024x768x32, why not bump it up? You get better graphics and still won't notice much of a difference. I'm not saying that the 8500 will top the GF4s by any means in 1600x1200, but they might be closer than you think, and do better than you would think.

The Windows Gods demand money to appease the BSOD! - Rev. Bill Gates
April 18, 2002 7:13:44 PM

Jedi Knight II has all kinds of graphics problems. Notablly beautiful shadows that show up every-fricking-where incuding the sky. Real nice when your fighting a droid and its shadow is cast 500 ft up on the sky instead of 6 ft down on the walkway where its supposed to be. Notice how the top 4 or 5 cards were all dead even right under 100fps? can we say "bottleneck". I'd utterly discard the JK2 test... Another note is, all these tests were done in max detail settings. At max settings in 1600x1200 alot of these games wouldnt even run on some if not most of the tested cards. Lately i've been testing with Renegade and i can make my 1.7ghz TI 4600 system crawl on its knees if i go high enough. It IS interesting to see how the KryoII boards etc turned out to be such horribly bad options for gaming-)
One last point is, yea, i know the 8500 tends to show off more at higher resolutions but so does the TI500 and TI4000 series, especially the 4400/4600 models.

In reply to the gentleman (er, yea, right) who decried these tests as a waste of bandwidth. I AGREE!

"OOOOO Shiny things. I can make powerful equipment with shiny things!" - The imp Butler from BGII
April 18, 2002 7:48:07 PM

Obvious fact, anyone who buys a GF4 4600 gets screwed big time.

here the 4600 is about 150$ more than 4400. All this more for only maybe 2-8 fps more.

And the gf4 mx is really a scam to call this card a gf4. really misleading.
April 18, 2002 8:09:16 PM

Here are my complaints with this review:

- lack of games (especially the popular OpenGL games)
- only one resolution is tested. (the R8500 would probably pull ahead of the Ti500 and approach Ti4200 performance at 1280*1024 and up).
- only a single system tested (what about people with a less top-of-the-line processor?)
- no extra features tested (isn't anyone interested in comparing the FSAA and anisotropic filtering of each card?)
- no OpenGL benchmark (GLExcess is getting popular but Tom's Hardware hasn't started using it yet)

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
Anonymous
a b U Graphics card
April 18, 2002 9:24:41 PM

i don't know but the scores weren't right I think (for 3Dmark 2001) I have XP 1700+ and a radeon 8500 and i have a score of 8749 how is it possible that a xp 2000 doesn't get the higher, or at least same results?
April 18, 2002 9:51:02 PM

Tom's Hardware has trouble setting up drivers. :tongue:

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 19, 2002 3:38:15 AM

Poster: rog
"i would like to congradulate tomshardware on putting out another wonderfully useless article. i mean those graphs were soo unexpected! especially by the end! wow another triangle of tapering scores! good thing it covered so many different resolutions and detail levels, ohh wait it didnt! nevermind the complete lack of practical use for this limited stupid information. god damn why not try something useful."

I really liked the article, I was looking forward for some benchs where all the cards appear. Althougth I think AMD MAN is right in some of his complaints, like " -lack of games (especially the popular OpenGL games) -only one resolution is tested."
Anyway I loved to see all the cards in a single benchmark.
April 19, 2002 4:59:40 AM

my guess would be that this article was made for reference purposes...
well...it kinda, sorta served its purpose...

<b><font color=red>ATI</font color=red>'s drivers are like a broken faucet, they both keep on leaking...</b> :cool:
April 19, 2002 6:45:59 AM

wow, a lot of people found a lot of problems with this article. i actually enjoyed it, i remmeebr when those older card (tnt 2 and radeon(orig)) came out and that were state of the art. but now look how they compare, we have come a long way in so little time.

oh yeah, why wasn't my video card tested? since i own the radeon 8500 AIW, it is the ONLY video card in existence, so why wouldn't it be tested?(j/k) it's not that big of a deal.

repeat after me, we are all individuals!
April 19, 2002 7:26:23 AM

I agree, I liked seeing how much of an improvement the latest cards are over the ones from just a few years ago. I remember getting my Riva 128, playing my first 3D game and thinking how incredible it was. That Riva 128 wouldn't even register on the charts.

Heck, I remember playing NHL96 on my friend's Gateway with its super powerful 75MHz Pentium processor and thinking how great the 2D graphics looked, with the reflections on the ice and everything. Then he said he read that they were coming out with 3D graphics cards, and I was like, "Huh? What's that going to do?" We have come a long way.

<i>Money talks. Mine always likes to say "goodbye." :smile: </i>
April 19, 2002 11:42:07 AM

Quote:
- lack of games (especially the popular OpenGL games)

- only one resolution is tested. (the R8500 would probably pull ahead of the Ti500 and approach Ti4200 performance at 1280*1024 and up).

- only a single system tested (what about people with a less top-of-the-line processor?)

- no extra features tested (isn't anyone interested in comparing the FSAA and anisotropic filtering of each card?)

- no OpenGL benchmark (GLExcess is getting popular but Tom's Hardware hasn't started using it yet)

- If you had to choose 5 benchmarks to use, with the exception of replacing quake 3 with rtcw, this would be the list I used.

- The 8500 LOSES ground as you increase resolution last I checked, so I would also like to see higher resolution tests.

- One system was used to maintain equality across the board, if you added just one other system you double the benchmarks needed, remember they tested 30 CARDS, thats 20x3x30=1800 minutes JUST FOR 3DMARK2001 ALONE.(20 mins per test, 3 times for accuracy times 30 cards, you can cut corners and do the test once or twice only, but then the results could be skewed by oddities, and integrity must be maintained.) You all need to stop complaining.

- See comment about time, if you feel like spending 100 hours benchmarking 30 cards for all different brands of features (fsaa aniso etc) then go for it.
Also, most of the cards tested do not support those features, the benchmark had to be kept at a level universally completeable by all cards.

- The majority of the benchmarks ran in ogl IIRC(actually it was 2/2 and 3dmark2001), every additional test would consume HOURS AND HOURS of time, assuming 10 minutes for a test, (which is a low guess especially since you have to run the test multiple times for accuracy) each additional benchmark would add 300 minutes, or 5 hours! Most likely more like 10 hours(due to triplicate testing), PER BENCHMARK!.

-(sidenote) 3dmark2001 uses direct3d IIRC, in both other d3d tests the radeon lost handily, which shows me maybe ati is spending a bit too much time optimising for benchmark apps, I would much rather have ditched 3dmark2001 and used a real game (ogl or d3d dosent matter to me) in its place.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 19, 2002 11:59:23 AM

Quote:

- The 8500 LOSES ground as you increase resolution last I checked, so I would also like to see higher resolution tests.

I've seen the opposite.


Quote:

- One system was used to maintain equality across the board, if you added just one other system you double the benchmarks needed, remember they tested 30 CARDS, thats 20x3x30=1800 minutes JUST FOR 3DMARK2001 ALONE.(20 mins per test, 3 times for accuracy times 30 cards, you can cut corners and do the test once or twice only, but then the results could be skewed by oddities, and integrity must be maintained.) You all need to stop complaining.

Hmm, never thought of that but they could get a small group of programmers to design an app that would run all the benchmarks automatically and record the results.

Quote:

-(sidenote) 3dmark2001 uses direct3d IIRC, in both other d3d tests the radeon lost handily, which shows me maybe ati is spending a bit too much time optimising for benchmark apps, I would much rather have ditched 3dmark2001 and used a real game (ogl or d3d dosent matter to me) in its place.

The R8500 is about 40% faster, IIRC, than the GF3Ti500 in Star Trek Armada II (after you apply the patch for both), which is a DX8.1 game.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 19, 2002 1:50:02 PM

Quote:
I've seen the opposite.


Which is why I would like to see higher resolution tests :-)


Quote:
Hmm, never thought of that but they could get a small group of programmers to design an app that would run all the benchmarks automatically and record the results.


You would still have to swap out the cards, but that would help slightly. Still not feasable for thg to do 30 card roundups with 10 tests with aa and aniso on and off at varying levels.

Quote:
The R8500 is about 40% faster, IIRC, than the GF3Ti500 in Star Trek Armada II (after you apply the patch for both), which is a DX8.1 game.


If star trek armada 2 is anything like 1 it is cpu and hdd limited, also what was it before the patch, also what drivers were the gf3 using at the time. Too many variables(also some links to scores would be nice).

That having been said, I find 40% faster than a ti500 NEARLY IMPOSSIBLE to believe in a r8500. We are talking a 40fps difference if the ti500 does 100fps, no benchmark I have ever seen between the ti500 and 8500 has shown a disparity even close to that one.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
April 19, 2002 2:54:05 PM

My bad, here's the <A HREF="http://www2.tweaktown.com/imagebank/ap64dh_test6.gif" target="_new">link</A>. Only 27% faster, not 40%, my bad.

Apparently, before the patch, Star Trek Armada II would by default turn off all buffers on ATI Radeon cards due to a bug in the very early driver releases. That left the frame rates stuck at ~60fps. Now ATI has fixed the bug and there is a patch for Armada II that runs the game at full speed under ATI cards. I don't doubt that many games used to take workarounds on ATI cards due to buggy drivers, but now, all these bugs are getting fixed so we may start seeing many game patches for ATI cards under the 6043 and up especially.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 19, 2002 3:36:03 PM

Thanks for you input :) 

Time is really a critical factor here... and don´t forget: I want to upgrade those charts frequently. So I can´t choose too many resolutions and benchmarks etc. I also have to take a look at driver updates etc. I will also include a slower system in future.

The VGA Charts are thought as a kind of "upgrade guide". So 1024x768 is the resolution to go. We have a lot of articles where you can see how modern VGA cards perform at higher resolutions (like our latest 4200 preview: http://www.tomshardware.com/graphic/02q2/020409/index.h... ).

Lars "Borsti" Weinand
Editor Tom's Hardware Guide
April 19, 2002 3:46:12 PM

Wow! A real THG editor here! WOW!! I think it would be better to get a custom title from Fredi, so people don't think you're an imposter.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor
April 19, 2002 6:32:33 PM

i thought writers of THG dont come to the forum...
but this is pretty interesting...

<b><font color=red>ATI</font color=red>'s drivers are like a broken faucet, they both keep on leaking...</b> :cool:
April 20, 2002 5:58:30 AM

Quote:
One last point is, yea, i know the 8500 tends to show off more at higher resolutions but so does the TI500 and TI4000 series, especially the 4400/4600 models.

The point is that the radeons tend to take less of a performance hit at higher res.

Yes, some cards wouldn't be able to handle it, but it would be nice to see.

As for JK2, strange that anyone would use that as a test if the game didn't render correctly, and this is the third review site I've seen use it to test cards. I would think that would have been mentioned somewhere.

The Windows Gods demand money to appease the BSOD! - Rev. Bill Gates
April 22, 2002 11:07:02 AM

I think that´s every editors very own descision :) 

Lars
April 22, 2002 12:43:56 PM

Kyro 2 cards are only fully direct7, and were meant to stand up against the MX-200/400 Nvidia's & Radeon VE range... that the GF3/4 & Radeon8500 beat it into the ground is no surprise (the only surprice was how badly the chipset did in JK2).
I personally didnt expect the Kyro2 to remain good for more than a year but when it came out its price/performance was the best around

There are no stupid questions... just lots of inquisitive idiots...
April 22, 2002 6:31:38 PM

Didn't the Radeon SDR become the 7200 and the Radeon VE become the Radeon 7000?

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